EPA-650/2-74-016
December 1973
Environmental Protection Technology Series
-------
EPA-650/2-74-016
SAMPLING INTERFACE
FOR QUANTITATIVE
TRANSPORT OF AEROSOLS
by
Madhav B . Ranade
IIT Research Institute
10 West 35th Street
Chicago, Illinois 60616
Contract No. 68-02-0579
Project No. 26AAM
Program Element No , 1AA010
EPA Project Officer: Dr. Kenneth T. Knapp
Chemistry and Physics Laboratory
National Environmental Research Center
Research Triangle Park, North Carolina 27711
Prepared for
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
WASHINGTON, D.C. 20460
December 1973
-------
This report has been reviewed by the Environmental Protection Agency
and approved for publication. Approval does not signify that the
contents necessarily reflect the views and policies of the Agency,
nor does mention of trade names or commercial products constitute
endorsement or recommendation for use.
11
-------
SAMPLING INTERFACE FOR QUANTITATIVE TRANSPORT
OF AEROSOLS
ABSTRACT
A sampling probe was designed, fabricated, and evaluated
for quantitative transport of aerosols through a conduit from
a source to a sensor. The probe consists of a porous metal
tube encased in a manifold through which transpiration air
was passed inward to provide a moving clean air sheath that
minimized particle deposition on the walls. In Phase I, the
quantitative mass transport of aerosols was investigated, and
in Phase II, the preservation of size distribution of the
transported aerosol was studied. The 178 cm (70 in.) long
by 1.27 cm (% in.) ID probe required only 14.2 1pm (0.5 cfm)
of transpiration air to virtually eliminate deposition of
particles in the 0.05 to 15 urn size range. Particles as
large as 70 urn required as much as 283 1pm (10 cfm) to pre-
vent deposition losses at low sample flow rates. A statis-
tical analysis of the data conclusively demonstrates the
effectiveness of the porous probe sampling concept. Tests
at selected conditions show that the porous probe is effec-
tive in the preservation of size distribution.
Optimization of the sample and transpiration flow ratio
is necessary for a given size range to obtain the most effec-
tive use of the porous probe concept.
iii
-------
TABLE OF CONTENTS
Page No
1, INTRODUCTION ................. 1
PHASE I - SAMPLING INTERFACE FOR QUANTITATIVE MASS
TRANSPORT OF AEROSOLS:
2. SAMPLING PROBE DESIGN AND OPERATION , „ , . . 5
2.1 Prototype Design ..„,.....„.. 5
2.1.1 The Porous Tube ......... 5
2,1.2 The Sampling Probe Assembly . , „ 6
2.1.3 The Test Layout ......... 8
2.2 Sampling Probe Operation ..„,.... 12
3. EXPERIMENTAL PROCEDURES FOR PHASE I ..... 17
3.1 Aerosol Generation ........... 17
3.1.1 KC1 Aerosol, 0.05 ym ....... 17
3.1.2 1-MAAQ Aerosol, 1.6 urn ...... 18
3.1.3 Uranine Aerosol, 50 um 23
3.2 Test Procedure ............. 24
3.3 Chemical Analysis ............ 28
4. TEST DESIGN, DATA, AND OBSERVATIONS ..... 29
4.1 Test Design ............... 29
4.2 Test Data and Observations 29
4.2.1 Tabular and Graphic Data 29
4.2.2 Photographic Proof of Effectiveness 37
5. STATISTICAL ANALYSIS OF PHASE I TEST DATA . . 42
5.1 Definition and Coding of Variables ... 42
5.2 Experiment Design ............ 44
5.3 Data Base ................ 46
5.4 Candidate Terms of the Multiple-Regression
Performance Model . 46
5.5 Methods of Data Analysis ........ 48
5.6 The Particle Deposition Equation .... 49
5.7 Residuals ................ 51
5.8 Plots of Functional Relationships Between
Particle Deposition and the Experimental
Factors ................. 53
5.9 Discussion and Conclusions ....... 53
PHASE II - PRESERVATION OF SIZE DISTRIBUTION OF
AEROSOL IN TRANSPORT THROUGH THE SAMPLING INTERFACE:
6. EXPERIMENTAL PROGRAM FOR PHASE II ...... 66
6.1 Tests with KC1 Aerosol ......... 66
6.2 Tests with Uranine Aerosol ....... 82
6.3 Tests with Flyash ............ 94
6.4 Tests with 1-MAAQ Aerosol 94
7. RESULTS ..... 106
7.1 Tests with KC1 Aerosol ......... 106
7.2 Tests with 1-MAAQ Aerosol ........ 106
7.3 Tests with Uranine and Flyash Aerosols . 115
8. CONCLUSIONS ................. 118
Appendix A -- AN ELECTRIC MOBILITY METHOD OF SIZING
0.01 TO 1.0 urn POLYDISPERSE AEROSOLS. 122
IV
-------
LIST OF FIGURES
Page No
1. Sampling Probe Transpiration Air Manifold
Details 7
2. Sampling Nozzle Details 9
3. Downstream Probe Seal and Filter Housing
Transition Details 10
4. Flow Schematic of Porous Probe Sampler as
Evaluated for Mass Transport of 0.05-50 ym
Aerosols . , 11
5. The Sampling Interface Test Facility 13
6. Transpiration Air Pressure Drop Across the
Porous Tube Wall 15
7. Test Aerosol Size Distribution from Electrical
Mobility Data 19
8. Photographs of the Aerosol Generators .... 20
9. Photomicrograph of the 1-MAAQ Test Aerosol . . 21
10. 1-MAAQ Test Aerosol Size Distribution .... 22
11. Photomicrograph of the 55 ym Uranine Test
Powder 25
12. Uranine Test Powder Size Distribution .... 26
13. Deposition of 50 ym Uranine Particles in the
1.27 cm Diameter by 178 cm Long Porous
Sampling Probe 33
14. Deposition of 1.6 ym 1-MAAQ Particles in the
1.27 cm Diameter by 178 cm Long Porous
Sampling Probe 34
15. Deposition of 0.05 ym KC1 Particles in the
1.27 cm Diameter by 178 cm Long Porous
Sampling Probe 35
16. Particle Deposit on Glass Fiber Filter Showing
Laminar Flow with Clean Air Sheath, Test 5 . . 38
17. Photomicrograph of Particle Deposit on Glass
Fiber Filter of Test 5 38
18. Photographs of Filter Deposits of 1-MAAQ
Aerosols which Confirm Effectiveness of
Boundary Layer Principle for Aerosol
Transport 39
19. Plot of Observed vs. Calculated Values of the
Dependent Variable Y 54
20. Plot of Cumulative Percentage of Residuals on
Normal Probability Paper 55
21. Deposition of Small, Medium, and Large
Particles vs. Transpiration Flow Rate with Low
Particle Concentration and Low Sample Flow
Rate 56
22. Deposition of Small, Medium, and Large
Particles vs. Transpiration Flow Rate with
High Particle Concentration and Low Sample
Flow Rate 57
v
-------
LIST OF FIGURES (continued)
Page No
23. Deposition of Small, Medium, and Large
Particles vs. Transpiration Flow Rate with
High Particle Concentration and Medium Sample
Flow Rate 58
24. Deposition of Small, Medium, and Large
Particles vs. Transpiration Flow Rate with
High Particle Concentration and High Sample
Flow Rate 59
25. Deposition of Particles in Relation to
Particle Size at Intermediate Levels of
Particle Concentration, Sample Flow Rate, and
Transpiration Flow Rate 60
26. Nebulizer for KC1 Aerosol Generation ..... 68
27. Nebulizer for Test K-5 70
28. Size Distributions for Test K-l 71
29. Size Distributions for Test K-2 72
30. Size Distributions for Test K-3 73
31. Size Distributions for Test K-4 74
32. Size Distributions for Test K-5 75
33. Log-Probability Plot of Size Distributions for
Test K-l 77
34. Log-Probability Plot of Size Distributions for
Test K-2 78
35. Log-Probability Plot of Size Distributions for
Test K-3 79
36. Log-Probability Plot of Size Distributions for
Test K-4 80
37. Log-Probability Plot of Size Distributions for
Test K-5 81
38. Histograms for Run U-l 84
39. Histograms for Run U-2 85
40. Histograms for Run U-3 86
41. Histograms for Run U-4 87
42. Histograms for Run U-5 88
43. Size Distributions for Test U-l 89
44. Size Distributions for Test U-2 90
45. Size Distributions for Test U-3 91
46. Size Distributions for Test U-4 92
47. Size Distributions for Test U-5 93
48. Histograms for Run F-l 96
49. Histograms for Run F-2 97
50. Histograms for Run F-3 98
51. Histograms for Run F-4 99
52. Histograms for Run F-5 100
53. Size Distributions for Test F-l 101
54. Size Distributions for Test F-2 102
55. Size Distributions for Test F-3 ! 103
56. Size Distributions for Test F-4 ....... 104
57. Size Distributions for Test F-5 105
58. Size Distributions for Test M-l , 108
VI
-------
LIST OF FIGURES (continued)
Page No
59. Size Distributions for Test M-2 JUJ9
60. Size Distributions for Test M-3 110
61. Size Distributions for Test M-4 Ill
62. Size Distributions for Test M-5 . 112
vii
-------
LIST OF TABLES
Page No
1. Aerosol Mass Transport Test Data for Phase I
Experiments with the IITRI Boundary Layer
Sampling Probe ................ 30
2. Experimental Data .............. 45
3. Set of Candidate Predictive Variables for the
Sampling Tube Performance Equation ...... 47
4. Particle Deposition Equation ,,„..„... 50
5, Observed and Calculated Values of the
Dependent Variable and Residuals .,.„,.. 52
6. Phase II Experimental Program ........ 67
7. Size Distribution Data for Tests with KC1
Aerosol ...... ,.,«..,..,... 76
8. Size Distribution Data for Tests with Uranine. 83
9, Size Distribution Data for Test with Flyash
Aerosol ................... 95
10. Size Distribution Data for Test with 1-MAAQ
Aerosol .....,..„„,,.„„.... 107
11. Size Distribution Parameters for KC1 Aerosol . 113
12. Size Distribution Parameters for 1-MAAQ
Aerosol ................... 114
13. Size Distribution Parameters for Tests with
Uranine and Flyash Aerosols ......... 116
VI11
-------
ACKNOWLEDGMENTS
This report was prepared by the Fine Particles Research
Section of the Chemistry Division at IIT Research Institute,
Chicago, Illinois, under EPA Contract No. 68-02-0579.
The work was divided into two phases. Phase I involves
fabrication and demonstration of a sampling interface for
quantitative mass transport of aerosols from a source to a
sensor, Mr, Don Werle served as the project director for
Phase I. Dr, Fred Bock designed the factorial test plan
and made the statistical analysis of the data, Flame emission
analyses were performed by Ms, G, Marks, K, Walanski and
L, llartin assisted in some of the tests conducted under
Phase I.
Phase II of this contract involves preservation of size
distribution in the sampling interface. Dr. Madhav Ranade
served as the project director for Phase II. Dr. Earl Knutson
direct the use of mobility analyzer for the Phase II tests.
George Yamate and Jean Graf assisted in the tests. Dr. Ken
Knapp acted as the EPA project officer.
ix
-------
SAMPLING INTERFACE FOR QUANTITATIVE TRANSPORT
OF AEROSOLS
1. INTRODUCTION
Conventional stack sampling probes used to transport
particulate material from a source through a conduit to a
collector or sensor can easily result in biased measurements
due to particulate deposition on and/or interaction with the
probe walls. Such interaction and deposition can result in
a significant loss of particle mass and alteration of the
particulate size distribution, especially as measured by
continuous monitors. The goal of this research program is
to develop and demonstrate through controlled laboratory
tests a sampling interface system which will prevent or
minimize aerosol deposition. The system selected is based on
the use of a porous walled probe through which clean filtered
air passes radially inward to provide a boundary layer of
particle free air.
The program is divided into two phases. Phase I is
concerned with the mass transport of source aerosols. Phase II
is concerned with the preservation of the source aerosol size
distribution during transport to a sensor. The sampling
interface requirements call for the following:
1, Accommodate aerosol concentrations from at least
10 particles/cm3 to at least 10 particles/cm3.
2. Accommodate aerosols from 0.05 ym to 50 ym as a
minimum size range.
3. Sample rate of from 7.1alpm(0.25 acfm) to
28.3 alpm (1.0 acfm).
4. Permit dilution of the aerosol with a clean air
stream where the dilution ratio is at least 10:1.
-------
The following tasks are involved in Phase I:
Task (A) Prepare detailed design and engineering
specifications for a sampling interface to
meet the requirements for quantitative
transport.
Task (B) Fabricate the system according to the design
and specifications agreed upon with the
Project Officer.
Task (C) Design and conduct laboratory tests to demon-
strate the effectiveness of the sampling
interface in achieving the above stated
goals for Phase I.
Task (D) Prepare detailed report on Phase I including
operating and maintenance instructions.
Review with Project Officer before proceeding
to Phase II,
Phase II is concerned with the preservation of source
aerosol size distribution during transport to a sensor, The
technical effort under Phase II consists of the following
tasks.
Task (E) Develop design and engineering specifications
for the sampling interface to meet the require-
ments for preservation of particle size dis-
tribution as well as quantitative transport.
Task (F) Fabricate the system according to the designs
and specifications agreed upon with the
Project Officer.
Task (G) Design and conduct laboratory tests to demon-
strate the effectiveness of the sampling
interface in achieving the above stated goals
for Phase II.
-------
Task (H) Prepare a detailed Phase II interim report
including operating instructions and main-
tenance manuals,
The sampling interface designed and tested under Phase I
was considered .to be adequate for Phase II. A test plan with
various particles representing a wide size range was suggested
to EPA and was executed.
This report describes the technical effort under both
Phase I and Phase II of the contract.
-------
PHASE I
SAMPLING INTERFACE FOR QUANTITATIVE
MASS TRANSPORT OF AEROSOLS
-------
2. SAMPLING PROBE DESIGN AND OPERATION
The sampling interface conduit utilized in the Phase I
experiments consists of 178 cm (70 in.) long by 2,54 cm
(1 in.) ID straight stainless steel porous tube obtained
from the Mott Metallurgical Corporation of Farmington,
Connecticut. While a tube with a 90° bend at the inlet end
was originally ordered,, the manufacturer encountered fabri-
cation problems which delayed the order and it became expedi-
ent to delete the bending operation to obtain a reasonable
delivery. The design philosophy of the prototype interface
hardware was necessarily oriented toward conduct of the lab-
oratory tests to demonstrate the effectiveness of the hard-
ware. Therefore, quick disassembly and assembly of the proto-
type was given priority to facilitate the analytical require-
ments of the experimental program. The hardware as it is
now fabricated is not intended for direct use in stack samp-
ling since it has temperature sensitive components (rubber
0-ring seals, lead solder connections, and epoxy seals) not
compatible with hot stack gases. The lack of a 90° inlet
bend is also not compatible with most stack sampling opera-
tions, but the straight configuration was ideal for rapid
disassembly of the interface and interpretation of the test
results. Thus, the hardware must be considered as the first
prototype model to demonstrate feasibility and effectiveness
rather than a prototype ready for limited production for
field use,
2,1 Prototype Design
2,1,1 The Porous Tube
Originally a porous ceramic tube was considered for the
sampling interface. However examination of several ceramic
specimens revealed two serious disadvantages in the proposed
application. First, a long small ID ceramic tube would be
very susceptible to breakage. Second, the firing process used
to sinter green ceramic compacts often leaves a rough surface
-------
where ceramic dust droppings fuse to the surface, A rough
surface would introduce objectionable boundary layer flow
disparities which are antagonistic to the goal of minimizing
wall losses.
