United States
Environmental Protection
Agency
Industrial Environmental Research
Laboratory
Research Triangle Park NC 27711
EPA-600/7-80-014
January 1980
Development Study
of a Novel Continuous-
flow Impactor
Interagency
Energy/Environment
R&D Program Report
-------
RESEARCH REPORTING SERIES
Research reports of the Office of Research and Development, U.S. Environmental
Protection Agency, have been grouped into nine series. These nine broad cate-
gories were established to facilitate further development and application of en-
vironmental technology. Elimination of traditional grouping was consciously
planned to foster technology transfer and a maximum interface in related fields.
The nine series are:
1. Environmental Health Effects Research
2. Environmental Protection Technology
3. Ecological Research
4. Environmental Monitoring
5. Socioeconomic Environmental Studies
6. Scientific and Technical Assessment Reports (STAR)
7. Interagency Energy-Environment Research and Development
8. "Special" Reports
9. Miscellaneous Reports
This report has been assigned to the INTERAGENCY ENERGY-ENVIRONMENT
RESEARCH AND DEVELOPMENT series Reports in this series result from the
effort funded under the 17-agency Federal Energy/Environment Research and
Development Program. These studies relate to EPA's mission to protect the public
health and welfare from adverse effects of pollutants associated with energy sys-
tems. The goal of the Program is to assure the rapid development of domestic
energy supplies in an environmentally-compatible manner by providing the nec-
essary environmental data and control technology. Investigations include analy-
ses of the transport of energy-related pollutants and their health and ecological
effects; assessments of, and development of, control technologies for energy
systems; and integrated assessments of a wide range of energy-related environ-
mental issues.
EPA REVIEW NOTICE
This report has been reviewed by the participating Federal Agencies, and approved
for publication. Approval does not signify that the contents necessarily reflect
the views and policies of the Government, nor does mention of trade names or
commercial products constitute endorsement or recommendation for use.
This document is available to the public through the National Technical Informa-
tion Service, Springfield, Virginia 22161.
-------
EPA-600/7-80-014
January 1980
Development Study of a Novel
Continuous-flow Impactor
by
E.F. Brooks, N. Gat, M.E. Taylor,
T.E. Chamberlain, R.J. Golik, and R. Watson
TRW Systems and Energy
One Space Park
Redondo Beach, California 90278
Contract No. 68-02-2165
Task No. 12
Program Element No. EHE624
EPA Project Officer: D. Bruce Harris
Industrial Environmental Research Laboratory
Office of Environmental Engineering and Technology
Research Triangle Park, NC 27711
Prepared for
U.S. ENVIRONMENTAL PROTECTION AGENCY
Office of Research and Development
Washington, DC 20460
-------
ABSTRACT
The development study involved feasibility verification for a novel
type of particle impactor in which the impaction surface is the inter-
face between two opposing jets. Particles which would impact a solid
surface in a standard impactor cross the interface between the aerosol
laden gas and a previously particle free gas, are entrained in the latter
and conveyed out for analysis. Work consisted of an initial literature
search and analysis to determine the likelihood of success, followed by
design, fabrication, and test of a laboratory unit. A good particle
separation capability was demonstrated. Upon completion of the laboratory
testing, a design effort showed the feasibility of a staged in situ par-
ticle monitoring subsystem to give semicontinuous (nominal one minute cycle
time) output of particle size distribution, among other applications.
This work was completed in May of 1979 under Task 12 of EPA Contract No.
68-02-2165.
ii
-------
CONTENTS
Page
Abstract ii
List of Figures m
List of Tables v111
1 Conclusions 1
2 Recommendations 2
3 Introduction 3
4 Subtask 1 - Analysis and Literature Search 5
4.1 Background 5
4.2 Impinging Jet Stability 6
4.3 Pre-Design Literature Search 11
5 Subtask 2 - Design and Fabrication of Test Unit 16
5.1 Design Parameters Affecting Stability 16
5.2 Variation of Stability Parameters 18
5.3 Other Design Considerations 20
5.4 Fabrication of Test Unit 23
6 Subtask 3 Laboratory Testing 24
6.1 Test Setup 24
6.2 Test Conduct 30
6.3 Test Results 35
6.4 Interpretation of Test Results 48
7 Subtask 4 Additional Particle Separation Testing 60
7.1 Hardware Modifications 60
7.2 Test Description 64
7.3 Test Results 67
111
-------
CONTENTS (Cont'd)
Page
8 Subtask 5 - Instrument.Utilization 74
8.1 Staged Continuous Flow Impactor 74
8.2 Additional Configurations 84
8.3 Cost Estimates 86
9 Discussion of Results 90
10 References 92
11 Glossary 93
12 Appendices 95
A. Stability Data 95
B. Optical Monitoring of Continuous Flow Impactor 106
Output Streams
1v
-------
FIGURES
Number Page
la. Schematic of Standard Particle Impactor (From Reference 4) 6
Ib. Schematic of Impinging Jet Impactor 6
2a. Illustration of Separation Streamline Distortion
Instability 8
2b. Illustration of Unequal Flow Instability 8
3a. Subsonic Jet Flowfields 9
3b. Impinging Jet Flowfields 9
4. Schematic of Luna's Test Setup for Impinging Jet Stability
Testing (From Reference 11 10
5a. Illustration of Large Particle Non-Collectfon Due to Large
Orifice 12
5b. Illustration of Large Particle Non-Collection Due to Small
Orifice 12
5c. Illustration of Local Flowfield Phenomena Which Permit
Undesired Migration of Gases and Small Particles Across
the Interface 13
6. Anticipated Continuous Impactor Efficiency Performance
Curve, Based on Known Hardware Limitations 15
7. Continuous Flow Impactor Assembly Drawing 17
8. Impactor Splitter Plate Designs 19
9. Impactor Contoured Nozzle 21
10. Impactor Conical Nozzle 22
11. Test Setup Schematics 25
12. Particle Sampling Detail 29
13. Turbulence Profiles for Conical Nozzle 36
14. Turbulence Profiles for Contoured Nozzle 37
15. Flow Control Schematics 38
-------
FIGURES (Cont'd)
Number Page
16. Data Trends for Particle Collection Efficiency Testing,
all Configurations 41
17. Particle Collection Efficiency as a Function of Particle
Size, Configuration A 42
18. Particle Collection Efficiency as a Function of Particle
Size, Configuration B 43
19. Particle Collection Efficiency as a Function of Particle
Size, Configuration C 44
20. Particle Collection Efficiency as a Function of Particle
Size, Configuration D 45
21. Particle Collection Efficiency as a Function of Particle
Size, Configuration E 46
22. Particle Collection Efficiency as a Function of Particle
Size, Configuration F 47
23. Schematic of Flowrate Variation Technique for Biased Inlet
Flow Cases, Configurations D, E, and F 49
24. Comparison of Continuous Impactor Data With Anderson
First Stage 50
25. Calculated Free Streamline Shape (From Reference 1} 53
26. Theorized Efficiency Curves for Continuous Flow Impactor,
Based on Accumulated Test Data 55
27a. Effect of Sampling Tube When Sampling Aerosol Side 58
27b. Effect of Sampling Tube When Sampling Clean Side 58
28. Schematic of Revised Exhaust Flow Setup 61
29. Impactor Outlet Detail-Configuration G 62
30. Impactor Test Setup-Configuration G 63
31. Sketch of Particle Counter Sampling Tube 1n
Impactor Exhaust Line 65
32. Particle Collection Efficiency as a Function of Particle
Size for Size Range 4-16 Microns, Configuration G 68
vi
-------
FIGURES (Cont'd)
Number Page
33. Particle Collection Efficiency as a Function of Particle
Size for Size Range .5-5 Microns, Configuration G 70
34. Complete Fitted Separation Efficiency Curve for
Configuration 6, Showing Data Points in Overlap Region 71
35. Comparison of Separation Curves for Configuration C, 6,
and the Anderson Impactor First Stage 72
36. Flow System Schematic and Pressure Hierarchy for Staged
Continuous Flow Impactor 75
37. Staged Continuous Flow Impactor Installation 77
38. Staged Continuous Flow Impactor Module 78
39. Staged Continuous Flow Impactor Assembly - First Two Stages 80
40. Schematic Diagram of Optical Concept 82
41. Filter Holder Subassembly 88
-------
TABLES
Number Page
1. Royco 225 Particle Size Discrimination Characteristics 28
2. Operating Conditions for Particle Separation Tests 40
3. Nominal Size Ranges for Staged Continuous Flow Impactor
Applications 71
4. Continuous plow Impactor Production Cost Estimates 74
-------
SECTION 1
CONCLUSIONS
The concept of using two impinging jets for the purpose of inertial
particle separation has been shown to be technically viable.
f In addition, preliminary design studies indicate that a staged
impactor can easily be integrated into a sampling probe to perform
in situ particulate size distribution measurements.
t Feasibility of a simple optical system compatible with the in situ
sampling hardware has been shown analytically.
t The continuous flow impactor has a potentially greater accuracy
capability than standard Impactors since it inherently avoids the
traditional problems of particle bounce and reentrainment.
Potential applications for the continous flow impactor include
control device evaluation, continuous monitoring of particulate
emissions for regulatory purposes, and collection of particles in
an inert atmosphere to quench chemical reactions.
A basic understanding of the continuous flow impactor operational
characteristics has been attained. This can serve as a background
for development of a commercially viable instrument.
-------
SECTION 2
RECOMMENDATIONS
The demonstrated performance and potential usefulness of the
continuous flow impactor argue strongly for development and
testing of a field worthy prototype instrument. This develop-
ment should be pursued.
-------
SECTION 3
INTRODUCTION
The purpose of the work described in this report was to investigate
the feasibility of an impactor design in which the impaction surface
would be the interface between two opposing jets. Instead of hitting and
sticking to a solid surface, the particles would cross over the interface
and be entrained in a previously particle free gas. This second stream
could then be analyzed for particle content, optical techniques being the
most obvious. The proposed instrument would be used in stationary source
sampling work. Among the most obvious attributes would be elimination of
particle bounce and reentrainment problems associated with standard impac-
tors. Its continuous operation capabilities, inherent in the design,
would avoid the problems of stage overloading at locations such as control
device inlets, as well as the present need for long sampling times due to
low grain loadings at places like ESP outlets. The continuous operation
feature would make the device a viable one for long term regulatory moni-
toring as well.
Work was divided into five subtasks:
1. Analysis and Literature Search this was done to determine the
likelihood of success prior to making a hardware commitment.
A decision was made to proceed with hardware design mainly in
the strength of thesis work performed by R. E. Luna, whose test
setup employed two impinging jets.
2. Design and Fabrication of Test Unit the laboratory test unit
was primarily designed to promote stability of the jet/jet
interface, and to allow for insertion of diagnostic instrumen-
tation. Several physical variables were designed into the unit
so that their effect on performance could be evaluated.
3. Initial Laboratory Testing - this consisted of stability testing and
particle separation testing. An initial flow instability was
eliminated through a simple, acceptable flow control addition.
-------
Particle separation testing revealed an aerosol withdrawal problem
which was handled in the next subtask.
4. Particle Separation Testing - this was an add-on subtask for the
specific purpose of sharpening the separation efficiency, and
dealt with changes in handling of the gas stream downstream of the
virtual impaction region.
5. Instrument Utilization - preliminary design work has shown that
a four stage device with a total of five output streams can be
packaged in a probe capable of insertion through a standard
four inch pipe nipple, thus allowing for in situ measurements.
Calculations show that these input streams can be adequately
monitored (also in situ) by simple optical units. Additional
applications and first cut price estimates are also presented.
Results of the work performed are highly encouraging. Although some
additional investigative testing will be required, there are no unresolved
questions regarding concept feasibility. There is no real doubt that the
instrument will work - the issue which will need to be resolved is how well
the continuous flow impactor will work for a given system cost. Prelim-
inary indications are that performance should be good - the device is s1mple
in operation, contains no moving parts or delicate or exotic components.
Flow control, which is the key to successful operation, will be accomplished
in a production device through use of sonic orifices, which are very simple
and reliable. Components most likely to need servicing (pump, light source
electronics) will be located ex situ to facilitate access, without degrading
performance.
The initial R&D task has been successfully completed, its objectives
accomplished. The next step should be development and field demonstration
of a prototype field unit. Characteristics established during performance
of the task indicate that the continuous flow impactor will be a very useful
device for both short and long term stationary source testing, since it has
capabilities not found in any other device currently in use.
-------
SECTION 4
SUBTASK 1 - ANALYSIS AND LITERATURE SEARCH
The first task objective was to determine the feasibility of using
impinging jets for particle size classification. This was accomplished
through a review of the technical literature. The second part of the
task involved another review to obtain impactor design parameters for
the purpose of designing the laboratory test unit.
4.1 BACKGROUND
The concept which is the subject of this report is illustrated in
Figure 1. A standard impactor configuration is shown in Figure la.
It consists of a nozzle to direct a particle laden flow in a prescribed
manner, and an impaction surface which the flow, and consequently some
of the particles, strikes. Ideally, all of the particles which strike
the surface are retained. The current task has been concerned with
improving two characteristics of the standard impactor: limited time
operation, and problems (particle bounce and reentrainment) associated
with the impaction surface. Standard impactor operation is intermittent
because the impaction surface must be physically removed and weighed to
determine the mass of the collected particles. The improvement technique
which has been selected is shown in Figure Ib. The gas/solid impaction
interface has been replaced by a gas/gas interface so that particles of
interest are transferred from the particle laden stream to a particle-
free stream. This interface eliminates the traditional bounce and re-
entrainment problems, but produces some new ones, described below. The
primary advantages of the opposing jet technique are:
1. It produces particle size fractionated output streams which
can be examined on a continuous basis to determine total
particulate mass emissions and size distribution.
2. It can readily be used to simultaneously fractionate and
collect particles in a chemically inert atmosphere.
There is a present and future need for devices with these capabilities,
and it is this need which has been the reason for performing the task.
-------
FLOW
FLOW
STREAMLINE
V////////////////Y//T7
IMPACTION
SURFACE
JET NOZZLE
SMALL PARTICLE
TRAJECTORY
111111111111111 n i
LARGE PARTICLE
TRAJECTORY
Figure la. Schematic of standard particle Impactor
(from Reference 4)
JET NOZZLE
STAGNATION
POINT
URGE PARTICLE
TRAJECTORY
STREAMLINE
SEPARATION
STREAMLINE
SPLITTER
PLATE
,. SMALL PARTICLE
'/ TRAJECTORY
AEROSOL
LADEN
FLOW IN
Figure Ib. Schematic of Impinging jet Impactor
-------
4.2 IMPINGING JET STABILITY
The key to proper operation of an impinging jet impactor is stabil-
ity of the gas/gas interface. The desired flowfield is illustrated in
Figure Ib. Unacceptable flowfields resulting from instabilities are
shown in Figure 2. Unstable operation would result in passage of signif-
icant amounts of gas from one or both nozzles through the plane of the
desired interface. This would render the device incapable of separating
particles in the desired manner. The first task objective was to deter-
mine whether the likelihood of stability was great enough to justify
building and testing a laboratory model.
