United States EPA-600/7-82-036
Environmental Protection
Agency May 1982
&EPA Research and
Development
SAMPLING AND DATA HANDLING
METHODS FOR INHALABLE
PARTICULATE SAMPLING
Prepared for
Office of Air Quality Planning and Standards
Prepared by
Industrial Environmental Research
Laboratory
Research Triangle Park NC 27711
-------
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. Soctoeconomic 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-600/7-82-036
MAY 1982
SAMPLING AND DATA HANDLING METHODS FOR
INHALABLE PARTICULATE SAMPLING
by
Wallace B. Smith, Kenneth M. Gushing, Jean W. Johnson,
Christine T. Parsons, Ashley D. Williamson, and Rufus R. Wilson,Jr.
Southern Research Institute
2000 Ninth Avenue, South
Birmingham, Alabama 35255
Contract 63-02-3118
Project 4181-37
Report SoRI-EAS-81-245R
EPA Project Officers D. Bruce Harris
Industrial Environmental Research Laboratory
Office of Environmental Engineering and Technology
Research Triangle Park, NC 27711
U.S. ENVIRONMENTAL PROTECTION AGENCY
OFFICE OF RESEARCH AND DEVELOPMENT
WASHINGTON, D.C. 20460
-------
ABSTRACT
The objectives of a research program on the sampling and
measurement of particles in the inhalable particulate (IP) size
range in emissions from stationary sources were reviewed and
methods and equipment required are described. A summary is given
of research programs conducted in support of the development of
new and improved techniques for sampling IP emissions, instruments
that are especially suited for IP measurements, and data-handling
methods.
A computer technique was developed for analysis of data on
particle-size distributions of samples taken with cascade impac-
tors from industrial process gas streams. The plot of cumulative
mass concentration vs particle size could be extrapolated from the
limit of the impactor (the first stage cut diameter of 10 ym) to
15 ym by using a third degree polynomial of first order osculation.
To estimate the accuracy of the method, simulated unimodal and bi-
modal log-normal particle-size distributions were sampled accord-
ing to the impactor stage collection efficiency curves. Errors
inherent to the method were found to be negligible. Concentrations
of particles < 15 ym in diameter could be calculated from existing
data to within a factor of 3 and probably within the sampling error
if the impactor data were corrected for nozzle and precollector
effects. Impactor data were also subjected to a modal analysis in
which the data were fitted with multi-component log-normal distri-
butions by a simplex minimization method. The results were
comparable to those obtained with the polynomial: IP concentra-
tions could be estimated within a factor of two or better.
Research on sampling systems for IP matter included the con-
sideration of concepts for maintaining isokinetic sampling condi-
tions, necessary for representative sampling of the larger parti-
cles, while flowrates in the particle-sizing device were constant.
Techniques and devices that were considered were nozzles with
cross-sectional areas that could be varied mechanically, split-
stream probes and shrouded probes, and gas recycle systems. All
the techniques involve the difficulty of manipulation of equipment
from outside the stack.
Laboratory studies were conducted to develop suitable IP
sampling systems with overall cut diameters of 15 ym and conform-
ing to a specified collection efficiency curve. Cascade impactors,
.cyclones, and horizontal elutriators were used as collectors. Lab-
oratory experiments were conducted with monodisperse aerosols on
iii
-------
commercially available impactors and buttonhook nozzles to charac
terize them for measurements in the IP range of particle sizes.
Collection efficiencies were similarly measured for a hori-
zontal elutriator. The measured values agreed well with values
calculated from theory, allowing the preparation of design nomo-
graphs for inlet precollectors in sampling systems.
Design parameters were calculated for horizontal elutriators
to be used with impactors, operated at 14.2 5,/min at 149 °C; the
EPA Source Assessment Sampling System, operated at 185 SL/min at
204°C; and the EPA Fugitive Ambient Sampling Train, operated at
5,282 Vmin and 23°C.
Two cyclone systems were designed and evaluated: a cyclone
to be used as a precollector for impactors, and a system of two
cyclones and a filter in series, to be used as the primary system
for measuring IP and fine particulate matter concentrations. Both
systems were designed for high-temperature service in process gas
streams .
Tests on the Andersen Size Selective Inlet, a 15-um precol-
lector for hi-vol samplers, showed its performance to be within
the proposed limits for IP samplers.
A stack sampling system was designed in which the aerosol is
diluted in flow patterns and with mixing times simulating those
in stack plumes. The system was thus designed to characterize
aerosols formed from condensable vapors in stack gas as the plume
is diluted and cooled by the atmosphere. Tests with the system.
on emissions from a domestic oil-fired furnace indicated that
condensation of organic chemical vapors took place in the system
as expected.
Procedures for the use of impactors for measuring particle-
size distributions in industrial process gas streams were modified
in order to measure IP concentrations. The modifications included
refinements in preparation, calibration, and operation of the im-
pactors and in processing the data.
This report was submitted in fulfillment of contracts No.
68-02-2131 and 68-02-3118 by Southern Research Institute under the
sponsorship of the U.S. Environmental Protection Agency. This
report covers the period November 1, 1978 to December 31, 1980.
IV
-------
CONTENTS
Abstract
Figures - « viii
Tables xiv
1. Introduction and Background 1
Consultant Workshop 3
General Comments on the EPA Program 3
Development of an IP Sampler Performance
Curve 4
State of the Art in Sampling 7
Ambient Samplers 7
Fugitive Emission Samplers 7
In-Stack Sampling. 8
Design and Sampling Considerations for IP
Measurement 8
Precollector Efficiency Curves 8
Isokinetic Sampling 9
Large Particles 10
Condensable Material 10
Fugitive Emissions .... 11
Calibration 11
Recommended Research Program 12
For Immediate Action 12
Six Months Program 12
Long Term Research and Development 16
Summary of Studies Completed 18
2. Development of Methods for Sampling and Data Analysis 20
Data Analysis for Cascade Impactors 20
Summary 20
Previous CIDRS 20
Proposed CIDRS 22
Extending Accuracy in the Mass Size Distri-
bution Calculated Beyond the First Stage
D5o 27
The Osculating Polynomial 28
Accuracy of the Extrapolation Technique. . . 31
Testing with Unimodal Particle-size
Distributions 33
Variation of the Particulate Mass
Median Diameter 33
-------
Variation of the Geometric Standard
Deviation of the Particulate Size
Distribution 33
Variation of the Geometric Standard
Deviation (Slope) of the Collection
Efficiency Curve 34
Variation of the Assumed Largest
Particle Diameter in the Distribution 34
Testing with Bimodal Particle-size Distri-
butions . 37
Variation of Geometric Standard
Deviation of the Particle-size
Distribution 37
Variation of the Mass Median Diameter
of the Second Mode 37
Variation in Fraction of the Total
Mass Contained in Each of the Modes. 40
Effect of Operating Problems 40
Isokinetic Sampling by Constant Flowrate
Particle-sizing Devices 51
Sampling Concepts 52
Variable Area Nozzles 53
In-stack Split Stream Probe 61
Gas Recycle Concept 67
In-stack Probe Shroud 71
3. Design, Fabrication, and Testing of Prototype Sampling
Systems for Inhalable Particulate Matter 73
Summary 73
Evaluation of Commercial Cascade Impactors and
Buttonhook Nozzles—Effect on Measurement of
Inhalable Particulate Matter 74
Experimental Procedures 74
Results and Evaluation 79
Buttonhook Nozzles 79
Impactors. 79
The Horizontal Elutriator as an IP Precollector. 85
Background 85
Experimental Procedures 91
Results 93
Collector Design 98
The FAST Elutriator Inlet 98
Horizontal Elutriator Inlet for Fugitive
Emissions Sampler 98
Cyclones as IP Collectors 104
Background 104
Calibration of the Brink Cyclone 105
Calibration of the Sierra Cyclone 105
Calibration of the Southern Research
Institute Five-Stage Cyclone 105
An Inhalable Particulate Precollector and
Sampling Train 109
Development and Fabrication 109
vi
-------
Calibration .....
Results 115
Size-selective Inlet for the Andersen Hi-Vol
Sampler 122
A Stack Dilution Sampling System ... 122
Design and Testing of Laboratory Prototype
Sampler ..... 125
Prototype Design 125
Performance Characterization of
Laboratory Prototype Sampler .... 127
Domestic Furnace Test. 132
Results of Dilution Sampling Measurements . 133
Gravimetric Measurements . . . . . . .133
Temporal Behavior 134
Effect of Dilution Ratio on Particle
Size Distribution 137
Conclusions 138
4. Operation and Sampling Parameters for the Inhalable
Particulate Sampling System 140
Cascade Impactor Calibration and Operation . . . 140
The Inhalable Particulate Sampling System. . . . 140
The Dual-Cyclone Train 141
Cleaning, Inspection, and Assembly . . 146
Sampling Parameters 147
Sampling 152
Sample Retrieval and Weighing. .... 155
Data Analysis and Reports 157
The Stack Dilution Sampling System 159
Design and Operating Procedures. . . . 159
System Description 159
Assembly 166
Sampling Parameters 167
Sampling 169
Sample Retrieval and Weighing 171
Data Analysis and Reports 171
References 171
Appendices
A. Modifications to CIDRS Cumulative Mass Curve Fitting
Programs for Inhalable Particulate Determination. . . 175
B. Modeling of SRI Impactor Data and Calculation of IP
Concentration 191
C. Design Drawings of the Inhalable Particulate Sampling
Train 267
D. Design Drawings of the Inhalable Particulate Precutter
Cyclone 276
E. Metric System Conversion Factors 281
VII
-------
FIGURES
Number Page
1 Calibration of the hi-vol and dichotomous sampler
inlets 5
2 Recommended IP sampler collection specifications. . . 6
3 Cumulative size distribution from raw impactor data . 23
4 Start of development of interpolated points between
first and last Dso 23
5 Continued generation of interpolated points ..... 24
6 Continued generation of interpolated points 24
7 Generation of interpolated points on parabola which
includes DMAX 25
8 Generation of interpolated points on hyperbola
through Dso (1) and DMAX 25
9 Generation of interpolated points on osculating
polynomial and zero slope line at total mass
concentration 26
10 Fitting the cumulative mass distribution for
D(1)
-------
Figures (Continued)
Number
14 Ratio of recovered to true IP concentration versus
assumed largest particle diameter using a unimodal
log-normal particle-size distribution ....... 36
15 Ratio of recovered to true IP concentration versus
aerosol geometric standard deviation of a bimodal
log- normal mass distribution ............ 38
16 Ratio of recovered to true inhalable particle concen-
tration versus second mode mass median diameter of
a bimodal log- normal particle-size distribution . . 39
17 Ratio of recovered to true IP concentration versus
fractional contribution of mass from each mode of
a bimodal log-normal mass distribution ....... 41
18 Cumulative particle-size distribution at inlet to
electrostatic precipitator ........... . . 43
19 Differential particle-size distribution at inlet to
electrostatic precipitator ............. 44
20 Cumulative particle-size distribution at outlet of
electrostatic precipitator .......... ... 45
21 Differential particle-size distribution at outlet of
electrostatic precipitator ............. 46
22 Calculated IP concentration vs. assumed D at
electrostatic precipitator inlet. Coal-fired power
plant ....................... 47
23 Calculated IP concentration vs. assumed D.., at
electrostatic precipitator outlet. Coal-fired
power plant ............... ..... 48
24 Calculated IP concentration vs. assumed DMWV at
MAX
electrostatic precipitator inlet. Cement kiln. . . 49
25 Calculated IP concentration vs. assumed D x at
electrostatic precipitator outlet. Cement kiln . . 50
26 Conceptual design of variable area nozzle attached
to an impactor ................... 54
ax
-------
Figures (Continued)
Number Page
27 A variable area nozzle attached to a cyclone ..... 55
28 A diamond- shaped nozzle ............... 56
29 Conceptual drawing of an elliptical nozzle ...... 58
30 A multiple nozzle revolver configuration ....... 59
31 An iris nozzle .................... 62
32 In-stack split-stream probe ............. 63
33 Split- stream probe: sli ding-cone probe ....... 65
34 Split-stream probe: adjustable flap probe. ..... 65
35 A split-stream nozzle ................ 66
36 A hot gas recycle concept .. ............ 67
37 Cool gas recycle concept ............... 69
38 In-stack probe shroud ._. ...... . ....... 72
39 Configuration of apparatus used in evaluation of
cascade impactor and nozzle ............ "75
40 Andersen Mark III cascade impactor situated in
isokinetic sampling apparatus ........... 76
41 Isokinetic sampling apparatus adjacent to vibrating
orifice aerosol generator ............. 77
42 Isokinetic sampling apparatus adjacent to vibrating
orifice aerosol generator, ............. 78
43 Calibration data for the first three stages of the
University of Washington Mark V cascade impactor. . 84
44 Particle collection efficiency for the impactor ... 85
45 Settling velocity in air for unit density spheres . . 88
46 Velocity profile and particle trajectory between
parallel plates .................. 89
-------
Figures (Continued)
Number Page
47 Zone of 100% particle collection 90
48 Theoretical collection efficiency by particle settling
in rectangular and circular tubes 91
49 Apparatus used to measure the collection efficiency
of the settling chamber 92
50 Profile of the air velocity immediately upstream from
the blower of the settling chamber before the plates
were positioned 94
51 Profile of the air velocity upstream from the plates
of the settling chamber 95
52 Theoretical and experimental collection efficiencies
for a horizontal elutriator with rectangular cross-
section, plate length 38.1 cm, average gas velocity
70 cm/sec 96
53 Theoretical and experimental collection efficiencies
for a horizontal elutriator with rectangular cross-
section, plate length 20 cm, average gas velocity
40 cm/sec 97
54 Relationship of design parameters for horizontal
elutriators with DSO cutpoints of 15 \m aerodynamic
diameter used as precollectors for in-stack cascade
impactors 99
55 Relationship of design parameters for horizontal
elutriators with DSO cutpoints of 15 ym aerodynamic
diameter to be used with SASS trains 100
56 Relationship of design parameters for horizontal
elutriators with D5o cutpoints of 15 ym aerodynamic
diameter to be used with FAST trains 101
57 Laboratory set-up for testing FAST horizontal
elutriator inlet 10
58 Collection efficiency versus flowrate for a Brink
precollector cyclone sampling 15 ym aerodynamic
diameter ammonium fluorescein particles at 22, 100,
and 150°C 106
XI
-------
Figures (Continued)
Number Page
59 Flowrate versus viscosity to maintain a Dso=15 ym
aerodynamic for Brink precollector cyclone 107
60 Calibration curves for the five-stage cyclone system.
Flowrate 1.0 ft3/min, temperature 22°C 109
61 Schematic of a cascade impactor-precollector cyclone
system 112
62 Schematic of two-cyclone system 112
63 Summary of cyclone dimensions 113
64 Calibration system for heated aerosols 114
65 Collection efficiency of Cyclone IX versus flowrate
for particles of 15+0.6 um aerodynamic diameter . . 116
66 Collection efficiency of Cyclone X versus flowrate
for particles of 15+0.6 vim aerodynamic diameter . . 117
67 Collection efficiency versus aerodynamic particle
diameter for Cyclone III at 22°C and 11.3 £/min (°)/
93°C and 19.8'£/min (o) , and 150°C and 22.7 A/min
(A) . . 118
68 Calibration data for Cyclone IX at 14 jl/min and
150°C .
69 Schematic of size-selective inlet 123
70 Block diagram of dilution sampling system 126
71 Concentration profile of salt aerosol across diluter. 128
72 Temperature dependence of cross-section of diluted
aerosol plume 129
73 Particle-size distribution of glycerol smoke aerosol. 131
74 Time variation of particle number concentration per
unit flue gas as measured after dilution. Dilution
ratio=19.6 (standard) 135
75 Time variation of particle mass concentration per
unit flue gas as measured after dilution. Dilution
ratio=19.6 (standard) 136
xii
-------
Figures (Continued)
Number Page
76 Particle mass concentration per unit flue gas as
measured at different dilution ratios. All
concentrations measured eight minutes after
beginning of combustion cycle 139
77 Schematic drawing of inhalable particulate sampler. . 142
78 The critical internal dimensions of a cyclone .... 145
79 Nomograph for selecting proper sampling duration. . . 148
80 Recommended sampling points for circular and square
or rectangular ducts. 151
81 Gas flowrate versus viscosity at Dso=15 ym aerodynamic
diameter for IP Cyclone SRI-X 153
82 Nomograph for selecting nozzles for isokinetic
sampling, using cyclone samplers 154
83 Diagram of Stack Dilution Sampling System 160
84 Sample probe and heated hose for the Stack Dilution
Sampling System 162
85 The dilution air line of the Stack Dilution Sampling
System 162
86 Inlet assembly for the dilution chamber 163
87 Hi-vol filter and impactor assembly of the Stack
Dilution Sampling System 164
88 Pressure and flow measurement module 165
89 Heater and flow control module 166
Kill
-------
TABLES
Number Page
1 EPA Needs and Priorities for Data on Emissions of
Inhalable Particulate Matter from Stationary Sources 2
2 Recommended Program to Develop Methods for Measurement
of Inhalable Particulate Matter 13
3 Techniques Recommended for Use in Obtaining Data to
Set IP Emission Factors for Industrial Sources... 14
4 Systems Available for Measuring Particle Size and Mass 17
5 Summary of Studies that have been Performed at
Southern Research Institute Towards Completion of
the IP Program 19
6 CIDRS Programs and Output 22
7 Sampling Parameters from a Hypothetical Sampling Test
Using the Multiple Nozzle Revolver Configuration. . 60
8 Change in Stage DSO'S for an Andersen Impactor as a
Function of Moisture Content of Sampled Gas .... 70
9 Evaluation of Buttonhook Nozzles 79
10 Evaluation of Commercial Cascade Impactors Using
Fifteen-Micron Aerodynamic Diameter Particles ... 80
11 FAST Horizontal Elutriator Inlet Test Data 102
12 Calibration Results for the Brink Cyclone 108
13 Sierra Cyclone Calibration Data 108
14 Laboratory Calibration of the Five-Stage Cyclones Dso
Cut Points 110
15 Operating Parameters of SRI Cyclones 121
16 Size-selective Inlet Calibration Data 124
xiv
-------
SECTION 1
INTRODUCTION AND BACKGROUND
The U.S. Environmental Protection Agency expects to set air
pollution standards for emission of inhalable particulate (IP)
matter from stationary sources. Inhalable particulate matter is
defined in terms of particle size, since the extent of penetra-
tion of inhaled particles into the lungs depends on their size.