A porous type 316 stainless steel tube of 2 ym nominal
porosity and 1.27 cm (% in.) ID by 1.9 cm (3/4 in.) long was
ordered and obtained from- the Mott Metallurgical Corporation.
Inspection of the tube interior on delivery revealed some
surface roughness as well as a shallow bend. The internal
surface roughness was readily corrected by lightly polishing
with a 240 grit aluminum oxide abrasive cloth, and the bend
was easily straightened. The actual OD is close to 2 cm
(13/16 in,) and the length is actually 177.6 cm (69-15/16 in.)
rather than the nominal 1,9 cm (3/4 in.) and 178 cm ( 70 in.)
dimensions, respectively. The ID at the ends is within
0.0013 cm (0.0005 in.) of 1.27 cm (% in.). The opening at
the forward end is centered within + 0.008 cm (0.003 in,)
while the opening at the downstream end is noticeably eccen-
tric, The wall thickness at the downstream end varied from
a minimum of 0.325 cm (0,128 in,) to a maximum of 0,406 cm
(0.160 in.),
2.1.2 The Sampling Probe Assembly
The details of the transpiration air manifold for admis-
sion of the boundary layer air to the porous tube are given
in Figure 1, A two-stage manifold is used to minimize the
incoming air velocity and objectionable jetting which could
prevent a balanced distribution of the air. The penetration
of the inlet air jet through the porous wall opposite the jet
had been a problem in diffusion experiments conducted earlier
by Dr. Wasan's graduate students. Thus, the inlet air is
first introduced into an outer 5.08 cm (2 in.) diameter copper
pipe before passage through the second inner manifold. The
inner manifold has a series of twenty 1.27 cm (% in.) dia-
meter holes in a 23 cm (9 in.) long section near the center
of the probe. Not shown in the drawing are two pressure
-------
20-1,27 Dia, Holes
90° Apart
k-8l.
25,4
20.5
5,08 Rigid Type L Copper Pipe
3,175 Rigid Type L
Copper Pipe
Porous Pyrex Tube
Air InletvTub\ \ Transition
A X
Adapter
All Dimensions in Centimeters
Figure 1, Sampling Probe Transpiration Air Manifold Details
-------
taps for measuring the air pressure in the center of the mani-
fold and at the extreme downstream end of the manifold. At
all flow conditions used identical pressures were observed
at these two locations.
The details of the sampling nozzle construction and as-
sembly are shown in Figure 2, The 45° sampling nozzle shown
was fabricated but was not used in the Phase I tests due to
the non-ideal shape for isokinetic sampling. Instead a
1.27 cm (% in.) ID Western Precipitation standard nozzle was
modified for attachment to the porous sampling probe and was
used in all of the Phase I tests.
The transition from the downstream end of the porous
sampling probe to the collection filter is detailed in
Figure 3. A 1.27 cm (% in.) ID glass tube in the transition
allows visual observation of the aerosol passage and in many
cases revealed the existence of the particle-free boundary
layer immediately after the porous tube, Turbulent breakup of
the aerosol stream is minimized or prevented by bringing the
glass tube to within 0,63 cm (% in.) of the filter surface
with the result that the filter became a witness plate for
the existence or absence of a particle-free air sheath.
2,1,3 The Test Layout
A flow schematic of the test layout is shown in Figure 4,
Three aerosol generators, described in Section 3,1, produced
any of three test aerosols in the desired size and concentra-
tion. Instead of running the tests in a wind tunnel as ori-
ginally proposed, a much more suitable system was developed,
which in effect reduced the size of the wind tunnel to the
ID of the sampling probe so that a 1007o aliquot was taken of
the aerosol flow. Thus, the aerosol mass sampled by the
porous probe was determined directly by adding the amount
collected on the downstream filter to the amount deposited on
the porous tube --an exact and direct measurement of the
aerosol mass sampled as well as the amount deposited upstream
of the filter,
8
-------
1.9 ID x 0.12
0-Ring 20 TPI x 3.02 OD
Porous
Tube
1.27
2,4 2.06' 2.
T " I
'•f-
45° Sampling Nozzle Attachment
7.78
Interchangeable Standard Nozzle
(Western Precipitation, Modified)
2.86
Mating Collar Adapter
All Dimensions in Centimeters.
Figure 2. Sampling Nozzle Details
-------
Copper
Pipe
Porous Tube
GO
I—
10.2 cm Filter
Holder
Epoxy Seal
1.9 ID x 0,12
0-Ring
1,27 ID x 11,11 Long
Pyrex Transition
All Dimension in Centimeters
Type A
Glass Fiber
Filters (2)
To Pump
Filter
Support
Figure 3, Downstream Probe Seal and Filter Housing Transition Details
-------
Flowmeters
1415, 5664, 28,300 1pm
Aerosol
/ Generators
Dilution
Air
Transpiration
Filter Air Supply
Vent
Excess
When
Diluted
Probe
Assembly
Protective
Filter
Laminar
Flow
Element
Air
Pump
Manometers
GF Filter
HD-
Vent
Figure 4. Flow Schematic of Porous Probe Sampler as Evaluated for
Mass Transport of 0.05-50 urn Aerosols
-------
The filtered transpiration air supply is metered by
three panel-mounted Dwyer flowmeters with 23, 90, and 470 1pm
(0.8, 3.2, and 16.6 cfm) capacities, A flexible hose con-
nects the filter holder to a Cambridge absolute filter
(#F-599), which protects the Meriam 50MW20-l-% laminar flow
element. The pressure drop across the laminar flow element
is proportional to velocity. The pressure drop is monitored
by a Dwyer Model No, 421-10 single column manometer, A 566 1pm
(20 cfm) Leiman #195-2 Type G vacuum pump with 1% HP, 230
volt, 3 phase, direct drive pulls the required air plus sam-
ple volume through the interface. The Leiman pump is pro-
vided with a by-pass and appropirate valves for easy manual
control of the interface flow rate,
A composite photograph of the laboratory test facility
used is shown in Figure 5, The entire facility is mounted
in a 5 meter long laboratory fume hood. The aerosol gener-
ators and the interface sampling nozzle are located in the
far left hood compartment. The main body of the sampling
interface is in the center left hood and extends partially
into the third compartment. Also located in the third com-
partment are the glass fiber filter collector, the protective
absolute filter, and the laminar flow element. The vacuum
pump is on the far right.
2,2 Sampling Probe Operation
The operation of the sampling probe is relatively simple.
The basic difference compared with a conventional stack sam-
pling probe is that there are two flows which must be moni-
tored and that the sample flow rate (SFR) is not measured
directly but is the net difference between the exhaust flow
rate (EFR) and the transpiration air flow rate (TAFR),
SFR = EFR - TAFR
The exhaust flow rate is monitored by means of the Meriam
laminar flow element. The transpiration air rotameters were
12
-------
u>
Figure 5, The Sampling Interface Test Facility
-------
calibrated with the laminar flow element The actual flow
rate was found to agree with the manufacturers indicated
settings. Since the rotameters are factory calibrated for
standard conditions, a correction must be applied to the ob
served flowmeter reading (FR in 1pm) to obtain either the
standard or absolute volume at the flow conditions, e.g.,
slpm = FR
where P and T are the standard pressure and temperature,
s s
and P, and T-, are the actual pressure and temperature, res-
pectively. It was found convenient to first set the exhaust
flow to the proper level as indicated by the laminar flow
AP. Next, the transpiration air (TA) flow was adjusted to
the desired value for the volume of TA desired. Then the
laminar flow level, if disturbed by the TA, was readjusted
to the target setting. The sample flow rate (SFR) was then
confirmed by wet-test meter attached directly to the sample
nozzle. Fine tuning of the SFR, if needed, was obtained
with a slight adjustment of the total exhaust flow. Once
the flow-rate was properly adjusted, the wet-test meter was
disconnected and aerosol was then introduced into the sampling
nozzle entrance.
For all of the test flow conditions used in the Phase I
tests, the axial pressure drop inside the porous tube was
determined as a function of distance from the entrance shown
in Figure 6,
Since the flow rate of transpiration air will vary in
proportion to the radial AP across the porous tube, it is
important to know the radial driving force at all points
along the length of the porous tube. Obviously, we could not
tolerate a situation where the transpiration air supply pres-
sure would be less than the pressure on the sample air side
of the porous tube. This situation could occur near the
14
-------
1.27 x 1.9 x 178 cm long MOTT 2 ym
porous SS tube preceded by a
1.27 x 7.78 cm long standard
sampling nozzle
28.3/283*
75
14,2/283
o 50
CM
s
o
* A/B
where A
B
25 -
slpm sample
slpm transpiration air
14,2/63.7
7,1/63.7
28,3/63.7
7,1/14.2
14.2/14.2
0
J_
0
25 50 75 100 150 180
Distance from Front of Porous Tube, cm
/14.2
210
Figure 6. Transpiration Air Pressure Drop Across
the Porous Tube Wall
15
-------
entrance of the probe at high sample flow rates and low
transpiration air flow rates if the porous tube is too per-
meable. In such a situation, some of the sample could by-pass
the inside bore in following a path of least resistance and
reenter the porous tube further downstream. Such a reverse
flow would defeat the purpose of the porous tube. The use
of a low permeability porous tube avoids this undesirable
situation by requiring an air supply pressure considerably
greater than the pressure at the inside upstream end of the
porous tube, An added advantage is a more even distribution
of transpiration air,
In all of the flow tests the transpiration air pressures
at the middle of the manifold and at the extreme downstream
end of the manifold were essentially identical, indicating
negligible pressure drop over the length of the transpiration
air manifold, Also, at all the parametric levels selected
for the experimental design, the air pressure in the trans-
piration air manifold was positive, indicating reverse flow
through the porous tube could not occur at these conditions.
Figure 6 shows the radial AP for the transpiration air pas-
sing through the wall of the porous tube as a function of the
axial distance from the upstream end of the porous tube.
Note that the porous tube was preceded by a 7,78 cm (3-1/16 in.)
long by 1.27 cm (% in.) ID sampling nozzle,
The pressure drop was invariably lowest at the upstream
end of the porous tube and highest at the downstream end.
The greatest percent change (nearly twice the AP at the
downstream end compared to the upstream end) in the radial
AP occurred when the highest transpiration air flow rate of
283 slpm (10 scfm) was used.
16
-------
3, EXPERIMENTAL PROCEDURES FOR PHASE I
Dependable aerosol generation in the desired concentra-
tion and size range was a vital part of the tests conducted
on the sampling interface. This section describes the methods
of aerosol generation, the aerosols produced, and the experi-
mental and analytical procedures,
3,1 Aerosol Generation
The experimental design of Phase I called for 0.05 ym,
1.6 ym and 50 ym aerosols, Accurate measurement of minute
amounts of aerosol deposition on the porous sampling probe
suggested the use of soluble aerosols which could be solvent
extracted from the porous tube and quantitatively analyzed
with a sensitive analytical test. KC1 was ultimately selected
for the generation of the 0.05 ym aerosols, 1-methylamino-
anthraquinone (1-MAAQ) for the 1.6 ym aerosols, and sodium
fluorescein (uranine) for the 50 ym aerosols, Flame emis-
sion, colorimetric absorption, and fluorescence analytical
techniques were used to assess the extracts and samples,
3.1,1 KC1 Aerosol. 0.05 ym
The KC1 aerosol generator was selected as the alternate
choice over the MgCl2 aerosol generator initially attempted
when it was found that the magnesium salt was much too sus-
ceptible to hydrolysis and erratic output. In contrast, the
KC1 aerosol generator proved to be quite stable and predict-
able over test periods which ran as long as six days of con-
tinuous operation. The KC1 aerosol generator operates by
vaporization from a plug of the salt in an electrically heated
nichrome coil. At an air flow of 7.1 1pm (0,25 cfm) and a
voltage of 21,8 volts across the 5 ohm coil (#12 BWG
Nichrome V wire), a satisfactory test aerosol was generated,
The condensation nuclei counter showed a concentration greater
than 10^ particles/cc while illumination of the aerosol in
an intense beam of light showed no light scattering despite
17
-------
the high concentration, A sample collected with the TSI
Electrostatic Aerosol Sampler (Model 3100, Thermosystem
Incorporated, St. Paul, Minn.) was examined and sized in the
electron microscope (EM). An approximate size distribution
from a limited number of particles showed that the aerosol
was in the desired size range with a count median diameter
(CMD) of 0.038 ym and a a of 1.63. The size distribution of
the KC1 aerosol as measured by EM examination was confirmed
in particle electrical mobility tests with an aerosol analyzer
similar to the Whitby Aerosol Analyzer sold by TSI (Appen-
dix A). Dr. Earl Knutson, a former student of Dr. Whitby,
made the measurements and reported a CMD of 0,032 ym,
Figure 7.
Flame emission measurements of the potassium concentra-
tion was used to analyze the samples in the sampling inter-
face tests with the KC1 aerosols.
3,1.2 1-MAAQ Aerosol, 1,6 ym
The 1,6 ym aerosol was readily generated by controlled
vaporization and condensation of the 1-MAAQ. Figure 8A is
a photograph of the combined KC1 and 1-MAAQ aerosol generator
which is operated at a reduced voltage (18 volts) to generate
the condensation nuclei needed for controlled condensation of
the 1-MAAQ vapor. The KCl-nucleated air stream at 0 25 cfm
passes into a second three-necked flask with the 1-MAAQ above
a hot glycerine bath. A lab jack raises the bath and immerses
the flask whenever aerosol generation is desired. The vapor-
laden air stream then passes through a heated glass tube
followed by the condensation tube. A portion of the aerosol
formed in the condensation tube was sampled by the interface
probe. The aerosol was sampled by an IITRI moving slide
impactor and sized from photomicrographs, Figures 9 and 10.
18
-------
10
0.7
0.5
OA
0.3
E 0.2
n
Q)
•U
VO
10"
.07
.05
.04
.03
.02
10
O.I
KC1 Test Aerosol
Count Median Diameter
CT
g
Mass Median Diameter
0.032 urn
1.56
0.058
5 10 50 90 95 99 99.9
% Less Than Stated Size
Figure 7. Test Aerosol Size Distribution from Electrical Mobility Data
-------
A, The combined KC1 and 1-MAAQ aerosol generators
for 0.05 urn KC1 or 1.6 ym 1-MAAQ aerosols,
B. The miniature venturi dust feeder for 50
uranine aerosol generation.
Figure 8. Photographs of the Aerosol Generators
20
-------
Figure 9. Photomicrograph of the 1-MAAQ Test Aerosol
21
-------
NJ
10
7
5
4
i 2
01
I I0<
0>
o0.7
•H
4J
£ 0.5
0.4
0.3
10
-1
0
O
Mass Median Diameter = 1.6
Count Median Diameter = 1.27
an = 1.35
O.I
IO 50 9O 95 99
7o Less Than Stated Size
99.9
-------
3,1,3 Uranine Aerosol, 50 ym
The uranine aerosol generator utilized the motorized
gear drive mechanism of a Wright dust feeder to rotate a
grooved aluminum plate, Four V-shaped grooves at 1, 2, 3,
and 4 in, radii are machined in the face of the aluminum
plate, Interchangeable gears also provide the capability
for changing the dust feed rate. The 270 to 325 mesh frac-
tion of uranine particles was freshly sieved before each test
and 250 mg of the powder was placed in one of the grooves,
The powder was subsequently aspirated into the venturi disr,
penser as the groove rotated past the aspiration tube. The
venturi dispenser is miniaturized so that it operates effir-
ciently at a total output (primary and aspirated air) of less
than 7,1 1pm (0,25 cfm) the smallest sample volume used in
the statistical tests. The system, Figure 8B, works very
well with the easily deagglomerated large uranine particles.