The structure of a jet of the type being considered is shown in
Figure 3a, and the impinging jet flowfield is shown in Figure 3b. In
standard impactor designs, the jet-to-plate spacing is such that part of
the inviscid core remains by the time the flow hits the plate. The most
useful literature information on the problem came from Reference 1, a
PhD dissertation by R. E. Luna. His test setup is shown schematically
in Figure 4. It consisted basically of two jets exhausting to room air,
the jet flow rates being held constant through upstream pressure regula-
tion and sonic orifices. He found that the flowfield and location of the
interface were stable for jet spacings between 0.5 and 1.5 jet diameters, and
that gross instabilities occurred for spacings greater than two diameters.
For this application, a working definition of stability may be given as
follows:
The impactor flow is stable if, when
disturbed, the interface returns to
its original equilibrium position.
In addition to stability of the flow, flow steadiness is also an important
feature. While stability is largely a property of the concept itself,
steadiness is more related to the actual hardware configuration. As
stated above, the impactor flow is stable if it returns to its equilibrium
position after a disturbance. In addition, it will be steady if no dis-
turbances occur. The most likely source of unsteadiness for this device
-------
STREAMLINE
SEPARATION
STREAMLINE
JET NOZZLE
AEROSOL
LADEN
FLOW IN
SPLITTER
PLATE
Location of separation streamline allows flow to cross through orifice
Figure 2a. Illustration of separation streamline
dislocation Instability
STREAMLINE
JET NOZZLE
CLEAN-
FLOW
IN
^S^3CCEECSS2l
t^*
SEPARATION
STREAMLINE
AEROSOL
LADEN
FLOW IN
h^dt^^A^^A^,.
SPLITTER
PLATE
Higher flow rate through aerosol nozzle spills over to clean side
Figure 2b. Illustration of unequal flow instability
8
-------
NOZZLE,
FLOW
INVISCID CORE
MIXING LAYERS
Figure 3a. Subsonic jet flowflelds
FLOW
MIXING LAYER
NOZZLE
FLOW
FLOW
DIFFUSION
LAYER
INVISCID CORE
FLOW
Figure 3b. Impinging jet flowflelds
-------
HIGH PRESSURE
FILTERED AIR
Key feature:
PRESSURE
REGULATOR
FLOWMETER
JET NOZZLE
HIGH PRESSURE
FILTERED AIR
EXHAUST TO ROOM
The high pressure air source, regulator, and
sonic orifice metering system guaranteed equal
flow rates through the two nozzles
Having both nozzles exhaust to the open room
insured a steady external pressure field in
the vicinity of the interface
Figure 4. Schematic of Luna's test setup for impinging
jet stability testing (from Reference 1)
-------
would be flow rate variations caused by changes 1n feed pressure, pump
characteristics, etc. Even in a stable system, such variations would
cause momentary changes in the shape and location of the interface,
adversely affecting particle separation properties. In Luna's testing,
acceptable steadiness was achieved by precise pressure regulation and
metering of the flow to each jet, in addition to having the jets exhaust
freely to (steady) room conditions.
Review of material in Reference 1 and of a similar study in Refer-
ence 2 led to the conclusion that flow stability could probably be
achieved. A decision was made at this point to continue with hardware
design, fabrication, and test.
4.3 PRE-DESIGN LITERATURE SEARCH
The literature review led to an early decision to work solely with
round jets, rather than rectangular ones, for the following reasons:
Impinging jet background data are available only for circular
jets
Theoretical calculations show sharper cutoffs for circular
than for rectangular jets (Reference 3)
Most commercially available impactors use circular jets
(Reference 4)
The most Important physical design parameters for a circular jet impactor
are nozzle to plate spacing and nozzle geometry. For the impinging jet
impactor, the following additional parameters must also be considered:
Jet relative concentricity and angularity (relative jet
alignment)
Size of the interface orifice
Shape of the Interface orifice edge
Relative flowrates of the two jets
The size of the interface orifice represents a particularly Interesting
engineering tradeoff, as illustrated in Figure 5. Figure 5a shows that
1f the orifice is large, a large particle crossing the interface near
11
-------
SPLITTER
PLATE
PARTICLE
TRAJECTORY
SEPARATION^
STREAMLINE
STREAMLINE
AEROSOL
LADEN
FLOW IN
Figure 5a. Illustration of large particle non-collection
due to large orifice
SPLITTER
PLATE
PARTICLE
TRAJECTORY
SEPARATION
STREAMLINE
STREAMLINE
AEROSOL
LADEN
FLOW IN
Figure 5b. Illustration of large particle non-collection
due to small orifice
12
-------
SPLITTER
PLATE
AEROSOL
LADEN
FLOW
IN
DIFFUSION REGION
MIXING REGION
Figure 5c. Illustration of local flowfield phenomena which
permit undeslred migration of gases and small
particles across the Interface.
Transfer of gas across the interface will occur by diffusion 1n
the diffusion region. Transfer of particles 1n this region will
occur by 1nert1al means.
The mixing region is the merging of the mixing layers in Figure 3B.
Active turbulent mixing will occur here, resulting in transfer of
both gases and small particles.
Flow unsteadiness (not illustrated) will also cause momentary
transfer of gas and particles across the interface by convection.
13
-------
the axis of the jets may be swept back across the interface and not
collected. Figure 5b shows that if the orifice is small, a large
particle near the jet wall will not have an opportunity to be collected.
Figure 5c, derived from Figure 3b, shows that at the interface, a mixing
layer is set up between the two jets. As the mixing layer thickens along
the interface, the mixing process carries small particles across the
interface to the collection side. Thus some small particles will always
be collected along with the larger particles. The larger the interface
orifice, the more small particles will be collected. Thus for the
impinging jet impactor we should expect to see an efficiency curve sim-
ilar to that in Figure 6 - some small particles will always be collected,
while some large particles will not be. Engineering tradeoffs would
involve optimizing the efficiency curve for a particular application.
Identification of the primary physical variables for the device
allowed us to proceed to the design and fabrication subtask discussed in
the following section.
14
-------
t
COLLECTION
EFFICIENCY, %
100
IDEAL BEHAVIOR
CONTINUOUS IMPACTOR
EFFICIENCY CURVE
NON-COLLECTION
DUE TO ORIFICE
SIZE LIMITATION
UNDESIRED COLLECTION
DUE TO UNSTEADINESS,
JET MISALIGNMENT,
DIFFUSION
IDEAL BEHAVIOR
PARTICLE DIAMETER
Figure 6. Anticipated continuous Impactor efficiency performance
curve, based on known hardware limitations
15
-------
SECTION 5
SUBTASK 2 - DESIGN AND FABRICATION OF TEST UNIT
The philosophy behind the design of the prototype continuous flow
impactor was to provide the capability to vary those parameters considered
to be of possible importance to jet-jet interface stability, since the
stability of impinging axisymmetric jets under stack sampling constraints
had not been previously investigated. Flexibility in impactor configura-
tion was, therefore, deemed necessary to identify configurations possess-
ing stable flow which might be suitable for particle separation work.
The impactor, as shown in Figure 7, is seen to be basically a
cylindrical housing, surrounded by two manifolds, with a ring shaped
center plate to which different splitter plates may be attached. Cylind-
rical nozzle holders, to which various sized and shaped nozzles can be
attached, are inserted into each end of the impactor and are positioned
relative to the splitter plate by means of micrometer heads attached to
the nozzle holders.
5.1 DESIGN PARAMETERS AFFECTING STABILITY
Parameters identified as being of probable importance to interface
stability were associated with three aspects of the impactor configuration-
nozzle - splitter plate separation, splitter plate orifice size and edge
shape, and nozzle. Each of these aspects is discussed below.
5.1.1 Nozzle - Splitter Plate Separation
It was felt that nozzle - splitter plate separation might affect
stability in two ways. First, there is the separation distance between
the nozzle and the splitter plate, and between the two nozzles, witze
and Dwyer (Reference 2) showed that the steadiness (in the mean properties\
of the flowfield produced by two impinging axisymmetric jets is dependent
upon the distance between the nozzle exit planes. Luna (Reference 1)
found similar results. Second, whether the nozzles are spaced at equal
or unequal distances from the splitter plate might also affect stability.
5.1.2 Splitter Plate Orifice
The splitter plate orifice, it was thought, might affect stability
in two ways, the first being the diameter of the orifice. This diameter
16
-------
-1
^J
O
o
O
c
CO
o
I
o»
4/1
cr
^*'
Q.'
O»
Jia
-------
sets the size of the interface. A standard impactor, which is stable,
may be considered to have an orifice diameter of zero. The bulk of the
work in References 1 and 2 was done with no splitter plate, i.e., an
infinite orifice. Our case lies somewhere in between. The second area of
concern was the shape of the orifice edge. It was not known, at this
stage of the study, whether the sharpness of the orifice edge would affect
the ability of the interface to attach itself to the orifice edge, which
determines whether or not there is flow across the orifice.
5.1.3 Nozzles
Aspects of nozzle design that were felt to have a possible effect on
interface stability were nozzle diameter, which, in part, determines the
flow Reynolds number, and nozzle shape.
5.2 VARIATION OF STABILITY PARAMETERS
Design features permitting variation of the parameters discussed
in Section 5.2 are detailed below.
5.2.1 Nozzle-Splitter Plate Separation
The impactor was designed so that each nozzle-nozzle holder assembly
could move independently with respect to the splitter plate. Micrometer
heads attached to each assembly allow a nozzle to be positioned with
respect to the splitter plate to within 0.001-inch. Therefore, any
desired nozzle-splitter plate separation distance could be examined to
determine its effect on interface stability. Since each nozzle-nozzle
holder assembly moves Independently of the other, the effect of both
equal and unequal nozzle-splitter plate separation distances upon inter-
face stability could be investigated.
5.2.2 Splitter Plate Orifice
The impactor center plate, shown in Figure 7, was designed so that
various splitter plates, with orifices of different sizes and edge shapes
could be fitted to it. The splitter plate design is shown in Figure 8.
Splitter plates were manufactured with orifice diameters of 1.0, 2.0, and
3.0 centimeters. For each orifice size, three different orifice edge
shapes were used, yielding a total of nine different splitter plates avail.
able for stability testing.
18
-------
(O
c
"5
n>
00
Ol
o
n>
-J
o
(V
> 4 S
INCHICAIC
-3
D/M(
.394
.767
I.IBZ
.O/6
.0/6
.O/6
.OX ML R
wcui.«
.O3ZFUU.K
.OKFUU.R.
.
-------
5.2.3 Nozzles
The effects of nozzle size and shape upon Interface stability were
examined by butldtng different pairs of nozzles. Each nozzle assembly
consisted of two parts, a nozzle and a nozzle holder, shown in Figures 7,
9, and 10. Only one set of nozzle holders was built, as each nozzle was
designed to be compatible with the nozzle holders.
The nozzles were butlt in two different shapes, contoured and coni-
cal, as shown in Figures 9 and 10. Marple and Willeke (Reference 4) have
shown that the more uniform the flow is at the nozzle exit, the sharper
the cutoff will be in a standard impactor. This result should also be
applicable to a continuous flow impactor. On the basis of this result,
contoured nozzles, using the nozzle destgn of Smtth and Wang (Reference 5)
to achieve uniform flow at the nozzle exit, were built. A pair of coni-
cal nozzles was also built. It was hoped that a conical nozzle impactor
would perform well enough, tn comparison to a contoured nozzle impactor,
to permit the use of the less expensive conical nozzles in any commercial
application of the instrument. The conical nozzles had a cone angle of
60°, this being the same cone angle used in experiments by Ranz and Wong
(Reference 6). A nozzle dtameter of 1,0 cm for both contoured and conical
nozzles was chosen for the fnttial studies of stability.
While it is belteved that concentricity of the impactor nozzles with
each other and with the splitter plate orifice is important to interface
stability, it is felt that in a production model, tolerances normally
encountered in machining would give adequate concentricity. Therefore,
provisions to vary nozzlemozzle and nozzle-splitter plate orifice conceru
tricity were not included in the design of the Impactor. Each nozzle-
nozzle holder assembly rides on three precision shaftings resting in
grooves in the Impactor housing to assure concentricity of the nozzles
and the splitter plate.
5.3 OTHER DESIGN CONSIDERATIONS
The impactor is connected to a vacuum pump through manifolds on each
side of the splitter plate. There are eight equally sized and spaced
circular inlets into each manifold from the Impactor. The inlets and the
cross-sectional area as seen by the flow inside the manifolds are large
20
-------
id
c
(D
VO
o
c+
o
ro o
_J O
(D
CL
=t
O
N
n>
-------
» « s i
MCHKAU
-s
"A"
. 78 7
-0930
OTY REOD HR ASSV
XJOZZCE
BUgiggfettea
T
IA|
<3.OOO OW
s-
TRW
UO2ZLE,PARTICLE /MR '.
COX BUT «L
11982
-------
compared to the nozzles, to prevent large pressure drops across the Inlets
and Inside the manifolds. This assures symmetrical radial flow outward
after impingement of the jets, rather than having more flow outwards toward
the side of the manifold that is connected to the pump.
Access ports into the impactor, on either side of the splitter plate,
were provided to allow hot wire anemometry measurements of interface
stability and selective sampling of aerosols in the particle separation
phase of testing.
5.4 FABRICATION OF TEST UNIT
All fabrication was performed in TRW machine shops. Aluminum was
selected as the material for construction since the unit would only be
operated at room conditions. No unusual problems were encountered during
the fabrication effort.
Since there were significant doubts about the stability characteristics of
the device during the early stages of the task, a large portion of the task
budget was put into design and fabrication. Attention was concentrated on
alignment and pressure distribution to maximize flow uniformity and elimin-
ate disturbances and asymmetries as much as possible.
23
-------
SECTION 6
SUBTASK 3 - INITIAL LABORATORY, TESTING
6.1 TEST SETUP
Initial testing of the prototype continuous flow impactor was performed
in the Engineering Sciences Laboratory nitrogen flow room. The basic test
setup, common to both the stability and particle separation phases of
testing, was as described below and shown in Figure 11.
Each of the impactor's two manifolds was connected by 1/4-inch stain-
less steel tubing to a rotameter flowmeter. The flowmeters could measure
flow rates in the range from zero to approximately 0.65 1/s (1.5 cfm).