Adequate characterization of a source of IP matter requires
measurement of stack or fugitive emissions from the source and
also background levels of ambient atmospheric particles in the
relevant size ranges. The concentration and particle-size dis-
tribution of the suspended particulate matter and, in some in-
stances, its chemical composition and biological properties must
be determined. The first step in obtaining information of this
kind is usually sampling of the gas stream or atmosphere for part-
icles of the appropriate sizes.
To assi-st the EPA in planning research on the IP sampling
and analysis problem, a workshop was held at Research Triangle
Park, NC, in December 1978. The workshop was attended by consult-
ants and other investigators experienced in aerosol sampling and
characterization.
At that time, it was generally recognized that there were no
methods in use that were suitable for measuring inhalable part-
iculate matter per se. Thus a need was recognized for informa-
tion on methods that could yield or could be adapted to yield IP
data, and also on the research that would be necessary for the
development of instruments and procedures for that purpose.
Table 1 lists the needs of the Environmental Protection
Agency for data on IP emissions from stationary sources, as seen
by the EPA at the time, and priorities assigned by the agency for
filling them.
-------
TABLE 1 EPA NEEDS AND PRIORITIES FOR DATA ON EMISSIONS OF
* INHALABLE PARTICULATE MATTER FROM STATIONARY SOURCES
P r i o r i ty E leme n t _ Description
1 la Total mass £ 15 ym aerodynamic diameter
Mass- cut to match ambient monitor
Mass- Dso = 15 jim +_ 2 ym
Mass- slope to match ambient monitor
2 Ib Condensable fraction of total mass
or
2c Discrete size fractions <_ 15 urn
3 2a Size distribution - continuum <_ 80 urn
4 2b Size distribution - continuum £ 15 ym
5 3 Chemical composition of material in
program elements 1 and 2
6 4 Bioassay of material in program
elements 1 and 2
The present report contains (1) recommendations on EPA
research prepared in the workshop and (2) a detailed summary of
research done at Southern Research Institute on the IP sampling
problem under contract with the EPA. Except as noted in the text
and in the references at the end of the text, the information in
the report has not been published.
-------
CONSULTANT WORKSHOP*
The purposes of the workshop were:
(1) to discuss the EPA objectives,
(2) to discuss any restrictions or ground rules in
developing or applying measurement techniquesr
(3) to define the current state of the art,
(4) to identify ongoing programs and establish lines
of communication, and
(5) to lay the ground work for a position paper re-
commending methods and research programs on the
development of methods.
At the workshop, elements 1 and 2 of the EPA program (Table
1) were addressed in order of priority, The following is a sum-
mary of the comments and recommendations of the workshop partic-
ipants, as recorded by Southern Research Institute staff members
attending.
General Comments on the EPA Program
The workshop participants agreed that the objectives of the
EPA program appeared to be logical and reasonable. It was under-
stood that the definition of inhalable particulate matter and the
"standard'' particle-size curve had resulted from prior decisions
by the EPA and were not subject to review or revision. All
particles smaller than 80 ym, including condensable materials,
were to be considered of interest. Condensables were included
because they were recognized as constituting most of the mass of
emissions from many sources. On the other hand, any emphasis on
*EPA personnel who participated or attended were: John
Backman, Fred Miller, Dennis Drehmel, John Nader, Bruce Harris,
Jim Abbott, Ken Knapp, Jack Wagman, Charles Rhodes, and Frank Noonan.
Consultants and other investigators who attended were:
Wallace Smith (Southern Research Institute), Kenneth Whitby
(University of Minnesota), Richard Parker (Air Pollution Techno-
logy), Virgil Marple (University of Minnesota), Dale Lundgren
(University of Florida), Thomas Mercer (University of Rochester),
Carl Peterson (Environmental Research Corporation), Michael Pilat
(University of Washington), Walter John (California Dept. of
Health Services), Morton Lippmann (New York University Medical
Center), Robert Heinsohn (Pennsylvania State University), Richard
Flagan (California Institute of Technology), Joe McCain (Southern
Research Institute), Ray Wilson (Southern Research Institute),
and Henry Kolnsberg (The Research Corporation of New England).
-------
particles larger than 20-30 ym was not considered to be justified
for most sources. It was argued that very large particles are
typically not emitted in large quantities and that they do not
travel far from the source. Frequently, samples of large parti-
.cles in ambient aerosols have been found to consist primarily of
dust generated by the wind, automobiles, or farming, rather than
being emitted from stationary sources. Another practical problem
associated with large particles was seen in the difficulty in-
volved in sampling them. It was recognized that great care would
be required in developing sampling systems and protocols to ensure
that representative samples would be obtained.
The workshop participants agreed that the greatest problem
associated with the EPA's proposed program would be the lack of
time available for the development of adequate measurement methods.
Although it appeared to be reasonable to design and evaluate sur-
vey systems for in-stack, ambient, and fugitive emissions in 6
months to a year, a number of the systems would have to be fab-
'ricated and procedures written before they could be put to wide-
spread use. It was expected that 3 to 5 years would be required
before reliable methods would be available that could be used to
obtain all of the detailed data needed. The only practical
course appeared to be the use of survey techniques to obtain less
comprehensive data as soon as possible while research programs
were initiated to develop more advanced and accurate methods for
future use.
Development of an IP Sampler Performance Curve
Figure 1 shows some calibration data for inlets for the hi-
vol sampler and dichotomous sampler. The data approximate the
"standard" curve for IP samplers. It was strongly recommended
that these data be confirmed and extended by independent research.
It was recommended that the EPA adopt a standard mathematical per-
formance curve for samplers used in measuring IP matter, with
specified tolerances. Any newly developed devices would have to
be calibrated and shown to fit the curve within the tolerances
specified. Suggested performance criteria for IP samplers are
shown in Figure 2. The performance curves of all samplers would
be required to fall within the shaded area. The shaded area is
bounded by log-normal curves drawn through the (13 ym, 50%) and
(17 ym, 50%) points with geometric standard deviations (a ) of
1.0 and 1.7. After allowing 10% for wall losses of smallgpart-
icles and 10% penetration of large particles, the area is com-
pletely defined as shown. After more thorough characterization
of existing and newly developed sampling systems, it might be
possible to place more stringent tolerances on the standard curve.*
*Subsequent to the workshop, the curve in Figure 2 was indeed
adopted as the standard IP curve by the EPA.
-------
80
60
40
~ 20
>
O
10
iu
i
TT
O 2 Km/hr
• 8 Km/hr
O 24 Km/hr
A 2 Km/hr
SIZE SELECTIVE HI-VOL
SIZE SELECTIVE MEMBRANE
4 8 Km/hr
O 2 Km/hr BECKMAN DICHOTOMOUS INLET
I
I
I . I
4 6 8 10
AERODYNAMIC PARTICLE DIAMETER,
20
40
4181-338
Figure 1. Calibration of the hi-vol and dichotomous sampler inlets.
-------
CTV
4 6 8 10 20
AERODYNAMIC PARTICLE DIAMETER, jum
Figure 2. Recommended IP sampler collection specifications.
60 80 100
4181-30
-------
State of the Art in Sampling
The following techniques were those available at the begin-
ning of the research program.
Ambient Samplers--
Suitable inlets were currently under development for the
dichotomous sampler and the hi-vol sampler. The performance
characteristics of the devices other than the inlets were pre-
sumed to be well-characterized and satisfactory. It was expected
that after thorough characterization of the new inlets they would
be available for acquiring IP mass data.
Standardized instruments for acquiring detailed information
on particle size were not available. It was expected that dif-
fusion batteries and electrical aerosol analyzers could be made
available within 6 months to a year for measuring very fine part-
icles, but that measuring particle sizes up to 80 ym would re-
quire extensive research.
No real-time monitors were currently available for measuring
mass or particle size on a routine basis.
Condensable materials were not expected to be present in
ambient and fugitive aerosols as vapors, and hence were not con-
sidered to have a high research priority.
Fugitive Emission Samplers—
The hi-vol and dichotomous samplers were presumed to be
suitable for sampling fugitive emissions. The FAST system, a
200 ft3/min sampler, would require the development of a new IP
inlet before it could be used. Prototype FAST units were pro-
jected as being available in about 6 months. The FAST instrument
yields limited information on particle size (as does the dichot-
omous sampler) as well as on the concentration of organic chem-
ical vapors.
The comments made above with reference to particle size and
real-time analysis were considered to apply to fugitive emissions
as well as to ambient sampling.
The greatest problem in sampling fugitive emissions was con-
sidered to be the development of suitable sampling strategies.
There was no protocol available for selecting sampling methods
or locations to characterize fugitive emissions.
-------
In-Stack Sampling—
It was expected that both cyclones and impactors could be
used as in-stack devices to approximate the IP performance curve.
Nevertheless, it was recognized that additional calibration and
perhaps modification would be required before practical systems
could be assembled.
In measuring the condensable fraction, it is important to
simulate the dilution and cooling process that occurs at the
stack-ambient interface. It was noted that a system under de-
velopment would allow the operator to mix the sample with filtered
ambient air before it reached the filter. This method appeared
to be suitable for measuring mass, but it was recognized that it
would require refinement before the dilution sufficiently repre-
sented the stack-ambient interaction to allow accurate measure-
ment of particle size. In addition, the collectors that would be
used would have to satisfy the requirements of the IP performance
curve.
It was noted that some advanced types of data analysis had
not been exploited efficiently to obtain in-stack information on
particle-size distribution. Specifically, it was noted that
methods had been developed in treating ambient sampling data to
extrapolate particle-size distribution curves to larger particle
sizes, thus yielding data on IP concentrations. This was thought
to be a technique that was immediately available for making pro-
gress in the search for IP data on stationary sources. The tech-
nique would not provide direct information on the condensable
fraction, but if the gas composition were known, it might be pos-
sible to make some general inferences.
It was pointed out that any method that involves extrapola-
tion must be carefully examined and verified before extensive
application to stack gas, especially at the outlet of a control
device, where the original particle-size distribution would have
been distorted by the removal of particles in a non-linear fash-
ion.
Design and Sampling Considerations for IP Measurement
Precollector Efficiency Curves—
The mass of inhalable particulate matter that is measured
is relatively insensitive to the standard deviation, ag, of the
inlet collection efficiency curve for most samplers, but the par-
ticle-size distribution may be greatly affected. However, if the
complete particle-size distribution smaller than 80 prn (or even
20-30 ym) was to be measured, there appeared to be little value
in having any stage in the sampler match a standard IP performance
curve. It was thought that all the information could be retrieved
-------
from the measured particle-size distribution. If the material
was to be used for bioassay, however, a .stage that matched the
standard IP curve would be desirable.
Another point that was considered concerning the interpre-
tation of data obtained from ambient samplers was the problem of
recovering the portion of the sample containing particles larger
than 15 ym, so that the ratio of the inhalable fraction to the
total could be obtained. It was recognized that theoretically
the information could be obtained by operating a standard hi-vol
sampler at the same location as the IP sampler and attributing
the difference in the amounts collected to particles larger than
15 ym. However, this would be a secondary measurement and might
lead to undesirably large errors; therefore, it was recommended
that some thought be given to recovering the larger particles.
It was noted that IP precollectors for in-stack samplers
might be larger than desirable and that size constraints might
force the operation of sampling trains at lower flowrates than
desired. For example, with a cyclone precollector, the Method 5
trains might be restricted to a maximum sampling rate of 0.5 ft3/
min.
A problem that had recently been observed and accepted as a
fundamental limitation of impactor performance was particle bounce
at large particle sizes or high impaction velocities. Although
it is not evident in Figure 1, a typical collection efficiency
curve for an. impactor exhibits very low values for particles much
larger than the Dso value. Thus, it was recognized as imperative
that all proposed inlets be calibrated with dry particles much
larger than the Dso value,
Isokinetic Sampling—
It was recognized that in stack sampling when precollectors
such as cyclones or impactors are used with a mass train, it is
necessary to maintain a fixed constant flowrate, in order to
avoid changes in the sampling characteristics of the precollector.
Any alternative to using a fixed flowrate would involve the. devel-
opment of new, complicated sampling "systems. It would not be
expected that errors introduced by limited traverses and non-
isokinetic sampling would significantly affect the quality or
reliability of the IP data obtained.
In ambient and fugitive sampling systems, the inlet would
have to be designed to minimize the effects of varying wind speed
and direction. The magnitude of errors due to non-isokinetic
sampling would have to be determined by careful calibration.
-------
Large Particles—
Because of their high inertia/ large particles are extremely
difficult to sample in a representative fashion. Also their con-
centration is usually low, and so counting statistics might be
poor. Large particles have to be sampled isokinetically and can-
not be transported around bends or through long sampling lines.
It was judged that special systems involving manual operation and
perhaps microscopic analysis would be required to measure size
distributions of large particles. It appeared doubtful that real-
time analysis could be achieved in the near future.*
Condensable Material—
It was recognized that accurate measurement of the amount of
condensable material existing in a process stream and determina-
tion of its physical and chemical properties in the atmosphere
presented perhaps the most important and challenging task of the
IP research program.
Many process streams contain materials in the vapor state
which condense to form homogeneous or heterogeneous aerosol par-
ticles. Some "condensable" vapors do not actually appear as par-
ticles in the air downwind of the source since their equilibrium
vapor pressure is high enough that they will not condense, or
condensed vapors may re-evaporate as dilution is increased. In
order to measure the properties of the aerosol emitted to the
atmosphere, the condensation in the sampling system would have
to simulate plume behavior. Dilution ratios on the order of
1000:1 might be required. Lower dilution rates might maximize
the super saturation and overestimate the contribution of con-
densed vapors to the IP matter.
It was noted that the rate of dilution might be important.
Studies of the aerosols produced by diesel engines (Kittelson,
University of Minnesota) have shown that rapid dilution of 1000:1
can prevent condensation of organic vapors.
It was also pointed out that condensed vapors can contribute
significantly to mass emissions under some circumstances. The
example of a coal-fired boiler burning coal with 10% ash was
given. If all the ash appeared in the flue gas and if 99% of this
fly ash was collected, 1% of the fly ash would be omitted from
the stack. The fly ash emissions would amount to 0,1% of the
mass of the original coal. If the coal contained 1% sulfur, and
if 5% of the sulfur was emitted from the stack, it would amount
to 0.05% of the mass of the coal. Furthermore, since the sulfur
is emitted as sulfur oxides or sulfuric acid, the mass contributed
to the emissions would be multiplied 2-or 3-fold. Thus it could
*Subsequent to the workshop, the measurement of very large part-
icles was relegated to a lower priority by the EPA.
10
-------
equal or exceed fly ash emissions.
In other processes, notably smelting, the fraction of the
emissions emitted as vapors and condensing at the stack-ambient
interface could far exceed the amount existing as solid particles
within the stack.
It was noted that it would be preferable in the sampling to
dilute and cool the entire aerosol as it exists in the stack.
However, because of the unacceptably high losses in transporting
large particles through sampling lines and dilution systems, the
workshop participants recognized that it would almost certainly
be necessary to collect and size particles larger than 2-3 ym in
the stack. The smaller particles would then pass to the dilution
system outside the stack for further characterization. Removing
particles from the sample aerosol would remove surfaces on which
condensation would likely occur upon cooling and dilution. Thus,
to some extent, the result would be biased and the mass concen-
tration of the aerosol and its particle-size distribution in the
diluter would be distorted.
In view of these complications, it was recognized that exper-
imental and theoretical studies had to be performed to determine
the magnitude and nature of the errors associated with this com-
promise procedure.
Fugitive Emissions—
The development of realistic sampling protocols was seen as
perhaps the most difficult problem in measuring fugitive emis-
sions. It was noted that a great deal of research would be
required for the development of suitable sampling strategies.
In addition to deploying networks of fixed sensors, mobile lab-
oratories and the use of chemical tracers were considered to be
methods that could be applied to isolating and measuring fugitive
emissions.
Calibration—
It was recognized that instruments and systems must be cal-
ibrated with standards and that the development of calibration
procedures would require considerable time and effort. As a
minimum, the following calibration parameters would have to be
considered:
- particle type - wet, dry, monodisperse, polydisperse
- velocity of gas
- particle diameter - aerodynamic, standard set for
verifying size
- sampling efficiency vs. particle size
- wall loss vs. particle size
11
-------
- aerosol concentration - it could affect the operation
of some devices
- temperature - it would have to simulate sampling
conditions
- wind direction
- gas loading and reentrainment
- frequency of calibration
- number of devices calibrated - initial prototypes, or all
The workshop participants noted that in the early stages of
a research and development program, it would be important to have
the opportunity of calibrating devices at more than one facility,
so that results could be compared and verified.
Recommended Research Program
The research plan outlined in this section was developed
within the context of the three-phase program illustrated in
Table 2. The participants pointed out other work being done at
several laboratories that would ultimately benefit the IP program,
Table 3 summarizes the research program recommended by the work-
shop participants.
For Immediate Action—
The only activities (other than initiating a comprehensive
research and development program) that appeared profitable for
immediate action were:
(1) Additional analysis of existing data. Two tech-
niques were to be evaluated: the University of
Minnesota modal analysis and the Southern Research
Institute curve-fitting procedure.
(2) Calibration of sampler inlets. It was recognized
that the new inlets for the hi-vol and dichotomous
samplers needed to be calibrated and their perfor-
mance verified, and that a total mass sampler (lo-
vol) with the same flowrate and inlet as the di-
chotomous sampler was needed for field comparisons
between total mass samplers and the dichotomous
sampler.
Six Months Program—
The picture projected at six months appeared somewhat
brighter. It was envisioned that systems could be made avail-
able to acquire much of the survey data required for the IP
program. Nevertheless, the participants noted that the sampling
program could not be fully implemented, adequate data obtained,
data reviewed, and the method accepted by the scientific com-
munity before several (3-5) years would have elapsed.