The electrostatic charge distribution on the uranine
aerosol particles was determined by collection in a modified
Wesix Ion Spectrometer operated vertically at 500 volts, The
charge from the collected particles was collected on a capaci-
tor and the voltage increase was monitored with a Keithly
electrometer, The average uranine particle carried 1,4 x
/ ^
10 negative charges and 6,5 x 10 positive charges. Kunkel (1)
found charges as high as 3 x 10 on air dispersed particles
in the size range 0,5 to 30 urn. He also noted that asymmetric
charging occurred in a heterogenous system with negative bias
for the large particles. Our measurements agree with Kunkel's.
He also observed that humidity did not affect charging. Thus,
differences in charging due to humidity changes of different
days during the test series should not be a significant fac-
tor, Charge measurements were not made on the KC1 and 1-MAAQ
condensation aerosols because of the insignificant charge
levels known to exist on such condensation aerosols. The
(1)
Kunkel, W, B., J. Appl. Physics, 21, 820 (1950),
23
-------
effect of the presence of charged particles in ' V~e deposition
on conduit walls tends to increase with '-he charge level
Figure 11 shows a photomicrograph and Figure 12 shows
the size distribution of the sieved uranine powder , The par-
ticles, while irregular in shape, are narrowly sized and dis-
persed readily as single particles with few associated fines,
Some attrition and fines formation undoubtedly occurs during
aerosol generation and transport, but the collected aerosol
samples show less than ~270 of the mass as fines
3,2 Test Procedure
The following procedure was followed in the Phase I tests:
(1) Two Gelman Type A glass fiber filters were inserted
in the filter holder (as shown in Figure 3) prior to assembly
in the sampling interface. Double filters were used to estab-
lish whether quantitative collection was obtained on the first
filter, a concern in the tests with the 0 05 _.m aerosol.
(2) The solvent-extracted clean porous tube was partly
inserted in the probe sheath from the rear. Rubber gloves
were used during all handling operations to a\oid potassium
contamination from perspiration, The rear 0-ring was placed
on the end of the porous tube and the. tube was then inserted
the rest of the way, The downstream filter unit (as shown in
Figure 3) was then screwed tightly with the spanner wrench
into the probe sheath, automatically positioning the porous
tube. The forward 0-ring was then placed on the porous tube
and the mating collar adapter was tightly fastened before the
standard nozzle was attached (Figure 2)
(3) The transpiration air and exhaust air flows were
turned on and adjusted to the desired levels as described pre-
viously in Section 2.2, During these adjustments, a protec-
tive inlet filter prevented contamination of the porous probe
interface.
24
-------
* *
* * / v*
**
...
+ I
Figure 11. Photomicrograph of the 55 ym Uranine Test Powder
25
-------
Ni
1000
700
500
400
300
Mass Median Diameter =
Count Median Diameter =
Average Aspect Ratio =
58
55
1.27
1.15
O.I
IO 5O 9O 95 99
% Less Than Stated Size
99.9
-------
(4) When the aerosol generator was ready for the ini-
tiation of the test, the protective inlet filter was removed
from the nozzle and the aerosol was injected directly into the
sampling nozzle for a test period which for the series ranged
from 1% minutes to as long as six days,
(5) At the end of the test the aerosol generator was
detached from the sampling nozzle and the protective inlet
filter was reattached to the nozzle.
(6) In order not to disturb any deposited particles by
sudden changes in flow conditions during shutdown, the exhaust
flow and the transpiration air flow were reduced by steps,
i.e., the exhaust flow was reduced until the sample flow ap-
proached zero, followed by a reduction in the transpiration
air flow until the sample flow increased to the original test
value. This cycle was repeated as many times as necessary
until a complete shutdown was reached,
(7) The sampling nozzle was carefully removed and inter-
nally deposited particles were solvent extracted and quanti-
tatively analyzed.
(8) The porous tube was removed after insertion of a
small stopper in the upstream opening. The porous tube with
stopper was inserted in a pyrex extraction column with the
stoppered end down, Again, rubber gloves were used since
skin fluids were found to be a significant source of potassium
at the trace levels that were measured.
(9) The appropriate solvent (distilled water for KC1 or
uranine and acetone for 1-MAAQ) was poured inside the porous
tube so that the solvent passed through the tube walls.
Quantitative extraction was achieved with 1-3 liters of sol-
vent. The porous tube extract was quantitatively analyzed.
(10) The glass fiber collection filter was photographed,
solvent extracted, and quantitatively analyzed.
27
-------
(11) The porous tube was washed with an additional
4 liters of solvent, followed by a final acetone rinse before
air drying in preparation for the next test,
3,3 Chemical Analysis
Uranine was analyzed by colorimetric and fluorometric
techniques, A Bausch and Lomb Spectronic 20 Colorimeter was
used to measure uranine solution optical density at 485 nm
and a Turner Model 111 Fluorometer was used for more sensi-
tive detection at trace levels.
1-MAAQ was analyzed colorimetrically at 500 nm, Solutions
were concentrated when necessary to obtain tneasureable levels
of optical density,
Potassium was analyzed by flame emission with a Jarrell
Ash Model 82-528 Atomic Absorption/Flame Emission Spectropho-
tometer, Most of the porous tube extracts were concentrated
10/1 but, because of the low mass concentration of the 0.05 ym
test aerosol and the very low deposition, many of the deposi-
tion levels are reported as upper limits. The actual amounts
deposited could be less than the reported levels.
28
-------
4. TEST DESIGN. DATA. AND OBSERVATIONS
4,1 Test Design
3
A balanced % replicate of a 2 x 3 factorial experiment
in which sample flow rate, transpiration flow rate, and par-
ticle size each have three levels and particle concentration
has two levels was designed. Seven added combinations were
subsequently added to the design to show comparative results
with the equivalent of a standard EPA stack sampling probe
with a 1.27 cm (% in.) ID pyrex tube in place of the porous
metal tube. The experiment design is described in greater
detail in Section 5 of this report,
4.2 Test Data and Observations
4,2,1 Tabular and Graphic Data
Table 1 summarizes the test data obtained. The percent
probe deposition was calculated independently of the nozzle
deposition since the standard nozzle used for these tests
would not be used in a final prototype interface, Thus, the
percent probe deposition in Table 1 is for the porous tube
only. The aerosol concentration could not be fixed at precise
levels in the experiment design but was either "high" or "low"
on a relative basis for a given size aerosol, The number
concentration of the test aerosols ranged from a low of 1.7/cc
for the 50 ym uranine to a high of 8 x 10 /cc for the 0.05 ym
3
KC1. The mass concentration varied from a low of 0,08 mg/m
3
for the KC1 to a high of 5,120 mg/m for the uranine aerosol,
The percent deposition on the porous tube as a function
of transpiration air flow and sample flow rate are shown in
Figures 13-15, The points are plotted with an adjacent + or -
sign to indicate the experiment design and concentration of
high or low, respectively. Concentration was expected to have
a very low order effect on deposition (confirmed in the sta-
tistical analysis of Section 5 in this report), allowing a
realistic graphic representation of the experimental data
which was a fractional design with respect to concentration,
29
-------
Table 1. AEROSOL MASS TRANSPORT TEST DATA FOR PHASE I EXPERIMENTS WITH THE IITRI BOUNDARY LAYER SAMPLING PROBE
Test Sequence
Combination
Sample, 1pm
Transpiration, 1pm
Particle Size, pm
Concentration
Probe
Aerosol Samples
(Filter + Probe), mg
Aerosol ,
Concentration,* mg/m
Aerosol Concentration,
Particles/cc
Aerosol Deposited
on Probe, mg
Aerosol Deposited
on Nozzle, mg
Probe Deposition,** %
Clean Air Sheath Noted
Filter Deposit Dia., cm
1
27
28.3
283.0
44-55
High
2 ym SS
212.1
5120.0
51.2
0.120
0.012
0.057
No
9.5
IB
-
28.3
0
44-53
High
Pyrex
199.9
4840 . 0
48.4
0.112
0.009
0.056
No
9.5
2
21
28.3
14.2
44-53
Low
2 pm SS
211.0
975.0
9.8
0.0362
0.010
0.017
No
9.5
3
7
7.1
283.0
0.05
High
2 yin SS
2.33
0.76
5.9xl07
<0.004
0.02
<0.2
Yes
2.5
3B
-
7.1
0
0.05
High
Pyrex
2.18
0.53
8-OxlO7
0.050
0.02
2.3
NV
4
19
28.3
14.2
0.05
High
2 urn SS
1.33
0.08
5.9xl06
<0.004
<0.02
<0.3
NV
5
3
7.1
14.2
44-53
High
2 urn SS
97.8
2300.0
23.0
96.4
70.7
98.9
Yes***
6.0
5B
-
7.1
0
44-53
High
Pyrex
31.0
749.0
7.5
29.9
109.7
96.5
No
9.5
6
9
7.1
283.0
44-53
Low
2 vim SS
57.4
272.0
2.7
0.0248
98.7
0.043
No
9.5
7
1
7.1
14.2
0.05
Low
2 vim SS
2.18
0.23
l.SxlO7
<0.004
<0.02
<0.2
Yes
6.2
8
25
28.3
283.0
0.05
Low
2 vim SS
2.02
0.08
5.9xl06
0.114
<0.02
5.7
NV
9
18
14.2
283.0
44-53
Low
2 ym SS
191.1
795.0
8.0
0.0671
0.036
0.035
No
9.5
(continued)
-------
Table 1 (continued). AEROSOL MASS TRANSPORT TEST DATA FOR PHASE I EXPERIMENTS WITH THE IITRI BOUNDARY LAYER SAMPLING PROBE
Test Sequence
Combination
Sample, 1pm
Transpiration, 1pm
Particle Size, ym
Cone entr a tion
Probe
Aerosol Samples
(Filter + Probe) , mg
Aerosol ,
Concentration,* mg/m
Aerosol Concentration,
Particles/cc
Aerosol Deposited
on Probe, mg
Aerosol Deposited
on Nozzle, mg
Probe Deposition,** %
Clean Air Sheath Noted
Filter Deposit Dia., cm
10
12
14.2
14.2
44.53
High
2 urn SS
156.9
2750.0
27.5
155.0
19.7
98.8
Yes
6.8
10B
14.2
0
44-53
High
Pyrex
30.0
689.0
6.9
29.4
22.0
97.9
No
9.5
11
6
63.7
63.7
44-53
Low
2 ym SS
39.1
173.0
1.7
37.5
95.4
96.0
Yes
2.8
12
5
7.1
63.7
1.6
Low
2 urn SS
28.4
91.2
3.0xl04
0.0168
0.024
0.059
Yes
2.9
13
16
14.2
63.7
0.05
High
2 ym SS
3.68
0.23
l.SxlO7
-0.020
<0.01
0.5
Yes
NM
14
22
28.3
283.0
0.05
Low
2 ym SS
2.48
0.08
5.9xl06
<0.010
<0.02
<0.4
NV
15
24
28.3
63.7
44-53
High
2 ym SS
177.0
3950.0
39.5
0.0195
0.004
0.011
No
9.5
16
23
28.3
63.7
1.6
High
2 ym SS
176.5
160.0
5.2xl07
0.0335
0.020
0.019
Yes
«« 5 QA)fe"A""Ar
17
4
7.1
63.7
0.05
High
2 ym SS
3.74
0.50
3.8xl07
<0.008
0.002
<0.02
Yes
5.0
18
11
14.2
14.2
1.6
High
2 ym SS
85.7
276.0
9 . IxlO4
0.0252
0.016
0.029
Yes
7.3
19
14
14.2
63.7
1.6
Low
2 ym SS
38.3
90.2
3 . OxlO4
0.0158
0.009
0.041
Yes
3.9
(continued)
-------
Table 1 (continued). AEROSOL MASS TRANSPORT TEST DATA FOR PHASE I EXPERIMENTS WITH THE IITRI BOUNDARY LAYER SAMPLING PROBE
Test Sequence
Combination
Sample, 1pm
Transpiration, 1pm
Particle Size, urn
Concentration
Probe
Aerosol Samples
(Filter + Probe) , mg
Aerosol „
Concentration,* mg/m
Aerosol Concentration,
Particles/cc
Aerosol Deposited
on Probe, mg
Aerosol Deposited
on Nozzle, mg
Probe Deposition,** %
Clean Air Sheath Noted
Filter Deposit Dia., cm
20
17
14.2
283.0
1.6
Low
2 ym SS
34.2
78.0
2 . 6xl04
0.0063
0.018
0.018
Yes
-4.0
21
20
28.3
14.2
1.6
High
2 ym SS
80.7
168.0
5 . 5xl04
0.0567
0.024
0.070
No
9.5
22
13
14.2
283.0
0.05
Low
2 ym SS
1.37
0.09
7 . IxlO6
<0.004
<0.02
<0.2
Yes
5.0
23
10
14.2
14.2
0.05
Low
2 ym SS
9.07
0.08
5.9xl06
<0.020
0.02
<0.2
Yes
7,6
24
26
28.3
283.0
1.6
Low
2 urn SS
25.7
47.1
l.SxlO4
0.2963
0.009
1.15
No
9.5
25
15
14.2
63.7
44-53
High
2 ym SS
134.9
3090.0
30.9
80.0
22.6
59.7
Yes
3.0
26
8
7.1
283.0
1.6
High
2 ym SS
14.0
198.0
6.5xl04
0.0032
0.007
0.022
Yes
-2.5
27
2
7.1
14.2
1.6
Low
2 ym SS
9.91
56.1
1 . SxlO4
0.0158
0.004
0.16
Yes
6.1
27B
-
7.1
0
1.6
High
Pyrex
32.6
219.0
7 . 2xl04
1.015
0.020
3.1
No
9.5
18B
-
14.2
0
1.6
High
Pyrex
87.6
193.0
6.3xl04
1.223
0.029
1.4
No
9.5
21B
-
28.3
0
1.6
High
Pyrex
90.9
117.0
3.8xl04
0.734
0.034
0.81
No
9.5
* At upstream end of porous tube.
** Porous tube only.
*** 0.3 pm to 12 pm fines transported with no deposition (see filter photo).
**** Not evident in photograph but estimated from visual appearance of eroded heavy deposit.
NV not visible
NM not measured
-------
10
u>
7 .1 1pm Sample
28.3 1pm Sample
14.2 1pm Sample
Pyrex Tube us
ed for 0 Level
0
0.1
Figure 13.
50 90 95 99 99.9
% Deposition
Deposition of 50 ym Uranine Particles in the 1.27 cm Diameter
by 178 cm Long Porous Sampling Probe
-------
GJ
-P-
14.2 1pm
Sample
0
28.3 1pm Sample
Pyrex Tube used for 0 Level
0.1
"Figure
10
50
% Deposition
90 95
99
99.9
of 1 . 6 vim 1-MAAQ Particles in the 1.27 cm Diameter
-------
10
ui
8
1
§ 5
•H
03
•H
8- *
CO
£
0
28.3 1pm Sampl
Pyrex Tube
used for 0 Level
0.1 1
Figure 15.
10
90 95
99
50
% Deposition
Deposition of 0.05 urn KC1 Particles in the 1.27 cm Diameter
by 178 cm Long Porous Sampling Probe
99.9
-------
The experiments with 50 ym uranine aerosols, Figure 13,
shows that when the sample flow rate is 7„1-14.2 1pm (0.25-
0,50 cfm), the use of 283 1pm (10 cfm) transpiration air re-
duces deposition from about 98 percent to less than 0,1%. At
28,3 1pm (1 cfm) sample flow rate, deposition was less than
0.1% regardless of the amount of transpiration air used. This
indicates that the greatest advantage of the porous probe in-
terface is in the quantitative transport of these large par-
ticles when the sample flow rate is low.