Needle valves downstream of the flowmeters permitted adjustment of the
flow rate through each manifold. Downstream of the needle valves, the
two flows were merged into a 1/2-inch stainless steel line by means of a
tee connector. The 1/2-inch line ran to a vacuum pump located in the room
adjacent to the nitrogen flow room. The pump was located outside the test
area to prevent the possibility of particle separation test data errors
due to oil vapor droplets from the vacuum pump. The pump used in testing
was a Welch Model 1403 Duo Seal Vacuum Pump, with a pumping capacity of
1.6 1/s (3.5 scfm).
Pressure taps were installed in the impactor housing tube to monitor
the difference in impactor chamber static pressures on either side of the
splitter plate, and installed In each nozzle, near the nozzle exit, to
measure the pressure drop across each nozzle. The impactor housing pres-
sure taps were connected to the ports of a MKS 77H-10 Pressure Head which
was connected to a MKS Type 77 Baratron Pressure Meter. The pressure taps
from each pair of nozzles were connected to one port of a MKS 77H-1 Pres-
sure Head through a five-way switching valve. The second port was left
open to ambient air and the pressure head was connected to a second MKS
Type 77 Baratron Pressure Meter. This arrangement allowed the measurement
of only one nozzle pressure drop at a time, but the use of the five-way
valve permitted rapid changing of the connector between pressure head and
nozzle pressure taps.
24
-------
ROOM
AIR
a. Stability test - Initial
HIGH PRESSURE
AIR
b. Stability test-modified
M| METERING VALVE
71 FLOWMETER
ROOM
'AIR
R| REGULATOR
HIGH PRESSURE
AIR
c. Particle separation test
Figure 11. Test setup schematics
25
AEROSOL LADEN AIR
PARTICLE
COUNTER
-------
6.1.1 Stability Test Setup
In addition to the basic test setup for the impactor, additional
equipment was used in the stability phase of testing. A Thermo Systems
Inc. (TSI) Model 1010 Constant Temperature Anemometer and a TSI Model
1220 hot film sensor probe were used to measure fluctuations in one jet
with the splitter plate orifice blocked off. A Hewlett Packard Model
41OC Multimeter was used to measure the RMS (root mean square) value of
the signal fluctuations from the TSI probe, and a Tektronix oscilloscope
was used to estimate the peak-to-peak value of the fluctuations in the
signal. A Leeds and Northrup Model W Speedomax Strip Chart Recorder was
used to record the pressure drops across the impactor nozzles and the
static pressure difference in the impactor across the splitter plate.
During the stability phase of testing, It was decided that, in addi-
tion to controlling the flow downstream of each manifold, flow rate
control was also needed upstream of one of the two nozzles. A plate was
made to fit into the inlet of the nozzle-nozzle holder assembly, leakage
being prevented by use of two 0-rings set in grooves in the plate. The
plate was fitted with a 1-1/2 inch pipe fitting which was reduced down to
a 1/4-inch tube fitting. The plate was connected by 1/4-inch flexible
tubing to a rotameter flowmeter. A needle valve was located upstream of
the flowmeter. The valve was connected by 1/4-inch flexible tubing to a
regulated nitrogen bottle or trailer. This upstream flow control was
used in the remainder of the stability test phase, as shown in Figure lib
6.1.2 Particle Separation Test Setup
The setup for the particle separation phase of testing consisted of
the basic setup described above and the following equipment, as shown
schematically in Figure lie.
Control of flow into one nozzle, using the end plate-flowmeter-needle
valve combination described in Section 6.1.1, was used during the entire
particle separation phase of testing. A^regulated nitrogen bottle or
trailer was used as the gas source. The nozzle with the upstream flow
control provided the clean jet, i.e., the jet into which particles were
not injected.
26
-------
A Berglund Liu Model 3050 Monodisperse Aerosol Generator, built by
Thermo Systems, Inc., was used as the source of monodisperse particles
that were injected into the impactor. The aerosol generator was equipped
with a B&K Model E-310B Sine/Square Wave Generator and a Harvard Apparatus
Model 975 Compact Infusion Pump. The pump used B-D Plastipak Disposable
plastic syringes as a source for the solutions used in the aerosol gener-
ator. The particles generated were introduced into the impactor by posi-
tioning the aerosol generator outlet hose at the inlet to the nozzle-nozzle
holder assembly that had not been modified for upstream flow control.
The aerosol generator functions in the following manner. A solution
of non-volatile solute in a volatile solvent is pumped by the infusion
pump through a filter fnto the aerosol generator. The solution is forced
through a small orifice, with a diameter of five, ten, or twenty microns,
in the form of a liquid column. The orifice is in a disc that is vibrated
by a piezoelectric crystal connected to the signal generator. The oscillat-
ion of the orifice causes the column of solution to break up into uniformly
sized droplets. The droplets pass through a second, larger orifice. Gas
also flows through this orifice, preventing the droplets from combining
to form larger droplets. The droplets are then entrained in gas flowing
upward through the aerosol generator's vertical drying column. This gas
flow serves two purposes. First, it evaporates the solvent, leaving only
the solute as a liquid or a solid particle, depending on whether the solute
is a liquid or a soltd. Second, tt carries these solute particles through
the outlet hose at the top of the drying column to the point of application
of the particles. The gas source used during testing with the aerosol
generator was a regulated nitrogen bottle or trailer. Knowing the solution
feed rate into the aerosol generator, the frequency of oscillation of the
orifice, and the concentation of the solution used, the size of the parti-
cles produced can be calculated accurately.
A Royco Model 225 Particle Counter Mainframe with Sensor, equipped
irfth a Model 518 digital display plug-in readout meter to display particle
counts, was used to count particles in samples drawn from either side of
the impactor splitter plate. The samples were drawn out through 1/4-inch
stainless steel tubes inserted through ports in the impactor housing. The
sampling tubes were located on either side of the splitter plate at a
27
-------
radial distance of two nozzle diameters from the impactor centerline and
at a distance of approximately 3/4 of a nozzle diameter from the face of
the splitter plate, as shown in Figure 12. The tubes were placed as close
as possible to the interface area, without being so close as to affect
the flowfield, in order to avoid sampling in an area where loss of particles
to the wall could occur. The sampling tubes were connected, alternately,
to the particle counter sensor by a minimum length of flexible tubing.
The sensor was connected by flexible tubing to a pump located in the counter
mainframe. This pump was used to draw the samples from the impactor.
The Royco Model 225 particle counter uses a near forward scattering
optical system in its sensor to detect particles. It can detect particles
with diameters of 0.5 microns or greater. The counter can handle particle
3
concentrations of up to 3 x 10 per cubic centimeter, and can sample at
o
two flow rates, 50 and 5 cm /sec. The counter mainframe is equipped with
a built-in calibration circuit and meter to calibrate the counter whenever
desired. The Model 518 readout module has a five channel memory into which
the particle counts In five adjustable size ranges may be simulataneously
stored. As used In the particle separation tests, the 518 module stored
particle counts In the following size ranges:
Table 1. Royco 225 Particle Size Discrimination Characteristics
Module Channel
1
2
3
4
5
Size Range
All particles > 0.5 microns
All particles > 0.7 microns
All particles > 1.4 microns
All particles > 3.0 microns
All particles > 5.0 microns
The module can be set to store counts taken over one minute or ten minute
periods, or the sample time duration can be controlled externally. Tne
counter was calibrated by a Royco employee during the test phase and
found to be In proper working order.
28
-------
SAMPLING
TUBE
2 CM
\
SPLITTER PLATE ORIFICE
a. Overhead view
SAMPLING TUBE
JET NOZZLE
SPLITTER PLATE
0.75 CM
b. Side view
Figure 12. Particle sampling detail
29
-------
6.2 TEST CONDUCT
6.2.1 Stability Testing
The first phase of testing dealt with investigation of the gas
dynamic properties of the impactor. The intended diagnostics techniques
were hot film anemometry and pressure measurement. The instability mode
which was considered to be of greatest concern prior to testing was oscil-
lation of the interface, which would allow some of the flow from each side
to cross over to the other side. That mode turned out not to be the one
which was actually observed during the test. The type of instability which
did occur, and is described in detail in Section 6.3.1, resulted in major
alterations of the originally proposed stability test sequence. The
stability testing as it actually occurred is described below.
6.2.1.1 Standard Impactor Configuration Investigation
It was decided that data would be taken with a standard impactor
configuration (Figure la) to serve as baseline data. The data obtained
consisted of hot film anemometer velocity and turbulence traverses in the
vicinity of the jet. A splitter plate with no hole was the impactor sur-
face. The purpose was to provide quantitative information about the jet
behavior which would be compared to data taken from the opposing jet con-
figuration. Data were taken for both conical and contoured nozzles.
Results are presented in Section 6.3.1.
6.2.1.2 Opposing Jet Configuration Investigations
Upon completion of the hot film testing described above, the opposing
jet configuration was set up to repeat the hot film measurements. Nozzle
differential pressure readings made it very evident that the desired and
intended equal flow through each of the two nozzles was not being obtained
The test setup at this point was as shown in Figure lla. The nozzle flow-
rates were observed to vary randomly over a range of ± 60% or more from
the nominal desired flow, even though the flowrate through each outlet
manifold stayed constant.
After enough pressure data were obtained to substantiate the nature
of the instability, a meeting was held to discuss the problem and the
configuration shown in Figure lib was proposed. Stability was verified
30
-------
for this configuration by means of pressure and flow measurements, as des-
cribed 1n Section 6.3.1. At this point, a decision was made to proceed direct-
ly to particle separation testing, due to budget and schedule constraints, so
no hot film work was performed on the opposing jet configuration.
6.2.2 Particle Separation Testing
6.2.2.1 Impactor Configurations Tested
Specifications for the impactor configurations tested in the particle
separation phase of testing, such as nozzle-nozzle spacing and splitter
plate orifice size, are given in Table 2 (p. 40). The variations made 1n
configuration were suggested by the results of particle separation for pre-
viously tested configurations. Two comments, concerning nozzle-nozzle
spacing and splitter plate orifice edge shape, should be made. The
nozzle-nozzle spacing was fixed at 1.5 nozzle diameters, with equal spac-
ing between each nozzle and the splitter plate, giving a nozzle-splitter
plate spacing of 0.75 nozzle diameters. Unequal spacing was not tried
because It was felt that this would be equivalent to having unequal flow
rates through the Impactor nozzles, which was tested. The nozzle-nozzle
spacing used was chosen for two reasons. First, Marple and Willeke
(Reference 4) have shown that for a standard Impactor with a round jet,
the 50* cutoff size drops rapidly for nozzle-to-plate spacings less than
0.5 nozzle diameters. This means that for small nozzle-to-plate spacings,
almost all particles are collected Instead of particle separation Into
different size classes betng achieved. Therefore, the nozzle-nozzle
spacing had to be greater than 1.0 nozzle diameters. Second, Luna (Refer-
ence 1) demonstrated that the flowfleld produced by two impinging circular
jets Is unstable if the nozzle-nozzle spacing 1s greater than 2.0 nozzle
diameters. So, as a compromise between these two limits, a nozzle-nozzle
spacing of 1.5 nozzle diameters was used. This 1s the same spacing
preferred by Luna. The splitter plate orifice edge shape was not varied
due to a lack of time.
5.2.2.2 Aerosol Generator Operation
The size of the particles generated by the Berglund Liu Model 3050
aerosol generator depend on three factors: 1) solution volume flow rate,
2) frequency of oscillation of orifice, and 3) solution concentration.
To accurately know the size of the particles produced, each of these
factors must be known accurately.
*5 I
-------
The solution volume flow rate was determined by flowing distilled
water from the infusion pump into a graduated cylinder and measuring the
time in which a certain volume of water flowed. The pump was run at an
intermediate pump speed setting. The relationship between flow rates for
two different pump speed settings was known, so the flow rate at the recom-
mended pump speed setting could then be determined. The infusion pump was
calibrated with two different sized syringes, 20 cc and 10 cc, which were
used with the 20 micron and 10 micron orifice plates, respectively.
The signal generator supplied with the aerosol generator was tested
by the Metrology Department, which concluded that the generator met the
manufacturer's specifications. The operating frequency range for the
aerosol generator depends on the orifice size of the vibrating plate. The
procedure used to determine the frequency range is described later in this
section.
Solutions of Di Octyl Phthalate (OOP) in reagent grade isopropyl
alcohol, prepared by the Chemistry Department, were used in the
aerosol generator. The concentrations, by volume, of the solutions pre-
pred were 1:1000, 1:500, 1:250, 1:125, and 2:125.
The 10 micron and 20 micron orifice plates were used in the aerosol
generator. The solution flow rate is higher, but the operating frequency
range is lower for the 20 micron orifice than the 10 micron orifice. The
operating manual provided with the aerosol generator lists the proper
infusion pump speed setting for each size orifice, with which one can then
determine the solution flow rate for each size orifice, as described above
and the nominal operating frequency range for each size orifice. The
actual frequency range is determined in the following manner. Solution
is flowed through the orifice at the proper rate. The solution column is
deflected by gas flowing through a nozzle at a rate of 700 cc per minute.
Beginning at the lower end of the nominal frequency range, the signal
generator frequency setting is increased until a nonuniformly sized parti..
cle stream is observed. The frequency at which this occurs, f is the
maximum frequency at which uniform droplets can be produced for the given
orifice end solution flow rate. The minimum recommended frequency is
0.5 fmax, so the operating frequency range is 0.5 fmax to fmax. The
manufacturer claims that the best performance is obtained at a frequency
of 0.6 f .
max 32
-------
The sizes of particles used in the particle separation phase of test-
ing could be calculated once the solution volume flow rate, frequency range,
and solution concentration were known. The particle size (diameter) range
used in the testing was 2.44 to 11.56 microns.
The gas flow rate through the aerosol generator is controlled by two
sets of rotameter flowmeters and needle valves built into the aerosol
generator. During the particle separation phase of testing, this flow
rate was always kept greater than the flow rate through the nozzle into
which the particles were introduced. The resultant excess flow, which
exited from the impactor through the nozzle holder inlet, was designed to
prevent particles in the room air from being introduced into the impactor.
6.2.2.3 Particle Counter Operation
The Royco Model 225 particle counter, as noted in Section 6.1.2, can
3
sample at two different flow rates, 50 and 5 cm /s, and for durations of
one and ten minutes. Durtng the particle separation testing, a sample
flow rate of 50 cm /s and a sample duration of one minute were used. The
shorter sample duration was used to avoid errors in the measured separ-
ation efficiency due to possible changes over a long period in the output
from the aerosol generator caused by a slow drift in the pressure at
which nitrogen was supplied to the aerosol generator.
6.2.2.4 Particle Separation Test Procedure
The procedure for the particle separation phase of testing is listed
below:
(1) Set up the Impactor in the configuration to be tested.