12
-------
TABLE 2 RECOMMENDED PROGRAM TO DEVELOP METHODS FOR
MEASUREMENT OF INHALABLE PARTICULATE MATTER
Phase I -
Phase II -
Develop Program Outline
Set Specifications
Assess Methods Available
Identify R&D Areas
Set Milestones
Contact Consultants and Subcontractors
for Phases II and III
Short Term Tasks
Identify Methods Where Specifications
May Be Met By Changing Only Sampling
Protocol
Identify Methods Where Specifications
May Be Met By Minor Hardware and
Procedure Modifications
Bread Board Hardware and Test
Demonstrate/Test as Time Allows
Draft New Protocols
Phase III - Research and Development
Identify Systems That Can Meet All
Program Ojectives
Develop Prototype Systems
Develop Calibration and Test Protocols
and Systems
Calibrate and Demonstrate Systems
Write Technical Reports, Operator's
Manuals, and Procedures Manuals
-Start
2 months
i
6 months
Initial Milestones
to be Set in Phase I
-------
TABLE 3. TECHNIQUES RECOMMENDED FOR USE IN OBTAINING DATA
TO SET IP EMISSION FACTORS FOR INDUSTRIAL SOURCES
TIME SCALE
Immediate
Six Months
Longer Term:
Nine Months
Two Years
PROGRAM ELEMENT
1a. TotalMass <15 /urn
1b. Condensable Mass
2c. Size Distribution
<15 jum,, Discrete
2a. Size Distribution
<80 jum
2b. Size Distribution
X15 iim, Continuous
1a.
tb.
2c.
2a.
2b.
1a.
1b.
2c.
2a.
STACK
Re-Analyze Data
Cyclones + Impingers,
for la, 1b, 2c
Also SASS Train
Cyclones, Impactors
15 jum, 2.5 jum Cyclones
Followed by Dilution
Run in Parallel with
Method 17
15 /urn, 2.5 jzm Cyclones
Dilution, Out of Stack
Sizing. (Impactor, Optical,
Diffusional, Electrical)
Body Impactor for 2-80 jum
AMBIENT
FUGITIVE
Lo-Vol, Hi-Vol, Dichotomous
(Similar inlets requiring more calibration and
performance verification)
1b, N/A FAST Train, New Inlet
Deposition on Suitable Substrate, with
Automatic or Manual Image Analysis.
Rotating, Body Impactor,
Optical, Diffusional, Electrical,
Centrifugal
Mobile Labs
-------
For in-stack use, a train was suggested that would include
in-stack cyclones with D50 values of 15 ym and 2.5 ym. This
would be a modified EPA Method 5 train with a conventional filter
box and a modified impinger train. A study of impinger solutions
would be required to reduce artifact formation. Dilution would
be preferred over impingers as a. means of promoting condensation
in collecting volatile components of the aerosol.
It was appreciated that isokinetic sampling of stack gas
would not be possible with fixed flowrate sampling systems unless
a very large number of nozzle sizes were available and the nozzles
were changed at each sampling point of a stack traverse. There-
fore a simplified (compared to EPA Method 5) sampling procedure
was suggested. In the light of the accuracy of stack sampling
procedures, no significant loss of accuracy is expected to result
from the simplification.
It was recommended that two simultaneous samples should be
obtained using four sampling sites positioned at 90-degree inter-
vals along the centroid of the gas flow in a circular duct (or
the equivalent in a rectangular duct) consisting of:
(1) a total particulate mass sample obtained using
a standard EPA Method 5 or Method 17 sampling
train at the isokinetic flowrate,
(2) IP samples obtained with the eyelone-impinger
or .cyclone-diluter train described above.
With respect to the sampling velocity, a reasonable com-
promise was judged to be operation at + 20% of isokinetic using
a fixed sampling rate and a set of 14 nozzles ranging from 2 to
20 mm in diameter in steps of 1.2 (20% diameter increase). This
arrangement would allow + 20% isokinetic rate sampling at 0.3 £/
sec and a gas velocity range of 1-100 m/sec, which would be ex-
pected to cover almost all stack-sampling situations.
Cascade impactors, cascade cyclones, diffusion-condensation
nuclei counters, and electrical mobility analyzers would be used
for detailed investigations of the in-stack aerosol (<_ 15 ym,
without condensables).
For ambient work, it was suggested that more calibration of
the new inlets to the hi-vol and dichotomous samplers was needed,
in addition to the development of the lo-vol instrument mentioned
above. In order to size particles up to 80 ym, particle collec-
tion by settling (dust fall) or a rotating impactor was considered
practical, with microscopic analysis for sizing.
The ambient systems would also be used for sampling fugitive
emissions, but in addition sampling strategy was noted to be a
15
-------
very important problem requiring strong emphasis. Also, the FAST
equipment would require a new inlet. A horizontal elutriator was
envisioned for this use.
' Continued development and adaptation of the more complex
data analysis techniques was seen as needed to ensure that the
maximum amount of information would be obtained from the data.
Long-Term Research and Development—
It was recommended that long-term research activities should
consist of refining and documenting the procedures that were de-
veloped in haste to make surveys of IP sources, developing stand-
dardized methods for obtaining more detailed information, and per-
forming specialized experiments to obtain data on unusual sources.
The techniques that were suggested for widespread application
based on current technology are indicated in Table 3. Additional
methods that might be used are given in Table 4.
In-stack sampling—It was recommended that a stack sampling system
with dilution that simulates the cooling-dilution process in a
plume be developed for measuring the particle-size distribution
of the diluted aerosol after condensation occurs. This should
be coupled to the development or refinement of a dynamic, theo-
retical model of aerosol behavior during cooling and dilution.
Standard hardware systems and procedures would need to be
developed for measuring the particle-size distribution of the
diluted aerosol.
For collecting samples for chemical analysis, cascade im-
pactors were favored, although diffusion batteries and cyclones
could also be used. In many instances, real-time data acquisi-
tion would be desired, and optical, diffusional, or electrical
methods would have to be used. Optical counters would have to
be calibrated for aerodynamic sizing, using cyclones or impactors
on the inlet.
For measuring very large particles, up to 80 ym, a body
impactor, consisting of a number of cylinders or rectangular
j areas of diminishing width, was considered most practical. The
performance of the device would have to be determined, and sam-
pling and analyzing procedures developed.
It was recognized that chemical analysis must be coupled with
particle sizing in all sampling programs, and that these tech-
niques required additional research.
It was also recognized that it would be possible to elimi-
nate conflicting requirements for fixed flowrates and isokinetic
16
-------
TABLE 4» SYSTEMS AVAILABLE FOR MEASURING PARTICLE
SIZE AND MASS
Mass
Particle Size
Ambient Sampling
Filter
Hi-Vol
Other
Piezoelectric
Nephelometer
Beta Gauge
Fugitive Sampling
Filter
Hi-Vol .
Other
Piezoelectric
Nephelometer
Beta Gauge
Lidar
Stack Sampling
Filter
Charge Transfer
Beta Gauge
Piezoelectric
Optical
Transraissometer
Lidar
Other
Impactor
Hi-Vol
Mega Vol
Other
Dichotomous Sampler
Diffusion Battery
Electrical Mobility
Analyzer
Optical Counter
Piezoelectric
Impactor
Centrifuge
Cyclone
Centrifuge
Filter
Hi-Vol
Mega-Vol
Other
Dichotomous Sampler
Diffusion Battery
Electrical Mobility
Analyzer
Optical Counter
Piezoelectric
Impactor
Centrifuge
Cyclone
Fugitive Assessment
Sampling Train (FAST)
Centrifuge
Impactor
Cyclone
Optical Counter
Electrical Mobility
Analyzer
17
-------
sampling if recirculation loops or variable inlets were used.
Research on the practicality of both approaches was recommended.
Also recommended was the study of the errors associated with
sampling IP matter in turbulent streams with nozzles of different
sizes and at non-isokinetic flowables.
An adequate theory of cyclone performance under stack con-
ditions was considered to be needed more than ever. Further ap-
plication of advanced data analysis techniques, if not completed
under the short-term study, were recommended for further invest-
igation.
Ambient sampling'—Continued refinement of the rotating impactor
method for sampling large particles and advanced methods of data
analysis were recommended for further study.
Real-time systems in mobile laboratories were recommended as
an important means of characterizing emission sources. Mobile
vans were viewed as more versatile than fixed sampling networks,
and less expensive for a large survey program. It was noted that
chemical analysis of particulate matter would be of great value
in relating ambient aerosol compositions to stack emissions or
non-stationary sources (.especially for large particles) and suit-
able procedures would have to be established for incorporating
the analyses into the survey program.
It was noted that research would be required to develop
methods of sampling large particles accurately and for transpos-
ing real-time particle size and concentration data to an aero-
dynamic basis.
Fugitive emissions-"It was suggested that in addition to the
methods used in ambient sampling, techniques that employed remote
sensing or balloon-lofted instruments might be especially applic-
able to sampling fugitive emissions. Here also, mobile labora-
tories would be valuable in characterizing fugitive emissions
rapidly and efficiently.
An extensive research program was seen as needed to develop
sampling strategies. Decisions would have to be made on how to
sample in space and time, and on what meteorological data would
be required. Statistical techniques and modelling would be used
to optimize the procedures, and chemical analysis would help in
discriminating between source and non-source aerosols.
Summary of Studies Completed
The remainder of this report summarizes the work that has
been done at Southern Research Institute under contract with the
EPA on the IP program. A list of the studies included in this
18
-------
report that originated from the recommendations described in the
previous section is shown in Table 5. This table also lists the
page number corresponding to the section of the report in which
each topic is discussed. References to published information
resulting from these studies are listed following the text.
TABLE 5. SUMMARY OF STUDIES THAT HAVE BEEN PERFORMED
AT SOUTHERN RESEARCH INSTITUTE TOWARDS COM-
PLETION OF THE IP PROGRAM
_____
Data handling methods for determination of inhalable par-
ticulate matter in previously obtained"data 20
Isokinetic sampling by constant flowrate sizing devices 51
Evaluation of commercial cascade impactors and button-
hook nozzles - effects on measurement of inhalable par-
ticulate matter 74
The horizontal elutriator as an IP precollector 35
The FAST elutriator inlet 98
Calibration of available cyclones at stack conditions ^05
An IP precollector cyclone 109
An IP cyclone train (D50 values 15 ym and2.5ym) 109
Calibration of a hi-vol sampler size-selective inlet 122
A stack dilution sampling system for measurement of con-
densation aerosols 122
Procedures for cascade impactor operation in process
streams 14Q
Operation of the IP sampling, system
Operation of the stack dilution sampling system 159
19
-------
SECTION 2
DEVELOPMENT OF METHODS FOR SAMPLING AND DATA ANALYSIS
This section contains information on (1) a computer technique
that allows the extrapolation of particle-size data from cascade
impactors, and (2) techniques and devices for maintaining iso-
kinetic sampling with particle-sizing equipment that is operated
at constant flowrates.
DATA ANALYSIS FOR CASCADE IMPACTORS
Summary
Because of the increased interest in the health effects
of particles up to 15 ym diameter, it has become desirable to
accurately extrapolate particle-size data above the effective
limit of the first stage D50 (approximately 10 ym) of a cascade
impactor. To do this, a first-order osculating polynomial is
used in conjunction with the Computer Impactor Data Reduction
System1 for fitting the cumulative mass curve between the first
stage D50 and the maximum particle size. The function is a third
degree polynomial which uses the known characteristics of the
cumulative mass curve for its solution over the range of particle
sizes. Tests of this technique on a number of theoretical uni-
modal and bimodal size distributions demonstrate a high degree
of accuracy in recovering the true cumulative particle concen-
tration up to 15 ym. This technique, as described below, can
be used for recovering inhalable particulate concentrations from
existing impactor data within a factor of about 3 if no information
is available on the type of nozzle or precollector that was used.
If the effects of these devices are known, the errors in the
IP data are probably within the experimental error of sampling.
Further testing of the data reduction technique is recom-
mended to evaluate the effect on its accuracy of random errors in
the measured size distribution.
Previous CIDRS
EPA Report 600/7-78-042, "A Computer-Based Impactor Data
Reduction System"1 described a series of five computer programs,
known by the acronym CIDRS, which are designed to reduce the
20
-------
field data taken by commercially available round jet cascade
impactors. An outline of the CIDRS programs is illustrated in
Table 6 with a listing of the program names and their functions.
In brief, the sampling hardware information, sampling condi-
tions, and particulate catch information are used as input to
CIDRS to determine effective cut sizes of the various impactor
stages and the particle concentration at each of these stages.
This cumulative mass concentration
-------
TABLE 6. CIDRS PROGRAMS AND OUTPUT
Averages
With
Confidence Bars
MPROG
Cumulative Mass Concentration
-------
o
5
UJ
D
II I I i i I r i l I l l | i i i i i i i I I I I
Dso(6) D50(5) D50(4) D50(3) D50(2) D50(1) DMAX
PARTICLE DIAMETER
Figure 3. Cumulative size distribution from raw impactor data.
5
<
o
CO
1
H FIRST INTERPOLATION
PARABOLA
O
INTERPOLATED POINTS
I I 1,1 M , 1 1 1 I I II II 1 1 1 . i . . .
Dgo'6) D50(5) D50(4) D50{3) D50(2) D50(1I DMAX
PARTICLE DIAMETER s4isi-7
Figure 4. Start of development of interpolated points between first and last
23
-------
(9
5
3
O)
i
Ul
§
SECOND
INTERPOLATION
PARABOLA
INTERPOLATED POINTS
j I
4-
D50<6) D50<5> D50(4) D50(3) D50(2) D50<1)
PARTICLE DIAMETER
Figure 5. Continued generation of interpolated points
DMAX
(9
2
5
<
o
5
LU
1
INTERPOLATED POINTS
i t i i i i i i
THIRD INTERPOLATION PARABOLA
D50<6) D50(5) D50I4) D50{3) D50(2)
PARTICLE DIAMETER
Figure 6. Continued generation of interpolated points
DMAX
S4181-8
24
-------
o
o
Ul
O
INTERPOLATED POINTS ON
FINAL PARABOLA
FINAL INTERPOLATION
PARABOLA
4-
1 1 1
D50<6> °50<5>
D50<3> °50<2> D50<1'
PARTICLE DIAMETER
Figure 7. Generation of interpolated points on parabola
which includes DMAX.
DMAX
S4181-9A
O
5
D
1
00
I — ,-J
+
' '.' "
SLOPE = 0
HYPERBOLA AND
HYPERBOLIC
INTERPOLATION POINTS
BETWEEN
D50 (1) and DMAX
D50(6) D50'5'
Dso<3> °SO<2> D50(1)
PARTICLE DIAMETER
Figure 8. Generation of interpolated points on hyperbola through
ni~n'11 and DMAX.
DMAX
S4181-9
25
-------
CD
Z
(4
1
Ui
>
5
u
INTERPOLATION
POINTS ON AN
OSCULATING
POLYNOMIAL
j L
i i
O— f
INTERPOLATION
POINTS ON LINE
(SLOPE=0) AT
TOTAL MASS
LOADING
i
D50(4) D50(3) D50(2) D50(1)
PARTICLE DIAMETER
DMAX
S4181-10
Figure 9. Generation of interpolated points on osculating polynomial and zero
slope line at total mass concentration.
the osculating polynomial approaches the total mass loading,
its first derivative goes to zero. For log diameters greater
than this zero slope point (ZSPT) the cumulative mass loading
is at its total value. The functional form of the osculating
polynominal is discussed below. Its importance lies in the fact
that it is a better suited function than the hyperbola for recov-
ering the inhalable particulate matter, i.e., cumulative mass
concentration <15.0 ym which lies in this fitting region beyond
the first stage DSO (-10.0 van).
The set of interpolated points and original DSO points is
used in fitting a series of continuous, second-degree polynomi-
als. The fitting coefficients along with their boundary points
(the set of cumulative mass concentration vs. particle size used
in making the fit) are stored in files. Using these coefficients,
the cumulative mass curve and the mass and number size distribution
curves may be recovered for use in any subsequent program.
26
-------
Extending Accuracy in the Mass Size Distribution Calculated
Beyond the First Stage Dso
CIDRS has proven to be an accurate and time saving tool
for impactor data reduction over the size region of approximately
0.25 urn up to 10 um or the first stage Dso.2 Because of the
present construction of cascade impactors, there is a large span
of particle size in the distribution. Since an extrapolation
to 15.0 ym extends the data only slightly beyond the first stage
Dso, calculating IP mass concentration appears feasible if proper
extrapolation techniques are used.
The known or obtainable properties of the distribution are:
(1) The first (largest) DSO is known.
(2) The cumulative mass at the first Dso is known.
(3) The slope of the cumulative mass curve at this D50
can be calculated.
(4) The largest particle diameter is known or can be
estimated.
(5) The total cumulative mass at the largest particle
diameter is known.
(6) The first derivative of the cumulative mass curve
at the largest particle diameter =0.0.
(7) The first derivative of the cumulative mass distribu-
tion is non-negative.
As mentioned previously, the function that has been used
in CIDRS for fitting data beyond the first stage Dso has been
a hyperbola of the form given in Equation 1.
Properties I, 2, 4, and 5 are automatically satisfied by
this two point fit (Equation 1). In addition, the form of the
hyperbolic function generates a close approximation to properties
6 and 7 in that the slope of this function for real data is never
negative and approaches zero at large D.
However, the hyperbolic function has drawbacks which limit
its usefulness for extrapolation to inhalable particle sizes
above the first stage Dso. First, property 6 is never rigorously
satisfied. While the slope of the hyperbolic function approaches
zero for large values of the maximum particle size, it will not
be zero for a finite maximum particle size as is required. More
seriously, there is no correlation between the slope of the hyper-
bola at the first stage D50 and the slope determined by the
spline fitting routine. This discontinuity in slope violates
property 3 of the true cumulative mass distribution and thus
the hyperbolic function will not define proper mass concentrations
for diameters slightly above the first stage Dso.
-------
The Osculating Polynomial
The new technique employs a polynomial of first-order oscula-
tion to fit cumulative mass concentration from the first stage
Dao to the maximum particle size.3 It not only passes through
the two end points/ but also may be constructed to have zero
slope at the maximum particle size and have a slope at the first
stage D50 equal to that of the spline fit through the first stage
Dso. The osculating polynomial may also be constrained to have
a non-negative slope over the specified range of particle sizes.
In other words, all seven of the known or obtainable properties
of the mass distribution in this particle range can be satisfied
by the proposed technique.
A polynomial of first-order osculation is used as an approxi-
mation to a given function. By definition it matches both the
function and its first derivatives at a finite number of points.