The tests with the 1.6 ym 1-MAAQ aerosols, Figure 14,
shows that deposition was virtually eliminated with only
14.2 1pm (0.5 cfm) of transpiration air at all sample rates.
The downward deposition trend continued for the 7.1-14.2 1pm
(0.25-0.50 cfm) sample rates as the transpiration air in-
creased to 283 1pm (10 cfm). An exception to this continuing
trend is noted with the 28.3 1pm (1.0 cfm) sample rate curve
where after dropping 3% to 0.07% deposition at 14,2 1pm
(0.5 cfm) of transpiration air, deposition increased to 1-2%
with further increases in transpiration air„ This phenomenon
is believed to be related to the jetting action of the air as
it issues from the individual micropores in the porous tube.
Microturbulence certainly exists close to the tube wall before
the air jets disappear and merge as a relatively smooth flow
farther away from the wall. At the tube entrance the in-
coming aerosol is quite near the wall where microturbulent
deposition can occur. As the aerosol stream progresses down
the tube it tends to be squeezed inward and accelerated down-
stream due to the influx of transpiration air. At the higher
sample flow rate, a longer distance of travel is needed to
squeeze the aerosol stream away from the microturbulent re-
gion near the wall; this accounts for the higher deposition
of 1.6 ym particles at the higher sample flow rate. Deposition
generally appeared to be concentrated in the first foot of
travel down the porous probe.
36
-------
The same phenomenon occurred in the tests with the
0.05 vim KC1 aerosol, Figure 15. After a virtual elimination
of deposition with only 14.2 1pm (0.5 cfm) of transpiration
air at all sample flow rates, the deposition at the 28.3 1pm
(1 cfm) sample flow rate significantly increased to 6% with
increasing transpiration air flow. At sample flows of 7.1-
14.2 1pm (0.25-0.50 cfm) the deposition remained below 0.5%
when the transpiration air flow was in the range of 14.2-
283 1pm (0.5-10.0 cfm).
4.2.2 Photographic Proof of Effectiveness
As stated in Section 2.1.2, the glass fiber collection
filter often served as a witness plate for confirmation of
the existence of a particle-free air sheath surrounding the
aerosol stream. Photographs were taken of the filters after
tests with one of the colored aerosols, i.e., uranine or
1-MAAQ, Figures 16 and 18, and measurements were taken of the
filter deposit diameter, Table 1. Photographs were not taken
of the KC1 deposits on the collection filters due to insuf-
ficient contrast, but in many tests with the KC1 the deposit
could be seen visually when viewed at a shallow angle. The
KC1 deposits were similar in size and shape to those in com-
parable tests with 1-MAAQ, confirming the similar probe
deposition behavior for the two aerosols.
Examination of the glass fiber filter sample from test
5 revealed an unexpected wealth of quantitative data on the
effectiveness of the porous probe at these flow conditions
for particles smaller than 50 ym. Generation of the 50 urn
uranine aerosol unavoidably produced a very small amount of
secondary small particle aerosol associated with the coarse
particle fraction. Thus, while test 5 was a failure with
respect to transport of 50 ym particles, it was an unquali-
fied success in transporting smaller particles. Figure 16
shows the fine particle deposit on the glass fiber filter
completely surrounded by a particle-free clean air sheath.
37
-------
CO
oo
Figure 16. Particle Deposit on
Glass Fiber Filter Showing
Laminar Flow with Clean Air
Sheath, Test 5
Figure 17. Photomicrograph of
Particle Deposit on Glass
Fiber Filter of Test 5
-------
test £7
0.25 CM Sample
0 50 CFM Transpiration
Test 12
0.25 CFM S
2.25 CFM Tra-
St 18
50 CFM Sample
50 CFM Transpiration
Test 19
0.50 CFM Sample
2.25 CFM Transp
rion
Test 21.
1.0 CFM Sample
0.50 CFM Transpiration
Test 16
1.0 CFM Sample
2.25 CFM Transpiration
Test 26
0.25 CFM Sample
10.0 CFM Transpiration
Test 20
0,50 Sample
10.0 CFM Transpiration
Test 24
1.0 CFM Sample
10,0 CFM Transpiration
Figure 18. Photographs of Filter Deposits of 1-MAAQ Aerosols
which Confirm Effectiveness of Boundary Layer Principle
for Aerosol Transport
39
-------
At these flow conditions 7.1 1pm (0.25 cfm) sample volume
and 14.2 1pm (0.5 cfm) transpiration air, the flow was lam-
inar such that the aerosol filament of fine particles per-
sisted downstream of the porous probe right up to the face
of the filter. The outline of the deposit was displaced
slightly downward from the center of the filter. Nonetheless,
particles in the size range observed on the filter were trans-
ported with essentially no deposition on the porous probe
with 14.2 1pm (0.5 cfm) of transpiration air. Figure 17
shows a photomicrograph of a portion of the particle laden
area of Figure 16. Particles on the filter ranged in size
from 0.3 ym to 15 ym. Since no uranine particles in this
size range were observed in the clean-air-sheath portion of
the filter, it follows that uranine particles in this size
range will not deposit on the sampling probe at these flow
conditions once the sheath has been formed.
The 1-MAAQ filter deposits are shown in Figure 18, Note
the tendency for the deposit to spread as the sample flow
rate is increased from 7.1-28.3 1pm (0.25-1.0 cfm), A-B-C,
D-E-F, and G-H-I of Figure 18. The extreme spread at 28.3 1pm
(1.0 cfm) sample flow is apparently related to the microtur-
bulent diffusion effect near the probe wall as discussed in
Section 4.2.1.
With a constant sample flow rate and increasing trans-
piration air the deposit tends to reduce in size, A-D and
B-E of Figure 18. At 283 1pm (10 cfm) of transpiration air,
G and H of Figure 18, the gross deposit covers a larger area,
but close visual examination of the deposit shows that the
triangular shape of the deposit is caused by the axial dis-
placement of the aerosol filament from the geometric center
of the probe opening. The displacement of the aerosol stream
is in turn caused by the eccentricity of the bore in the down-
stream portion of the porous tube. The comparable test with
0.05 ym KC1 at 7.1 1pm (0..25 cfm) sample and 283 1pm (10 cfm)
transpiration air revealed a deposit the size of a quarter.
40
-------
Because of the low aerosol mass, the KC1 deposit did not
smear and was sharply outlined.
In test 23 with 0.05 ym KC1, a six day test was conducted
with 14.2 1pm (0.5 cfm) of transpiration air and 14.2 1pm
(0.5 cfm) of sample volume to obtain a visible deposit on
the collection filter. The filter deposit from this test
showed clearly that the KC1 particulates were completely con-
fined within a clean air sheath. The deposit was circular in
shape with a diameter of 76 mm on the 95 mm diameter open face
of the filter. At no point did the deposit come closer than
8 mm to the outside portion of the filter sealed by the re-
taining ring. If no diffusion of the aerosol had occurred
during transport through the probe, the area of the deposit
would have covered 50% of the available area of the collection
filter (50-50 volume ratio of aerosol/transpiration air). The
deposit actually covered 64% of the filter area, suggesting
that despite some diffusion toward the probe wall a lower
transpiration air flow rate would have been effective for
0.05 um particles. At 14.2 1pm (0.5 cfm) the average face
velocity of the existing TA is 3.4 mm/sec, well in excess of
the actual diffusion velocity of 0.9 mm/sec for the 0.05 ym
particles in this test.
The existence of an aerosol filament surrounded by a
clean air sheath was observed directly in test 26 as well as
in many other tests with 1-MAAQ. A pencil-lead thin red aero-
sol filament was clearly visible in the pyrex tube of the
filter holder, Figure 3. The filament observed was invari-
ably stable with no visible spreading over the distance ob-
served. The existence of a stable aerosol filament at 283 1pm
(10 cfm) of transpiration air flow is surprising in view of
the high Reynolds Number (-32,000) and is a phenomenon res-
tricted to a region within 1-2 mm of the center of the opening
in the porous tube. Closer to the porous tube wall turbulent
eddies exist, increasing in scale as the wall is approached,
but the very center of flow at the tube axis appears to be
laminar at least for the 10 tube diameters from the downstream
end of the porous tube to the face of the collection filter.
41
-------
5, STATISTICAL ANALYSIS OF PHASE I TEST DATA
The effects of four controlled variables -- sample flow
rate, transpiration flow rate, particle size, and particle
concentration -- on mass percent particle deposition in tubu-
lar sampling probes have been investigated experimentally
with a fractional factorial design, The results of statisti-
cal analysis of the data are presented here and the principal
conclusions are stated.
5.1 Definition and Coding of Variables
Dependent variable. The property (dependent variable)
measured in each test is the mass percent particle deposition
in the tube, denoted by PD. Prior to the statistical analysis
of the data a mathematical transformation was made, yielding
a working dependent variable, Y, which is functionally re-
lated to PD as follows (forward and backward transformations):
Y = log1Q (PD/(100-PD))
PD = 100 (10Y)/1 + 10Y)
PD is restricted to the range of values 0
-------
of the values physically. The levels are expressed below
both in physical units and corresponding orthogonally coded
values. The coded (X) variables are especially well suited
computationally for accurately estimating the effects of the
controlled factors. For the three factors tested at three
or four levels there are coded variables representing possible
quadratic (curvilinear) effects as well as linear effects.
Sample flow rate (SFR) was tested at three levels, evenly
spaced on a logarithmic scale.
Coded Variables
SFR, 1pm
7.1
14.2
28.3
Linear
Xl
-1
0
+1
Quadratic
X1Q
+1
-2
+1
The defining formulas for X-, and X,Q are:
Xx = (log10(SFR) - 1.523)/1.523
2
X1Q = 3X1 ~ 2
Transpiration flow rate (TFR) was tested at four levels.
Coded Variables
TFR, 1pm
0
14.2
63.7
283.0
Linear
x2
-3.000
-0.933
+0.933
+3.000
Quadratic
X2Q
1.000
-1.032
-1.032
1.000
The defining formulas for X2 and X2Q are:
X2 = (Iog10 (TFR + 4.389) - 1.5506)/0.30273
X9n = X^/4 - 1.25
^x *•
43
-------
The constant K = 0/155 in the formula for X2 was chosen so
that the four values corresponding to the four levels of TFR
are, after addition of K, as nearly equally spaced as possi-
ble on a logarithmic scale,
Particle size (PS) was tested at three levels, nearly
evenly spaced on a logarithmic scale,
Coded Variables
PS, ym
0,05
1.6
50.0
Linear
x3
-1
0
+1
Quadratic
X3Q
+1
-2
+1
The defining formulas for X~ and X~Q are:
X- - (login (PS) - 0
•j
-------
Table 2. EXPERIMENTAL DATA
Combination
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
SFR, 1pm
7.1
7.1
7.1
7.1
7.1
7.1
7.1
7.1
7.1
14.2
14.2
14.2
14.2
14.2
14.2
14.2
14.2
28.3
28.3
28.3
28.3
28.3
28.3
28.3
28.3
28.3
28.3
7.1
7.1
14.2
14.2
28.3
7.1
TFR , 1pm
14.2
14.2
14.2
63.7
63.7
63.7
283.0
283.0
283.0
14.2
14.2
14.2
63.7
63.7
63.7
283.0
283.0
283.0
14.2
14.2
14.2
63.7
63.7
63.7
283.0
283.0
283.0
0.00
0.00
0.00
0.00
0.00
0.00
0.00
PS
0.05
1.6
50.0
0.05
1.6
50.0
0.05
1.6
50.0
0.05
1.6
50.0
0.05
1.6
50.0
0.05
1.6
50.0
0.05
1.6
50.0
0.05
1.6
50.0
0.05
1.6
50.0
50.0
0.05
50.0
50.0
1.6
1.6
1.6
PC
Low
Low
High
High
Low
Low
High
High
Low
Low
High
High
Low
Low
High
High
Low
Low
High
High
Low
Low
High
High
Low
Low
High
High
High
High
High
High
High
High
Percent
Deposition
<0.2
0.16
98.9
<0.2
0.059
96.0
<0.2
0.022
0.043
<0.2
0.029
98.8
<0.2
0.041
59.7
0.5
0.018
0.035
<0.3
0.07
0.017
<0.4
0.019
0.011
5.7
1.15
0.057
0.056
2.3
96.5
97.3
1.4
0.81
3.1
45
-------
2
than a % replicate of a 2 x 3 x 4 factorial consisting of 72
possible combinations of factor levels
5 3 Data Base
The levels of the four controlled variables in each of
the 34 test combinations, and the observed percent deposition,
are given in Table 1, The percent deposition ranges from a
low of 0.011 to a high of 98,9, In seven tests involving the
smallest particle size the percent deposition could not be
exactly measured; however, an upper limit for PD was estab-
lished in each such test as noted in the table (combinations
1, 4, 7, 10, 13, 19, and 22)c
5,4 Candidate Terms of the Multiple-Regression
Per formance Moder
The framework for the statistical analysis of the data
is provided by a model equation that incorporates a set of
terms considered capable of representing the ways in which
the percent particle deposition might be affected by the con-
trolled factors of the experiment. The dependent variable
of the model is Y, as defined above. The candidate indepen-
dent variables of the model are given in Table 3. These are
the X variables defined above together with some simply de-
rived variables. There are 14 candidate terms altogether,
including the constant term. The variable components of the
linear and quadratic terms are exactly as defined above. The
variable components of the six interaction terms are the
pairwise products of the linear variables,
The complete model has the form
Y - b0XQ + b^ + . . . + b^ + b1QX1Q
+ b3QX3Q + b!2X!X2 + " • + b34X3X4 + e
A
= Y + e
Where Y stands for the observed values of the dependent vari-
A
able, Y stands for the corresponding values of the dependent
46
-------
Table 3. SET OF CANDIDATE PREDICTIVE VARIABLES FOR
THE SAMPLING TUBE PERFORMANCE EQUATION
Constant Term
1, XQ = 1
Linear Terms
2. X, Coded sample flow rate
3. X« Coded transpiration flow rate
4. X~ Coded particle size
5. X, Coded particle concentration
Quadratic Terms
6. X,Q Coded sample flow rate
7. X2Q Coded transpiration flow rate
8. XOQ Coded particle size
Interaction Terms (Product Variables)
9.
10.
11.
12.
13.
14.
a\
1 /
/ v \
\"**1 /
/ V \
V*^l /
(X2)
(X2)
(X3)
(X2)
(X3)
(X4)
(x3)
(X4)
(X4)
47
-------
variable computed from the expression on the right, the X's
are the coded values of the independent variables, the b's
are the regression coefficients to be estimated from the ex-
perimental data, and e represents the differences between
observed and computed values of the dependent variable due
to residual variation or "experimental error" in the observations
5.5 Methods of Data Analysis
The experimental data were analyzed by the method of
least squares in accordance with the model, resulting in the
actual performance equation that is presented in the next
section.
The computer program used, BMD-02R, performs stepwise
multiple regression -- i.e., the equation is built up, term
by term, by introducing at each step that candidate term which
will result in the greatest reduction in the sum of squared
deviations between the observed values of the dependent vari-
able and the values computed from the regression equation (the
error sum of squares), A cutoff point for this process can
be set by the user of the program through choice of a thres-
hold "F" value. The F value associated with the coefficient
of a term in a regression equation is the square of the ratio
of the coefficient, to its standard error. In other words,
no candidate term is introduced into the equation unless the
value of the coefficient of that term is a specified multiple
of its standard error. This excludes from the equation terms
with coefficients of a magnitude that could readily arise due
merely to the inevitable background or residual variation
between measurements (experimental error), In the present
analysis the threshold F value was set to correspond to a
probability of 10%, i,et) no term was introduced if the re-
gression coefficient was so small as to have that probability
or higher of occurring due to random variation in the data.