(2) Start vacuum pump, nitrogen flow through clean side of impactor
(i.e., side into which particles are not injected), and set
flow rates through Impactor.
(3) Aerosol generator is started. Due to high initial liquid flow
rate needed to start generator, must wait several minutes for
liquid flow rate to decrease to proper rate before taking
sample from Impactor.
(4) Take sample from sample tube on one side of splitter plate.
33
-------
(5) Record particle counts on all five readout meter channels
(recording counts on all channels serves as a check on whether
the aerosol generator 1s producing monodisperse particles).
(6) Connect particle counter to sample tube on other side of
splitter plate.
(7) Take sample.
(8) Record particle counts on all readout meter channels.
(9) Repeat steps (4) through (8) two to four times. I
(10) After every three to four pair of measurements, where steps
(4) through (8) constitute a pair of measurements, check the
particle counter calibration using its built-in calibration
circuit and meter.
(11) Change particle size by changing frequency setting of signal
generator.
(12) Repeat steps (4) through (10).
(13) Repeat steps (10) through'(11) until measurements have been
made at all desired frequency settings.
(14) Turn off Infusion pump. Leave aerosol generator nitrogen
flow on.
(15) Take particle count measurements on each side of the splitter
plate. (This step is used to check whether the nitrogen
used 1s clean.).
(16) Turn off aerosol generator nitrogen flow and signal generator.
(17) Turn off vacuum pump and Impactor clean side nitrogen flow.
(18) Change aerosol generator vibrating orifice plate if desired.
(19) Flush aerosol generator with reagent grade isopropyl alcohol.
(20) Change solution in aerosol generator 1f desired.
(21) Repeat steps (2) through (20).
34
-------
(22) Repeat step (21) until Impactor configuration has been tested
at all desired aerosol generator settings.
(23) Repeat steps (1) through (22) until all desired Impactor
configurations have been tested.
6.3 TEST RESULTS
6.3.1 Stability Testing
6.3.1.1 Standard^ Impactor Configuration Investigation
Turbulence profiles obtained using a hot film anemometer are presented
in Figures 13 and 14 for the conical and contoured nozzles, respectively.
The data trends show relatively low turbulence levels in the jet itself.
As the nozzle to plate spacing increased, the measured turbulence level in
the jet also increased due to the effect of mixing. Turbulence was seen
to increase rapidly 1n the region where the flow was being turned (up to
a distance of about 5 nozzle radii) and then it started to drop off after
completion of the turn.
It was the original Intent to repeat the measurements in the opposing
jet configuration, but the stability problems described in the following
paragraph required too much attention, and no addition hot film data were
taken. The data in Figures 13 and 14 indicate the probable existence of
rapidly increasing turbulence along the interface between the jets, sug-
gesting that some turbulent mixing will occur along the interface which
would cause small particle to be convected across the interface.
6.3.2.2 Opposing Jet Configuration Investigations
It is useful to consider flow through the Impactor as consisting of
four separate streams - two inlet streams and two outlet streams. Con-
servation of mass requires that the sum of the inlet flowrates be equal to
the sun of the outlet flowrates. In the test setup used by Luna, shown 1n
Figure 15a, each of the two inlet flowrates was actively and separately
maintained constant, while the outlet flowrates simply merged Into a single
flow. This type of setup was excellent for Investigating the basic stabil-
ity of opposing jets, but is not applicable to the source sampling problem.
In the present Investigation, it was recognized that the aerosol laden
stream must be altered as little as possible prior to impaction and analysis,
35
-------
1
2
2.9
SYMBOL
2 46 8 10
r, DISTANCE FROM NOZZLE CENTERLINE, NOZZLE RADII
Figure 13. Turbulence profiles for conical nozzle
36
-------
SYMBOL
24 6 8 10
r. Distance from Nozzle Centerllne, Nozzle Ra
-------
HIGH PRESSURE
FILTERED AIR
PRESSURE
REGULATOR
FLOWETER
JET NOZZLE
HIGH PRESSURE
FILTERED AIR
EXHAUST TO ROOM
a. Luna's test configuration
(from Reference 1)
TO PUMP
FLOWETER AND
METERING VALVE
ROOM AIR
ROOM AIR
b. Stability - original
REGULATED AIR
HIGH PRESSURE
TO PUMP
FLOfcMETER AND
METERING VALVE
ROOM AIR
c. Stability - modified
Figure 15. Flow control schematics
38
-------
In addition, the two output streams had to be kept separate for purposes of
analysis. We therefore selected the configuration shown 1n Figure 15b for
the laboratory setup. Flow control was performed only on the outlet streams
each of which was maintained at a flowrate of 470 cm3/s. No controls were
Initially established on the inlet streams other than to insure that the
Inlet geometries were identical. The wisdom of hindsight reveals that
identical geometry was not sufficient to promote identical flows through
the two inlets. It was observed during the early stability testing that
Wh1le the sum of the inlet flowrates remained constant, the individual
flowrates varied over as much as t60% range 1n a random manner. This was
a very gross Instability, worse than the interface instability which was
the original concern and which Luna investigated in detail.
The solution to the problem was to actively meter the "clean side"
inlet flow in addition to the two outlet flows. This provided the neces-
sary stabilizing constraint -metering of the outlet flows fixed the
total flowrate, and metering of one of the inlet flows therefore passively
regulated the other inlet flow. The final configuration is shown in
Figure 15c. As is shown in Section VII, this approach is compatible with
source sampling requirements.
Visual observation of pressures and flows for this configuration in
Figure 15c confirmed flow stability. Pressure traces were obtained and
are presented 1n Appendix A. The traces show pressure (and consequently
flow) stability to the performance limits of the regulator used on the
clean side inlet.
6.3.2 Particle Separation Testing
Results for this phase of the test work are presented as efficiency
curves - separation efficiency as a function of particle size. Calibra-
tions for six separate physical configuration/flow combinations were
obtained. The variety of conditions has helped to provide an understanding
of how this Instrument functions, and to identify the critical design and
operational parameters. Operating conditions are summarized in Table 2,
and the efficiency curves are shown in Figures 16 through 22. Figure 16
Is a summary plot of the data trends for each configuration, while the
39
-------
Table 2. OPERATING CONDITIONS FOR PARTICLE
SEPARATION TESTS
CONFIGURATION
A
B
C
D
E
F
FIGURE
17
18
19
20
21
22
FLOWRATES, CM3/S
CLEAN SIDE
IN
470
470
470
420
420
520
OUT
470
470
470
470
470
470
AEROSOL SIDE
IN
470
470
470
520
520
420
OUT
470
470
470
470
470
470
SPLITTER PLATE
ORIFICE DIAMETER,
CM
3
3
2
3
2
3
NOZZLE
SHAPE
Conical
Contoured
Contoured
Conical
Countoured
Conical
For all configurations* nozzle/nozzle spacing was 1.5 cm.
-------
90
80
70
60
50
o
o
t-<
o
0 40
30
20
10
I
2 4 7 10
PARTICLE DIAMETER, MICRONS
Figure 16. Data trends for particle collection
efficiency testing, all configurations
41
CONFIGURATION
D
-------
O
II
Ll_
LU-
LU
o
i<
^-
o
o
o
80
70
60
50
40
30
20
10
_L
O
O
2 4 7 10
PARTICLE DIAMETER, MICRONS
15
Figure 17. Particle collection efficiency as a
function of particle size, Configuration A
42
-------
o
Ul
o
o
80 r-
70
60
50
40
30
20
10
I
_L
2 4 7 10
PARTICLE DIAMETER, MICRONS
15
Figure 18. Particle collection efficiency as a
function of particle size, Configuration B
43
-------
a*
o
Ul
II
CJ
o
UJ
o
o
80
70
60
50
*^ * *h
u. 40
30
20
10
O
O
2 4 7 10
PARTICLE DIAMETER, MICRONS
15
Figure 19. Particle collection efficiency as a
function of particle size, Configuration C
44
-------
100
.
to
o
o
90
80
» 70
>-
60
50
40
30
2 4 7 10
PARTICLE DIAMETER, MICRONS
15
Figure 20. Particle collection efficiency as a
function of particle size, Configuration D
45
-------
o
*-l
o
90
80
70
60
50
40
30
20
J_
_L
2 4 7 10
PARTICLE DIAMETER, MICRONS
15
Figure 21. Particle collection efficiency as a
function of particle size, Configuration E
46
-------
60
50
~ 40
u.
U-
LU
O
UJ
O
O
30
20
10
24 7 10
PARTICLE DIAMETER (MICRONS)
15
Figure 22. Particle collection efficiency as a
function of particle size, Configuration F
47
-------
remaining figures show the individual data points obtained. Although dif-
fering numbers of data points were obtained for the various configurations
in each case a large enough number of points was obtained with sufficiently
narrow scatter to establish the data trends. Thus the occurrence of
decreasing efficiency with increasing particle size, for example, which
occurred for two of the configurations, is considered to be a real phenom-
enon and not due to measurement errors. As Table 2 shows, the parameters
varied were nozzle shape, orifice size, and relative flowrate through the
nozzles. The nature of the flowrate variation is shown schematically in
Figure 23. For all configurations, exhaust flowrates were maintained
equal, while the clean side nozzle flow was regulated to provide equal or
unequal flows through the nozzles.
The efficiency curves for four of the configurations are compared with
that of the first stage of an Anderson impactor (Reference 4) in Figure 24
and show comparable sharpness of cutoff for the portions of the curves
above 40% efficiency. The problems associated with unwanted collection of
small particles are discussed below.
6.4 INTERPRETATION OF TEST RESULTS
The possibility of the flowrate instability observed early in the test
program was anticipated, but it was originally felt that good control of
the inlet geometry would result in stable operation. When this proved not
to be the case, appropriate time and effort had to be devoted to finding
a solution to the problem. The solution of metering the clean side Inlet
flow was very successful, but the effort involved in achieving it precluded
(from a budget and schedule standpoint) any further investigation of the
flow properties in the region of the interface itself. Some information
has been inferred about the interface region from the particle separation
test data, discussed below. The stability testing showed a definite need
to control three of the four flowrates involved, and it is felt that this
can be done in a production Item without introducing great hardware com-
plexity. We can now proceed with interpretation of the particle separation
data.
The first check performed on the data was to compare data with a
theoretical calculation of the 50% efficiency point. The S0% efficiency
48
-------
470
CLEAN
SIDE
FLOWRATES FOR CONFIGURATIONS D AND E, CM°/S
470 A A 470
CLEAN
SIDE
520
AEROSOL
SIDE
420
FLOWRATES FOR CONFIGURATION F, CMJ/S
Figure 23, Schematic of flowrate variation
technique for biased Inlet flow
cases, Configurations D, E and F,
49
-------
90
80
70
& 60
ii
o
HH
u_
u_
LU
850
o
o
o
40
30
20
ANDERSON
IMPACTOR
FIRST
STAGE
(REF 4)
2 4 7 10 15 20
PARTICLE DIAMETER, MICRONS
Figure 24. Comparison of continuous impactor data
with Anderson first stage
50
-------
particle diameter was calculated from Figure 12 of Reference 4, which
showed that at SQ% efficiency, we have
J5TR" = 0.474
By definition
p C V D2
STK = PA '° P = Stokes number
rjet
where
p = particle density, * 1.1 g/cm3 for OOP
C = Cunningham slip correction factor, * 1.02 in the region
of interest
V = mean jet (nozzle) velocity
D s particle diameter
y = gas viscosity, 1.781 x ICf1* g/cm-s
r,.f s nozzle radius, 0.5 cm
jet
For our case, we had a flowrate of 1 CFM, or 472 cm3/s, which gives
(1.1) (1.02) (601) D2,
V = 601 cm/s. Thus we have
o
STK
(18) (1.781x10'*) (.5)
STK = 4.21 x 105 D*
P
so that
Dn * °-474 _ = 7.31 x lO'^cm = 7.Sly
P50 V4.21 x 105
The observed 50% efficiency point for configurations A, B, and C, Figures
17-19, were 7.6, 8.4, and 9.3u, respectively. Since the conical nozzle
used in Configuration A most approximates the configuration used to obtain
the background data for the calculated point, agreement between calculation
and experiment was considered to be excellent, and confirmed that basic
operation of the device was not significantly different from that of a
51
-------
standard impactor. The unequal inlet flowrates in Configurations D-F were
a sufficient deviation from the standard impactor that a comparison with
standard cutoff calculations would not be meaningful.
Figure 24 is illustrative of problems during this early devel-
opment phase of the task. Sharpness of cutoff above 50% efficiency is
seen to be comparable to that of one of the most successful impactor
designs, which is good. Unfortunately, the sharpness is seen not to con-
tinue into the low efficiency part of the curve -a significant percentage
of small particles was collected in all cases. Such collection has
definite implications as to what is occurring in the vicinity of the
interface. Configurations D, E, and F in Table 2 were used for the specific
purpose of investigating possible improvements to the top and bottom ends
of the efficiency curve through deliberate flow biasing.
The results for configurations D and E (Figures 16 and 17) are easily
understood. The higher flow rate through the aerosol nozzle (by convention
"aerosol nozzle" will refer to the nozzle through which the particle gen-
erator output flows, while "clean nozzle" will refer to the nozzle through
which the initially particle free gas flows) dictated that some of the
aerosol laden gas would exhaust on the clean side of the splitter plate.
Since this gas would carry its entrained particles along with it, we would
expect a leftward shift of the efficiency curve compared to the equal flow-
rate case. The major difference between configurations D and E was the
larger orifice size for D, which, as Figure 5 suggests, resulted in higher
efficiencies for a given particle size for configuration D.
The "leveling out" of collection efficiencies at around 30% for con-
figurations A, B, and C and the unusual shape of the efficiency curve for
configuration F were initially puzzling. The following scenario, based on
the test data, was hypothesized upon conclusion of initial testing. The
hypothesis was confirmed by subsequent tests, as described in Section VII
It became clear that the key to the unusual portions of the effi-
ciency curves is as illustrated in Figure 25. This figure, adapted from
Reference 1, shows the calculated free streamline shape for the configura-
tion tested. The nozzle I.D. was one centimeter, and the distance from
52
-------
Free streamline shape was calculated for the
case of an infinite interface (no splitter
plate). Splitter plate edge detail is shown
separately to indicate the relative jet and
splitter plate thicknesses. All dimensions
are to same scale.