In general, there exists a unique (2n-l) degree polynomial fit
to a set of n points. For the case of impactor data, let par-
ticle diameter be represented by the variable D, the cumulative
mass loading by M(D) , and the osculating polynomial approxima-
tion of this function by P(D); then properties 1-6 require
P(Di) =M(Di), and P'(Di) = M' (D^ (2)
for i = 0, 1 where the first point corresponds to the first stage
Dso and the second to the maximum particle diameter. For the
approximating equation to be physically realistic it is also
necessary that property 7 be satisfied, therefore
P'(D) > 0 for D0 _< D _< Dj. (3)
To first order osculation, the polynomial P(D) can be
described by Hermite's formula,3
n n
P(D) = D U. (D)M(D.) + £ V. (D)M'(D.) (4)
i=0 x x i=0 x i
where M(D.) and M' (D^) are the known values of mass concentration
and its derivatives at D0 and Dx? thus n has values of 0 and 1
only. The functions U.(D) and V. (D) can be expressed in terms
of the Lagrange multipliers L. (D) and their derivatives L.'(D)
such that * i
U.(D) = [l-2L.' (Dj) (D-D.)] [L.(D)J
(5)
28
-------
V^D) - (D-Di) [^(0)1 (6)
where
n
n "Dj _ i-= 0,1,2, ---- ,n. (7)
j=0 (D.-D.)
For the two points i = 0, i = 1 we have from Equation 7
T - D ~
L
l " D, - D. (9)
and their derivatives are
Lo'(D) = 1
DO ~ DI (10)
Li'(D) = 1 . (11)
DI - D0
For the points i = Of i = 1 Equation 4 becomes
P(D) =U0(D)M(D0) +01(D)M(D1) +V0(D)M'(D0) + Vx (D)M' (Dl ) .
(12)
Combining Equations 5 and 6 with 8, 9, 10, and 11, the osculating
polynomial then becomes
- D
-------
+ f(D ~ Do) (D - Pi)2 M'(Do)
L (Do - DI)*
+ f(D " Pi) (D - Dn)2 M'(D i)
"*" L (Di - Do) 2 (13)
Equation 13 can be put into a simpler form by defining the
following constants:
u - M(D0)
kl - -
2M(Dn) „ M(D,
(Do - DJ * ' 2 " (Dfl - D
,) , M1 (Dn) k = M' (Pi)
= ' :
(14)
a3
k3 - (2 D! + D0 ) (k. + ks ) - (2D0
= [(k2 + ks) (Df + 2 D0Di) + (k^ + k 6) (Df
k
6
- 2k1D1 -
a* = [kiD? -i- k3D§ - D0Df(k2 -i- k5) - D§DX (k^ + ks)] .
In terms of these constants, the first order osculating polynomial
can be written as a cubic equation in particle diameter D such
that
P(D) = axD3 + a2D2 + a3D + a^
and its first derivative is
P'(D) = 3aiD^ + 2a2D -)- a3.
( J-O )
30
-------
Property 7 above for impactor data limits the acceptable
solutions to 15 and 16 to those for which
p' (D) > 0, D 0 < D
(17)
For some combinations of impactor data, particularly those
for which the slope M1(Do) of the cumulative mass distribution
at the first D50 is large, property 7 places an upper limit on
the range (D0 to D!) of permissible particle diameters. For
those cases, physically acceptable solutions can be found for
a reduced range of particle diameters wherein the maximum size
is somewhat smaller than the original Dx or assumed maximum par-
ticle size. Figure 10 illustrates this. The first fitting
osculating polynomial has the proper cumulative mass loading
value at the first D50 and at the maximum particle diameter.
The first derivative of the function at these two particle sizes
is also correct, i.e., the first derivative of the osculating
polynomial with respect to log diameter is the same as that for
the spline fit at the first Dso and is zero at the maximum par-
ticle size. However, since the osculating polynomial does have
negative first derivative values for Do^D^Di (where Dj, here
equals DMAX), the Di is "stepped" to a smaller value, and a
second osculating polynomial is fitted. This second osculating
polynomial is then tested for negative first derivative values
for Do^D_
-------
o
5
<
o
V)
1
Ul
I
o
1ST FIT (THROUGH
ORIGINAL DMAX)
FINAL (CRITICAL)FIT
(2SPT)
D,
(DMAX)
PARTICLE DIAMETER
S4181-11A
Figure 10. Fitting the cumulative mass distribution for Df1) < D 0 for DQ
-------
of the particulate distribution which were individually varied
were the mass median diameter MMD, the geometric standard devia-
tion cr , and for bimodal size distributions, the fractional distri-
bution9of mass in each mode. Also varied were the assumed largest
particle diameter DMAX and the geometric standard deviation of
the stage efficiency curves a . Since there were no experimental
y o
errors or uncertainties, the size distributions calculated from
the modified CIDRS program are a true test of the extrapolation
technique.
Testing with Unimodal Particle-size Distributions
For unimodal testing a log-normal size distribution was
used with MMD = 5.0 urn and a - 2.5 unless otherwise specified
in the text. Also, unless otherwise specified in the text, DMAX
was input as 100.0 ym, and the masses were collected using log-
normal stage efficiency curves with or value of 1.06 to simulate
g&
impactor behavior using greased substrates, and with a a of
1.3 to simulate impactor behavior using glass fiber substrates.
Variation of the Particulate Mass Median Diameter—
The polynomial used in conjunction with the modified spline
technique in CIDRS shows a high degree of accuracy in recovering
the cumulative inhalable particulate concentration (<15.0 urn)
for a wide range of MMD values. Figure 11 illustrates this with
a plot of the ratio of the recovered inhalable particulate to
the true inhalable particulate/ IP_/IPT/ vs. MMD for stage effi-
ciency curves with a values of 1.3 and 1.06. For MMD's of
gs
1.0, 2.0, 5.0, 10.0, and 20.0 pm, the recovery is near a perfect
1.00 except at one point. For a _ of 1.3 the IP_,/IP_ ratio is
y S j\ X
at its highest value of 1.06 at an MMD of 20.0 urn. This greater
error in IP recovery may be caused by assuming a largest particle
diameter (100.0 \m here) which is too small with respect to the
20.0 ym MMD.
Variation of the Geometric Standard Deviation of the Particulate
Size Distribution—
To test the sensitivity of this fitting technique to different
values of the aerosol geometric standard deviation, a , unimodal
particle-size distributions having a values of 1.5 to 3.5 were
tested. Figure 12 shows the ratio of recovered over true inhalable
particulate IPR/IPT vs. these values of cr for stages having
collection curves with geometric standard deviations of 1.3 and
1.06. The technique shows good accuracy in recovering the true
IP concentration. The highest ratio value here is 1.04 and a
relatively sharp mass distribution with standard deviation of
1.5. The accuracy increases as the particle-size distribution
broadens.
33
-------
a.
LU
D
OC
o7
tECOVERED
I.VO
1.05
1.04
1.03
1.02
1.01
0.99 i
1 1 1 T
ASSUMED DMAX = 100.0 fjan
_ ag = iS _
LEGEND
O Ogs " 1.3 -
H 0gj = 1.06
—
-
-OB
i i • i
1.0
2.0
5.0 10.0 20.0
MASS MEDIAN DIAMETER (MMD), urn
100.0
S4181-12
Figure 11. Ratio of recovered to true IP concentration versus mass median
diameter of a unimodal log.~normaf particle-size distribution.
Variation of the Geometric Standard Deviation (Slope) of the
Collection Efficiency Curve—
The geometric standard deviation of the collection efficiency
curve of the sampler, a, ranges from a near perfect cut value
of 1.01 to 1.7 in this test. As shown in Figure 13 the variation
of a has little effect on an accurate recovery of the inhalable
particle concentration. The maximum error for IP recovery occurs
for a sampler efficiency a of 1.01. Here the recovered over
true inhalable particulate ratio, IPR/IPT/ is only 1.02.
Variation of the Assumed Largest Particle Diameter in the Dis-
tribution—
In this test the assumed largest particle diameter, DMAX,
is varied from 20.0 ym up to 999.0 ym for the same unimodal log-
normal size distribution. The results, as seen in Figure 14, show
34
-------
2: 1.04
•D
oc
i 1.03
O
UJ
CC
> 1.02
0
0
IU
cc
1.01
1.00
0.99
I i i i 1 1 1
ASSUMED DMAX = 100.0 urn
MMD = 5.0 Mm
• LEGEND
O ffgs " 1-3
0 • *>, -LOB
- H —
ODD
i i i i i i i
0,0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.
GEOMETRIC STANDARD DEVIATION, a_ S4isi-i:
Figure 12. Ratio of recovered to true IP concentration versus aerosol geometric
standard deviation of a unimodal log-normal particle-size distribution.
01
D
CC
£ 1.02
Q
LU
CC
UJ
o
111
cc
1.01 —
1.00
0.99 11
1 1
A
- A '
1 t
1 I 1
ASSUMED DMAX = 100.0 jum
MMD = 5.0 urn
0g = 2.5
A A A
I I I
1.01 1.06
1.3 1.5 1.7
STAGE EFFICIENCY STANDARD DEVIATION, oa
S4181-14
Figure 13. Ratio of recovered to true IP concentration versus stage efficiency standard
deviation using a unimodal log-normal particle-size distribution.
35
-------
a.
Ul
CC
a
Ul
cc
Ul
o
0
Ul
cc
I.U/
1.06
1.05
1.04
1.03
1.02
1.01
1.00
0.99
1C
f III 1
MMD = 5.0 (Jim
0g - 2.5
LEGEND "~
O ag$ = 1.3
• «„-,. 06
"
• a -c
a a o
i i t i i
1.0 20.0 50.0 100.0 200.0 500 100C
ASSUMED LARGEST PARTICLE DIAMETER, DMAX, urn 54181-15
Figure 14. Ratio of recovered to true IP concentration versus assumed largest
particle diameter using a unimodal log-normal particle-size distribution.
36
-------
that accurate IP recovery is relatively insensitive to assumed
DMAX unless it is extremely under-valued. In this distribution
the ratio of recovered to true inhalable particulate IPR/IPT
has its highest values of 1.07 (for a = 1.06) and 1.06 (for
y "
a = 1.3) for an assumed DMAX of 20.0 ym. However, for this
gs
size distribution having a mass median diameter of 5.0 ym, an
assumed DMAX of 20.0 ym is unreasonably low. The other IPR/IPT
values are on the order of 1.01 to 1.02. It is concluded then
that only an extremely low guess of DMAX might cause large errors
in IP recovery. This data reduction method still produces good
IP recovery if the assumed DMAX is over approximated.
Testing with Bimodal Particle-size Distributions
For bimodal testing a log-normal size distribution is used
having mass median diameters of 2.0 ym (MMDi) and 15.0 ym (MMD2)
and geometric standard deviations, a, and
-------
Ul
CC
£ 1.00
o
Ul
ee
Ul
o
Ul
cc
0.95
0.90
0.85
I I I
ASSUMED DMAX = 100.0 n™
MMDi = 2.0 Aim MMD2 = 15.0 urn
FRACTIONAL CONTRIBUTION OF MODES: 0.50/0.50
LEGEND
1.3
1.06
1.0
1.5 2.0 2.5 3.0
GEOMETRIC STANDARD DEVIATION (org1 = ag2)
S4181-16
Figure 15. Ratio of recovered to true IP concentration versus aerosol geometric
standard deviation of a bimodal log-normal mass distribution.
38
-------
1.05
a.
UJ
3
CC
a
a
UJ
cc
UJ
01 nn
l.UU
O
UJ
CC
0.95
I I I I I
ASSUMED DMAX = 100.0 jum
MMDf = 2.0 jum
_ A ft
FRACTIONAL CONTRIBUTION OF MODES: .50/.50
— O -
LEGEND
0 crg$ = 1-3
g ffgs = 1.06 •
a
_
| o .
1 1 1 1 I
1.0 5.0 10.0 15.0 20.0 30.0 100.0
SECOND MODE MASS MEDIAN DIAMETER, MMDo, l*m S4181-17
Figure 16. Ratio of recovered to true inhalable particle concentration versus second
mode mass median diameter of a bimodal log-normal particle-size
. distribution.
39
-------
5.0 to 30.0 ym. The highest value of IPR/IPT of 1.05 occurs
for an MMD of 30.0 ym. This may be attributed to the large amount
of mass in the second mode which falls beyond the first D50.
Variation in Fraction of the Total Mass Contained in Each of
the Modes—
In this test the ratio of mass contributed by each of the
two modes in the bimodal distribution was varied. Beginning
with a 25% contribution of mass in the first mode and a 75%
contribution of mass in the second mode, the ratio was changed
on each trial with a greater fraction of mass being contributed
by the first mode in each successive trial. Figure 17 shows
that the recovery of inhalable particulate less than 15.0 urn
of this bimodal distribution is acceptable regardless of the
ratio of modes. The ratio of recovered to true inhalable particu-
late IP-./IP-, is at its lowest value of 0.931 for glass fiber
K J.
substrates where the ratio of first mode mass to second mode
mass is 0.25/0.75. As mass becomes less concentrated at sizes
greater than the first stage D50, i.e., as the mass mode ratio
approaches 0.75/0.25, IPR/IPT approaches a perfect 1.00.
Effect of Operating Problems—
Further tests were conducted in an attempt to estimate the
effect of operating problems. Some questions that might arise
are: (1) were the impactors properly calibrated and operated?
(2) was the largest particle diameter known? and (3) was a
buttonhook nozzle or improperly calibrated cyclone precollector
used?
Although it is not possible to determine exactly how large
the errors are that might be introduced by the three parameters
mentioned above, it is possible to determine the sensitivity
of the calculated IP concentration to them, and thus establish
probable bounds for the errors.
For the sensitivity tests, data from two sources were chosen.
One was a coal-fired power plant and the other a cement kiln.
Both sources employed electrostatic precipitators as pollution
control devices.
Size distributions were calculated for the aerosols at the
control device inlet and outlet at each source using theoretical
stage constants and constants determined at Southern Research
Institute by calibration. A range of values for DMAX was used.
It was further assumed that the first, and possibly the second,
stages of the impactors were interfered with by collection in
a buttonhook nozzle or improperly calibrated cyclone. To elimi-
nate the latter as an interference, the extrapolation was started
from the second and third stages, ignoring the first or second
as appropriate. This of course, results in a loss of resolution
as a compromise to eliminate possible errors.
40
-------
1.00
a
ec
s
a
UJ
oc
> 0.95
O
O
UI
cc
0.90
II I I I
0 •
0 •
g ASSUMED DMAX = 100.0 urn
O MMD-j = 2.0 urn MMD2 = 15.0 jum
O ag1 = 2.0 ag2 » 2.0
"" • ""
LEGEND
II I I I
0.25 0.35 0.50 0.65 0.75
0.75 0.65 0.50 0.35 0.25
FRACTIONAL CONTRIBUTION OF FIRST MODE. Fi »
FRACTIONAL CONTRIBUTION OF SECOND MODE. F2
S4181-18
Figure 17. Ratio of recovered to true IP concentration versus fractional
contribution of mass from each mode of a bimodal log-normal
mass distribution.
41
-------
Figures 18-21 are typical results from the study. In this
series of graphs, impactor stage constants determined by calibra-
tion at Southern Research Institute were used for the impactors
and the DMAX was chosen to be 100 \m at the inlet and 20 ym for
the aerosols at the outlet of the precipitator. As shown,
extrapolations were made from the 1st, 2nd, and 3rd D50's of
the impactors.
Similar tests were made with theoretical impactor stage
constants, and a range of DMAX. In all tests, the variations
in calculated IP are less at the control device outlet than the
inlet as the parameters are changed, as expected.
Figures 22-25 summarize the results from the sensitivity/
accuracy analysis. In these figures the calculated IP concen-
tration is plotted versus the assumed largest particle diameter.
Data are shown for uncalibrated impactors, calibrated impactors,
and with the first and second DSQ'S omitted from the curve fit.
Although these results are also source specific to some extent,
the emissions from coal-fired boilers and cement kilns are quite
different and allow some estimate of the variation from source
to source.
Referring to Figure 22, note that the difference between
the IP concentrations calculated assuming calibrated and uncali-
brated impactors is small. However, omitting the first stage,
or first two stages, from the fit makes a large difference.
It can be seen here, as in the theoretical challenge, that the
choice of DMAX does not introduce large errors.
In Figure 23, outlet data of the same size are treated.
Since there is little mass above the largest Dso, the calculated
IP concentration is rather insensitive to all of the test para-
meters.
Data taken at a cement kiln are shown in Figures 24 and 25.
Since this source has a much larger size distribution it is not
surprising to see a stronger variation in the calculated IP con-
centration at the inlet to the precipitator, as the parameters
are changed. At the outlet, there is little variation in the
calculated IP concentration because the control device removes
most of the larger particles.
The results of this study indicate that, although it is
desirable to make an accurate estimate of DMAX, it is not vital
and should not greatly affect the accuracy of IP concentrations
calculated from data stored in the EPA's Fine Particle Emissions
Information System (FPEIS). Also, it is indicated that valid
data can be recovered, even in tests where the impactors were
uncalibrated. By far the largest uncertainty, other than improper
operation and unidentified experimental errors, is introduced
by the lack of knowledge with respect to the types of nozzles
42
-------
t
"-*
PO
r
(V
TJ g
H *
M
£
S K
CJ
.
§
Q>
•t
M
n
jf
Co
3
CJ.
•§;
G
8 -
SI-
CUMULATIVE MASS LOADING (MG/ACM)
e-
1—(—i
MINI
•i—i—i i 111ii
~3
CUMULATIVE MASS LOADING (GR/ACF)
-------
1O4
103--
a
a
102-
101
Q ALL DBQ'S FITTED
O 1st Dgo OMITTED
& 1st & 2nd DSQ OMITTED
I I I Hill 1—I I I HIM 1—I I I Hll|
1CT1 10° 101 102
PARTICLE DIAMETER (MICROMETERS)
4181-340
Figure 19. Differential particle-size distribution at inlet to electrostatic precipitator.
44
-------
en
"
6
q-
p-
CUMULATIVE MASS LOADING (MG/ACM)
9,
H 1 1 t I I |
H 1—I—I I I |
1
1
-k.
-U
1 1 1 1 1 T 1 I I
q
us
O
3
m
O
1 1
1 ( — t — I — r— J
H-
q
ID
CUMULATIVE MASS LOADING (GR/ACF)
-------
101--
10°-
D ALL DBQ'S FITTED
O 1st DSQ OMITTED
A 1st & 2nd DSQ OMITTED
\J
^—i i i 111H 1—I M HIM 1—II MM H
10"1 10°
PARTICLE DIAME-
101 10s
(MICROMETERS)
4181-342
Figure 21. Differential particle-size distribution at outlet of electrostatic precipitator.