An elaboration of the stepwise fitting procedure was
made necessary by the occurrence of the tests in which upper
48
-------
limits on percent deposition were obtained instead of point
values. Such outcomes are called "censored observations" in
statistical terminology. An iterative method was used to ob-
tain the least-squares estimates of the regression coefficients
taking into account the limit values resulting from the cen-
sored tests in conjunction with all point values. The step-
wise regression procedure was performed at each iteration.
In the initial iteration the limit values from the cen-
sored tests were treated as if they were point values. Each
limit value was then compared with the corresponding value
computed from the fitted equation. If the computed value was
smaller than the observed limit values, a trial value smaller
than the limit value was substituted for the latter, to be
used in the next iteration. If the computed value was no
smaller than the limit value, the latter was retained. The
iterative process was continued until a solution was reached
at which, 1) all limit values smaller than corresponding com-
puted values were retained as data points, and 2) all trial
values from the previous iteration were equal to the corres-
ponding computed values from the current iteration.
Information provided as computer output in conjunction
with the fitted equation includes: an overall analysis of
variance with respect to the terms in the equation and the
residual degrees of freedom, the residual standard deviation,
the multiple correlation coefficient, the coefficient of
determination, the computed value and the standard error of
each regression coefficient, the F ratios for all terms in
the equation and all candidate terms not in the equation,
and individual residuals for all observations.
5.6 The Particle Deposition Equation
The final particle deposition equation is presented in
Table 4. Of the 14 candidate terms (Table 3) nine are in
the equation; the remaining five were excluded in the fitting
process because of lack of statistical significance at the
49
-------
Table 4, PARTICLE DEPOSITION EQUATION
Dependent Variable:
Data Base:
Number of Tests:
Number of Terms in Equation:
Residual Degrees of Freedom:
Residual Standard Deviation:
Coefficient of ~
Determination (R ):
Independent Variable
Constant Term
XQ = 1
Linear Terms
X
Ll
[2
Y - Iog10 (PD/C100-PD)); PD is
mass percent deposition
All test results
34
9
25
1,0659
0,7505
Regression
Coefficient
b
-2,2889
X,
X,
Sample Flow Rate,
Coded
Transpiration Flow
Rate, Coded
Particle Size, Coded
Particle Concentration,
Coded
Quadratic Term
XOQ Particle Size, Coded
Interaction Terms
(xt) (X2)
(xx) (X3)
(x2) (x3)
-0 4960
-0.2002
0,8011
0=3327
0,3246
0,2941
-1,3172
-0,2355
Standard Variance
Error Ratio
0-2258
0,0933
0,2381
0 2105
0,1292
0.1046
0,2888
0„1142
4,83
4,60
11,32
2,50
6.31
7 ,,91
20,80
4,25
50
-------
10% probability level. All four candidate linear terms ap-
pear (the term for particle concentration is of marginal sig-
nificance) . Of the three candidate quadratic terms only the
term referring to particle size is present. Of the six can-
didate interaction terms three are present, representing in-
teractions between 1) sample flow rate and transpiration
flow rate, 2) sample flow rate and particle size, and 3) trans-
piration flow rate and particle size, The F values associated
with all the terms in the equation except the term involving
X, (coded particle concentration) are sufficiently large that
the effects are clearly of importance in understanding and
predicting particle deposition in porous or non-porous
sampling tubes.
The effects represented by the individual terms of the
equation can be examined separately. For instance, the linear
term for particle size has a relatively large F value, indi-
cating quite a strong effect, and the regression coefficient
is positive, so that as particle size increases the rate of
deposition also increases. The quadratic term for particle
size is definitely significant and the coefficient is posi-
tive, so the curve associated with the quadratic effect is
concave upward, and so on. To understand the net effect on
particle deposition (taking into account all the terms of
the equation) as the four controlled factors, or any subset,
are varied, tables and families of curves can be constructed
by solving the equation for specified combinations of factor
levels. Curves of this nature are presented below.
5.7 Residuals
The observed and calculated values are listed in Table 5
for both the transformed dependent variable Y and the percent
deposition PD. The calculated PD values were obtained by
backward transformation of the calculated Y values. The Y
residuals (observed minus calculated values) are also given.
As noted, in five of the seven cased in which the recorded
51
-------
Table 5. OBSERVED AND CALCULATED VALUES OF THE
DEPENDENT VARIABLE AND RESIDUALS
/-i ___1_ 4
Combi-
nation
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
YOBS
<-2.698
-2,795
1.954
<-2.698
-3.229
1.380
<-2.698
-3.657
-3.366
<-2.698
-3.537
1.916
<-2.698
-3.387
0.171
-2.299
-3.745
-3.456
<-2.522
-3.155
-3.769
<-2.396
-3.721
-3.959
-1.219
-1.934
-3.244
-3.252
-1.628
1.440
1.557
-1.848
-2,088
-1.495
YCAL
-3.678
-2,314
1.664
-3.495
-3.236
-0.364
-4.030
-3.592
-1.874
-3.131
-2.419
-0.424
-3.065
-3.458
-1.237
-2.327
-3.871
-2.803
-1,919
-3.189
-3.177
-1.970
-3.014
-2.776
-1.289
-3.485
-3.069
-2.219
-2.478
3.172
0.477
-2.005
-3,383
-0.627
Residual
Y -Y
*OBS XCAL
•>'<
-0.482
0.290
*
0.007
1.744
*
-0.065
-1.494
i<
-1.119
2.340
*
0.070
1.408
0.028
0.127
-0.653
-0.603
0.034
-0.592
-0.427
-0.707
-1.183
0.070
1.551
-0.175
-1.033
0,850
-1.732
1.080
0.157
1.295
-0.868
Percent
PDOBS
<0.2
0.160
98.9
<0.2
0.059
96.0
<0.2
0.022
0.043
<0.2
0.029
98.8
<0.2
0.041
59.7
0.500
0.018
0.035
<0.3
0.070
0.017
<0.4
0,019
0.011
5.70
1,15
0.057
0,056
2.30
96,5
97.3
1.40 •
0.810
3.10
Deposition
PDCAL
0,021
0,483
97.9
0.032
0.058
30,2
0.009
0.026
1.325
0.074
0,380
27.4
0.086
0.035
5.48
0.469
0.013
0.157
1.191
0.065
0.066
1,061
0.097
0.167
4.89
0.033
0,085
0,600
0,332
99,9
75,0
.0.979
0 , 041
1*9-. 11
* The observed value is an upper bound and the calculated
value is less than the upper bound.
52
-------
deposition is an upper bound the computed value is less than
that upper bound. In these five cases there is no meaningful
residual.
The estimate, s, of the standard deviation of residuals
is 1.066 (Table 4). The degree of determination, R2, is
0.75; in other wards, about 3/4 of the raw variability of Y
(as measured by sum of squared deviations from the mean) is
explained by the terms in the regression equation.
A plot of calculated vs. observed Y values (Figure 19)
shows no extreme outliers. One of the 29 points lies somewhat
outside the 2-sigma limits; this is test combination 12.
The cumulative percentage of residuals is plotted against
the values of the residuals on normal probability paper in
Figure 20. Considering the limited number of observations,
the approach to a straight line is reasonably good.
5.8 Plots of Functional Relationships Between Particle
Deposition and the Experimental Factors
Families of curves are presented (Figures 21-24), derived
from the particle deposition equation, to show the net ef-
fects of changes in the experimental factors. In each figure
the three separate lines represent percent deposition as a
function of transpiration flow rate for particles of size
0.05, 1.6, and 50 ym diameter. In Figure 21 particle concen-
tration is low and sample flow rate is low, 7.1 1pm (0.25 cfm) ,
In Figures 22, 23, and 24 particle concentration is high and
the sample flow rates are 7.1, 14.2, and 28.3 1pm (0.25, 0.5,
and 1.0 cfm), respectively. Figure 25 shows calculated per-
cent deposition as a function of particle size at interme-
diate levels of sample flow rate, transpiration flow rate,
and particle concentration.
5.9 Discussion and Conclusions
From the fairly small but carefully designed set of tests
that have been conducted, the major effects of sample flow
53
-------
-4h
-3 -2
-1 0
Y Observed
Figure 19. Plot of Observed vs. Calculated Values of the
Dependent Variable Y
(The line of perfect agreement and the + 2a lines are shown.)
54
-------
Ul
Ul
-2.0
i-H
a
o
-1.5
co
cu
-0.5
0.0
CO
£» 0.5
1.0
1.5
2.0
0.1
99.9
l 5 10 :>u yu
Cumulative Percentage of Residuals
Figure 20. Plot of Cumulative Percentage of Residuals on Normal Probability Paper
-------
ti
0)
u
M
0)
o
•I-l
4-J
•H
CO
O
fX
(U
Q
CD
,—1
O
•H
4-1
M
cfl
PH
99,9
99
90
50
20
10
5
1
0,5
0.2
0.1
0,01
0.001
7,1 1pm Sample Flow Rate
ym Particles
-1
-2
-3
-4
-5
14.2 63.7
Transpiration Flow Rate, 1pm
28.3
Figure 21. Deposition of Small, Medium, and Large Particles
vs. Transpiration Flow Rate with Low Particle Concentration
and Low Sample Flow Rate
56
-------
99.9
99
QJ
O
M
0)
Pk
o
•H
4J
•rl
0)
O
P
0)
H
O
•H
4J
M
efl
PM
90
50
20
10
5
1
0.5
0,2
0.1
0.01
7.1 1pm Sample Flow Rate
0
Vim Particles
Y
3
14.2 63.7
Transpiration Flow Rate, 1pm
283.0
0
-1
_2
-3
-4
-5
Figure 22, Deposition of Small, Medium, and Large Particles
vs. Transpiration Flow Rate with High Particle Concentration
and Low Sample Flow Rate
57
-------
4J
a
CU
o
^
a>
•H
CO
O
ft
0)
a
0)
iH
O
•H
4-J
M
CO
PH
80
50
30
10
5
1
0.5
0.2
0.1
0.01
0
14.2 1pm Sample Flow Rate
)0 ym Particles
0.05 ym Particles
1,6 ym Particle
_L
-1
-2
-3
-4
14.2 63.7
Transpiration Flow Rate, 1pm
283.0
Figure 23. Deposition of Small, Medium, and Large Particles vs
Transpiration Flow Rate with High Particle Concentration
and Medium Sample Flow Rate
58
-------
O
H
0)
PL,
o
•r)
•U
•rl
CO
O
(X
(1)
P
(1)
H
O
•H
4J
80
50
30
10
5
1
0.5
0.2
0.1
0.01
28.3 1pm Sample Flow Rate
0.05 ym Particles
1.6 ym Particles
Y
1
50 ym Particles
0
-1
-2
-3
-4
_L
1472 6377
Transpiration Flow Rate, 1pm
Z8J ,U
Figure 24. Deposition of Small, Medium, and Large Particles
vs. Transpiration Flow Rate with High Particle Concentration
and High Sample Flow Rate
59
-------
o
M
0)
Pk
4-)
•rl
W
O
P-.
0)
Q
O
50
10
1
0.5
.
cti
P4 0,05
14.2 1pm Sample Flow Rate
28.3 1pm Transpiration Flow Rate
Y
0
-1
-2
-3
4-4
U.05
_L
"1.6
Particle Size, ym
Figure 25. Deposition of Particles in Relation to Particle
Size at Intermediate Levels of Particle Concentration,
Sample Flow Rate, and Transpiration Flow Rate
60
-------
rate (SFR), transpiration flow rate (TFR), particle size (PS),
and particle concentration (PC), on the mass percent particle
deposition (PD) in porous sampling tubes have been clearly
brought out. A non-porous sampling tube was included as a
control. A single predictive performance equation was de-
veloped from the combined data incorporating the significant
relationships among the variables,
The analysis was facilitated by application of suitable
mathematical transformations to the variables. The indepen-
dent linear variables X-^, X2, X~, and X, are orthogonally
coded logarithmic transforms of the physical variables SFR,
TFR, PS, and PC, respectively, X-^, X20, and XOQ similarly
are quadratic variables for SFR, TFR, and PS. The dependent
variable Y, which is of the form log (p/(l-p)) where P is
a proportion, expands the PD scale with endpoints of 0 and
100% to a scale with unlimited range in both directions. The
transformations have the desired properties of simplifying
the representation of the systematic effects of the controlled
factors and providing approximate uniformity of variance for
the measure of performance. Results can be readily trans-
lated back into the original scale of physical measurement.
The statistical analysis of the data has been made within
the framework of a multivariate regression model. The model
includes candidate terms representing all the first and
second order effects, including interaction effects, of the
independent variables on which the experiment furnishes
evidence. The actual equation, developed from the data by a
stepwise least-squares method, includes nine of the 14 can-
didate terms. All four controlled factors are represented
in the equation. Graphs of some of the functional relation-
ships implicit in the equation have been constructed, Other
graphs and tables can be prepared if desired by solving the
equation for specified combinations of values of the indepen-
dent variables, including combinations not included in the
experiment.
61
-------
Some principal substantive conclusions are as follows:
(1) Of the four factors investigated, particle concen-
tration has the least effect on percent deposition. The
linear term in X^, which is marginally significant statisti-
cally, has a positive regression coefficient, indicating some
increase in percent deposition as particle concentration
changes from low to high, The candidate quadratic term of
this factor did not come into the equation, nor did any in-
teraction term involving the factor. Comparison of Figure 22
with Figure 21 shows the contrast between high and low par-
ticle concentration, i.e., the shift upward of all three
curves relating deposition to transpiration flow rate.
(2) Deposition depends strongly on the size of the par-
ticles being sampled. The equation includes both a linear
term in X., and a quadratic term in X~Q to represent the ef-
fect of this factor when the other factors are held constant
at their intermediate levels, and particle size is involved
in interactions with both sample flow rate (X~ times X,) and
transpiration flow rate (X~ times X2)„ The calculated curve
of percent deposition vs. particle size at intermediate levels
of SFR, TFR, and PC is shown in Figure 25, The curve com-
bines the linear effect associated with the variable X~ and
the quadratic effect associated with XOQ. Very little dif-
ference between 0.05 and 1 = 6 ym particles is indicated. The
50 ym particles, however, have a substantially higher deposi-
tion rate: between 5 and 10% as contrasted with the range
0.1 to 0.5% for the smaller particles. The three-way inter-
actions involving particle size, sample flow rate, and trans-
piration flow rate are illustrated by Figures 22, 23, and
24 and discussed below under point 5,
(3) Sample flow rate and transpiration flow rate are
both important factors with respect to the proportion of the
particle mass trapped in the sampling tube. The main effects
of both SFR and TFR are captured in the equation by linear
62
-------
terms -- in X, and X^, respectively; the absence of the quad-
ratic terms means that there is no evidence of a departure
from linearity in either case when other factors are held
constant. There is an interaction between SFR and TFR (term
in X, times X~) and an interaction of each with particle size
(terms in X, times Xn and H^ times X~). There is a certain
degree of parallelism in the effects of SFR and TFR as evi-
denced by the similarity in the terms brought into the equa-
tion and the fact that the coefficients of like terms have
the same sign.