FREE
STREAMLINE
.081 CM
SPLITTER
'PLATE
EDGE
en
CO
DIVIDING
STREAMLINE
1 CM
(edge of 2 cm
orifice)
1.5 CM
(edge of 3 cm
orifice)
Figure 25. Calculated free streamline shape (from Reference 1)
-------
the jet exit to the interface was 0,75 cm. Figure 25 shows the predicted
thickness of the jet along the interface. As noted in Table 2, two orifice
diameters were used - 2 and 3 cm. At the edge of the orifice, the calcu-
lated flow heights were 0.125 cm and 0.084 cm, respectively, for the 2 cm
and 3 cm orifices. The thickness of the splitter plate edge was 0.081 cm,
which means that the local thickness of the flowing region was of the same
order as the splitter plate thickness (i.e., quite thin). It is now clear
that the unusual data trends were due to passage of this thin layer of gas
across the interface. Such passage would be promoted for one or more of
four major reasons: instability, unsteadiness, misalignment or sample tube
placement, in addition to the minor sources of diffusion and mixing.
In the following discussion, keep in mind that the jet thickness near
the edge of the splitter plate is only about one millimeter. A momentary
pressure pulse on the aerosol side of the splitter plate could cause the
thin jet to cross over to the other side of the plate, carrying entrained
particles with it. This gives particles a second mechanism (the first
being inertia) by which to cross from the aerosol side to the clean side.
Large particles with sufficient inertia resist the turning drag force as
the flow turns normal to the original direction, and pass across the Inter-
face to be collected on the clean side. The smallest particles will have
insufficient inertia to resist the drag force and will go wherever the f^q^
they are entrained in goes. If the flow crosses the interface, the small
particles go with it. Now consider particles of an intermediate size.
Many of these particles will not have sufficient inertia to cross the
interface by their own momentum. They will then turn and flow along the
thin layer next to the interface. If the jet now oscillates back and forth
across the splitter plate, some of these intermediate particles will have
enough inertia to continue in a nearly straight line and stay on the
aerosol side, while others, particularly the smaller ones, will stay with
the flow and cross over the interface. In this kind of situation, we
expect relatively high collection efficiencies for large particles, a
collection efficiency for the smallest particles which would be dependent
on the amount of gas transfer across the splitter plate, and a minimum
collection efficiency at some in-between size. For a calibration over a
wide range of particle sizes, then, we would expect an efficiency curve
as depicted in Figure 26, which is a refinement of Figure 6. W£ would
54
-------
100
f
COLLECTION EFFICIENCY, %
LARGE PARTICLE PATEAU
(DUE TO ORIFICE SIZE)
SMALL PARTICLE PLATEAU
(DUE TO GAS TRANSFER
ACROSS INTERFACE)
MINIMUM EFFICIENCY POINT
(DUE TO BALANCE BETWEEN
DRAG AND INERTIAL FORCES)
PARTICLE SIZE
Figure 26. Theorized efficiency curve for continuous flow,
Impactor, based on accumulated test data
55
-------
expect plateaus above zero and below 100% efficiency for the smallest and
largest particles, respectively, and a minimum efficiency somewhere in
between. Sections of this basic curve apply to the obtained efficiency
curves for all configurations tested. Configurations A, B, and C would
fit around the middle of the proposed curve. The aerosol flow biased
configurations, D and E, are shifted to the left of A, B, and C by the
bias, while the clean flow biased configuration, F, is shifted to the
right of the basic configuration.
The conclusion was eventually reached, then, that the calibration
curves for configurations A, B, and C would have the basic shape of
the curve in Figure 26 if tested over a wide enough range of particle
sizes.
The above is a theory of what happens if there is significant gas
transfer across the interface. Now we need to consider what could cause
such transfer. A misalignment of the nozzles would be expected to produce
a flow pattern as shown in Figure 2a, resulting in the undesired transfer
of gas. In our case, great care was taken in both initial design and
during setup to avoid alignment problems. After each installation, the
nozzle alignment was verified by inserting a close fitting steel pin
through both nozzles to insure axial co-linearity. This leaves us with
three identified remaining mechanisms: instability, unsteadiness, and
the effect of the sampling tube. Although the problem was not pin-
pointed with certainty until later in the program, we felt that there was
a strong argument for considering the sampling tube to be the culprit.
A possible case may be made for unsteadiness since the rotameter floats
downstream of the splitter plate were seen to vibrate constantly, as such
floats are notorious in doing. In a subsonic stream such as occurred
during the test, such fluctuations would feed back upstream and affect
the flow in the vicinity of the interface. The feeling at that time was
that unsteadiness observed during testing was not of sufficient magnitude
to cause the gas transfer inferred from the separation data.
Oscillation of the interface was considered at the beginning of the
task to be the most likely instability to occur, which is why nine dif-
ferent splitter plates were designed (Figure 10). We had hoped to
56
-------
Investigate the effect of various splitter plate edge configurations on
stability. Unfortunately, other problems took precedence and we were not
able to perform this investigation.
The shape of the jet and the location of and flowrate through the
sampling tube now combine, unfortunately in retrospect, to give the correct
explanation of the initially obtained particle separation data. Pertinent
facts were:
The sampling tube was located well away from the thin sheet
flowing radially outward along the splitter plate
The flowrate through the sampling tube was 10% of the jet
flowrate
Separation data were generated by drawing samples from each
side of the splitter plate and comparing readings.
The reason for choosing the sampling location used (Figure 12) was to be
near as possible to the interface in order to avoid sampling after sig-
nificant wall losses may have occurred. The laboratory unit was designed
primarily to promote, and to provide a capability to investigate, stability.
Particle sampling considerations by necessity were then reduced in impor-
tance. Test data now strongly show that the sampling tubes should have
been located further out radially and closer to the splitter plate.
Current feelings on the effect of the sampling tube are shown in
Figure 27. Figure 25 shows the limiting streamline for operations without
sample withdrawal to the particle counter. Figures 27a and 27b show the
effect was to produce a pressure gradient normal to the splitter plate,
which produced a local detachment of the jet on the side being sampled.
This locally caused gas from the other side to be drawn across the inter-
face and into the probe. Particle trajectories shown in Figure 27 indicate
only the general curve shape shown in Figure 26 was obtained:
Large particle when sampling was done on the aerosol side,
large particles which had crossed the interface would likely
have enough inertia to avoid being turned around and drawn
back over, so sampling on the aerosol side would not produce
an error. However, the clean side gas would be diluted with
57
-------
LARGE
PARTICLE
TRAJECTORY
**%*>
^^ f r *~ **-
CLEAN
FLOW
IN
X^
LARGE
PARTICLE
TRAJECTORY
SAMPLING
TUBE
SMALL
PARTICLE
TRAJECTORY
"v
<^:
AEROSOL
LADEN
FLOW
IN
SMALL
PARTICLE
TRAJECTORY
Figure 27a. Effect of sampling tube when sampling aerosol side.
SAMPLING
TUBE
SHALL
PARTICLE
TRAJECTORY
LARGE
PARTICLE
TRAJECTORY
LARGE
PARTICLE^
TRAJECTORY
SMALL
PARTICLE
TRAJECTORY
AEROSOL
LADEN
FLOW
IN
Figure 27b. Effect of sampling tube when sampling clean side,
58
-------
gas from the aerosol side which is now relatively free of large
particles, and the dilution would show up as a low efficiency
for large particles
§ Medium size particles -near the 50% cutoff point, the number
density would be roughly the same on either side of the splitter
plate, so the sampling error would be small. Passage of gas
across the interface would, however, artificially lessen the
sharpness of the separation curve.
Small particles small particles will stay entrained in the gas.
During sampling on the aerosol side, dilution with the clean gas
(relatively free of small particles) would artificially lower the
small particle concentration. On the other side, diverting the aero-
sol gas across the interface would raise the small particle
population. The combination of the two would show significant
collection of small particles.
Nothing in Luna's work or in our observations of steadiness of the flows
{see traces 1n Appendix A) suggested the presence of Instabilities or'
unsteadiness capable of producing the high collection efficiencies for
small particles which were observed, especially for configurations B and
C. The sampling tube theory Illustrated in Figure 27 seems highly plaus-
ible, both qualitatively and quantitatively. Present feeling is that
proper sample withdrawal technique would result in much more normal
efficiency curves.
The original task was sufficiently low on funds that additional
laboratory testing to eliminate the anomaly was not possible immediately.
Initial testing was completed in early 1977, and a final report draft
was issued in April of 1977. Fortunately, a funding increase was obtained
for more testing in 1978. Thts additional test work centered around the
sample extraction problem, and was completed in October, 1978. The testing
confirmed that the above postulated scenario was indeed correct, and
greatly improved results were obtained, as described in the next section.
59
-------
SECTION 7
SUBTASK 4 - ADDITIONAL PARTICLE SEPARATION TESTING
The specific objective of this subtask was to improve the shape of
the collection efficiency curve. This was done successfully by changing
the manner in which gas was withdrawn from the impactor, which minimized
transfer of gas across the interface. The testing also made it clear
that we have gone as far as we can with the development model impactor,
and that further work can best be performed with improved hardware. The
additional testing added substantially to our knowledge about this type
of particle separation technique, and has made us confident that good
field worthy systems can be produced at reasonable cost,
7.1 HARDWARE MODIFICATIONS
An inadequate means of withdrawing samples for the particle counter
was cited as the primary reason for the relatively poor particle efficiency
curves obtained during initial testing. The purpose of the hardware modi-
fications performed during this subtask was both to improve performance
and to withdraw gas in a manner compatible with preliminary designs for
second generation hardware. Since the previously used technique involved
creation of a non-symmetric flow ffeld in the vicinity of the splitter
plate orifice, the most important single consideration was to obtain a
symmetric flow field.
The original withdrawal technique Is shown in Figures lie and 12.
The revised system tested during this subtask is shown schematically in
Figure 28. Seven of the eight manifold holes (Figure 7) on each side of
the splitter plate were taped up to minimize gas recirculation. The
remaining hole on each side was used as an entry hole for the new half
diameter exhaust tubes, which were positioned opposite each other and
pointing toward the splitter plate as shown in Figure 28. The tube to
spacing was selected to provide a steadily increasing gas velocity from
the plenum to the exhaust tube. These tubes were then connected to flexlbi
plastic tubing by means of quick - connects, as shown in Figures 29 and
30, which then led to the particle counter and to the flowmeters used to
60
-------
OUT
CLEAN
FLOW IN
1/2" O.D. OUTLET
TUBE
AEROSOL LADEN
FLOW IN
Figure 28. Schematic of revised exhaust flow setup
-------
Figure 29. Impactor outlet detail - Configuration G
-------
PRESSURE
GAGE
EXHAUST LINES
CLEAN AIR INLET
LINES SWITCHED
HERE TO SAMPLE
ALTERNATE SIDES
1 PARTICLE
GENERATOR
OUTLET HOSE
PARTICLE COUNTER
Figure 30. Impactor test setup - Configuration G
63
-------
adjust the outlet flow. Recall that the relatively low sample flow rate
for the Royco 225 counter (10% of the nominal flow through each nozzle)
necessitated the splitting of the exhaust flow on the side being sampled
The configuration shown in Figures 28-30 represents several improve-
ments over the initial configuration. The most important one, of course
is that in the revised configuration, a symmetric exhaust gas pattern was
achieved within the impactor, so that the flow pattern in the vicinity of
the splitter plate orifice would no longer be disturbed. The flow to the
particle counter was split off from the main flow by means of the isokinefi
sampling nozzle shown in Figure 31. In the previous work, the exhaust flow
had to be adjusted every time the sample location was changed form one s1<|
of the plate to the other. In the revised configuration, use of the quick-
connects made this unnecessary. This resulted in less data scatter than
was obtained in the initial work.
The modifications made, while not particularly costly or complex,
i very dramatic effect on impactor performanc*
remained as it was during the previous test work.
had a very dramatic effect on impactor performance. The rest of the system
7.2 TEST DESCRIPTION
During the year and a half between completion of subtask 3 and the
start Of the test work for this subtask, the impactor itself was not idle
it made several trips between the east and west coasts. Somewhere along
the way, the contoured nozzles and two of the three splitter plates were
lost, stolen, or strayed, leaving us with the conical nozzles and the
splitter plate with the 2 cm diameter orifice. We decided to run a full
test with this configuration before making any decision to fabricate
replacement hardware. As it turned out, additional fabrication was not
required.
Since the original test setup had been completely torn down, it was
necessary to re-acquire support equipment (valves, flowmeters, pump,
particle counter, etc.) and floor space to set up for the additional
testing. The only particular problem encountered was in getting the
Vibrating Orifice aerosol generator to function properly. The difficulty
was traced to scratches below the orifice place, which kept the 0-ring
64
-------
150 KINETIC SAMPLING TUBE
FROM
IMPACTOR
JT
o»
en
TO ROYCO
PARTICLE
COUNTER
47 sec/sec
TO
PUMP
425 sec/sec
Figure 31. Sketch of particle counter sampling tube 1n
impactor exhaust line
-------
from sealing properly, and other gremlins in the system. Leakage was
solved by polishing the sealing surfaces, and the gremlins were banished
by benevolent gnomes supplied by the factory. Once the setup was complete
with all equipment working properly, the test work was completed in a
short time.
In the context of the previous work, the test configuration may be
referred to as configuration G, and had the following conditions:
Configuration : G
Orifice Diameter: 2 cm
Nozzle Shape : Conical
Flowrates : 470 cm3/s, all flow streams
Nozzle/nozzle spacing: 1.5 cm
A collection efficiency curve was produced with the use of two
aerosol sources - the TSI Vibrating Orifice Generator for the range
4-15.5 y, and room air (combined with the sizing capability of the
counter) in the range .5 - 5 u. As was the case with the previous work
the aerosol generator used OOP (in alcohol) for the aerosol. Basic
procedure was as follows (see Section 6.2.2.4 for detailed procedure):
Start the clean side flow and adjust for proper flow rate
Start the aerosol generator and particle counter
After the generator has stabilized (^ 10 minutes),
adjust clean side inlet flow and outlet flows to
achieve proper balance
Take particle count data on each side of the splitter plate
t Change generator operating conditions as required for next
particle size
t For room air testing, turn off aerosol generator and
disconnect generator outlet hose. Balance the flows and
take particle counter readings
66
-------
7.3 TEST RESULTS
7.3.1 Aerosol Generator Testing
The collection efficiency curve from the testing with the aerosol
generator is shown in Figure 32. Data scatter is considerably less than
that for the similar configurations A, B, and C (Figures 17-19), and the
separation is much improved. We were unable to go higher than 15.5
microns in particle size due to internal losses - not enough particles
made it all the way through to the counter to obtain a large enough
sample. We were limited at the low end to 4 y because the smallest orifice
for the generator adamantly refused to function properly. The high end
limitation was a fundamental problem of the hardware configuration itself -
the large plenums on either side of the splitter plate, which were designed
specifically to promote jet stability and allow access for diagnostic
probes, also provided a large volume which permitted dropout of the large
particles. A more appropriate physical configuration will be required to
investigate separation efficiency for the largest particles.
7.3.2 Room Air Testing
It was important to extend the separation curve to particle sizes
below 4 y in order to get information about gas transfer across the orifice.