46
-------
214 , 1 1 p
17.0
O THEORETICAL STAGE CONSTANTS
A CALIBRATED STAGE CONSTANTS
D CALIBRATED STAGE CONSTANTS
(1st 050 OMITTED IN FIT)
V CALIBRATED STAGE CONSTANTS
(1st, 2nd DBQ OMITTED IN FIT)
x
'I
I
z
o
i= 13.0
DC
z
LU
u
o
o
fe 9.0 _
0 ° o a o
3
8 e »
O 0
5.0
F/flfw/ie 22. Calculated IP concentration vs. assumed DM AX at electrostatic precipitator
inlet Coal-fired power plant.
0 200 400 600 800 1000
LARGEST PARTICLE DIAMETER, Mnt
4181-343
47
-------
7.0
«*>
O
< 6.0
CONCENT
0.
5.0
4.0
1 1 1 1 1
O THEORETICAL STAGE CONSTANTS
& CALIBRATED STAGE CONSTANTS
O CALIBRATED STAGE CONSTANTS
(1st D5o OMITTED IN FIT)
V CALIBRATED STAGE CONSTANTS
V (1st, 2nd Dgo OMITTED IN FIT)
o ;
— — .
0 X
— ~ o —
o
a
I I I I ?
0 20 40 60 80 100
LARGEST PARTICLE DIAMETER, fim
Figure 23. Calculated IP concentration vs. assumed
outlet. Coal-fired power plant.
4181-344
at electrostatic precipitator
48
-------
*o 7.0
X
m
~Si
£
2*
O
K 5.0
<
oc
1-
z
111
O
1
2: 3.0
1 0
I I
O
6
„ m,
a
V
^7
^7
_
D ^
_»__
a
a
A
0 ft o
I I
0 200 400
I I I
THEORETICAL STAGE CONSTANTS
CALIBRATED STAGE CONSTANTS
CALIBRATED STAGE CONSTANTS
(1st DSQ OMITTED IN FIT)
CALIBRATED STAGE CONSTANTS
(1st, 2nd DSQ OMITTED IN FIT)
_
57
>7 f^ _— HMMMM
V ^
° a ' o D a o
^^ fcji
I ° ? fi ?
600 800 1000
LARGEST PARTICLE DIAMETER, urn
4181-346
Figure 24. Calculated IP concentration vs. assumed
inlet Cement kiln.
at electrostatic precipitator
49
-------
22.5
21.5
O
IU
O
O
20.5
19.5
18.5
A
I
O THEORETICAL STAGE CONSTANTS
£ CALIBRATED STAGE CONSTANTS
D CALIBRATED STAGE CONSTANTS
(1st D5Q OMITTED IN FIT)
V CALIBRATED STAGE CONSTANTS
(1st, 2nd 050 OMITTED IN FIT)
a
o
o
a
O
a
0
a
40 80 120
LARGEST PARTICLE DIAMETER, urn
160
200
4181-345
Figure 25. Calculated IP concentration vs. assumed DM AX at electrostatic precipitator
• outlet. Cement kiln.
and precollectors that were used. The difference in calculated
IP when the first DSO point was included and omitted in these
studies was about a factor of two. It is therefore highly
desirable that the effect of using buttonhook nozzles and cyclone
precollectors is determined quantitatively before retrieving
IP data from FPEIS in order to minimize the probable error.
New data—It has been demonstrated in the previous section
that the modified CIDRS program can be used to accurately calcu-
late the IP concentration from raw impactor data if no experimental
50
-------
errors exist; that is, the program itself does not appear to
introduce large errors. If the sampling protocols outlined in
the following sections are used, CIDRS can be used routinely
to yield the IP concentration. Again, as new sampling systems
become available, no extrapolations will be required.
Impactor data were also subjected to a modal analysis at the
University of Minnesota, in which the data were fitted with multi-
component log-normal distributions by a simplex minimization method,
Data were extrapolated to 100 ym particle diameter by fitting the
portion of the size distribution for which data were available.
Results were similar to those obtained with the curve-fitting
polynomial procedure. IP concentrations could be estimated within
a factor of two or better. The results of this analysis are given
in Appendix B.
ISOKINETIC SAMPLING BY CONSTANT FLOWRATE PARTICLE-SIZING DEVICES
The research at Southern Research Institute on a sampling
system for inhalable particulate matter includes the consideration
of concepts for maintaining isokinetic sampling at constant flow-
rates in the sizing device. Isokinetic sampling is necessary
to obtain a representative sample of suspended particles, notably
those larger than 2 ym in diameter.1*
Various techniques for achieving this goal have been proposed
or used in previous work. Isokinetic sampling is maintained
in EPA Method 5 by varying the flowrate through the filter in
the sample train.5 A single fixed-diameter nozzle is generally
used throughout the sampling to avoid interrupting the tests
to change nozzles.
For a sizing device such as a cyclone or an impactor, the
particle sizing characteristics are a function of flowrate.
Consequently, these devices are generally operated at constant
flowrates.6 The velocity in the nozzle is set equal to the
average velocity in the duct at the point where the sample is
taken, or to the average velocity at points on a sampling traverse.
However, sampling at the average velocity of the gas stream is
not accurate enough to yield a good measure of the concentration
of inhalable particulate matter.
Several techniques and devices that were considered for
sampling isokinetically at a constant flowrate in the sizing
and measuring device are described in this section of the report,
and their advantages and disadvantages are discussed. These
items are: (1) variable area nozzles, (2) split-stream probes,
(3) combinations of (1) and (2), (4) gas recycle systems, and
(5) probe shrouds.
51
-------
All these techniques involve measurement of the gas velocity
in the flue, and they all require methods for producing, control-
ling, and measuring mechanical motion of parts of the apparatus
from outside the stack. This is probably the most difficult
problem to be solved in the design of such a device. The sampling
device must be able to withstand the harsh environment in the
flue or stack, which may contain corrosive gases, abrasive dusts,
and moisture, often mixed with condensed acids, all at high
temperatures and velocities.
The sampler must also be able to withstand the thermal shock
of being placed in and removed from this environment into an
ambient environment, which may consist of rain, snow, or ice
at freezing temperatures. Some mechanical designs may require
that all parts be made of materials that have the same.thermal
coefficient of expansion. The mobility of the parts and the
integrity of seals are limited or destroyed by accumulated dust,
especially in industrial process streams containing sticky or
abrasive particles.
The isokinetic sampling device must be constructed so that
small adjustments are predictable and accurate. In the case
of a variable area nozzle, for example, it is essential that
the sampling area is known exactly. As the nozzle is closed,
a small change in radius of a round nozzle or width of a rectan-
gular nozzle produces a large change in total area.
Three possible methods of producing mechanical motion inside
a flue at the end of a probe are a sliding rod or cable, a rotat-
ing rod or cable, and a fluid-pressure to mechanical-motion
transducer. A rod or cable attached to the device at the end
of the probe and extending outside the stack could operate the
device through a push or pull by the operator. Alternatively,
the rod could be twisted if a rotary motion were needed or perhaps
to give a finer adjustment. The use of a fluid under pressure
to move a small piston on the device would eliminate the need for
moving parts extending outside the probe. Such a fluid would have
to perform adequately at stack gas temperatures. In this case,
a separate measurement of the piston position would be necessary.
All of these -concepts have obvious and subtle problems and would
require further investigation to determine which is best.
Sampling Concepts
Sampling concepts which allow isokinetic sampling with con-
stant flowrate devices have been examined previously by Blake'
and Elder, Littlefield, Tillery and Ettinger.8 In this report,
the concepts are classified as those which require mechanical
motion at the probe such as the variable area nozzle and the
split stream probe, and those which do not, such as the probe
shroud and the gas recycle concepts.
52
-------
Variable Area Nozzles—
The sampling velocity at the nozzle is the sampling flow-
rate divided by the area of the open face of the nozzle. Since
the flowrate must be kept constant for inertial sizing devices
such as impactors or cyclones, the sampling velocity may be
changed by varying the area of the nozzle. The five techniques
discussed below all include the problem of mechanical motion
as a prime concern and the method will be a desirable technique
only to the extent to which this problem can be solved. The
five nozzles are the rectangular, the diamond or lozenge, the
round or iris, the ellipse, and the multiple nozzle revolver
configuration.
Rectangular nozzle—A variable area nozzle with rectangular
geometry is shown in Figure 26. The cross-sectional area of the
nozzle is varied by moving one side of the nozzle, thus varying
the height of the rectangle formed by the edges of the nozzle.
A screw provides the fine adjustment and is connected to a cable
which leads to a control knob outside the stack. A hose clamp on
the nozzle holds the cable and keeps it from twisting when the
screw turns. The screw threads are fine enough that a small
change in the nozzle area is made with each revolution of the screw.
For some sampling situations, a cyclone is used to remove
the large fraction of dust and thus decrease the loading on the
upper stages of the main sizing device. A variable area nozzle
attached to a cyclone is shown in Figure 27 with the cyclone
connected to a probe. The nozzle is attached to the inlet of
the cyclone and a push-pull motion is used to vary the nozzle area.
There is a possibility that in some sampling situations
dust may cause the movable nozzle wall to jam by wedging between
it and the two nozzle walls flanking it. If that should prove
to be a problem, a small clamping device could be used to press
those walls together against the movable wall, releasing them
only momentarily when the nozzle needed adjustment. A small
"lip" directly upstream of the movable wall or flap would help
keep dust out.
The nozzle would have a largest and smallest "realistic"
area. The largest area would be defined by the size of the
nozzle when it was fully opened. The smallest area would be
defined by the sharpness of the nozzle edges, their tapers, and
the ability of the mechanical mechanism to make small adjustments
in the area of the nozzle. For blunter nozzle edges larger nozzle
areas are required to reduce errors associated with impaction
around the nozzle edge and deviations of the flow streamlines.
Diamond nozzle—Figure 28 illustrates a diamond or lozenge-
shaped nozzle. The area is changed by moving one corner away
from or closer to its opposite corner. The area then changes
53
-------
AIR FLOW
LIP
m//.
STATIONARY NUTS
INSIDE OR INLINE
WITH TUBING
SCREW
••»
FLEXIBLE HOSE
\\\\\\\\\Vl-GAUGE
DIAL KNOB
GOOSENECK NOZZLE
FRONT OF IMPACTOR
REAR OF IMPACTOR
TO PROBE
AIR FLOW
4181-347
Figure 26. Conceptual design of variable area nozzle attached to an impactor.
54
-------
NOZZLE
1 SLI
L./
SLIDE
CYCLONE PRECUTTER
SAMPLING PROBE
VARIABLE AREA NOZZLE
TO SAMPLING PROBE
OR SIZING DEVICE
SLIDE
' MECHANICAL
LINKAGE
4181-348
Figure 27. A variable area nozzle attached to a cyclone.
55
-------
GAS
FLOW
GAS
FLOW
FIXED POINT
GAS
FLOW
4181-349
Figure 28. A diamond-shaped nozzle.
56
-------
according to the area of the parallelogram the nozzle describes
(area = base x height). A push-pull mechanical motion is used
with this concept. Since every corner moves, they must be con-
nected with hinges or similar connectors that allow motion. The
size of the connectors will determine whether they will interrupt
the flow streamlines appreciably. Perhaps the major problem
with this concept is designing a transform connection from the
movable to the fixed part of the nozzle.
Ellipse nozzle—The ellipse nozzle shown in Figure 29 is
made of thin spring metal. Two arms squeeze the metal on opposite
sides and deform the original circle into an elliptical or oval
shape. It is doubtful that the various cross-sectional faces
are actually ellipses but their areas may be approximated as
such for design purposes. Calibration of the nozzle would deter-
mine the true area. A nozzle might be made of shim stock of
alloys such as Inconel.
Multiple nozzle revolver configuration—The multiple nozzle
revolver configuration is not a continuously variable nozzle
but a set of variously sized nozzles which can be interchanged
while sampling. Figure 30 illustrates an example. Before sam-
pling, nozzles are selected which will allow isokinetic sampling
at a range of velocities around the average velocity in the flue,
and they are connected to a spoked wheel attached to the probe.
The operator then selects the correct sampling nozzle to match
the measured gas velocity by adjusting a mechanical linkage
extending outside the stack wall. Nozzle selections are made
to maintain isokinetic sampling within a certain percentage error.
For flues which do not have great changes in velocity, the nozzles
are selected so that anisokinetic errors are very low. For example,
for a sizing device sampling at a constant flowrate of 28.3 8,/nin
(1.0 ft3/min) in a flue with a gas velocity ranging from 6.2 to
12.3 m/sec (20 to 40 ft/sec) the nozzles selected would be 7.0,
7.5, 8.0, 8.5, 9.0, and 9.5 mm in diameter. The gas velocity is
given for each sampling point in Table 7. The operator adjusts
the nozzle set so that the 8.0 mm nozzle is the initial sampling
nozzle. After sampling the necessary time at point one, the
probe is moved to point two and the operator uses the mechanical
linkage to adjust the nozzle set so that the 7.0 mm nozzle is the
sampling nozzle. Remaining sampling points are sampled with the
appropriate nozzle indicated in Table 7. The adjustment occurs
essentially instantaneously and the seal between the sampling
head and the sampling nozzle is obtained by close tolerance machin-
ing. Since the nozzles are hollow, flow disturbance around the
sampling nozzle is minimized. As calculated by Reference Method
5 Subsection 6.11, the isokinetic variation is given in Table 7
and is well within the limits required of Method 5 tests
(90%^I^110%).2 Using this particular set of nozzles, the maximum
isokinetic variation in sampling at any velocity from 6.1 to 13.0
m/sec (20 to 43 ft/sec) is 107.5%. The use of more nozzles per
57
-------
TUBING
SHIM STOCK
RING'
CANTILEVER
H H
MOTION OF RING
NOZZLE FACE
s. /
.X"
— — \
. — __ ^
RING
CANTILEVER
4181-350
Figure 29. Conceptual drawing of an elliptical nozzle.
58
-------
SIDE VIEW
FRONT VIEW
4181-351
Figure 30. A multiple nozzle revolver configuration.
59
-------
TABLE 7. SAMPLING PARAMETERS FROM A HYPOTHETICAL SAMPLING
TEST USING THE MULTIPLE NOZZLE REVOLVER CONFIGURATION
Sampling Point
Nozzle Used
1 2 3 45 6 7 89
8.0 7.0 7.0 7.0 7.5 7.5 8.0 8.5 8.5
10
9.0
11
9.5
12
9.5
(diameter, in mm)
i
Sampling Velocity 9.28 12.13 12.13 12.13 10.56 10.56 9.28 8.22 8.22 7.34 6.58 6.58
(ra/sec)
a\
° Gas Velocity 9.9 11.7 12.3 11.9 11.1 10.5 9.7 8.6 7.9 7.2 6.7 6.2
(m/sec)
Error in Sampling -6.3 +3.7 -1.4 +1.9 -4.9 +0.6 -4.3 -4.4 +4.1 +1.9 -1.8 +6.1
Velocity (%)
Eer Cent 93.7 103.7 98.6 101.9 95.1 100.6 95.7 95.6 104.1 101.9 98.2 106.1
Isokineticity
-------
set (7, 8, or 9 instead of 6) would allow reduction of the iso-
kinetic variation or extension of the velocity range.
Iris nozzle—The round or iris nozzle is only mentioned
briefly here as a design concept. The nozzle consists of two
or more very thin strips of a suitable metal alloy held together
in an overlapping conical shape (see Figure 31). The nozzle
face would be opened or closed by adjusting the position of the
collar. As the collar is moved toward the face of the nozzle,
the metal leaves would spring outward and increase the face area.
The face area would decrease as the collar is moved in the oppo-
site direction.
The major difficulties with this concept are the design
and engineering problems. The overlapping leaves.may experience
metal fatigue due to thermal and mechanical stress and cease
operating properly. It is necessary that the leaves retain their
tendency to be flat so that they will spring out as the collar
is moved toward the face of the nozzle. The collar must be
designed so that it slides smoothly along the nozzle and does
not interfere appreciably with the gas flow at the nozzle face.
In-Stack Split Stream Probe—
An in-stack, split-stream probe would allow true isokinetic
sampling at the nozzle and yet maintain a constant flowrate
through the sizing device. It does not require the use of a
separate pump and cooling and condensing components. Instead, a
valving arrangement adds cleaned stack gas to the sample gas
stream at the rate needed to maintain a constant flowrate. Only
the nozzle would need to be modified, and it could be used with
any sizing device. The nozzle probably would not need to be
larger than the sizing device in cross-section and thus not
require a larger port than normally needed.
A schematic showing the general operation and necessary
components of the split-stream probe is shown in Figure 32. The
flowrate into the sizing device is set at a constant value. The
nozzle is chosen to be a size such that isokinetic sampling can
be maintained throughout the test. The minimum velocity that can
be sampled isokinetically is equal to the sizing device flowrate
divided by the.cross-sectional area of the nozzle. The practical
maximum velocity is that which does not cause unacceptable particle
deposition in the nozzle and on the control valve and which does
not lengthen sampling time unreasonably. As stack gas velocity,
pressure, and temperature change and as the dilution air flowrate
changes (due to stack gas changes or dilution filter load-up),
an automatic monitor/controller adjusts the control valve to
maintain isokinetic conditions.7'9'10 At the minimum stack gas
velocity, the control valve closes and no dilution air flows.
At the maximum stack gas velocity, the control valve opens so
that the dilution air is a substantial percentage of the total
flow into the sizing device.
61
-------
OVERLAPPING LEAVES
LEAF NO. 1
MECHANICAL
LINKAGE
LEAF NO. 2
COLLAR
LEAF NO. 3
STATIONARY PART
OF NOZZLE
4181-352
Figure 31. An iris nozzle.
62
-------
STACK GAS TEMPERATURE-
STACK GAS PRESSURE-
STACK GAS VELOCITY-
FLOW METER SIGNAL-^
IN-STACK
FILTER
CONTROL
VALVE
SIGNAL
PITOT
TO SIZING
DEVICE
•STANDARD NOZZLE
4181-353
Figure 32. In-stack split-stream probe, from Blake.?
63
-------
Several observations can be made about this concept. As
pointed out by Blake7, the design and development of the moni-
tor/controller would be expensive and the volume of gas sampled
is not directly measured. Sampling in dusty streams could cause
the pressure drop across the dilution filter to increase and
thus necessitate constant adjustment of the control valve, even
when the stack gas velocity is constant. For extremely dirty
streams and a large variation of stack gas velocity from the
constant gas velocity entering the sizing device, the dilution
air filter could become plugged and interrupt the test. Since
the control valve must restrict the sample flow in order to
increase the dilution air flow, there will be wall losses on the
control valve and other nearby parts. High temperature- and
corrosion-resistant components for the filter housing, the flow-
meter, and the control valve will be necessary. No separate
cooling system or pump is required, and except for a slightly
longer and larger nozzle, the sizing device and probe would not
be modified.