(4) Since the sign of the term in X-, (coded sample flow
rate) is negative, the simple effect of increasing SFR, hold-
ing other factors at their intermediate settings, is to de-
crease the deposition. The interactions of SFR with TFR and
particle size substantially modify the influence of this
factor: compare Figures 22, 23, and 24, which differ due to
increasing SFR. The interactions involving SFR, TFR, and
PS are discussed under the following point,
(5) The effect of transpiration flow rate on percent
particle deposition is represented in the equation by a linear
term in X? and cross-product terms involving sample flow rate
(X~ times X,) and particle size (X,-, times X-) . The coeffi-
cients of these terms are all negative. Therefore, the main
effect of increasing transpiration flow from zero (non-porous
tube) to the maximum level tested (10 cu ft/min), with the
other factors at intermediate levels, is to reduce the pro-
portion of the particle mass that is deposited on the walls
of the sampling tube. Taking into account the interactions
with sample flow rate and particle size, and also the inter-
action between the latter two factors, one sees a more com-
plex picture. The relationships are shown by Figures 22, 23,
and 24. At all levels of sample flow rate tested, increasing
the transpiration flow rate is effective in lowering the de-
position of large particles. The slope (Y vs. X2) is steepest
at the low sample flow rate and becomes progressively less
-------
steep while remaining negative. Under conditions of a high
concentration of large particles and a low sample flow rate
of 7.1 1pm (0.25 cfm) the estimated effect of increasing the
transpiration flow rate from zero to 283 1pm (10 cfm) is to
reduce sharply the deposition, from more than 99.9% to about
5%, At the higher levels of SFR the percentage deposition
of large particles is brought considerably lower than at low
SFR as TFR increases from zero to 283 1pm (10 cfm), but the
reference values at zero TFR are also much lower. At low
SFR, the porous tube with increasing rates of transpiration
flow reduces deposition over the entire range of particle
sizes tested (Figure 22). At the intermediate SFR the data
indicate a continued effect of transpiration on reducing
deposition of medium sized particles, but little or no effect
on small particles. At high SFR, the range of variation of
particle deposition rate is markedly damped. Increasing the
TFR is indicated to actually increase the rate of deposition
of intermediate sized, and especially small, particles.
64
-------
PHASE II
PRESERVATION OF SIZE DISTRIBUTION OF AEROSOL
IN TRANSPORT THROUGH THE SAMPLING INTERFACE
65
-------
6 EXPERIMENTAL PROGRAM FOR PHASE II
Phase II is concerned with the preservation of aerosol
size distribution. The experimental program included a
series of tests with aerosols covering particle size range
of 0.05-50 ym. Originally, the three aerosols KC1, 1-MAAQ
and Uranine used in the Phase I were to be used for these
tests. On suggestion by EPA, another aerosol, flyash,
was added to the experimental plan. Sample flow rates of
7.1 1pm (0.25 cfm) and 28.3 1pm (1 cfm) were used. The
transpiration flow rate also varied from 14.2 1pm (0.5 cfm)
to 283 1pm (10 cfm). In addition, on suggestion by EPA, tests
with a 1.27 cm (%") diameter pyrex tube were also planned.
The experimental plan is presented in Table 6.
6.1 Tests with KC1 Aerosol
Two aerosols were used to cover the nominal size ranges
of 0.01-0.1 ym. For the 0.01-0.1 ym aerosol the generator used
on Phase I was used. It operates by vaporization from a plug
of salt in an electrically heated nichrome coil. At an air
flow of 7.1 1pm (0.25 cfm) and a voltage of 31 volts across
the 5 ohm coil, a satisfactory test aerosol was obtained.
In test K-l, the aerosol was directly transferred to the
porous-probe at a rate of 7.1 1pm (0.25 cfm). A transpiration
flow rate of 14.2 1pm (0.5 cfm) was used, In test K-2, the
sample flow rate was increased to 28.3 1pm (1 cfm) by adding
clean air to the aerosol. The transpiration flow rate was
maintained at 14.2 1pm (0.5 cfm). Size distribution at the
inlet to the porous probe and at the outlet were measured with
an electric mobility analyzer. The principle and operation
of this device is described in Appendix A.
For test K-3 and K-4, the nebulizer shown in Figure 26
was used to cover particle size range of 0,1-1.0 pm.
66
-------
Table 6. PHASE II EXPERIMENTAL PROGRAM
Test
No.
K-l
K-2
K-3
K-4
K-5
U-l
U-2
U-3
U-4
U-5
F-l
F-2
F-3
F-4
F-5
M-l
M-2
M-3
M-4
M-5
Aerosol
0.01-0.10 ym KC1
ii n M
0.10-1.0 ym KC1
n ii n
0.01-1.0 ym KC1
1-50 ym Uranine
n n
n ii
M it
n n
1-50 ym Flyash
n n
n ti
n n
ti n
1.6 ym 1-MAAQ
ii n
n n
M n
n M
Sample
Flow Rate
(1pm)
7.1
28.3
7.1
28.3
7.1
7.1
28.3
7.1
28.3
7.1
7.1
28.3
7.1
28.3
7.1
28.3
7.1
28.3
7.1
7.1
Transpiration
(1pm)
14.2
14.2
14.2
14.2
0
141.0
141.0
283.0
14.2
0
141.0
141.0
283.0
14.2
0
14.2
14.2
63.7
63.7
0
Probe
Porous
"
Porous
"
Pyrex
Porous
ti
"
"
Pyrex
Porous
"
"
"
Pyrex
Porous
"
"
"
Pyrex
Size Assessment
Technique
Electrical mobility
it M
it n
n n
M M
Imanco Quantimet 720
M n ii
M n M
n it n
M M n
Optical microscopy
tt it
ii M
n n
n n
„
ti M
ti n
ti n
n n
-------
Compressed air
Dilution air
Aerosol out
No. 80 hole
Figure 26, Nebulizer for KC1 Aerosol Generation
68
-------
For run #K-3, a 10% w/v solution of KC1 in water
was nebulized at a flow rate of 3.5 1pm (0.124 cfm) and was
diluted by 3.6 1pm (0.126 cfm) of clean dry air. This was
fed directly to the nozzle of the porous probe. For run
#K-4, the KC1 solution was nebulized at 7.1 1pm (0.25 cfm) air
flow rate and was diluted by adding dry clean air to 28.3 1pm
(1.0 cfm) .
For run #K-5, a different nebulizer was used to cover
the entire range 0.01-1.0 ym. This nebulizer is shown in
Figure 27. The total sample flow rate was 7.1 1pm (0.25 cfm).
The size distributions of the KC1 aerosol at the inlet
and at the outlet of the porous probe, for all tests with
KC1 were measured by the electric mobility analyzer
described in Appendix A. The data reduction is also described
in Appendix A.
The data are presented in Figures 28 to 32. The Y axis
represents the frequency of particles in a small differential
interval of the logarithm of the particle size. This is
plotted against the logarithm of particle size represented
on the X-axis. The symmetrical shape of these curves
(Figures 28-32) indicates that the distribution is log normal.
The geometric standard deviation is obtained by dividing
the GNMD by the size representing 60% of the peak value.
In Table 7, the size distribution parameters for all
the KC1 tests are presented. From these parameters log-
normal plots were constructed for all the runs (Figures 33-37).
69
-------
Compressed air
No. 80 hole
Aerosol
KC1 solution
Figure 27. Nebulizer for Test. K-5
70
-------
o
o
23
T)
S 5 x 10C
w'
T3
^d
A \
A.fc KC1
T Sample Flow 7.1 LPM
Transpiration Flow
\\ 14.2 LPM
O Upstream
Downstream
I i i
i I i i I I
0.005 0.01 0.02 0.05 0.1 0.2
Particle Size,
Figure 28. Size Distributions for Test Kr-1
71
-------
o
o
10
•o 6
5 x 10
0
I . I
O Upstream
Downstream
KCl
Sample Flow 28.3 LPM
Transpiration Flow 14.2 LPM
.01
.02
.05
.1
.2
Particle Size, ^m
Figure 29. Size Distributions for Test K-2
-------
10'
1-1
I
o
o
5 x 10
0
I I I
O Upstream
^ Downstream
KC1
Sample Flow 7.1 LPM
Transpiration Flow 14.2
LPM
\
I I 'A I I
.02
.05
.1
.2
.5
1.0
Particle Size, ^m
Figure 30. Size Distributions for Test K-3
73
-------
o
o
5 x
10
0
.02
O Upstream
Downstream
KCl
Sample Flow 28.3 LPM
Transpiration Flow 14.2 LPM
i r> i
.05
.1
.2
.5
Particle Size, |jn
Figure 31. Size Distributions for Test K-4
74
-------
10 <
o
o
5x10" -
KC1
Sample Flow 7.1 LPM
Pyrex Tube
.02 .05 .1 .2 .5 1.0
Particle Size, |jm
Figure 32. Size Distributions for Test K-5
75
-------
Table 7. SIZE DISTRIBUTION DATA FOR TESTS
WITH KC1 AEROSOL
Sample Transpiration UPstream Downstream
Test Flow Rate Flow Rate GNMD GNMD
No. (1pm) (1pm) prn gg ym qg
K-5 7.1 0 0.045 2.8 0.045 2.8
K-l 7.1 14.2 0.021 2.0 0.025 2.0
K-3 7.1 14.2 0.07 1.8 0.05 2.5
K-2 28.3 14.2 0.016 2.5 0.016 2.5
K-4 28.3 14.2 0.045 2.6 0.06 2.2
76
-------
01
N
0)
r-4
O
•H
4-1
5
4
3
2
-2
10
5
4
3
2
Upstream
— Downstream
KC1
Sample Flow 7.1 LPM
Transpiration Flow 14.2 LPM
10
JJ_LL
X
x
X
/x"
X
1.
O.I I
Figure 33.
5 10 50 90 95 99
% Smaller than Stated Size
Log-Probability Plot of Size Distributions for Test K-l
99.9
-------
00
a
-------
10
N
•H
tn
-------
1O
oo
o
I
n
ro
i
Ni
O
Q>
N
•H
CO
0)
I-l
o
T-l
4J
M
tO
5
4
10
5
4
3 '
10
Upstream
Downstream
KC1
Sampler Flow 28.3 LPM
Transpiration Flow 14.2 LPM
O.I
5 IO 5O
% Smaller Than Stated Size
90 95
99
99.9
-------
00
5
4
E
a)
N
•'-| 1
W ~-l
-------
6„2 Tests with Uranine Aerosol
The uranine for these tests was obtained by sieving the
reagent grade uranine through a 270 mesh screen. The venturi
dispenser used to generate the uranine aerosol in Phase I
was used in these tests.
The aerosol at the inlet and the outlet of the porous
probe was sampled on 47 mm dia. Nuclepore^—•'filters with
a pore size of 1 urn. The particle size distributions were
measured directly from the filter, made transparent with
immersion oil of refractive index 1.590S by an optical
/TJ\
microscope interfaced with the Quantimet 720^ image analyzing
computer. The smallest size measured was 5 ym. The largest
size was 50 ym,
The size distribution data for all the tests are presented
in Table 8. Histograms and log-normal plots are also presented
in Figures 38-47.
After run #U-4, qualitative information on the location
of high deposition was obtained. At the end of the size
distribution test, the front end of the porous probe was
immersed vertically in a graduated cylinder„ The graduated
cylinder contained distilled water such that the level after
the immersion of the porous probe was 16 cm. The uranine
deposited was allowed to dissolve for five minutes. Next
the probe was immersed in another cylinder filled with distilled
water to a height of 60 cm. Approximately 2 mg of uranine
were found on the first 16 cm, compared to 1,5 mg for the
following 60 cm. This indicates that approximately 40-50%
of the total deposit was in the first 16 cm.
In the test #U-5 with the pyrex tube a visible deposit
in the bottom half of the tube was observed. This deposit
was heaviest at the inlet to the probe and tapered off in the
first 50 cm.
82
-------
Table 8. SIZE DISTRIBUTION DATA FOR TESTS WITH URANINE
Particle
Size
5-7
9
11
13
15
20
25
30
37
50
% Smaller than Stated Size
Test No. U-l
Upstream
26.3
47.3
58.5
72.5
78.9
86.6
92.3
95.2
96.8
100.0
Downstream
19.5
30.2
39.1
49.7
59.2
76.0
82.1
90.5
96.1
100.0
Test No. U-2
Upstream
21.5
42.3
57.7
71.1
83.0
94.4
98.9
99.3
99.5
100.0
Downstream
21.6
37.4
45.2
53.0
57.5
69.2
82.3
90.7
96.7
100.0
Test No. U-3
Upstream
17.4
34.1
49.1
59.8
71.6
87.8
93.0
95.7
97.3
100.0
Downstream
12.4
23.1
34.1
47.2
54.6
73.3
87.2
92.3
96.3
100.0
Test No. U-4
Upstream
18.5
37.4
54.1
68.7
78.7
89.4
95.4
98.6
99.2
100.0
Downstream
13.6
24.7
38.7
44.4
50.5
62.7
77 A
84.6
91.7
100.0
Test No. U-5
Upstream
24.0
46.6
61.7
71.7
81.1
90.6
94.9
98.0
99.4
100.0
Downstream
32.4
57.1
73.6
84.7
91.3
98.1
99.4
99.71
100.0
100.0
300-400 particles counted for each sample.
-------
50
40
Upstream
Downstream
30
cfl
fc
0)
.u
c
20
10
0
9 11 13 15 20 25
Size Interval, (jtn
Figure 38. Histograms for Run U-l
30 37 50
84
-------
50
40
Upstream
Downstream
3GT
-------
50
40
Cfl
fc
-------
50
I I
i I I i r
40
Upstream
Downstream
30
to
fc
20
10
0
579
11 13 15 20 25 30 37 50
Size Interval, (jm
Figure 41. Histograms for Run U-4
87
-------
50
40
30
20
10
0
' "1
1 - 1 - 1
\ - 1 - T
Upstream
Downstream
5 7 9 11 13 15 20 25 30 37 50
Size Interval, (jm
Figure 42, Histograms for Run U-5
88
-------
0)
CO
01
^1
P*
IOO
70
50
40
30
20
10
7
5
4
3
2
1.0
O Upstream
^ Downstream
X X
/ X
Uranine
Sample Flow 7.1 LPM
Transpiration Flow 141 LPM
O.I
50
90 95
^Smaller than Stated Size.
Figure 43. Size Distributions for Test U-l
99
99.9
-------
100
6
0)
N
C/3
0)
$
50
40
30
20
5
4
3
2
Upstream
Downstream
Uranine
Sample Flow 28.3 LPM
Transpiration Flow 141 LPM
G
ZV"
'\
O.I
IO 5O 90 95
7o Smaller than Stated Size
99
99.9
-------
Q>
M
to
-------
vO
ro
a
n
•i
-------
vo
U>
B
Q)
N
-H
CO
O
•H
4J
>-i
«
DL,
ioo
70
50
40
30
20
10
7
5
4
3
2
IP
O Upstream
A Downstream
Uranine
Pyrex Tube
Sample Flow 7.1 LPM
O.I
5 10 50 90 95
% Smaller than Stated Size
Figure 47. size Distributions for Test U-5
99
99.9
-------
6,3 Tests with Flyash
The flyash was prepared by sieving it through a 270 mesh
screen. The same venturi dispenser used with the uranine aero-
sol was used to generate the flyash aerosol. The samples,as
in the tests with uranine, were taken on a nuclepore filter.
The immersion oil used with uranine could not be used
for the flyash since the oil dissolved the particles. There-
fore, the deposit was transfered to a glass slide by washing
the filter with isopropyl alcohol and placing a drop of the
suspension on the glass slide, After drying the slide, a drop
of Aerochlor^ was used instead of an immersion oil.
We had hoped to measure the size distribution on the
Quantimet 720 image analyzer, but the wide variation of the
grey levels of the flyash particles coupled with the wide size
distribution was too much for the image analyzer to handle.
The counting had to be done manually with an optical microscope.