After investigating several ways of producing small particles, we decided
the simplest, but still acceptable in terms of accuracy, would be to use
the polydisperse distribution in ordinary unflltered room air. The
separation efficiency as a function of size could then be determined by
means of the diagnostic sizing capability of the Royco particle counter,
which, in the window mode, displays the following size ranges:
Channel Size Range, y
1 .5 - .7
2 .7 - 1.4
3 1.4-3
4 3-5
5 > 5
67
-------
COLLECTION
EFFICIENCY, %
100
90
80
70
60
50
40
30
20
10
0
I I I i
. . I
i i i
i i i
.2
.4
.7
4 7 10
PARTICLE DIAMETER, MICRONS
20
40
Figure 32, Particle collection efffcfency as a function of particle
size for size range 4-16 nicrons, Configuration 6
-------
Thus the first four channels allowed us to examine four bands of particle
sizes. Results are presented in Figure 33, where the horizontal lines
represent the net collection efficiency for particles in the designated
size range. The wider spread for channels 3 and, especially, 4, is
almost certainly statistical in nature rather than being representative
of the actual collection. The "clean side" counts for channel 4 were
typically in the range of 10 - 30, compared to 5000 and 1000 for channels
1 and 2, respectively. For each channel shown, a simple average of the
individual values may be considered as representative.
7.3.3 Comparison With Other Data
The full set of calibration data for configuration 6 is shown in
Figure 34. Note that there is good agreement in the overlap region for
the room air and aerosol generator data. Recall from section 6.4 that
the calculated 50% cut point for a standard impactor under our operating
conditions is 7.31 y. The 50% cut point for configuration G, from
Figure 34, was 9.9 y. The difference is due to the relatively small
orifice diameter (twice the nozzle diameter), as previously explained.
i
Configuration C, which had the same orifice diameter, had a 50% cut point
of 9.3 M, which 1s reasonably close to that for configuration G.
A comparison of configurations C, G, and the first stage of the
Anderson viable sampler is shown in Figure 35. The dramatic improvement
in performance between configurations C and G is quite evident, and, since
only the gas exhaust configuration was changed, the performance increase
clearly confirms the correctness of the hypothesis about the initially
poor performance of the continuous impactor. The configuration G data
also shows the presence of another as yet unresolved problem. The confi-
guration G curve is sharper than that for the Anderson impactor down to an
efficiency of about 20%, and it levels out at about 10%. This is strongly
indicative of a net transfer of about 10% of the "aerosol side" gas stream
to the "clean side" of the splitter plate. This problem was previously
masked by the large gas transfer tn the vicinity of the sampling tube.
The gas transfer in configuration G is a direct result of unsteadiness of
flow through the rotameters used as metering devices during the test. A
continuous oscillation of the rotameter floats was visually observable.
69
-------
COLLECTION
EFFICIENCY, %
100
90
80
70
60
50
40
30
20
10
0
AEROSOL: POLYDISPERSE ORDINARY ROOM AIR AEROSOL
DETECTION: ROYCO 225 PARTICLE COUNTER OPERATING IN "WINDOW" MODE
ROYCO COUNTER CHANNEL
2 I 3 I
BAND REPRESENTS NET
COLLECTION EFFICIENCY
IN DESIGNATED SIZE RANGE
I
.2
.4
Figure 33.
.7
4 7 10
PARTICLE DIAMETER, MICRONS
20
40
Particle collection efficiency as a function of particle
size for size range .5-5 mfcrons, Configuration 6
-------
COLLECTION
EFFICIENCY, %
100
90
80
70
60
50
40
30
20
10
0
I i i i i
I I 1
.2
.4
Figure 34.
.7
4 7 10
PARTICLE DIAMETER, MICRONS
20
40
Complete fitted separation efficiency curve for Configuration G,
showing data points in overlap region
-------
ro
COLLECTION
EFFICIENCY, %
100
90
80
70
60
50
40
30
20
10
0
1.5
CONFIGURATION
G
I I
I
ANDERSON
VIABLE SAMPLER,
FIRST STAGE
(LIQUID PARTICLES)
REFERENCE 4
7 8 9 10
15
20
30
40
PARTICLE DIAMETER, MICRONS
Figure 35. Comparison of separation curves for Configurations C, G,
and the Anderson Impactor first stage
-------
The rotameters were selected for use primarily for purposes of convenience.
It is now clear that future work will require better flow control elements,
and this is reflected in the recommended design in Section VIII.
7.3.4 Summary
The specific purpose of this task was to correct a deficiency in .the
previous test setup, and hopefully obtain better separation efficiency. The
deficiency was felt to be the way in which aerosol samples were removed from
the impactor and delivered to the particle counter. Test work during this
subtask showed that the diagnosis was correct. When we switched to a fully
symmetrical exhaust technique, a very dramatic improvement in performance
occurred. The only substantial remaining problem is a higher than desired
small particle carryover, and we presently feel that this is due to instabi-
lities in the feed and exhaust systems which have been due primarily to the
use of rotameters and their control values as measurement and control devices.
Sonic orifices and pressure regulators will be preferred for future work, as
described in the following section.
This subtask accomplished several important things. First, it constituted
a reasonable and successful conclusion to proof of principle testing. Keep in
mind that the laboratory model impactor design was primarily intended for
investigation of gas dynamic phenomena and promotion of a stable jet/jet
interface - particle separation characteristics were quite secondary. In that
context, then, the achieved separation efficiency can be considered to be
quite good. Further work should focus on redesign of the impactor for field
applications, including minimization of small particle carryover and internal
losses, and fabrication and test of a staged version of the device. The
laboratory model device tested in this task has served its purpose well, but
will not be appropriate for future work.
73
-------
SECTION 8
SUBTASK 5 - INSTRUMENT UTILIZATION
The laboratory test unit was designed specifically for basic inves-
tigatory testing. There was no intention for the unit to have any
resemblance to a field usable device, other than the basic operating
principle. After the concept was successfully demonstrated, as discussed
in the previous section, a brief study was made to investigate possible
packaging and monitoring techniques applicable to a field usable unit.
There are several distinct configurations of the continuous flow impactor
which could be of value in source testing. The discussion below will
concentrate on one of them -a staged impactor with five size fractionated
output streams, with an optical monitor on each stream. Other configura-
tions are then considered as modifications to this basic one. In each case,
preliminary design is for in situ operation.
8.1 STAGED CONTINUOUS FLOW IMPACTOR
8.1.1 Flow Related Components
Fluid mechanically, the proposed staged impactor will operate 1n the
same manner as the laboratory unit - outlet flows and the clean side inlet
flows will be actively metered, resulting in passive metering of the
aerosol laden flow. The flow schematic of a four stage device having a
total of five size fractionated output streams is shown in Figure 36, and
includes the pressure hierarchy for the system. The aerosol laden flow
(flue gas) enters the "aerosol side" of each stage, where it is met by
an equal clean flow. Sonic orifices to meter the inlet and outlet flows
are the simplest and surest way to control the flows. A similar system
was used with great success to produce highly stable flows in the
wind tunnel built by TRW to calibrate the Viking Meteorology Instrument
System wind sensors. As long as the pump provides a sufficiently low
pressure (below 260 Torr), any variation in pumping characteristics will
not be noticed by the impactor. It will be critical to provide a stable,
accurate pressure regulator for the clean side inlet flows, since it is
the combination of the orifice size, upstream pressure and temperature
74
-------
SAMPLING
NOZZLE
IMPACTOR
MODULE
FLUE
GAS IN
760
760
NUMBERS ARE LOCAL STATIC
PRESSURE IN TORR
NOMINAL FLOWRATE INTO EACH
SIDE OF EACH STAGE IS
470 SCC/S (1 SCFM)
>1700
SONIC
ORIFICE
759.8
(75978
755
1755
710
(710
520.
<260
SHOP OR
INSTRUMENT
AIR
PRESSURE
REGULATOR
PUMP
AS REQUIRED BY QUALITY
OF AVAILABLE AIR
TO
ATMOSPHERE
Figure 36. Flow system schematic and pressure hierarchy for
staged continuous flow impactor
75
-------
which provides the correct flowrate of clean air to each state. Monitoring
of the output streams will be performed between the clean side outlet and
respective sonic orifices, and between the final aerosol side outlet and
its orifice. The optical monitoring technique, discussed in the following
paragraph, will have no effect on the,pressure or flow distribution.
A sketch of the entire "installed system is shown in Figure 37. Key
features are as follows:
The heart of the system (impactor modules, metering orifices,
optics) is contained in the first 75 cm (29 inches) of the
sampling probe. It will be essential that this section be
maintained at stream temperature. The obvious way to accom-
plish this is by maintaining a minimum 75 cm insertion depth.
The more complex method for lesser insertions would be use of
a servo-controlled heating mantle.
The pressure regulator need not be in situ, nor would it require
temperature control.
Use of a pump such as a Gast Model 2565 would eliminate a need
for any conditioning of the stream ahead of the pump.
A laser is an integral part of the optical monitoring system.
It would be mounted ex situ with appropriate safety measures.
The electronic readout for the optical subsystem may be located
wherever desired.
Since the flowrate through the device will be held constant, a
pi tot probe (not shown) will be required to perform a prelim-
inary velocity measurement at the selected sampling point (or
points) in order to select an appropriate sampling nozzle I.D.
A pitot probe could be made an integral part of the assembly.
Preliminary design efforts have led to the conclusion that the
probe O.D. will be compatible with insertion through a four inch
Schedule 40 pipe nipple (10.3 cm I.D.).
A sketch of an impactor module is shown in Figure 38. The major difference
between this unit and the laboratory model is of course the smaller overall
size and smaller plenums. The arrangement of the exhaust tubing came about
76
-------
r
\
H
V
IN
. OPTICS MODULE
_/
-- f ii roi rn roi rn roi rri roiro-1
I *J 1 ! ! 1 S ! 2 ! ' 2 * ! 3 ! ! 3 I ! 4 j * 4 'i 5 '
f ^^.i J U_J L±_l UIJ U_J l__ J 1 1 C-Jt-_J
} \
IMPACTOR
MODULE
S
1
7
/
7
>
/
y
j IVM.L.
A
j 4- INCH, SCHEDULE 40
/ PIPE NIPPLE
-J«H
1 5 mW LASER
\ |-J
^~7\ PUMP
AND READOUT N U
ELECTRONICS \ \
J f
CLEAN OUT
AIR
IN
Figure 37. Staged continuous flow impactor installation
-------
SPLITTER
PLATE
NOZZLE/PLENUM PIECE
HINGE
00
SECTION A-A
PIN
CLOSURE
RETAINING
RING
INLET
TUBE
AEROSOL
LADEN
GAS IN
5^0^
^|^a
CLEAN GAS IN
AEROSOL SIDE
EXHAUST TUBE
OUT
*r/y/vffl
B
SECTION B-B
CLEAN SIDE
EXHAUST TUBE
AEROSOL SIDE
EXHAUST TUBE
CLEAN SIDE
EXHAUST TUBE
F-I ,-.,,-., -30
-------
primarily f«wi packaging requirements to stay within the four inch envelope.
The non-symmetry is presently considered to be a necessary evil imposed by
the packaging constraints. An installation sketch of the first two modules
ts shown in Figure 39. In this sketch, a slightly different treatment of the
clean gas inlet is shown. The purpose of this alternate approach is to
provide maximum space for opttcs components between the impactor stages.
The aerosol side interstage tube is kept as short and straight as possible
to mtnifliize interstage losses. The first stage needs to be as large in
diameter as possible due to the relatively low jet velocity. Other stages
can then be made smaller to allow for passage of tubing, light, etc.,
without adversely affecting performance. Other notable features are:
Components requiring critical alignment (nozzle and splitter
plate) are all axysytnmetrtc. Alignment is assured through close,
concentric fits easily achieved by machining on a lathe.
t Exhaust is accomplished very near to the splitter plate surface
on both sides. Thts 1s an important modification to the labora-
tory design to Insure the absence of pressure gradients normal
to the splttter plate which could cause disruption of the jet
interface and resultant gas transfer across the interface.
The design also minimizes recirculation regions within the module,
which helps to minimize interstage particle losses.
t The second, third, and fourth modules may be made identical except
for the nozzle and splttter plate orifice diameter, which helps to
keep production costs down.
Inlet and outlet sonic orifices will be located as close as
possible to the respective modules to insure that the proper
orifice temperature is maintained. Since the critical portion
of the probe will be fully inserted into the flow, proper tem-
perature will be maintained automatically by the flue gas inself
with no additional controls and no diagnostics required.
The direct coupling between the sampling nozzle and the first
stage minimizes aerosol losses.
79
-------
FIRST STAGE
FIRST STAGE OPTICS
AREA (OPTICS NOT SHOWN)
SECOND STAGE
PROBE SLEEVE
Figure 39. Staged continuous flow impactor assembly - first two stages
-------
8.1.2 Optical Monitor Components
During laboratory testing, a Royco particle counter was used to
monitor the impactor output streams. The particle counter accelerates
the aerosol laden flow through a nozzle to produce a high speed, small
diameter jet. The jet then passes through an optical viewing volume
which has a slightly larger diameter, and the particles are detected using
a light scattering technique. Particle size is obtained through pulse
height analysis of the scattered light pulse.
Optical monitoring is the most obvious technique to consider for the
staged continuous flow impactor. Use of optical techniques for a commercial
device such as that shown in Figure 37 implies the following constraints:
Optical components must be inexpensive, reliable, and not require
adjustments in the field.
With the exception of the light source selected, the components
must be capable of withstanding stream temperatures.
Each optical module must be able to be packaged in a small space.
The approach eventually selected from among several considered is single
particle imaging with modified pulse height analysis. Detailed background,
including supporting calculations, is presented in Appendix B. A sketch
of an optical subassembly is shown 1n Figure 40. The basic technique used
1s near-forward scattering of the incident light beam. Due to the con-
straints mentioned above, the proposed technique differs in several
particulars from standard optical counters. The main features of the system
shown in Figure 40 are as follows:
The optical path lengths involved are compatible with the required
envelope, although some folding will be needed.
Components are available to withstand the normal range of stream
temperature.
For each stage, the output electronics will only count particles
above a specified size by requiring a minimum detector output per
pulse.
81
-------
BEAN FROM
HE-NE CH LASER
00
ro
TRANSMITTED
BEAM TO
NEXT STAGE
NOT TO SCALE
EXHAUST FLOW
FROM IMPACTOR
LIGHT
TRAP 2
PARTICLE
SCATTERING
VOLUME
PHOTON
ECTOR
TO PARTICLE
COUNTING
ELECTRONICS
CM LASER:
LENS A:
LENS
LENS
LENS
B:
Br
B2:
DETECTOR:
0.5 MW HE-NE (0.63y)
(f/D)A =4 DA = 1/4-INCH
(f/D)
(f/D)
BI
B2
-
"
4
2
°B, "
°B2"
1/2- INCH
1/2-INCH
PIN PHOTODIODE.