Sliding-cone probe—One type of split-stream probe, shown
in Figure 33, uses a sliding cone to vary the amount of sample
gas pulled into the nozzle. As the cone slides in and reduces
the amount of sample gas, small holes are spaced around the wall
of the probe (or a small screen is used) to allow dilution air
to enter. One advantage is that the dilution air can form an
envelope around the sample air, thus minimizing wall losses beyond
that point. Of course, the wall losses at the cone where the
sample air enters an annular ring may be unacceptable at certain
sampling velocities. Considerable thought would have to be given
to the mechanical linkage that would adjust the position of the
slide.
Adjustable-flap probe—Another type of split- stream probe
is shown in Figure 34. It consists of a rectangular probe with
two "flaps" which adjust the ratio of dilution air to sample
air by opening and closing. A small bar which slides along the
length of the probe can be used to adjust the setting. The dilu-
tion air would form layers on two sides of the stack gas stream
instead of completely surrounding it. Otherwise, this design is
much like the. sliding cone probe.
Split-stream nozzle—A concept incorporating both the variable
nozzle and the split-stream probe concepts is illustrated in
Figure 35. The gas is sampled isokinetically by a fixed area
nozzle and then split into two gas streams. One stream is pulled
at a constant flowrate through a sizing device. The other stream
flowrate is adjusted so that the nozzle face velocity equals
the stack gas velocity. The excess gas is drawn out of the main
flow "isokinetically". Two sampling trains are used: one for
the sizing device and the other for the excess gas. The excess
gas is filtered and provides a bulk sample for chemical analysis.
The main advantages of this concept over the split-stream probe
64
-------
SAMPLE AIR
r
NOZZLE
DILUTION AIR
FLOWMETER FILTER
SLIDING CONE
MOVEMENT
Figure 33. Split-stream probe: sliding-cone probe.
DILUTION AIR
4181-354
I NOZZLE
tti
STACK GAS
FLOW
FILTER FLOWMETER
\
SAMPLE GAS
DILUTION AIR
DILUTION AIR
FLAP
SAMPLE GAS
FLAP
END VIEW
(MINUS NOZZLE)
SLIDING BAR
CONSTANT FLOW
TO SIZING
DEVICE
SLIDING BAR
4181-355
Figure 34. Split-stream probe: adjustable flap probe.
65
-------
GAS
STREAMLINES
r
NULL PROBE
CONSTANT FLOW RATE •
TO SIZING DEVICE I
AND FIRST SAMPLING
TRAIN I
EXCESS GAS
VARIABLE FLOW RATE
TO FILTER AND
SECOND SAMPLING
TRAIN
4181-356
Figure 35. A split-stream nozzle.
66
-------
are: there is no need for in-stack filters and flowmeters; there
is no particle loss due to the introduction of dilution air;
and the sampling time is reduced since the stack particle concen-
tration is maintained. The advantage this concept has over the
variable area nozzle is the location of flow disturbances caused
by mechanical linkages and sampling tubes away from the face
of the nozzle. Disadvantages include the need for a second sampling
train and the fact that the sample gas has been sampled "twice"
essentially, once from the stack gas stream and once from the
split. Also, it might be difficult to operate the movable flap
so that isokinetic sampling takes place in the probe.
Gas Recycle Concept—
Similar to the split-stream probe is the gas recycle concept.
Instead of separately filtering dirty stack gas to use as dilution
air, the gas exiting the sizing device is used as the dilution
air. Thus there is no separate filter to clog, because the sizing
device final filter is cleaning the gas of particles. The gas
can be introduced as an annular envelope around the stack gas
and since there are no restrictions in the line, wall losses
can be decreased. Since part of the gas entering the sizing
device is clean, the sampling time to collect a given amount
of mass would be lengthened.
Hot gas recycle—Figure 36 illustrates the method of recycling
the stack gas while it is still at stack conditions (temperature
and moisture). Stack gas entering the nozzle would be surrounded
by clean dilution air immediately prior to its entrance into
DILUTI
FAN
BLADES SHAFT
SEAL
TO SMALL
ELECTRIC MOTOR
NOZZLE
CONSTANT
FLOW RATE
SAMPLE
FLOW RATE
IMPACTOR, CYCLONE, ETC.
WITH FINAL FILTER
4181-357
figure 36. A hot gas recycle concept.
67
-------
the sizing device. Upon exiting the final filter, a portion
of the flow would be pumped back to the entrance of the sizing
device through a bypass line. This would necessitate that the
port be large enough for both the sizing device and the bypass
line to pass through. The pumping device would have to be capable
of high temperature operations and able to overcome the pressure
drop of the sizing device. The sizing device flowrate and the
sample flowrate are both monitored. Adjustments of the recycle
gas flowrate and the sample flowrate maintain a constant sizing
device flowrate and isokinetic sampling conditions.
Cool gas recycle—A cool gas recycle system is shown sche-
matically in Figure 37. This system incorporates components
that are more readily available than for any system discussed
so far. Similar to the hot gas recycle concept, the cool gas
recycle concept has less stringent requirements on the recycle
pump. The stack gas is sampled isokinetically at the nozzle.
Between the nozzle and the sizing device enough heated, dry air
is added to the sample air to maintain a constant flowrate through
the sizing device. After exiting the sizing device, the clean
gas flows through a regular sampling train where it is cooled,
dried, and measured. Part of the gas is then pumped through
a gas heater which raises it to stack temperature. The gas is
then used to provide the constant flowrate required by the sizing
device by combining with the sample gas flow immediately down-
stream of the nozzle.
The cool gas recycle concept involves one consideration
not present in the other systems. Since the dilution air is
dry, the sizing device flow will consist of a gas with varying
amounts of moisture, all of them less than the moisture in the
stack gas. How this will affect the operation of the sizing
device, and whether it will be a significant effect, is dependent
on what the sizing device is and how it works. An investigation
of the change in stage D50 for an Andersen impactor as a function
of gas moisture content for three different temperatures is
summarized in Table 8. The largest change in stage D50 for a
run in which the moisture content varies from 5 to 20 percent
is 3.4 percent for a gas temperature of 38°C (100°F), 2.7 percent
for 149°C (3.00°F) and 2.3 percent for 316°C (600°F) . The vari-
ation in stage D50 will be smaller for smaller variations in
moisture content.
The major advantage this concept has over the other concepts
is its lack of moving components in the stack environment. The
flow adjustment and pumping of the gas stream can be performed
on the cool gas thus allowing the use of conventional equipment.
All the other items used such as the gas heater, temperature
controller, and piping are stock items except for the flowrate
monitor/controller. Unless the sizing device has an easily
accessible, constant pressure drop component (such as a cyclone)
which can be monitored by a differential pressure controller,
68
-------
REMOTE ACTUATED
VALVE
THERMOCOUPLE
TEMPERATURE
CONTROLLER
MICRO PROCESSOR
4181-368
Figure 37. Cool gas recycle concept.
-------
TABLE 8. CHANGE IN STAGE D50'S FOR AN ANDERSEN IMPACTOR AS A FUNCTION
OF MOISTURE CONTENT OF SAMPLED GAS
Temperature, °C (°F)
Viscosity,
Moisture,
Ds o , Via
micropoise
percent
Stage
SI
S2
S3
S4
S5
S6
S7
S8
177
5
10.59
10.10
6.20
4.17
2.34
1.13
0.727
0.372
38 (100)
173
10
10.47
9.98
6.14
4.12
2.31
1.12
0.719
0.368
169
15
10.35
9.87
6.07
4.08
2.28
1.10
0.711
0.364
165
20 Largest
Difference
(Percent)
10.23 3.44
9.75 3.44
5.99 3.44
4.03 3.44
2.26 • 3.44
1.09 3.42
0.703 3.40
0.360 3.30
223
5
11.86
11.30
6.93
4.65
2.59
1.24
0.791
0.394
149
219
10
11.75
11.20
6.87
4.61
2.57
1.23
0.784
0.391
(300)
215
15
11.65
11.10
6.81
4.57
2.55
1.22
0.777
0.387
213
20 Largest
Difference
(Percent)
11.54 2.66
11.00 2.66
6.75 2.67
4.53 2.67
2.53 2.68
1.21 2.69
.770 2.70
.384 2.73
284
5
13.35
12.72
7.79
5.21
2.89
1.36
0.854
0.408
316
281
10
13.27
12.64
7.74
5.18
2.87
1.35
0.849
0.405
(600)
278
15
13.19
12.57
7.69
5.15
2.85
1.34
0.843
0.402
274
20 Largest
Difference
(Percent)
13.10 1.85
12.49 1.85
7.65 1.86
5.11 1.87
2.83 1.91
1.34 1.99
0.837 2.07
0.398 2.28
-------
a microprocessor will have to be used to calculate the flowrates
at stack conditions from the measured cool, dry gas flowrates.
For example, a commercial cascade impactor does not have any
accessible points at which the pressure drop across a stage could
be measured. However, a small series cyclone system would allow
the pressure drop across any of its components to be read with
a minimum of modification.
In-Stack Probe Shroud—
As shown in Figure 38, this design consists of a standard
sampling nozzle centered in a large shroud.7 Both the nozzle
and the shroud have the same velocity at the inlets. The concept
is based on the fact that the nozzle samples isokinetically from
the shroud gas stream and on the assumption that the part of
the shroud gas stream that enters the nozzle is representative
of the stack gas at that point. It is argued that this system
would work for stack gas velocities both higher and lower than
the shroud velocity due to the fact that the flow streamlines
of the gas entering the nozzle are not disturbed greatly. In
actuality, the flow streamlines entering the nozzle must be bent
but not, perhaps, at the immediate entrance to the nozzle. In
the case of the stack gas velocity being less than the sampling
velocity a larger gas volume is flowing through the nozzle area
than is flowing through an equal, coaxial area located a distance
upstream from the nozzle. This extra volume must come from the
gas immediately outside that area. Thus the flow streamlines
outside that area must bend and flow into that area to give the
higher velocity. In this case the bending may not be as abrupt
or sharp as when a nozzle is used without a shroud, so that the
effect of sampling anisokinetically is reduced. Blake reports
". . .a preliminary evaluation indicates that the concept is
valid over relatively wide stack/shroud velocity mismatches
(3:1 and more) for particles of 30 ym diameter or less."' The
extent to which anisokinetic effects are reduced would have to
be determined by calculation and each user would have to decide
whether the limits of error of the device would be compatible
with the goals of the sampling program.
Other considerations of this method are the relatively large
size of the probe, the need for a large, interchangeable cartridge
filter, a large pump or exhaust blower, a large capacity flowmeter
for the shroud flow, the need to place the in-situ sizing device
inside the shroud, and the need to remove the shroud "nozzle"
in order to remove the sizing device. Unless the shroud gas
is cooled by a cooling component and the moisture is removed
by a condenser, the blower would have to be capable of operation
at stack temperatures. A valving arrangement would be used in
conjunction with a large capacity flowmeter to set the shroud
velocity equal to the nozzle velocity and to correct for increas-
ing pressure drop across the shroud filter as it loads up.
71
-------
STACK WALL
IMPACTOR
STANDARD
SAMPLING NOZZLE
r /
CARTRIDGE FILTER
FLOW STREAMLINES
TO SAMPLING
TRAIN
COOLING AND
CONDENSER COMPONENT
FLOWMETER AND
VALVING COMPONENT
I EXHAUST BLOWER
OR PUMP
Figure 38. In-stack probe shroud, after Blake.?
4181-369
72
-------
SECTION 3
DESIGN, FABRICATION, AND TESTING OF PROTOTYPE
SAMPLING SYSTEMS FOR INHALABLE PARTICULATE MATTER
A variety of aerosol sampling systems have been studied for
use in measuring the concentration of inhalable particulate (IP)
matter.1l The system must be designed to separate particles
larger and smaller than 15 ym in aerodynamic diameter (i.e., have
a D50 value of 15 ym) according to a specified efficiency curve.
This section describes laboratory studies made at Southern
Research Institute to develop suitable IP sampling systems that
will fit EPA specifications. These systems are based on impactors,
cyclones, and elutriators as collectors. All of these devices
have features that limit their applicability. Impactors do not
always retain particles efficiently that are much larger than the
stage cut point (Dso). The particles bounce and are re-entrained
into the gas stream, resulting in contamination of the smaller
particle fraction. 12/13r llf Cyclones are not subject to that
problem, but they are larger and their performance is difficult
to predict.15 Virtual impactors can have significant losses of
small particles.16 Horizontal elutriators are bulky and their
performance is sensitive to orientation. However, re-entrainment
is not a problem in elutriators for reasonable velocities, and
the theory describing their performance is straightforward and
accurate.
SUMMARY
The significant results that were attained may be summarized
as follows:
Tests were conducted with monodisperse aerosols on several
commercially available cascade impactors and buttonhook nozzles
to characterize them for measurements in the IP range of particle
sizes.
Collection efficiencies were measured for a horizontal
elutriator. The measured values agreed well with values calcu-
lated from theory, allowing the preparation of design nomographs
for inlet precollectors in sampling systems. Design parameters
were calculated for horizontal elutriators to be used with cascade
73
-------
impactors, operated at 14.2 Jl/min and 149°C/ the EPA Source
Assessment Sampling System, operated at 185 H/min and 204°C, and
the EPA Fugitive Ambient Sampling Train, operated at 5,282 5,/min
and 23°C.
Two cyclone systems were designed and evaluated: a cyclone
to be used as a precollector for cascade impactors, and a system
of two cyclones and a filter in series, to be used as the primary
system for measuring IP and fine particle concentrations. Both
systems were designed for high-temperature operation in process
streams.
Tests on the Andersen Size Selective Inlet, a 15-ym precol-
lector for hi-vol samplers, showed its performance to be within
the proposed limits for IP samplers.
A sampling system was designed in which the aerosol is
diluted in flow patterns and with mixing times simulating those
in stack plumes. The system is thus designed to characterize
aerosols formed from condensable vapors in stack gas as the plume
is diluted and cooled by the atmosphere. Tests with the system
on a domestic oil-fired furnace indicated that condensation of
organic chemical vapors took place in the system as expected.
EVALUATION OF COMMERCIAL CASCADE IMPACTORS AND BUTTONHOOK NOZZLES-
EFFECT ON MEASUREMENT OF INHALABLE PARTICULATE MATTER
This part of Section 3 describes various tests that were
made on commercial cascade impactors and buttonhook nozzles to
determine their applicability to inhalable particulate measure-
ment. *»
Experimental Procedures
Using the apparatus shown in Figures 39 through 42, various
buttonhook nozzles and commercial impactors were allowed to sample
large monodisperse aerosols to determine their collection effi-
ciency. The buttonhook nozzles tested were 0.32, 0.64, and
0.95 cm (1/8, 1/4, and 3/8 in.) in diameter. The cascade impac-
tors tested were the Modified Brink, the Andersen Mark III, the
Sierra Model 226, the MRI Model 1502, and the University of
Washington Model V.
The particles used in this calibration were generated with a
vibrating orifice aerosol generator (VOAG) using a solution of
methylene blue dye in 1:1 distilled water and ethyl alcohol. The
particles were dried and lofted up through a 30-in. Plexiglas
column and were drawn by a high volume blower into a small Plexi-
glas wind tunnel via another smaller Plexiglas tube. The flow
rates of the blower and pump were set so that sampling could be
accomplished isokinetically.
74
-------
18"-
FILTER
-J
tn
HIGH VOLUME BLOWER
VIBRATING
ORIFICE
AEROSOL
GENERATOR
FREQUENCY
GENERATOR
SYRINGE PUMP LINE
41S1-221
Figure 39. Configuration of apparatus used in evaluation of cascade impactor
and nozzle.
-------
4181-363
Figure 40. Andersen Mark III cascade impactor situated in isokinetic sampling apparatus.
-------
4181-362
Figure 41. I so kinetic sampling apparatus adjacent to vibrating orifice aerosol generator.
-------
00
4181-361
Figure 42. Isokinetic sampling apparatus adjacent to vibrating orifice aerosol generator.
-------
The buttonhook nozzles were allowed to sample for about 1.5
hours in each run, corresponding to one 500-ml syringe of solution
being put through the VOAG. It was necessary to double the
sampling time for the impactors in order to collect a measurable
amount of material on each of the components and stages. In the
tests on the buttonhook nozzles, filters were installed after the
nozzles to collect all penetrating particles. After sampling,
each impactor or nozzle was disassembled and each component was
washed with a known amount of water, thus providing solutions
with concentrations, and hence amounts of dissolved material that
could be determined by absorption spectroscopy. When the compo-
nent was, or contained, a glass fiber filter, it was washed by
placing the filter in water in a container and leaving it in an
ultrasonic cleaner for several minutes. Any glass fibers sus-
pended in the resulting solution were removed by centrifugation.
Results and Evaluation
Buttonhook Nozzles—
The data from the nozzle evaluations are shown in Table 9.
Nozzles larger than 0.95 cm (3/8 in.) in diameter could not be
tested because higher air flow systems were not available. Gener-
ally it can be said that for the same flowrate, larger particles
penetrated the buttonhook nozzles better, probably a result of
higher inertia and subsequent bounce and re-entrainment.
TABLE 9. EVALUATION OF BUTTONHOOK NOZZLES
Nozzle Particle Flowrate, Inlet Velocity, Collection,
Diameter, in. Size, ym £/min m/sec
1/8
1/8
1/8
1/8
1/4
1/4
1/4
1/4
1/4
3/8
3/8
15.2
10.7
14.4
14.1
15.6
16.4
21.7
14.1
14.2
10.6
15.2
7.22
7.22
7.22
10.8
23.2
23.2
23.2
28.3
23.2
56.5
65.1
15.2
15.2
15.2
22.6
12.2
12.2
12.2
15.2
12.2
13.1
15.2
u
27. 6a
33.6
31.9
23.9
17.4
15.6^
2.9b
30.7
42.0
52.7
27.3
a) Difficulty with particle generator
b) Low collection efficiency probably due to large particle bounce
Impactors—
Test data for the cascade impactors are shown in Table 10.