The same size categories as uranine were used. The size dis-
tribution data are presented in Table 9 and Figures 48-57 in
the same form as for uranine.
In test No. F-5 with the pyrex tube, a deposition pattern
similar to the corresponding uranine test U-5 was observed.
6.4 Tests with 1-MAAQ Aerosol
The aerosol for these tests was generated with the genera-
tor used in Phase I except no KC1 nuclei were used. This
resulted in a wider size distribution much more meaningful
for the Phase II tests than the nearly monodisperse 1.6 ym
(Geometric mass mean diameter) aerosol. The aerosol had an GNMD
(count basis) of approximately 0.7 urn.
The aerosol was generated at a flow rate of 7.1 1pm
(0.25 cfm). When required, it was diluted with clean dry air.
The samples at the inlet and the outlet of probe were collected with
94
-------
Table 9. SIZE DISTRIBUTION DATA FOR TESTS WITH FLYASH AEROSOL
Particle
Size
5-7
9
11
13
15
20
25
30
37
>37
% Smaller than Stated Size
Test No. F-l
Upstream
33.1
63.0
75.4
83.1
88.7
95.0
97.8
98.9
99.2
100.0
Downstream
28.6
46.9
57.8
67.3
74.9
87.2
90.2
93.2
95.6
100.0
Test No. F-2
Upstream
33.9
65.8
81.1
90.6
93.6
97.5
98.9
99.4
99.7
100.0
Downstream
23.0
45.8
61.9
70.9
77.5
83.9
87.8
91.5
95.0
100.0
Test No. F-3
Upstream
28.7
74.3
88.4
91.6
94.6
96.8
98.4
99.2
99.5
100.0
Downstream
29.1
44.9
47.1
65.0
71.2
81.1
86.7
90.4
93.2
100.0
Test No. F-4
Upstream
27.0
40.3
64.9
76.7
84.5
93.0
98.0
99.2
99.4
100.0
Downstream
37.2
59.4
68.3
72.9
76.1
82.4
85.6
88.5
92.5
100.0
Test No. F-5
Upstream
32.0
60.2
76.2
83.7
87.8
92.0
93.9
95.3
97.0
100.0
Downstream
55.4
79.7
89.7
94.3
96.0
97.1
98.3
99.1
99.4
100.0
300-400 Particles counted for each sample.
-------
T 1 1 1 1 1 T
40
cfl
fc
30-
2C
10
Upstream
Downstream
7 9
25 30 37 50
11 13 15 20
Size Interval, |jin
Figure 48. Histograms for Run F-l
96
-------
50
40
30
CO
fc
-------
to
t
0)
c
•r-l
50
1 »
40
30
20
10
0
Upstream
Downstream
9 11 13 15 20 25 30 37 50
Size Interval, ^m
Figure 50. Histograms for Run F-3
98
-------
50
T 1 1 1 1
40
30
— Upstream
Downstream
cfl
20
1C"
0
9 11 13 15 20
Size Interval, ^
Figure 51. Histograms for Run F-4
25 30 37 50
99
-------
50
40
30
CO
I
-------
0
«t
0)
N
U
i-l
4-1
H
CO
IOO
70
50
40
30
10
7
5
4
3
2
10
o
Upstream
A Downstream
Fly Ash
Sample Flow 7.1 LPM
Transpiration Flow 141 LPM
O.I
5 10 50 90 95
% Smaller than Stated Size
Figure 53. Size Distributions for Test F-l
99
99.9
-------
s
6
01
N
i-l
CO
to
70
50
40
30
20
5
4
IO
O Upstream
A Downstream
Fly Ash
Sample Flow 28.3 LPM
Transpiration Flow 141 LPM
O.I
5 10 50
/«. Smaller than Stated Size
Figure 54.
90 95
Size Distributions for Test F-2
99
99.9
-------
o
u>
0)
N
t-l
CO
Q)
i-l
O
•H
CO
Cu
I OQ
70
50
30
20
10
7
5
4
3 '
2
10
O Upstream
A Downstream
Fly Ash
Sample Flow 7.1 LPM
Transpiration Flow 283 LPM
O.I
£
90 95
5 10 50
% Smaller than Stated Size
Figure 55. Size Distributions for Test F-3
99
99.9
-------
o
.£>
-------
o
in
e
0)
N
«H
OT
0)
»H
O
•H
4J
^
CCS
IOO
70
40
30
10
7
5
4
3
2
10
Upstream
A Downstream
Fly Ash
Pyrex Tube
Sample Flow 7.1 LPM
JJi
O.I
5 10 50 90 95
7o Smaller than Stated Size
Figure 57. Size Distributions for Test F-5
99
99.9
-------
the IITRI moving slide impactor on microscope slides. The
size distribution was determined by counting manually on an
optical microscope.
The size distribution data are presented in Table 10 and
Figures 58-62.
7. RESULTS
7.1 Tests with KC1 Aerosol
In Table 11, the size distribution parameters for tests
with KC1 aerosol are presented in the ascending order of
transpiration flow rates, and the ratios of the transpiration
velocity to the sample velocity at the inlet to the porous
probe. The size distribution changed very little in all
the tests. For runs K-5, K-l, K-2 the agreement is excellent.
For runs K-3 and K-4, the agreement is not as good. However,
this scatter is believed to be due to the nebulizer (Figure 26)
used for the generation aerosols. The size parameters of the
aerosol are very sensitive to the level of the solution in
the flask, and are subject to change with time.
The results indicate that for particles below 1 urn, the
pyrex tube and the porous sampling interface are both
effective in the preservation of the size distribution. These
results are consistent with the Phase I results which
indicate that the maximum deposition for KC1 aerosol with
0.05 ym nominal diameter is less than 3% on mass basis for all
of the above conditions.
7.2 Tests with 1-MAAQ Aerosol
As with KC1, the size parameters are grouped in Table 12
for 1-MAAQ. The agreement between the upstream and downstream
size distribution is excellent. Even for the pyrex tube, the
size distribution is well preserved. Examination of Figure
106
-------
Table 10. SIZE DISTRIBUTION DATA FOR TESTS WITH 1-MAAQ AEROSOL
Particle
Size
(ym)
0.5
1.0
1.5
2.0
2.5
>2.5
°L Smaller than Stated Size
Test No. M-l
Upstream
23.7
79.0
95.8
99.3
100.0
100.0
Downstream
17.9
60.7
85.7
95.7
98.6
100.0
Test No. M-2
Upstream
24.8
71.8
92.6
98.0
99.3
100.0
Downstream
28.9
70.4
88.9
97.0
99.3
100.0
Test No. M-3
Upstream
21.0
72.7
95.1
97.9
99.3
100.0
Downstream
21.7
73.4
96.5
99.3
100.0
100.0
Test No. M-4
Upstream
15.0
67.9
93.6
98.6
99.3
100.0
Downstream
36.5
82.5
97.8
99.3
100.0
100.0
Test No. M-5
Upstream
21.2
72.9
94.3
98.4
99.5
100.0
Downstream
25.4
73.9
94.9
100.0
100.0
100.0
150-250 Particles counted for each sample.
-------
10.0
o
00
Q)
N
i-l
CO
<0
iH
O
•H
•U
H
(0
5
4
3
2
.7
•5
.4
O.JO
O Upstream
^ Downstream
1-MAAQ
Sample Flow 28.3 LPM
Transpiration Flow 14.2 LPM
O.I
5 10 50 90 95
% Smaller than Stated Size
Figure58. Size Distributions for Test M-l
99.9
-------
N
1-1
wa
n
CO
so.o
7
5
4
.7
5
•
4
.3
•2
0. 10
O Upstream
A Downstream
1-MAAQ
Sample Flow 7.1LPM
Transpiration Flow 14.2 LPM
O.I
5 10 50 J , 90 95
% Smaller than Stated Size
Figure 59. Size Distributions for Test M-2
99
99.9
-------
10-0
0)
N
•H
CO
0)
t-l
O
•H
4J
5
4
3
2
1O
7
? •
2
0.10
O Upstream
A Downstream
1-MAAQ
Sample Flow 28.3 LPM
Transpiration Flow 63.7 LPM
o.i
10
i ~,,~^ fif)
5O 90 95 99
Smaller than Stated Sixe
Distributions for Test M-3
99.9
-------
0)
T-l
0)
tH
O
•H
CO
IO
7
5
A
3
2
UO
.7
5
4
.3
.2
o.io
Upstream
Dovmstream
1-MAAQ
Sample Flow 7.1 LPM
Transpiration Flow 63.7 LPM
0.1
5 10 50 90 95
% Smaller than Stated Size
Figure 61. Size Distributions for Test M-4
99
99.9
-------
10
|
0.
0.2
O.ll
O Upstream
Downstream
1-MAAQ
Pyrex Tube
Sample Flow 7.1 1pm
0.1
10 50 90 95
7o Smaller than Stated Size
99
99.9
-------
Table 11. SIZE DISTRIBUTION PARAMETERS FOR KC1 AEROSOL
u>
Upstream
Test
No.
K-5
K-l
K-3
K-2
K-4
Nominal Size
Range Covered
(ym)
0
0
0
0
0
.01-1.
.01-0.
.01-1.
.01-1.
.01-1.
0
1
0
0
0
Sample
Flow Rate
(1pm)
7
7
7
28
28
.1
.1
.1
.3
.3
Transpiration
Flow Rate
(1pm)
0
14.2
14.2
14.2
14.2
V Transp
V Sample
0
0.0034
0.0034
0.0009
0.0009
GNMD
ym
0.045
0.021
0.016
0.045
0.045
_!*_
2.8
2.0
2.5
2.6
2.6
Downstream
GNMD
ym
0.045
0.025
0.016
0.06
0.06
ft
2.8
2.0
2.5
2.2
2.2
-------
Table 12. SIZE DISTRIBUTION PARAMETERS FOR 1-MAAQ AEROSOL
Upstream
Test
No.
M-5
M-2
M-4
M-l
M-3
Nominal Size
Range Covered
(ym)
0.
0.
0.
0.
0.
5-2.25
5-2.25
5-2.25
5-2.25
5-2.25
Sample
Flow Rate
(1pm)
7.1
7.1
7.1
28.3
28.3
Transpiration
Flow Rate
(1pm)
0
14
63
14
63
.2
.7
.2
.7
V Transp
V
0
0
0
0
0
Sample
.0034
.016
.0009
.0039
GNMD
ym
0.7
0.72
0.8
0.63
0.72
0.
1.7
1.8
1.6
1.6
1.7
Downstream
GNMD
ym
0.7
6.72
0.6
0.82
0.72
fa.
1.7
1.8
1.6
1.7
1.7
-------
62 shows that the particles above 2 ym were lost. Even
though this end of the plot represents a very small number,
absence of the particles larger than 2 ym was obvious during
the microscopic examination.
This suggests that particles above 2„0 ym size are de-
posited by gravity at the downstream end of the pyrex tube.
These tests indicate that for particles below 2 ym, the
pyrex tube and the porous sampling interface are both
effective. These results are consistent with the Phase I re-
sults which show that the maximum deposition of 1.6 ym 1-MAAQ
on the mass basis is less than 370 in all the cases.
7.3 Tests with Uranine and Flyash Aerosols
The size parameters for uranine and flyash are presented
together in Table 13 as the nominal size range and test
conditions were identical.
The effect of gravitational deposition on the particle
size distribution during the transport through the pyrex
probe (Test U-5, F-5) is seen from Figures 42, 47, and 55,
The mean particle size has decreased. The standard
deviation, a , is not significantly affected. As expected,
the deposition of larger particles is more severe for the
heavier flyash. In tests U-l and F-l, the standard deviation
has not changed significantly and the shape of the upstream
and downstream curves are similar. The mean particle size
has increased downstream. In tests U-3 and F-3, the same
trend is observed. However, the standard deviation for F-3
has gone up significantly. In Test U-4, the mean particle
size has increased, and the shape of the distribution has been
significantly changed (Figures 41 and 46) . At the two
ends of the distribution, the curves are parallel, but in
the midsection a skewness is observed in the curve for the down-
stream sample.
115
-------
Table 13. SIZE DISTRIBUTION PARAMETERS FOR TESTS
WITH URANINE AND FLYASH AEROSOLS
o>
Test
No.
U-5
U-l
U-3
U-4
U-2
F-5
F-l
F-3
F-4
F-2
Nominal Size
Range Covered
(ym)
Uranine; 5-50
Uranine; 5-50
Uranine; 5-50
Uranine; 5-50
Uranine; 5-50
Flyash; 5-50
Flyash; 5-50
Flyash; 5-50
Flyash; 5-50
Flyash; 5-50
Sample
Flow Rate
(1pm)
7.
7.
7.
28.
28.
7.
7.
7.
28.
28.
1
1
1
3
3
1
1
1
3
3
Upstream
Transpiration
Flow Rate
(1pm)
0
141.
283.
14.
141.
0
141.
283.
14.
141.
5
0
2
5
5
0
2
5
V Transp.
V
0
0
0
0
0
0
0
0
0
0
Sample
.034
.069
.0009
.009
.034
.069
.0009
.009
GNMD
ym
9.5
10.0
11.0
10.5
10.0
8.0
8.0
7.4
9.0
8.0
a
_£_
1.8
1.9
1.8
1.7
1.6
1.7
1.8
1.4
1.7
1.4
Downstream
GNMD
urn
8.5
12.0
13.0
14.5
12.5
6.6
9.6
9.3
8.3
9.1
i
1.6
2.1
1.8
2.0
1.8
1.5
2.0
2.4
2.5
2.2
-------
For the flyash, in the corresponding test F-4, the skewness
is even more marked. Particles around 10 ym are being lost
in the flow through the porous sampling interface at a higher
rate than the very large particles. This trend is observed
for tests U-2 and F-2. The sample flow rate is high in all
those tests which showed the skewness. The transpiration
flow rates were moderate to moderately high. This effect
is believed to be due to the microturbulence at the inlet of
the porous probe where the sample and transpiration flow
meet. A qualitative justification of this reasoning was
obtained in test U-3. In a crude test, the amount of deposit
in the first 16 cm of the porous probe was found to be
approximately 40-5070 of the total deposit in the tube.
Large particles, once they are entrained, are disturbed
least by the microturbulence while the smaller particles are
carried by the eddies and are probably deposited on the walls.
117
-------
8. CONCLUSIONS
In Phase I, the quantitative mass transport of the
aerosols was investigated. The following conclusions were
reached.
(1) Of the four factors investigated, particle con-
centration has the least effect on particle deposition,
There was a slight increase in deposition with an increase
in concentration, but this effect is marginally significant
statistically.
(2) Deposition depends strongly on particle size.
Deposition was quite low for both 0.05 ym and 1.6 ym particles,
but 50 ym particles have a substantially higher deposition
rate at intermediate levels of transpiration flow rate,
sample flow rate, and particle concentration: between 5
and 10% as contrasted with the range 0.1 to 0.5% for the
smaller particles.
(3) Sample flow rate and transpiration flow rate have
a degree of parallelism in their general effect on reduction
of deposition with increase in flow. There is an interaction
between the two terms and an interaction of each with particle
size.
(4) The simple effect of increasing sample flow rate,
holding other factors at intermediate settings, is to decrease
the deposition.
(5) The main effect of increasing sample flow rate,
holding other factors at intermediate levels, is to
reduce the fraction of the aerosol particle mass that is
deposited on the walls of the sampling probe. At all levels
of sample flow rate tested deposition of large particles
(~ 50 urn) is reduced when the transpiration flow rate is in-
creased; and the reduction is greatest at the low sample flow
rate. For example, deposition of 50 urn particles from a high
118
-------
concentration aerosol sample at 7.1 1pm (0.25 cfm) is sharply
reduced with 283 1pm (10 cfm) of transpiration air from
more than 99.970 to about 5%. At the higher levels of sample
flow rate the percentage deposition of large particles is
brought considerably lower than at low sample flow rate as
the transpiration flow rate increases from zero to 283 1pm.