Figure 40^ Scneoatic diagram of optical concept.
-------
The viewing volume for each stage will be made as large as pos-
sible while avoiding the presence of more than one particle above
the minimum size in the viewing volume.
Minimum time response requirements dictate.that the viewing volume
cross-sectional area will be smaller than the stream cross-sectional
area.
Calculations in Appendix B show a good probability of being able
to resolve (count) particles down to 0.2y diameter. This will
need to be confirmed experimentally.
The major difference between the proposed optical technique and normal
particle counters is that the impactor optics will be working on size
fractionated streams. This allows important simplifications. In a
standard particle counter, an observed particle is assigned to a size
range by producing a signal large enough to trigger one channel but too
small to trigger the next larger one. In the continuous impactor, the
upper cutoff is accomplished mechanically in the separation process, so
the optics need only establish the lower threshold. The monitoring sub-
system can also ignore undesired small particles in each stage, which
helps to optimize system accuracy.
Limitations on input power (size, cost, and safety aspects of the
laser) and size of the collector optics (space constraints) limit the
minimum response time for detection. This means that the optics will
only be looking at a portion of each output stream, thus requiring uniform
particle distribution within the stream to insure accurate measurements.
It is important to maintain the flowrate as high as possible to avoid
particle losses to the walls through settling and diffusion so accuracy
would suffer if the flowrate were cut down in order to reduce the area.
As each individual flow leaves its respective impactor stage ahead of
the optics, it will undergo approximately a four to one area contraction
to provide a uniform flow past the optics. The fact that size fractiona-
tion will have occurred prior to the contraction will help to insure a
uniform particle distribution due to the relatively narrow size range
and substantial mixing which will occur at the stage outlet.
83
-------
The major effort to this point in the optics area has been to analy-
tically determine feasibility (Appendix B) and select appropriate hardware
components. The next step will be to demonstrate adequacy of the approach
and package the components in an acceptable manner. At this point, we feel
confident that an optical system to monitor five separate streams in an
in situ probe can be manufactured at reasonable cost, and have good accu-
racy and reliability.
8.1.3 Application
The primary use of the device ts expected to be determination of total
particulate mass emission rate and size distribution. The most likely
specific applications will be control device efficiency testing (inlet
and outlet measurements) and continuous regulatory monitoring. These two
applications would likely involve instruments with different cutoff ranges
nominal ranges are shown in Table 3. The control device evaluation instru-
ment will concentrate primarily on the smaller sizes to provide maximum
information on fine particle emissions, while the regulatory instrument
will cover a broader range to both provide information on fine particles
and cover the higher size ranges to more accurately determine the total
particulate mass emission rate. The use of a modular design will maximize
this type of flexibility without significantly impacting cost. Basic
readout will be the same for either device -counts per channel over a
prescribed sampling period.
The primary advantage of this device over a standard impactor for
control device evaluation is elimination of the normally large differen-
tial between Inlet and outlet sampling times. The continuous Impactor
is not concerned about overloading a stage, which is the normal Inlet
problem, nor will it need to sample for long periods of time at an outlet
due to the low outlet grain loading.
8.2 ADDITIONAL CONFIGURATIONS
A number of applications different from the mass monitor described
above have been postulated. Two will be described briefly here -a staqed
Impactor to collect samples in an inert environment, and a single stage
Impactor to selectively remove large particles from a stream which will
then go into another device, such as a standard Impactor.
84
-------
Table 3. NOMINAL SIZE RANGES FOR STAGED CONTINUOUS
FLOW IMPACTOR APPLICATIONS
00
OUTPUT
STREAM
1
2
3
4
5
NOMINAL PARTICLE SIZE RANGE, MICRONS
FINE PARTICLE MONITOR
>3.2
1,6-3.2
0.8-1.6
0.4-0.8
0.2-0.4
TOTAL PARTICULATE MASS MONITOR
>10
5-10
3-5
1-3
2-1
-------
8.2.1 Inert Atmosphere Collection
It is often desirable in process stream evaluation work to collect
a participate sample and immediately quench any chemical reactions which
may be occurring, so that subsequent analysis will reveal the particle
composition at the sampling point. In standard inertial separation
devices, the collected particles are continually exposed to the flue gas
during the sampling period. The continuous flow impactor offers a unique
capability to rapidly transfer particles from the flue gas to an inert gas
(nitrogen is the easiest). This will help to radially quench any reactions
Particle collection can then take place on a filter element within the probe
A prototype holder is shown in Figure 41. The filter element used would be
selected for compatibility with the stream temperature and subsequent analy-
tical techniques. The filter holder would be used in place of the optics
modules shown in Figure 37, which would make the device essentially a high
capacity staged impactor in which the particles are collected in an Inert
(quenching) atmosphere.
8.2.2 Single Stage Preconditioner
Several standard types of Impactors and trains employ a cyclone at
the inlet to remove particles to avoid overloading the subsequent device
The characteristics of the continuous flow Impactor make it an ideal devl
to perform this same function without the cyclone's cup capacity and orlen
tatlon limitations. For this application, only a single stage would be
needed, which would cut down the clean air and pumping requirements. The
cutoff would be selected in accordance with the characteristics of the
downstream device.
8.3 COST ESTIMATES
Rough estimates of hardware costs based on a production run of ten
units are presented 1n Table 4 for the three configurations considered'
a four stage Impactor with five sets of optical monitors, a four stage
Impactor with five filters, and a single stage preconditioning Impactor
It is emphasized that these are very rough estimates. The hardware
design was carried far enough to show that an 1n situ device 1s feasible
and that the probe diameter will be allow for Insertion through a four
86
-------
FROM
IMPACTOR
FILTER CANISTER
RISER
END PLATE (SOLID)
WITH GASKET
SONIC
ORIFICE
FILTER ELEMENT
SUPPORT TUBE,
SLOTTED OR
PERFORATED TO
ALLOW PASSAGE
OF GAS
I
TO
PUMP
Filter element (M1ll1pore filter, glass filter,
etc.) forms a cylindrical surface on the Inside
of the filter element support tube. Filter
element 1s recovered by unscrewing canister 11d
and removing end plate. Sonic orifice must be
located ahead of filter to avoid flowrate change
as filter loads up.
Figure 41. Filter holder subassembly
87
-------
Table 4. CONTINUOUS FLOW IMPACTOR PRODUCTION COST ESTIMATES
Configuration
4 Stage Impactor With
5 Channels of Optical
Monitoring
Subsystem
Impactor Modules
Optics
Processing & Readout
Electronics
Probe Body, Including
Alignment & Support
Hardware
Pump, Regulator, and
Air Cleaning Components
Sampling Nozzles (6)
Estimated
Cost
1600
2000
300
1000
800
150
Total: $ 5850
4 Stage Impactor With
5 Filters for Collec-
tion in an Inert Gas
Impactor Modules
Filter Assemblies
Probe Body
Pump, Regulator, and
Air Cleaning Components
Sampling Nozzles (6)
1600
400
700
800
150
Total: $ 3650
Single Stage Pre-
conditioning Impactor
Impactor Module
Pump, Regulator, and
Air Cleaning Components
Sampling Nozzles (6)
500
400
150
Total: $ 1Q50
Estimated cost is estimated production cost for a
production run of ten systems of the particular
configuration.
-------
inch pipe nipple. Optical design was primarily concerned with being sure
that the small particles (around .2 y) could be detected, and that there
were readily available components which would be inexpensive, able to
tolerate high temperatures, and be compatible with the selected probe
size. More accurate cost estimates cannot be made until actual produc-
tion drawings are made.
89
-------
SECTION 9
DISCUSSION OF RESULTS
All of the primary task objectives were accomplished. Test hardware
was fabricated after an initial literature search indicated feasibility of
the concept. Reasonable jet stability was demonstrated prior to particle
separation testing. Testing with the larger splitter plate orifice showed
good correlation between the calculated and measured 50% cut points, while
use of the smaller orifice shifted the BQ% cut point to larger particle
sizes. A reasonably good separation efficiency capability was demonstrated
during the final laboratory test phase, after some initial problems had
been diagnosed and overcome. Preliminary design work showed that it will
be reasonable to integrate multi-stage hardware into a probe capable of
being inserted through a standard four inch pipe nipple, with an optical
monitoring subsystem on each of five output streams. Important questions
to be answered durtng further development deal with instrument accuracy,
cost, reliability, and handling characteristics. Basic proof of principle,
including demonstration of a field acceptable hardware concept, has been
accomplished.
Luna's work was very helpful in providing a basic understanding of
the stability of impinging gas jets. Because the projected use for the
continuous flow impactor, process stream sampling, our hardware configura-
tion differed significantly from Luna's. It was one of these differences
lack of active control over the individual inlet flowrates, which was the
source of the major flow instability observed early in the test program.
This instability was the single largest surprise received during testing,
although the probability of its occurrence is obvious in retrospect.
The first phase of particle separation testing resulted in relatively
poor sharpness of cutoff, including carryover of a very substantial amount
of the small particle population. The problem was diagnosed after completion
of the intial test effort and corrected as part of the additional test work
The separation achieved in the final configuration was excellent except for
90
-------
some small particle carryover resulting from unsteadiness in the gas feed
and exhaust lines.
The final subtask was a brief study to insure that the instrument
could be packaged for in situ monitoring, and to investigate the use of
optical techniques to monitor the output streams. Based primarily on
laboratory test data, the current opinion is that packaging impactor
modules and optics in a probe to fit a standard four inch pipe nipple will
not present a problem, nor will it have a notably adverse affect on
accuracy. Loss of accuracy during an extraction process would likely be
greater than that which will occur due to compromises required to obtain
an in situ instrument.
The next step in the instrument development process should be
fabrication, laboratory test, and field test of a multistage version
of the Impactor. For this effort, ft would be most reasonable to use
particle collection in filters as the operating mode. This testing should
be followed by a demonstration of the optical subsystem, which could be
done mainly with available in-house components. The simplicity of the
continuous flow impactor components and assembly, its coordination of
many of the attractive features of traditional impactors and optical
particle counters while avoiding their primary drawbacks, and its
potential usefulness as a diagnostic and monitoring tool all strongly
recommend that the effort begun during this task should be continued.
91
-------
SECTION 10
REFERENCES
1. Luna, R. E., "A Study of Impinging Axisymmetric Jets and Their
Application to Size Classification of Small Particles,"
Ph.D. dissertation, Princeton University, 1965, University
Microfilms, Inc., Ann Arbor, Michigan
2. Witze, P.O. and Dwyer, H. A., "The Turbulent Radial Jet," Journal
of Fluid Mechanics, Vol. 75, Part 3, June 1976
3. Marple, V. A., and Liu, B. Y. H., "Characteristics of Laminar
Jet Impactors," Journal of Environmental Science and Technology
8: 648-654, July 1974
4. Marple, V. A., and Willeke, K., "Inertial Impactor: Theory,
Design and Use," Fine Particles: Aerosol Generation, Measure-
ment. Sampling, and Analysis. Academic Press, Inc., 1976
5. Smith, R. H., and Wang, C. T., "Contracting Cones Giving Uniform
Throat Speeds," Journal of the Aeronautical Sciences, pp 356-
360, October, 1944
6. Ranz, W. E., and Wong, J. B., "Impaction of Dust and Smoke
Particles," I & E Chem. 44: 1371, 1952
7. Davies, C. N. Ed., Aerosol Science. Academic Press, New York,
1966
92
-------
SECTION 11
GLOSSARY
SYMBOL
C
D
DP
f(e)
f
'o
J(e)
m
N
Nscatt
P
pscatt
t
V
e
0
USAGE
geometric cross-section of particle, cm2
Cunningham slip correction factor
lens diameter, cm
particle diameter, cm
scattering function, sr"
focal length of lens, cm
laser beam illumination Intensity, W/cm2
light Intensity scattered at angle 6, W/sr
refractive Index of particle
number of photons Incident on particle
number of photons scattered onto detector by particle
laser beam power, W
power of light scattered onto detector by particle, VI
impactor nozzle radius, cm
thickness of diffraction cylinder at test section center,
particle velocity, cm/sec
mean jet velocity, cm/sec
particle size parameter
energy per photon, ergs/photon
scattering angle, radians
wavelength of light Incident on particle, yM
93
-------
SYMBOL USAGE
p gas viscosity, g/cm-sec
p particle density, g/cm3
T transit time of particle through illuminated scattering volume, sec
n solid angle of scattered light subtended by light trap lens, sr
94
-------
APPENDIX A
STABILITY DATA
The purpose of this appendix is to present strip chart traces taken
during the stability test phase. The traces display three different
differential pressures:
1. Differential pressure across the splitter plate - this
indicates the difference in static pressure across the
splitter plate, and is an indication of how accurately
the pressure fields on either side of the interface
are matched. Ideally, this differential pressure will
be zero.
2. & 3. Right and left nozzle differential pressures - for each
nozzle, this will be the difference between the total
and the static pressure at the nozzle exit - the same
differential pressure as would be measured by a very
small pi tot-static probe located at the nozzle exit.
The nozzle flowrate is proportional to the square root
of the nozzle differential pressure. The right and
left nozzle differential pressures should be equal
for proper operation.
Segments 1 through 5 were obtained from the setup shown in Figure lib in
the text (the modified stability test configuration) while Segments 6
through 9 were obtained from the setup shown in Figure lla (the initial
stability test configuration). The traces were lined over with a felt
tip pen, as the original traces were too weak to reproduce properly.
flotation added to the traces shows the applicable pressure scales, the
time rate of motion of the chart paper, and identifies which parameter is
t>e1ng measured. Segments 1 through 5 form a continuous trace except for
the identified interuption between segments 4 and 5. Segments 6 through
g also form a continuous trace.
Segment 1 shows the differential pressure across the splitter plate.
-fv/o things are worthy of note. First, the effect of the drift of the
pltrogen regulator on this differential pressure is quite apparent. The
Affect was lessened during particle separation testing by using a group
of interconnected nitrogen bottles rather than the single one used when this
trace was taken. Second, the measured differential pressure (0 - 0.01
forr) was small compared with the measured nozzle differential pressures
(,16 - .17 Torr^ which is desirable.
95
-------
Segments 2 through 5 alternately show the right and left nozzles.
The observed variations were due to random unsteadiness in the flow. The
magnitude of the variations was about +. 3% of the measured differential
pressure, which means that the actual nozzle flowrate variations were about
+ 1.5%. This qualifies as stable operation.