Because of the design of the University of Washington Mark V, it
was considered to have the best chance of having upper stages
79
-------
TABLE 10. EVALUATION OF COMMERCIAL CASCADE IMPACTORS
USING FIFTEEN-MICRON AERODYNAMIC DIAMETER PARTICLES
Impactor
Orientation
Substrate
Nozzle
Nozzle Diameter (inches)
Inlet Velocity (m/sec)
Aerodynamic Particle
Size (microns)
Flow rate (1pm)
Nozzle
g Nozzle Holder
Total Wall Loss
Stage 0
Stage 1
Stage 2
Stage 3
Stage 4
Stage 5
Stage 6
Stage 7
Stage 8
Back-up
Sierra Model 226
Horizontal
Sierra glass fiber
Straight
5/32
8.53
14.5
6.40
% Of
Total
Mass % Eff.
27.8 27.8
9.8 13.6
37.6 37.6
6.8 10.9
7.6 13.7
17.0 35.5
16.7 53.9
7.1 50*
3.6* 50*
1.8* 50*
NA NA
NA NA
1.8* 100*
Comments 'Assumed
Particles observed
microscopically on
all stages
Sierra Model 226
Horizontal
Sierra glass fiber
Straight
5/32.
8.53
15.2
6.40
% of
Total
Mass % Eff.
28.9 28.9
16.9 23.7
45.8 45.8
4.7 8.7
7.4 14.9
26.5 62.8
8.5 54.0
3.6 50.6
3.6 100*
_ -
NA NA
NA NA
— —
* Assumed
Particles observed
microscopically on
each stage
Sierra Model 226
Vertical
Glass fiber
Straight
5/32
8.63
18.3 urn
6.40
% of Total
Mass
24.4
10.8
35.2
NA
3.8
13.2
40.0
5.5
1.9
0.5
-
% Effic.
24.4%
14.3
35.2
NA
5.9
21.7
83.7
70.1
79.7
100
-
Sierra Model
Vertical
Glass fiber
Straight
7/32
9.72
14.5 urn
14.2
% of Total
Mass
35.1
8.0
43.1
NA
4.6
19.6
18.2
9.1
3.0
2.3
••
226
% Effic.
35.1%
12.3
43.1
NA
8.1
37.6
55.8
62.9
56.5
100
«
Water spilled on
Stage 3 during
washdown
- Collected mass too little to measure
NA-stage does not exist in this model
-------
TABLE 10. (Continued)
CO
Impactor :
Orientation:
Substrate:
Nozzle:
Noz. Diara.: (inches)
Inlet Vel. : m/sec
Part. Diam. : ytn
Flowrate : ipm
Nozzle
Holder
Wall Loss:
Stage 0
Stage 1
Stage 2
Stage 3
Stage 4
Stage 5
Stage 6
Stage 7
Comments
MRI Model 1506
Horizontal
Glass fiber
Straight
1/4
8.23
14.6
15.6
% of
Total
Mass % Eff.
7.8 7.8
62.6 67.8
70.3 70.3
NA NA
4.7 15.8
18.1 72.3
6.9 100*
-
-
-
-
NA NA
* Assumed
MRI Model 1506
Horizontal
Grease-
Straight
.25
8.20
14.8
15.6
% of Total
Mass % Effic.
6.9 6.9%
42.3 45.4
49.2 49.2
7.1 13.9
26.5 60.5
14.8 85.8
2.5 100
-
-
-
-
Andersen Mark III
Hor izontal
Andersen glass fiber
Straight
0.24
8.84
14.2
15.6
% of
Total
Mass % Eff.
26.3 26.3
16.3 22.1
42.5 42.5
11.9 20.8
13.9 30.6
9.6 30.6
13.3 60.7
8.6 100*
-
-
- -
— ~
Inter-Stage o-rings
leaked during run.
* Assumed
Andersen Mark III
Horizontal
Andersen glass fiber
Straight
0.24
8.84
14.7
15.6
% of
Total
Mass % Eff.
23.6 23.6
18.9 24.7
42.5 42.5
5.4 9.5
19.5 37.5
25.4 78.1
7.1 100*
- -
-
_
-
- -
* Assumed
-------
TABLE 10. (Continued)
Impactor :
Orientation:
Substrate:
Nozzle :
Nozzle Diameter: (mm)
Inlet Velocity: m/sec
Particle Dianeter: \im
Plowrate: 1pm
Nozzle
Cyclone
Stage 0
Stage 1
Stage 2
Filter
Impactor:
Orientation :
Substrate:
Nozzle:
Nozzle Diameter: (mm)
Inlet Velocity: m/sec
Particle Diameter :ym
Flowrate: 1pm
Nozzle
Cyclone
Stage 0
Stage 1
Stage 2
Filter
Brink
Vertical
Grease
Straight
1.5
8.02
15.2
0.85
% of %
Total Efficiency
Mass
28.4% 28.4%
71.6 100
-
-
- -
Brink
Vertical
Grease
Straight
1.7
12.5
15.0
1.70
% of %
Total Efficiency
Mass
57.1% 57.1%
42.9 100
- -
- -
- —
Brink
Vertical
Grease
Straight
1.7
12.5
12.8
1.70
% of
Total
Mass
25.5%
37.2
34,5
2.9
—
Brink Brink
Vertical Vertical
Grease Grease
Straight Straight
1.5 1.5
8.02 8.02
12.6 9.5
0.85 1.13
% of % % of %
Total Efficiency Total Efficiency
Mass Mass
13.3% 13.3% 18.6% 18.6%
57.3 66.1 36.7 45.1
23.0 78.2 9.2 20.6
6.4 100 31.8 89.7
3.7 100
Brink
Vertical
Grease
Straight
1.7
12.5
9.3
1.70
% of
%
Efficiency
25.5%
49.9
92.2
100
""
%
Total Efficiency
Mass
29.9%
22.4
43.5
4.2
_
29.9%
32.0
91.1
100
_
82
-------
TABLE 10. (Continued)
Impactor :
Orientation:
Substrate:
Nozzle:
Nozzle Diam. (inches)
Inlet Velocity:
Particle Size:
Flowrate: 1pm
Nozzle
Nozzle Holder
Stage 0
Stage 1
Stage 2
Stage 3
Stage 4
Spacers
Filter
Wall Loss
m/sec
ym
% of
Total
Mass
37.9
6.3
18.8
16.3
18.0
2.1
0.2
0.5
-
44.2
Univ. of Wash.
Vertical
Grease
Straight
3/16
7.96
15.2
8.49
Collection
Efficiency
37.9
10.1
33.7
44.0
86.6
74.3
33.5
100
-
44.2
Univ. of Wash.
Vertical
Grease
Straight
3/16
7.96
16.4
8.49
% of Collection
Total Efficiency
Mass
43.6 43.6
10.7 18.9
15.6 34.1
19.9 66.1
10.2 100.0
- -
- -
- -
- -
54.2 54.2
Univ. of Wash.
Vertical
Grease
Straight
k
11.9
15.2
22.7
% of Collection
Total Efficiency
Mass
27.9 27.9
9.8 13.5
41.3 66.3
20.3 96.3
0.8 100.0
- -
- -
— —
— -
37.6 37.6
Comments :
Stage Zero inverted
in assembly.
Impactor :
Orientation :
Substrate:
Nozzle .
Univ. of Wash.
Horizontal
Grease
Straight
Nozzle Diam. (inches) %
Inlet Velocity
:m/sec 11.9
Particle Size:ym 14.5
Flowrate : 1pm
Nozzle
Nozzle Holder
Stage 0
Stage 1
Stage 2
Stage 3
Stage 4
Spacers
Filter
Wall Loss
22.7
% of Collection
Total Efficiency
Mass
24.8 24.8
8.2 10.9
40.7 60.8
25.0 95.3
1.2 100.0
_
-
-
- -
33.0 33.0
Univ. of Wash.
Horizontal
Grease
Straight
3/26
7,96
14.9
8.49
% of Collection
Total Efficiency
Mass
41.6 41.6
3.7 6.4
9.8 17.9
16.9 37.6
25.2 90.3
2.7 100.0
- _
-
- -
45.4 45.4
Univ. of Wash.
Horizontal
Grease
Straight
3/16
13.3
14.5
14.2
% of Collection
Total Efficiency
Mass
34.3 34.3
7.4 11.3
27.8 47.6
26.1 85.3
4.5 100.0
_ _
-
- _
— _
41.7 41.7
83
-------
which might have 15 yra aerodynamic diameter DSD'S at particular
flowrates, 22.7, 14.2, and 8.49 5,/min (0.8, 0.5, and 0.3 acfm) .
These results are shown graphically in Figure 43, where the stage
collection efficiency is shown as a function of the value of the
dimensionless impaction parameter /$. For both the zero stage
(nozzle) and the first stage, a /i|»so value could be clearly defined.
For stage zero this value is 0.19, and for stage one /^50 = 0.34.
The behavior of the zero stage is similar to results from previous
calibration studies. Using these values of /ijJTo" it can be calcu-
lated that stage zero will have a DSO of 15 ym aerodynamic diameter
at a flowrate of 15 l/min (0.53 acfm) and that stage one will have
the same DSO value at 9.35 Jl/min (0.33 acfm). At 9.35 £/min,
stage zero would have a DSO of 19 ym aerodynamic diameter.
100
u
UJ
u.
U.
Ul
z.
o
u
LU
O
o
50
-O r-D-
V I H
•a—a-
H V
A STAGE 0
O STAGE 1
D STAGE 2
H - HORIZONTAL ORIENTATION
V - VERTICAL ORIENTATION
0.50
1.00
1.50
4181-223
Figure 43, Calibration data for the first three stages of the University of Washington
Mark V cascade impactor.
Even though a 15 ym D50 stage does exist for this impactor,
overloading of large particles on either stage zero or stage one
could result in serious re-entrainment problems.
In addition to the studies done to evaluate in-stack cascade
impactors as IP collectors, tests were made of the Sierra Model
84
-------
236 high-volume cascade impactor. This impactor was evaluated
with glass fiber collection substrates and greased substrates
with the intent of using it as a precollector for the Fugitive
Ambient Sampling Train (FAST). The results of this evaluation
are shown in Figure 44. Because of severe particle bounce, the
Sierra impactor was rejected as an IP collector in favor of the
horizontal elutriator described below.
a?
o
ui
Ui
z
o
g
IU
O
o
IUU
90
an
70
BO
50
40
30
?0
10
0
O GREASED PLATE!
5
• GLASS FIBER SUBSTRATES
.—
-/
/
/
/
/
i
(
1
/
/
'
O
'"«v
* \°
\
\
I
\^
• •
I
1
1
1
•1
! 2 3 4 5 6 8 10
AERODYNAMIC PARTICLE DIAMETER, Mm
Figure 44. Particle collection efficiency for the impactor.
20
4100-13
THE HORIZONTAL ELUTRIATOR AS AN IP PRECOLLECTOR
Background
Horizontal elutriators have been used extensively as devices
for measuring and collecting particles and as apparatus for a
85
-------
variety of aerosol experiments.17"20 The British Medical Research
Council adopted the penetration curve of a particular horizontal
elutriator as the definition of respirable particles.Zl Mercer
has reviewed many of the practical applications of elutriators
for sampling respirable particles.22 Hamilton and Walton21 have
established the criterion that re-entrainment of deposited par-
ticles may occur if the ratio of mean air velocity to duct height
(rectangular geometry) reaches a critical value which lies between
240 and 650 sec."1.
The experiments at Southern Research Institute on horizontal
elutriators were performed to verify the theory of particle
collection by settling and to develop design criteria for inlet
collectors to be useid in conjunction with two sampling systems
developed for the Environmental Protection Agency: the Fugitive
Ambient Sampling Train (FAST) and the Source Assessment Sampling
System (SASS). Designs were also developed for the use of elu-
triators with in-stack cascade impactors.
The initial concept of the horizontal elutriator was the
result of studies to measure quantitatively the deposition of
aerosol particles on surfaces adjacent to, or suspended in,
moving gas streams flowing through laboratory apparatus.
Natanson23 and Thomas19 independently solved this problem theo-
retically for a circular horizontal tube of radius a. Although
the derivations are somewhat complex, the results can be
summarized and simplified in terms of the collection efficiency
(Eff) by the following equation:
Eff . i 2e
where
Vl - £2/3 - e 1/3 yi - e2/3 + arcsin e1/3
(18)
8aV
V is the settling velocity,
s
L is the length of tube,
a is the radius of tube, and
V is the average velocity of the gas.
The theory describing the performance of a rectangular flat
plate elutriator is considerably less complex and is re-derived
here to give the reader a better understanding of the particle
motion in the elutriator. The settling velocity of spheres in
a gas stream in laminar flow is given by the equation
86
-------
where
Vg is the settling velocity,
g is the acceleration due to gravity,
C is the slip correction factor,
p is the particle density,
d is the particle diameter, and
U is the viscosity of the gas.
In health-related studies, the aerodynamic behavior of the
particles is of interest. It is therefore convenient to relate
the settling velocity of all particles, whatever their shape or
density, to that of spheres having unit density. The aerodynamic
diameter is defined as the diameter of a unit density sphere
having the same settling velocity as the particle of interest.
The aerodynamic diameter is given from Equation 19 by setting
p = 1 g/cm3 (the units must be retained) . Also, for particles
larger than about 2 jam, the slip correction factor is approxi-
mately equal to unity. A graph of settling velocity vs. diameter
is shown in Figure 45.
For laminar flow between parallel plates, the velocity
profile of a gas is parabolic with the maximum velocity equal to
1.5 times the average and with zero velocity at the plates:
<20>
where V is the velocity parallel to the plates, at a point y,
«£ *
y is the displacement above the bottom plate,
h is the spacing between plates, and
V is the average velocity of the gas
The velocity profile and a typical particle trajectory are
illustrated in Figure 46.
The efficiency with which particles are collected is deter-
mined by the dimensions of the channel, the velocity of the gas,
87
-------
6.0 :;==; ;!,;;;
1.0 — ijffl |
:£::::::::::
§ EJjiiiiiiiii
v> till
w
3 0.1 Egi: :
uj c:::::::: .:::
O -J- IT
1 Ml
tn -T"^"3"i
4*ltfi..:S
SIS::!:
*Stl
7 '
0.01 £53|::|
ifti
piil
Siislt
[ ' I jj
, , i i • Jf
0.001 .^Tf^lfr
--pT-|--^|-'-l' -
„,
::::::::^::::::$:^:::: ::)::
i=E!Ei!|i=E!i!i:J|i :=EE
::±±::=_l_.iX.±:...l;: ,£.
iEEEiiiEEEEEEEEEIJIl |E
_-j p--.. .-4: 1:444 ,J.
[ j i j !
"iiiSz":!Si|:5
A i i *
:::::-ii"i::::j|:::: :|
i=ii)!i^ii=iliii!p
:::!:±:::-:::::^:::::::: :|
iZ:::|::-±::::l|p [g
It
1 1
I! In t
|::| : ::::::!:
*::::: t ::f ::::?
j . -- 1 -
;;;;!; Z::;;;;E
... ., It". . ~
|::::: I ::::::::r
f . ,..j | — _
T--J' • -
.. .- .. . ___.,...-
t — h- -- -- -
T || j
1|;:|::::I::E
i in — i i
r --£--..-
:::::::g £::::::;
::::::: It ::±±::-
1
I u *
-H-4----+-
H -it-
.
__IL
4
) j
-- Jl
..j....:.:..
lEEEEEElii:
======EE:;I
mm
||l l|l |
*::::|::::
tTT 1 T T i r
0.6 1 2 4 8 10 20
PARTICLE DIAMETER, /un AERODYNAMIC
}i.||j,Lm)ija
— _ I —
::: ::;!: ::::
IllJjflllllll
• ' --T • --T
.J, .g
... 11
It i .1
t... -+
::::tifi
• f • -~T- "^^
i 1 ! '
40
4181-86
Figure 45. Settling velocity in air for unit density spheres.
88
-------
the settling velocity of the particles, and the height at which
the particles enter the channel.
VX
IT-
I
1
y
^ \ ^y
\ /
w \ •— — — ^S
* o _^
f
< s -*•
4181-106
Figure 46. Velocity profile and particle trajectory between parallel plates.
Consider a particle which enters the channel at position y
as shown in Figure 47 and which has a trajectory of length L (the
length of the channel). All particles of this diameter or larger
entering at or below position yQ will be captured. All particles
of this diameter entering the channel at positions higher than
yQ will penetrate the channel.
Now, from Equation 20;
/ \
dx
but dy is given by dy =
-V dt; therefore
3
from which the settling velocity of the particle may be determined.
89
-------
4181-73
Figure 47. Zone of 100% particle collection.
If the particles are uniformly distributed within the gas,
the fractional collection efficiency of the particle illustrated
in Figure 46 is equal to the ratio of the volume of gas passing
belpw position y to the total volume; thus:
Eff *
W
dy
dy
where W is the width of the channel,
Now, from Equation 21,
E££ .
h V 18 yh V
(22)
The theoretical efficiency curves for a horizontal elutri-
ator with a circular cross-section (from Equation 18) and for
a horizontal elutriator with a rectangular cross-section (from
Equation 22), both designed to have a cut point of 15 urn aero-
dynamic diameter, are shown for comparison in Figure 48.
90
-------
u
UJ
o
iZ
UL
Ul
til
1.0
1.0
4.0 6.0 8.0 10 20 40
GEOMETRIC MEAN DIAMETER, micrometers
60 80 100
4181-126
Figure 48. Theoretical collection efficiency by particle settling in rectangular
and circular tubes.
The efficiency is found to be independent of the vertical
velocity profile and, as Fuchs17 observed, Equation 22 can be
derived more easily assuming plug flow.
Experimental Procedures
Experiments were conducted to verify Equation 22 and to set
design parameters for the FAST and SASS inlets. It should be
noted that .the systems tested here are designed so that the length
and width are large relative to the height of the channels. This
was done so that end and edge effects could be ignored.
Figure 49 is a schematic diagram of the experimental appa-
ratus used to investigate the performance of a horizontal elutri-
ator designed to have a cut point (Dso) of 15 urn aerodynamic
diameter. The settling chamber consists of 28 channels, each
7 mm high, 17.8 cm wide, and 38.1 or 20 cm long. A high volume
blower (Model 305, Sierra Instruments) connected in series with
a variable voltage transformer was used to supply the desired
air flow rate through the chamber. Monodisperse particles of
91
-------
ID
N)
COMPRESSED AIR LINE
REGULATOR
AND TRAP
REGULATOR
DRYER
ABSOLUTE FILTER
REGULATOR
VALVE,
r-c>
VIBRATING
ORIFICE
GENERATOR
ROTOMETER D|SPERS|ON A|R
I
DISPERSION AIR
'.. 1
FILTER
SETTLING CHAMBER
VARIABLE
VOLTAGE
TRANSFORMER
HIGH VOLUME
BLOWER
I
SYRINGE PUMP WATER MANOMETER
FUNCTION GENERATOR
4181-74
Figure 49. Apparatus used to measure the collection efficiency of the settling chamber.