(10 cfm), but the reference values at zero transpiration
flow rate are also much lower. At the low sample flow
rate the porous tube with increasing rates of transpiration
flow reduces deposition for all particle sizes tested. At
the intermediate sample flow rate, transpiration air continued
to reduce deposition of medium sized particles (~ 0.05 ym) ;
this reduced effect is of little practical importance due to
the low percentage deposition of the small and intermediate
particles. At the high sample flow rate, the range of varia-
tion of deposition is much reduced with a slight increase in
deposition of the intermediate, and especially small, particles
with increasing transpiration air, but, again, the magnitude
of deposition is only of the order of 0.1 to 5%.
(6) Tests with the 0.05 and 1.6 ym aerosols indicate
that only 14.2 1pm (0.5 cfm) of transpiration air is needed
to reduce deposition to virtually insignificant levels on
the order of 0.2% or less.
(7) The porous probe experiments were clearly definitive
in demonstrating the significant reduction in loss of aerosol
mass through the use of transpiration air. At 7.1-14.2 1pm
(0.25-0.50 cfm) of sample volume deposition of 50 ym particles
ranged from close to 100% at Orl4.2 1pm (0-0.5 cfm) of trans-
piration air to less than 0.1% a5 283 1pm (10 cfm) of trans-
piration air.
(8) A redesign of the probe inlet is recommended to
make the porous sampling probe even more effective for
large particles at reduced transpiration air flow rates.
If more of the transpiration air is introduced at the very
-------
Inlet of the porous probe, deposition of large particles
near the entrance, at the 7.1-14.2 1pm (0.25-0.50 cfm) sample
flow rates, will be greatly reduced at modest transpiration
air flow rates. Such a design should establish a clean air
sheath in a shorter distance and should also reduce depo-
sition of the less problematical intermediate and small
aerosol particles.
The conditions for Phase II were chosen on the basis
of results of Phase I experiments. Changes in the size
distribution of aerosols transported through the sampling
interface ranging from 0.01-50 um were investigated using
conditions that gave good transport efficiencies on the
mass basis. For comparison, a set of tests with a 1.25 cm
(% in.) diameter, pyrex tube was also conducted. The
following conclusions were reached.
(1) For the particles below 2 yna, the pyrex tube and
the porous sampling interface were equally effective.
(2) For particles around 2-2.5 um, the porous tube was
more effective than the pyrex tube.
(3) For the 5-50 ym particles, the pyrex tube data
showed that the mean particle size decreased. With the porous
tube the mean size generally increased.
(4) At low sample flow rate (7.1 1pm) the size dis-
tribution was similar in shape, At the higher sample rate
(28.3 1pm) the size distribution of the aerosol transported
through the porous probe was skewed. Particles around 10 ym
in size were deposited to a greater extent than larger
particles. This was believed to be due to the microturbulence
at the entrance to the porous probe.
(5) The porous probe was found to be effective in pre-
serving the size distribution of the aerosol during trans-
port in addition to the capability of preserving mass as
120
-------
demonstrated in Phase I. For particles up to a few micrometers
in size the flow rates had little effect on the size distri-
bution. For the particles such as flyash in the range of
5-50 ym, low sample flow rate and a transpiration flow rate
high enough to preserve mass are most effective for their
total quantitative transport.
(6) The feasibility of the porous probe concept has
been established under this contract. Design of a quanti-
tative sampling interface is now possible due to the in-
formation generated under this program. Factors such as
exposures to high temperatures and corrosive substance
expected in the actual stack sampling in fuel must be
considered for the design of the sampling interface prototype.
121
-------
Appendix A
AN ELECTRIC MOBILITY METHOD OF SIZING
0,01 to 1.0 ym POLYDISPERSE AEROSOLS
122
-------
AN ELECTRIC MOBILITY METHOD OF SIZING
0.01 to 1.0 ym POLYDISPERSE AEROSOLS
Of the indirect methods for aerosol size analysis,
Whitby and Clark's (Al) electric mobility analysis method
with unipolar charging has poor resolution above 0.3 ym.
Knutson's electric mobility method (A2) , with bipolar
charging, is limited to near monodisperse aerosols. Kudo
and Takahashi (A3) have explored another mobility analysis
method with bipolar charging, but its potential resolving
power seems limited.
In the following, another variation of the electric mobil-
ity method for aerosol size analysis is described. Bi-
polar charging is used. This method applies to polydisperse
aerosols, i.e., geometric standard deviation ^ 1.2. It is
suited for automation, since no manual microscopy is required.
No empirical calibration is necessary. As a penalty, the
method requires a complicated and lengthy calculation to
recover the size distribution from the raw data. These are
best carried out on a remote computer terminal.
APPARATUS
The apparatus required for this method of aerosol size
determination are:
1. A concentric cylinder electric mobility analyzer
described by Knutson (A2). This type has two inlet
air streams and two outlet air streams.
2. A bipolar charging device (charge neutralizer).
3. A suitable aerosol sensing device. This device
must give a response proportional to either the
number concentration charge, concentration or mass
concentration.
4 A precision high voltage power supply. A sketch
of the mobility analyzer is given in Figure Al
Condensation nuclei counter (number_measurement)
was used as the aerosol sensing device.
123
-------
Absolute Filter
Capillary
Flowmeter
Bipolar
Charger
Aerosol
1.23 1pm
High Voltag
Power suppl
y .
7.19 1pm
Critical"
Orifices
Clean Air
6.94 1pm
Electric
Mobility
Analyzer.
0.981pm
Condensation
Nuclei Counter
Vacuum pump
Figure A-l: System for Measuring Aerosol
Size in the Range 0.01-1.0 Microns
124,
-------
OPERATING PROCEDURE
The aerosol sample is continuously drawn through the
analyzer by the vacuum pump (Figure Al). The sample is split
into two streams. One stream flows through a capillary
flow meter and a bipolar charger for neutralization of charges
on the aerosol.. The other part of the sample stream is passed
through a glass fiber filter. This acts as a sheath for the
first sample stream in the annulus between the two coaxial
cylinders. A voltage differential between the two cylinders
is maintained with the help of the high voltage power supply.
At a given voltage setting, a stream of aerosol particles
having equal electric mobility is removed through a series
of holes at the other end of the analyzer. The particle con-
centration of this stream is measured by the nuclei counter
(Gardner Associates; Schenectady, N.Y.). This procedure
is repeated at various voltage settings ranging from 0-10,000
volts.
THEORY AND DATA REDUCTION
The starting point for the theory of this method of
aerosol size determination is the transfer function, S7, for
the mobility analyzer. This has been given by Knutson (A3).
The transfer function is most conveniently presented in
graphical form, with the product Z V as absissa and £2, as
ordinate, as in Figure A2. ft is the probability that a
particle with electric mobility Z will reach the mobility
analyzer sampling outlet flow, given that the voltage applied
to the center rod is V. Values for certain flow rates are
also needed, as are certain geometric dimensions. These
are indicated in Figure A2.
The number of particles entering the mobility analyzer
in unit time is C^, where CE is the aerosol concentration
rate. Of these, a fraction f(dp)ddp have size in the range
d to d + dd f is the unknown size distribution function.
P P P
125
-------
Min -f-
d,qs/qa)
Mobility Analyzer Flow Rates:
q = clean air inlet
Hc
q = aerosol inlet
Ma
q = main outlet
Hm
q = sampling outlet
s
Mobility Analyzer Dimensions:
b = I.D. of outer cylinder
a = O.D. of inner cylinder
L = distance between aerosol inlet and sampling outlet
Figure A2
TRANSFER FUNCTION FOR THE MOBILITY ANALYZER
126
-------
In turn, the fraction carrying k elementary charges is
), which is assumed
Number of particle of
p(k;d ), which is assumed known. Thus,
r- ' Ca"a P*> dp> f
-------
For each integral in Equation A2, change variable from
d to x, where x = k • Z, -V and k and V are held constant,
P 1
(A3)
dx
dd =
P
-^ f
The first of Equations A3 implicitly defines d in terms of
x. The second, in which ZJ is the derivative of Z-, , gives
the relationship between dd and dx. The integrals in
Equation A2 become
oo
f
n(x)p(k;d )f(d ) ^r- (A4)
u u J\- V *j -i
o f f i
Figure A2 shows that if q and q are small in comparison
S 3
with q and q , the function ft is non-zero only near
x = x* = (q + q ) • Jtn(b/a)/4irLc Hence, by the mean value
theorem for integrals, integral A4 may be approximated by
P(k;dn*)_._ .. 7 . . P(k;d^*) qg _
(A5)
P(k;dp*) g(d*)
In Equation A5 , d * is the particle size defined implicitly
by Equation A3, i.e., by
,J ,x x* (qm
7 (A ~k\ = 2—, - m
^p > kV -
Z^ * is the corresponding value of Z' „ The function g(d *) ,
p(l;d *)f(d *)
g(dp*) = Caqaqs /,f P - (A7)
has been introduced for later notational convenience
128
-------
With the approximation in Equation A5, Equation A2
becomes
p(k;d *)
p(l-d *) 8^d *) (A8)
where d * depends on both k and V. For a given V, the con-
centration C receives contributions from several discrete
S
particle sizes, the size increasing with k. The key to
solving Equation A8 for the corresponding discrete values
of g lies in a particular choice of the sequence of voltage
settings, V, for which C is measured. The first setting,
V, , must be so large that only singly charged particles con-
tribute significantly to C „ The remaining members of the
S
sequence C$2' ^v ^4' •••) must form a descending geometric
progression:
V = V./r
23 (A9)
with r = / 10
This sequence has the property that
0.9925 Vi_y for k = 2
0.9997 Vi_11 for k = 3
0.9847 Vi_14 for k = 4
1.0077 Vi_16 for k = 5
0.9898 V. 1Q for k = 6
i-lo
To good approximation, the numerical coefficients of the V's
on the right may be set equal to unity. Thus, the sequence
2V. duplicates the sequence V^ except that it is shifted
by1"seven places. Similar remarks apply to the sequences 3V±,
4Vi' 5Vi' and 6Vi'
Equation A8 may now be solved for g(dp*) as follows.
Let d *(1), d *(2), d *(3), ... be the sequence of values
129
kV. =
-------
generated from the sequence V,s V2 , V~ , . . , by Equation A6
with k = 1. Truncate the sum in Equation A8 after the sixth
term. Then
Cs(Vi)qs = gWp*(i-18)} +
p{2;d *(i-7)}
P . , . „,-. P f H * ( i. - 7 ) 1 + +
^^ g m ll /; ' ^
p{6;d *(i
8
This equation may be solved recursively for the g {d *(!)}»
recalling that g {d *(i)) = 0 for i *' ls by the choice of the
first voltage setting, V,o With these values for the g's,
the values C f {d *(i)l may be recovered from Equation A7 „
a p
Finally, C is determined by normalizing f (d *) „
a p
This discussion of the theory of the method has assumed
that the aerosol sensor measures number concentration „ If
a mass sensor is used, it is only necessary to interpret C
3.
and f(dp) as the mass-by-diameter distribution. If a charge
sensor is used, it is necessary to insert the factor Ke,
where e is the elementary unit of charge, within the summa-
tion sign in Equation A2 and carry it through the analysis.
The quantity C (V)q should then be interpreted as the elec-
s s
trie current due to the sampling outlet flow.
A computer program for the size distribution calculations
was developed. The output is plotted in terms of the
derivative of the number concentration of particles with
respect to the log of particle size. The symmetrical nature of the
curve represents a log-normal distribution. The GNMD is
given by the value corresponding to the peak of the curve.
The geometric standard derivative is given by ratio of the
size corresponding to the 6070 of the peak value and the GNMD.
130
-------
REFERENCES
Al. K.T. Whitby and W.E. Clark, "Electric Aerosol
Particle Counting and Size Distribution Measuring
System for the 0.015 to 1 y Size Range, Tellus XVIII,
573-86 (1966).
A2. E.G. Knutson, The Distribution of Electric Charge
Among the Particles of an Artificially Charged
Aerosol^Ph.D.Thesis, University of Minnesota,
Minneapolis, Minn. (1971).
A3. A. Kudo and K. Takahashi, "A Method of Determining
Aerosol Particle Size Applying Boltzmann's Law,"
Atmos. Envir., 6 (8), 543-50 (1972).
131
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
REPORT NOl
EPA-650A2-74-016
3. RECIPIENT'S ACC£SSION>NO.
4^TITLE AND SUBTITLE
Sampling.Interface for Quantitative Transport of Aerosc
5. REPORT DATE
Is December r 1973
6. PERFORMING ORGANIZATION CODET
7..AUTHOR(S) ..
Madhav-B. Ranade
8. PERFORMING ORGANIZATION REPORT NO
9. PERFORMING ORGANIZATION NAME AND ADDRESS
IIT Research Institute
10 West 35th Street
Chicago, Illinois 60617
10. PROGRAM ELEMENT NO.
1AA010
11. CONTRACT/GRANT NO.
68-02-05-79
12. SPONSORING AGENCY NAME AND ADDRESS
EPA, Chemistry & Physics Laboratory
_National Environmental Research Center
Research Triangle Parky N.Tj—277TT-
13. TYPE OF REPORT AND PERIOD COVERED
Final Report
14. SPONSORING AGENCY CODE
15. SUPPLEMENTARY NOTES
16. ABSTRACT
A-safflplfng-;probe_WcUS-desJ^ned^J:abrlcated.,-aad evaluaie.difor quantitative-
;transpor±-jof^^"aerosrjIsLthi^ugl^^conduit-from^a^souj^ce^to -a Benson* The-probes
consists of a porous metal- tube-encasecUin a jnaaifold-through- which- transpiration—
_air was-passed-inward io- prav:ide::a jnovlng-clean^air sheath-that minimized-particle-
-depo-sition-on-the-waU-s^- la Phasfcrl ^ the^quanti±atlve_jtias5_trarisp.ojLt JxfLaerosols
was investigated, and inrPhase_JI^th\e^pTeservatio^ of size distribution of the
.-transported aerosol was studied. - The 178 £m~ (70 in.):long by_ 1.27 cm (1/2 in.)
JD^roha required^only 14r2'-lpm (0.5cfm) of-transpiration air to virtually eliminate
deposition of partialesrin the~0.05-to 15^m^size range. Particles as large as
70 pm required asunuch asL 283 1pm (10-cfm)-to—prevent deposition—losses at low
sample flow.rates^r A-statistical analysis^f the data concl_us_ive4y demonstrates
the -effectiveness-of the::porous probe sampling concept. Tests at selected -
conditions show that the porous probe is effective in the preservation of size--
dis'tribution—
- Optimization-of-the-sample-and-trarvspiTation flow ration-^s necessary for a
giverrsize range, to obtain-the-most-effectiveruse-^of—the^porous- probe-concept/
1Z.
— KEY_WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b. IDENTIFtERS/OPEN ENDED TERMS
COSAIL Field/Group
Sampling Probe
Sampling Interface
Aerosol Transport
Quantitative Transport of AerosoL
Boundary^! ayer probe-
Pol 1utioniMoni toning
Sampling
Aerosols
Stack-Monitoring
Stationary Source Sampli
3; DISTRI BUTiON STATEMENT
Release _Uri"Kini ted
19. SECURITY^CLASS (This Report)
Unclassified
207SECURITY CLASS JThizpage)
22, PRICE
EPA Form 2220-T(9-73r"
132
------- |