Unstable operation is shown in segments 6 through 9. In this case, the
flowrate through the right nozzle was about a factor of 2.5 higher than the
flowrate through the left nozzle. For the trace shown, the differential
in flowrates stayed constant. During other runs using the initial con-
figuration, the right and left nozzle flowrates were observed to vary in
a random manner over a period of several minutes - first one would be
higher, then the other. The improvement in performance of the configuration
in Figure lib in the main text over that of the configuration In Figure 11a
is made dramatically clear through a comparison of Segments 2 through 5
(stable operation) with segments 6 through 9 (unstable ooeration).
96
-------
SWITCH TO
*o SEC.
spLirreR
ol .02.
.05-
-------
i
LEFT
CO
^o sec.
I
o
I
.100
i
,1*0
,1.00
-------
i
VO
sec.
I
.075-
I
.I5O
.7.00
-------
o
o
sec
I
o
.075"
1 I
.12.5-
I I
,175- .2.00
-------
PR.es 5
-------
o
IM
k I
I
.7
.8
r
-------
o
to>
3o sec
i
o
.3
PR6SJURE, TORR
.6
I
.7
.8
-------
LEFT
1
.3
n
.s
-------
o
01
i
c*
-------
APPENDIX B
OPTICAL MONITORING OF CONTINUOUS
FLOW IMPACTOR OUTPUT STREAMS
B. 1. INTRODUCTION
The purpose of this Appendix is to document analyses performed on
an aerosol detection device utilizing forward scattering of light from
a standard, low power laser. Table B-l summarizes the requirements and
performance of the system concept shown schematically in Figure B-l
(Figure 31 in text).
With a 5mW He-Ne cw laser, the system is designed to produce count
rates of individually resolved particles up to about 106 particles/sec
(0.2 ysec response time) corresponding to a flow velocity of 15 m/sec
aid particle concentrations of 7xl03 per cubic millimeter. Beam splitters
are used with a single laser source and 5 identical sets of optics to
monitor the particle flow in the center of 5 output streams in the
impactor system. The most sensitive monitoring channel would produce
readily detectable signals with particles down to 0.2-0.5y diameter,
using a photodiode detector.
All optical components are standard, commercial items. The cost of
the optical elements only (including the laser source, but without exter-
nal hardware or particle counting electronics) is estimated to be about
$1420 for 5 channels. The mechanical design and physical size will be
consistent with mounting the optical system within the probe shown in
Figure 29 in the main text.
B.2. LIGHT SCATTERING FROM PARTICLES
The relationships governing the light scattered from small particles
depend on the ratio of the particle size to the incident light wavelength.
The particle size parameter,
_ Tfd
a - JT ' (B-l)
106
-------
Table B-l. REQUIREMENTS AND ESTIMATED PERFORMANCE OF
LIGHT SCATTERING SYSTEM CONCEPT
Requirements
Performance
Count aerosol particle flow
at impactor exhuasts
Detect particle sizes down
to 0.2-0.5p size.
Concept utilizes forward scattering
of 5 mW He-Ne laser beam; 0.2-0.5y
diameter particles estimated to
produce S/N ^ 12. Individual par-
ticles resolved in center of flow.
All particles larger than stage
threshold size are counted.
Optics size compatible with
preliminary Impactor design
(package within ^ 10 cm O.D.)
Capability for remote readout
3 lens system concept with D £ 0.5",
1^6"; single laser with beam
splitter for 5 impactor outlets.
Power supplies for detectors and
laser mounted external to probe.
Photodiode responds to individual
particles above set threshold size;
remote indication of particle count
is straightforward.
107
-------
BEAM FROM
HE-NE CW LASER
NOT TO SCALE
EXHAUST FLOW
FROM IMPACTOR
if
BEAM SPLITTER
DIVERTED
JLASER BEAM
1
LENS B
ASSEMBLY
TRANSMITTED
BEAM TO
NEXT STAGE
LENS
LK;T
TRAP 2
o
03
PARTICLE
SCATTERING
VOLUME
PHOTON
DETECTOR
TO PARTICLE
COUNTING
ELECTRONICS
CW LASER:
LENS A:
LENS
LENS
LENS
B:
Br
B2:
DETECTOR:
0.5 MW HE-NE (0.63u)
(f/D)A - 4 DA - 1/4-INCH
(f/D)
(f/D)
Bi
B2
« 4
* 2
D
D
BI
B2
= 1/2- INCH
= 1/2-INCH
PIN PHOTODIODE.
Figure B-l. Schematic diagram of optical concept,
-------
describes the appropriate scattering regime, where d is the particle
diameter, and A the wavelength of the illuminating light. For the
smallest size particles considered here (d = 0.2y), and a He-Ne laser
(A = 0.63 ), a = 0.99. The scattering is strongly dependent on the
index of refraction of the particles and is outside the pure Rayleigh
regime (for which a < 0.3) and in the so-called intermediate or Mie
Rayleigh-Gauss regime (cf., Reference 7). Analytical formulations for
the intermediate regime require very detailed calculations, and since
the Rayleigh results are conservative and we are not far from the
Rayleigh regime, we shall use estimates based on Rayleigh scattering
functions.
Considering a beam of illumination density, IQ(W/cm2) incident upon
a single spherical particle, the intensity (W/sr) scattered at an angle
e is given by
J(e) = I0f(e) Ap , (W/sr) (B-2)
where f(e) is the scattering function (sr-1) from scattering theory, and
A is the geometric cross-section of the particle (cm2). The illumination
density is determined from the light source intensity and the optics
geometry.
For Rayleigh scattering, we have
f\
(l+cos2e) , (sr-M (B-3)
where m is the refractive Index, taken here to be 1.5 for graphite/carbon
particles. Thus, for 0.2 particles, we have* f(0°) = 0.0138 sr-1 and
equation (B-2) can then be written as
J(0°) = l.OSxlO-10 IQ dp2 , (W/sr) (B-4)
where d is the particle diameter in ym.** From Equation (B-4), we see that
the intensity of scattered light is proportional to the illumination density
at the scattering volume and to the square of the particle diameter.
*Graphical data 1n Reference 7 (Figure 7, p.321) shows that
f(0°) * 0.019 sr-1 for a = 1.0, d = 0.17p, and m = 1.5.
**The optical system of Figure 1 actually detects scattering at
angles somewhat different from 0°, but differing only by a
small amount.
109
-------
We shall now discuss the requirements for an optical system to provide
laser illumination onto the particles in the impactor exhaust stream.
B.3 OPTICAL SUBSYSTEM REQUIREMENTS
Assuming an optical system as shown in Figure B-l, we can formulate
a general expression for the power (or number of photons) scattered per
particle in the scattering volume. Since our scattering angles are set
to be small, we shall assume the scattering function is 0.0138 sr-1
throughout.
The instantaneous power scattered into the detector is
Pscatt " J(°0) ° W (B-B)
where 0(0°) is given by Equation (B-4) and fi is the solid angle of scat-
tered light subtended by the lens B. Neglecting off-axis effects,
Equation (B-5) may be written in terms of the f-number of lens B as
The illumination density, IQ, at the scattering volume is determined by
the size to which the diverted laser beam can be focused. This is limited
by diffraction; the thickness (diameter) of the diffraction "cylinder" at
the test section center is
t = 1.22 x(f/D)A « 3y . (B-7)
Thus, for a laser beam power of P, in the first stage, we have
'o-rji (w/cm2) (B-B)
The exit optics should be set to image a small region about the scattering
volume onto the detector; no critical requirement exists, however.
From Equations (B-6), (B-8), and (B-2), we obtain the power scattered
by a single particle as
"sc,tt/Pl f<°°> If) 4777DJ7
no
-------
6r the optics of Figure B-l, with (f/D)B = 4, then
0.75 x 10"* y 2
(B-10)
where p is the particle diameter in microns.
Along with detector threshold levels, Equation (B-10) defines the
limit of particle size detectability for a given diverted laser beam power.
For 0.2 particles, we would have Pscatt/P0 = 3xlO~6; for 0.5y particles we
would have 8.3xlO"6.
B.4 PERFORMANCE ESTIMATES
For reliable detection of the scattered light, we use the following
criteria for various detectors
Table B-2. PHOTONS REQUIRED FOR RELIABLE DETECTION
Detector
PIN photodiode
Avalanche photodiode
Photomultiplier
Scattered Photons/Particle
Required for
Reliable Detection
103 - 1
103
102
For cost and simplicity, we would prefer to use PIN photodiodes to detect
particles down to 0.2u diameter
Equation (B-10) can also be written as the ratio of the scattered
photons to incident photons, viz.,
N
T
. o.75 x 10-*
(B-ll)
The number of incident photons is the product of the photon rate (from the
laser beam) and the transit time, T, of a particle through the illuminated
scattering volume; this is
111
-------
N1 = P1 T/E . (
where e = hc/x is the energy per photon which for He-Ne 0.63y light is
0.314 x 10"11 ergs/photon. The transit time is
r-i^SJI" , (sec) (B.13)
where V is the particle flow velocity defined by the impactor stage design.
From Reference 1, we take V = 14 m/sec so that x ^ 0.2 ysec. Hence,
N, = 0.68 x 1012 PI f (B_M)
Figure B-2 is a plot of the calculations for photons scattered per
particle vs. the beam power of the final stage, for various diameter par-
ticles. As can be seen, a 5 mW laser source, with 50% diverted to the
final stage (P^ = 2.5 mW) would allow counting (with 0.2 psec response) of
0.5y particles with high certainty, and 0.2u particles with lower confidence.
An experimental determination of the size detectability of our optics concept
is discussed in Section B.6.
B.5 MECHANICAL DESIGN AND COST
The components of the optical system shown in Figure B-l are small and
can be installed in a small space. No detailed layout has been made here
but installation in the required 10 cm diameter tube should present no
major problems. Small plane mirrors can be used in the optical path to
"fold back" the beam to utilize the available space to full advantage.
Depending on the environment, some attention should be given to prevent-
ing buildup of dirt, etc., on the optics. The most desirable system would
be basically airtight; a minimum approach would at least allow access for
cleaning.
It is envisioned that the power supplies for the laser and photodlode
the laser head, and particle counting electronics (not discussed here) would
all be located outside the probe.
Approximate costs for the components are given in Table B-3. A more
detailed cost analysis was not performed since the mechanical design was
not laid out in detail.
112
-------
ly DIAMETER
PARTICLE
2 3
INCIDENT POWER, mW (FINAL)
Figure B-2.
Calculated photons scattered vs. incident power
for impactor stage.
113
-------
No trade-off analyses have been attempted here. It is likely that
some performance gain could be obtained by optimizing the geometry and
sizes of the system components. The configuration shown in Figure B-l
should be considered a preliminary design.
Table B-3. ESTIMATED COSTS FOR PRIMARY OPTICAL COMPONENTS
OF FIGURE B-l.
Item
No.
1
2
3
4
5
6
7
8
9
10
Description
(Refer to Figure B-l)
5 mW He-Ne laser with
power supply
beam splitter
lens A
window
light trap 1
light trap 2
lens B,
lens B2
photodiode
photodiode power supply
(batteries)
Approx.
Cost
(Each)
$650
70
30
5
5
5
10
10
15
5
Number
Required
1
5
5
10
5
5
5
5
5
1
Total
Total
Component
Cost
$650
350
150
50
25
25
50
50
65
5
$1420
B.6 RECOMMENDED SENSITIVITY EXPERIMENTS
Since there is some latitude in the parameters used for the performance
predictions, especially in values used for the scattering function, labora-
tory experiments should be performed to determine the threshold sensitivity
for particle detection for the system concept of Figure B-l. The procedure
would use the TSI vibrating orifice monodisperse aerosol generator used
for separation testing, to produce a known flow rate and, hence, concen-
tration of particles of precise diameter.
In the laboratory, a buildup of the optical equivalent of the Figure B-l
system would be set up to observe the flow of monodisperse particles through
an approrpiate test section at the aerosol outlet. Prior determination of
114
-------
the illuminated region would be made by observing the exposed pattern on
photosensitive material placed in the test section and subjected to laser
radiation.
With a calibrated, steady flow of known diameter particles, a series
of tests would be performed for various laser beam powers. For high power,
particles will be easy to detect. As the energy is lowered, illumination
levels will eventually become so low that the particles will not be detect-
able; that level is the threshold detectability for that particular size
of particles observed. From the laboratory threshold data, final design
parameters could be determined for a prototype system.
115
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
1. REPORT NO.
EPA-600/7-80-014
2.
3. RECIPIENT'S ACCESSION NO.
4. TITLE AND SUBTITLE
Development Study of a Novel Continuous-flow
Impactor
. REPORT DATE
January 1980
6. PERFORMING ORGANIZATION CODE
T. AUTHOR(S)
E. F. Brooks,-N. Gat, M.E.Taylor, T.E. Chamber lain,
R.J.Golik, and R. Watson
8. PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
TRW Systems and Energy
One Space Park
Redondo Beach, California 90278
10. PROGRAM ELEMENT NO.
EHE624
11. CONTRACT/GRANT NO.
68-02-2165, Task 12
12. SPONSORING AGENCY NAME AND ADDRESS
EPA, Office of Research and Development
Industrial Environmental Research Laboratory
Research Triangle Park, NC 27711
13. TYPE OF REPORT AND PERIOD COVERED
Task Final; 7/76 - 11/78
14. SPONSORING AGENCY CODE
EPA/600/13
T5. SUPPLEMENTARY NOTES TfDT nrnn ' I7T- ' ^[ ^ ' -, .. . ^
nio/c/fi OKC 1ERL-RTP project officer is D. Bruce Harris, Mail Drop 62,
-2557.
16. ABSTRACT
The report gives results of a development study involving feasibility veri-
fication of a novel particle impactor in which the impaction surface is the interface
between two opposing jets. Particles (which would impact a solid surface in a stan-
dard impactor) cross the interface between the aerosol-laden gas and a previously
particle-free gas , are entrained in the latter, and are conveyed out for analysis.
Work consisted of an initial literature search and analysis to determine the likeli-
hood of success, followed by design, fabrication, and testing of a laboratory unit.
A good particle separation capability was demonstrated. Upon completion of the
laboratory tests, a design effort showed the feasibility of a staged in situ particle
monitoring subsystem to give semicontinuous (nominal 1 minute cycle time) output of
particle size distribution, among other applications.
1?-
3.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b. IDENTIFIERS/OPEN ENDED TERMS
. COSATI Field/Group
Pollution
Dust
Aerosols
Impactors
Monitors
Particle Size Distribution
Pollution Control
Stationary Sources
Particulate
Particle Impactors
13B
11G
07D
131
14B
TiTDISTRIBUTION STATEMENT
Release to Public
19. SECURITY CLASS (ThisReport)
Unclassified
21. NO. OF PAGES
123
20. SECURITY CLASS (Thispage)
Unclassified
22. PRICE
Form 2220-1 (i-73)
------- |