-------
methylene blue were generated using a vibrating orifice aerosol
generator (VOAG) for particles larger than 4 pm aerodynamic
diameter. During each test, the particles were sampled and
checked frequently by optical microscopy to ensure constant mono-
dispersity. All particles entering the horizontal elutriator
were collected either by settling on the plates or on the
8 in. x 10 in. (20.3 cm x 25.4 cm) filter downstream from the
plates.
Upon completion of each test, the plates and filter were
washed separately with water. Samples from each wash were
centrifuged to remove debris, and the amounts of methylene -blue
collected on the plates and on the filter were determined by
absorption spectroscopy.
The velocity distribution through the settling chamber was
studied to determine the configuration necessary to obtain uniform
flow. The use of two filters and an extended flared inlet was
necessary to provide the desired velocity distribution. Figure
50 is a velocity profile measured using a thermal anemometer
immediately upstream from the blower before the plates were in
position. From these data, it was evident that the blower was
pulling uniformly across the rectangular opening where the
20.3 cm x 25.4 cm filter was positioned. Figure 51 shows the
velocity profile measured upstream of the collector plates.
The velocity profiles shown in Figures 50 and 51 were con-
sidered satisfactory, and the overall average velocity was used
to calculate the theoretical performance curves for the system.
The average velocity through the chamber, measured using the
thermal anemometer, was in agreement with previous calibration
of the blower.
Results
Two sets of experimental data were obtained from the testing
of the horizontal elutriator. The first set was acquired while
operating the settling chamber at an average gas velocity of
70 cm/sec and a plate length of 38.1 cm. The second data set was
obtained by operating the system at 40 cm/sec after shortening
the plate length to 20 cm. The theoretical curves of the collec-
tion efficiency versus aerodynamic particle diameter shown in
Figures 52 and 53 were developed from Equation 22 using the
following parameter values:
Figure 52 Figure 53
p = 1.35 g/cm3 p = 1.35 g/cm3
g = 9.8 m/sec2 g = 9.8 m/sec2
L = 38.1 cm L = 20 cm
y = 181 micropoise y = 181 micropoise
h = 0.701 cm h = 0.701 cm
V = 70 cm/sec V = 40 cm/sec
Reynolds No. = 315 Reynolds No. = 180
93
-------
'' V (m/sec)
/ *•-. 0.30
0.20
0.10
STATISTICAL DATA:
MEAN, x , = 0.24 m/sec
STANDARD DEVIATION,
Sx = 0.02 m/sec
COEFFICIENT OF VARIATION,
SX/JT = 8.2%
4181-107
Figure 50. Profile of the air velocity immediately upstream from the blower
of the settling chamber before the plates were positioned.
94
-------
STATISTICAL DATA:
MEAN, x . = 0.23 m/sec
STANDARD DEVIATION.
Sx = 0.02 m/sec
COEFFICIENT OF VARIATION,
Sx/x = 8.8%
vo
tn
4181-108
Figure 51. Profile of the air velocity upstream from the plates of the
settling chamber.
-------
as
98
95
90
as
u 80
LU
o 70
£ 60
LU __
Z 50
2 40
H
$ 30
10
5
2
1
imammm
1
— THEORY (RECTANGULAR)
> EXPERIMENT
4
s'
\
'
/
/,
/
^
"
/
II
/
/
/
/
a
/A
r
/
/
/
x
1 2 34 6 8 10 20 30 40 60 80 10
ACDftrtVM AKJUf OAO-|-I er r\i « « 4181-1C
Figure 52. Theoretical and experimental collection efficiencies for a horizontal elutriator
with rectangular cross-section, plate length 38.1 cm, average gas velocity 70 cm/sec.
96
-------
yy
98
95
90
£"
| 70
5 60
u.
£ 50
Z 40
O
g 30
2 20
O
° 10
5
2
1
•
THEO
EXPE
^
^n
RY
RIME
X
NT
f
r
/
j
I
*
y
\ 1
I
\
1 >
//
^2—
/
/A
//
f
/
/
1 2 3 4 6 8 10 20 40 60 80 100
AERODYNAMIC PARTICLE DIAMETER, pm 4181-75
Figure 53. Theoretical and experimental collection efficiencies for a horizontal e/utriator
with rectangular cross-section, plate length 20 cm, average gas velocity 40 cm/sec.
97
-------
The calculated values from theory for the collection effi-
ciency were found to be in excellent agreement with the measured
values. No corrections were made for end effects.
Collector Design
The experiments described above indicated that the theoreti-
cal equations can be used to predict particle collection by
horizontal elutriators to a high degree of accuracy. Consequently,
the equations were used to create design nomographs for inhalable
particulate precollectors to be used in conjunction with the
three systems of interest. Design parameters were calculated
for horizontal elutriators to be used with: (1) cascade impac-^
tors operated at 14.2 SL/min and 149°C, (2) SASS trains operated
at 185 Vmin and 204°C, and (3) FAST trains operated at 5,282
SL/rnin and 23°C.
Figures 54, 55, and 56 are the design nomographs for the
three trains. On the vertical axis is the open area required,
neglecting the thickness of the tube walls. The horizontal axis
is the length of the elutriator required to yield the inhalable
particulate performance at the specified flowrate and tempera-
ture. In constructing the graphs, it is assumed that the rectan-
gular channels have widths much greater than their heights, so
that the vertical walls will not have a significant effect on
the gas flow.
THE FAST ELUTRIATOR INLET
Upon completion of the theoretical and prototype study of
horizontal elutriators, an elutriator inlet was designed and
fabricated for the Fugitive Assessment Sampling Train. It was
designed to have a 15 ym aerodynamic diameter cut point at the
normal operating flowrate. This inlet section was calibrated
using ammonium fluorescein particles of 15 and 16 ym aerodynamic
diameter generated with the VOAG. The collection efficiencies
of the elutriator were 48% and 59%, respectively, for the two
particle sizes. This is well within the designed collection
efficiency fot these particle sizes. The data obtained during
this testing procedure are summarized in Table 11. Figure 57
depicts the experimental arrangement in the laboratory.
Horizontal Elutriator Inlet for Fugitive Emissions Sampler
The final application of this study of horizontal elutriators
was the design of an inlet to be used with high volume impactors
to measure the size and concentration of inhalable particulate
matter in fugitive emission plumes.
98
-------
vo
VD
RECTANGULAR ELUTRIATOR
2 3 45678
PLATE SEPARATION, mm
CM
III
oc
o
u
UJ
DC
U
CYLINDRICAL ELUTRIATOR
RECTANGULAR ELUTRIATOR
TUBE DIAMETER, mm
CYLINDRICAL ELUTRIATOR
2.0
2.0
3.0 4.0 5.0 6 7 8 9 10
LENGTH, cm
20
30 40 50 60 70
4181-124
Figure 54. Relationship of design parameters for horizontal elutriators with DSQ cutpoints of
15 urn aerodynamic diameter used as precollectors for in-stack cascade impactors.
-------
o
o
u
2
O
u
ul
CO
to
CO
O
cc
o
200
100
7 I 8 -1 PLATE SEPARATION
CYLINDRICAL ELUTRIATOR f
RECTANGULAR ELUTRIATOR
TUBE DIAMETER^ 2mm?
CYLINDRICAL ELUTRIATOR
40 50 60
20
3.0
4.0 5.0 6.0
20
LENGTH, cm
Figure 55. Relationship of design parameters for horizontal elutriators with DSQ outpoints
of 15 ju/77 aerodynamic diameter to be used with SASS trains.
-------
U
00
at
O
K
U
2mm
3mm 4mm
RECTANGULAR ELUTRIATOR
6 7 8 - PLATE SEPARATION
CYLINDRICAL ELUTRIATOR
RECTANGULAR
600
500
400
300
200
100
40
80 100
LENGTH, cm
200
300 400 500
700 1000
4181-125
Figure 56. Relationship of design parameters for horizontal elutriators with DQQ outpoints
of 15 urn aerodynamic diameter to be used with FAST trains.
-------
TABLE 11. FAST HORIZONTAL ELUTRIATOR INLET TEST DATA
o
to
Test No.
Aerodynamic Particle Diameter (jim)
Filter Pressure Drop (in mercury)
Sampling Duration (hr)
1
16
2
4
% of
Total Mass
Inlet Louvre Section
Screen
Screen to Elutriator Transform
Elutriator Plates
Plate Holder Section
Filter Transform and Filter
0.75*
0.0009*
3.63
56.2
0.5
38.9
.0
.0
% Collection
Efficiency
0.75
0.0009
3.65
58.8
1.27
100.0
2
15.0
2.2
2.5
% of
Total Mass
28.2
5.34
1.56
31.2
0.14
33.7
% Collection
Efficiency
28.1
7.4
2.4
47.8
0.4
100.0
*Questionable results due to inaccurate washing
technique.
-------
o
u>
4181-98
Figure 57. Laboratory set-up for testing FAST horizontal elutriator inlet.
-------
The laboratory sampling device illustrated in Figure 49 was
modified slightly by the addition of a cascade impactor between
the horizontal elutriator section and the filter. The resulting
system will be mounted on masts and arranged in a vertical, planar
grid for measuring the plume "profile".
CYCLONES AS IP COLLECTORS
Background
Although no theory exists to adequately predict cyclone
behavior, several experimental investigations have been performed
demonstrating their utility for separating and sizing small par-
ticles. 2"*-2S Smith et al.15 and John et al.27 have demonstrated
that the curves of particle collection efficiency for small
cyclones can be as steep as those of impactors, and significant
re-entrainment does not occur. Chan and Lippmann28 have shown
that experimental efficiency data for small cyclones can be
fitted well using the empirical relationship:
- 0.5 + 0.5 tanh
where
Q is the sample flow rate, cm3 /sec,
D is the diameter of the sampled particle, cm, and
A,B,K,n are empirical constants.
Also it was shown that the Dso vs. flowrate relationship is
given over a limited range of sampling conditions by:
D50 = KQn. (24)
Prom their own research and data reported by others, Chan
and Lippmann reported values of K ranging from 6.17 to 4591, and
values of n from -0.636 to -2.13. Smith et al. reported values
of K from 14.0 to 44.6, and n from -0.63 to -1.11. Although the
trend in the data is for n to have larger values for small cyclones
(Dso's) and K to be larger, the correlation is not consistent
enough to be predicted from the cyclone geometry, and no data are
reported for temperatures other than ambient. Furthermore, there
are definite discontinuities and hysteresis effects in the rela-
tionship given in Equation 24, even for individual cyclones, as
104
-------
the flow is increased and decreased. The discontinuous and
hysteresis effects are generally attributed to transitions from
turbulent to laminar flow, or the reverse, in the outlet tube of
the cyclone as the flow is changed.25"28
Smith et al.15 found the D50 of cyclones to increase linearly
as the gas temperature and viscosity were increased; but again
the rate of increase was not predictable from the cyclone geome-
try. Certainly a modification to Equation 24 would be required
to predict any temperature dependence. Lacking an adequate
theory for predicting the performance of cyclones before they
are designed and calibrated, it was found necessary in this study,
as in previous work, to extrapolate the dimensions for a new
cyclone to give the desired performance from those of similar
cyclones of known performance. Several cyclones were calibrated
to determine a basis for extrapolation for the new IP Sampler
cyclone dimensions.
However, because of these difficulties, and the requirement
that the inhalable particulate cyclones always be operated to
yield D50's at 15 + 2 ym, it was still considered necessary to
calibrate them over a range of temperatures and flowrates similar
to those expected in field operation.
Calibration of the Brink Cyclone
The Brink precollector cyclone (Southern Research Institute
design) was calibrated at ambient temperature and at 100 and
150°C. Ammonium fluorescein particles of 15 ym aerodynamic
diameter were generated by a vibrating orifice aerosol generator,
heated, and then sampled by the cyclone. The flowrate was varied
to determine the flowrate at which the cyclone collected 15 ym
particles with 50% efficiency. The results of these tests are
shown in Figures 58 and 59 and in Table 12. The Brink cyclone
has a Dso of 15 ym at 0.51 5,/min at ambient temperatures,
0.79 Vmin at 100°C and 1.1 A/min at 150°C.
Calibration of the Sierra Cyclone
The first cyclone in the Sierra Instruments, Inc. Model 283
Cyclade (TM) four-stage instack cascade cyclone sampler was
calibrated at ambient temperature with 15 ym aerodynamic diameter
ammonium fluorescein particles. The cyclone was found to have a
Dso of 15 ym at 5 £/min. The calibration data are presented in
Table 13.
Calibration of the Southern Research Institute Five-Stage Cyclone
The individual cyclones of the Southern Research Institute
Five-Stage Cyclone System were calibrated in the laboratory under
conditions similar to those frequently encountered in the field.
105
-------
o
crv
Figure 58.
*!8
&B.J?
.^ 31?.
l^s
s?l
a2 §'
-* Qi §
§ll
•°f^
s^
flowrate for a Brink precollector cyclone
c diameter ammonium fluorescein rticles
COLLECTION EFFICIENCY, %
8
ACTUAL FLOW RATE, £/min
413
28
pa
-------
e
1
%
O
LU
oc
i
U.
1
1.5
150
200
VISCOSITY
250
, micropotse
300
4181-127
Figure 59. Flowrate versus viscosity to maintain a
Brink precollector cyclone.
- 15 ton aerodynamic for
107
-------
TABLE 12". CALIBRATION RESULTS FOR THE BRINK CYCLONE
D = 15 ym Aerodynamic
Temperature,
°C
150
150
150
100
23
21
Flowrate,
A/min
0.79
1.5
1.1
0.79
0.60
0.51
Collection Efficiency,
37
62
50
53
65
51
TABLE 13. SIERRA CYCLONE CALIBRATION DATA
Flowrate,
Temperature,
°C
Aerodynamic
Particle Diameter, urn
Collection
Efficiency,
7
7
14
14
14
14
21
21
21
23
23
23
23
23
23
23
23
23
15.1
15.0
14.5
14.3
9.9
7.1
7.1
10.3
15.4
96.1
95.6
64.6
58.0
27.4
14,
10,
40,
1
1
1
71.9
108
-------
The Dso cut points of these cyclones at various operating condi-
tions are given in Table 14. For laboratory conditions (tempera-
ture 22°C, flowrate 28.3 H/min, and particle density 1.0 g/cm3)
the cut points are 5.6, 2.1, 1.4, 0.63, and 0.33 um, as shown by
the calibration curve of efficiency vs. aerodynamic particle
diameter in Figure 60.
100
u
z
ui
O
u.
UJ
O
U
IU
8
111
1.0 ACFM
22°C
I CYCLONE V
CYCLONE IV
CYCLONE III
CYCLONE II
CYCLONE I
L_
0.2
0.3 0.4 0.6 0.81.0 2.0 468 10.0
AERODYNAMIC PARTICLE DIAMETER, jun
20.0 30.0
4181-360
Figure 60. Calibration curves for the five-stage cyclone system. Flow rate 1.0 ft^/min,
temperature 22°C.
AN INHALABLE PARTICULATE PRECOLLECTOR AND SAMPLING TRAIN
Development and Fabrication
The objective of this phase of the program on the develop-
ment of samplers for inhalable particles was the design and
evaluation of two systems: (1) a cyclone to be used as a
precollector for cascade impactors, and (2) a system, consisting
of two cyclones and a filter in series, to be used as the primary
system for measuring the total particulate concentration, the
inhalable particulate concentration, and the fine particle con-
centration (<2.5 ym). Both systems are to be used in process
streams where the particle concentration and temperature are
generally much higher than in the atmosphere. The precollector
for impactors and the first cyclone in the series train must have
109
-------
TABLE 14. LABORATORY CALIBRATION OF THE FIVE-STAGE CYCLONES
D5o CUT POINTS
Flowrate ,
H /min
7
9.7
14
20
21
21
28
19.8
28
14
19.4
28
28
Temperature ,
°C
22
22
22
22
22
22
22
93
93
151
151
151
204
Dso Cut joints, urn
Cyclone I Cyclone II
_
_ _
8.2 3.6
3.1
2.8
2.3
5.6 2.1
— —
6.4 3.1
_ _
- _
3.4
8.6 4.1
(p = 1.0_gm/cm3)
Cyclone III Cyclone IV Cyclone V
2.6 1.6
3.1
2.3 1.4 0.84
_ _ _
_ _ _
0.65
1.4 0.63 0.33
2.5
1.8
5.0
2.5
2.9
3.1
-------
efficiency curves satisfying the specifications for inhalable
particulate samplers shown in Figure 2. The second cyclone in
the series train must be designed to have a DSO of 2.5 + 0.5 ym
aerodynamic.
The nominal operating conditions used in designing the
systems were flowrates of 14 £/min and 21 4,/min at 150°C,
respectively, for the precollector and cyclone systems. With
these operating conditions as a goal, the dimensions of the
cyclones were extrapolated from those previously evaluated.15
Figures 61 and 62 are schematic illustrations of the two new
systems. The precollector cyclone is designated SRI-IX and the
large cyclone in the series train, SRI-X. Cyclone SRI-III, which
had been previously evaluated as part of the SRI/EPA 5-stage
cyclone train, was found to be adequate for the smaller cyclone
in the new system.
The critical dimensions of all three cyclones are given in
Figure 63.
Calibration
In order to calibrate the cyclones at elevated temperatures,
the heating arrangement shown in Figure 64 was used in conjunc-
tion with the vibrating orifice aerosol generator described above.
Tests were made at temperatures of 23°C, 93°C, and 150°C. Ammo-
nium fluorescein particles were used for calibrating the cyclones.
Samples were taken frequently of the heated aerosol to ensure
that the calibration system was stable and that the particles
were spherical and of the proper size.
In this study, the primary objective was to determine the
operating conditions under which the cyclones satisfied the
design criteria. For this purpose, it was sufficient to use an
abbreviated calibration procedure and thus to reduce the extensive
amount of testing that would be needed for complete calibration.
At a specific temperature, monodisperse particles having nominal
aerodynamic diameters of 15 ym were generated and sampled. Tests
were made at a variety of flowrates to determine the sampling
rate required to yield a D5o of 15 ym at the given temperature.
Only limited data were taken to determine collection efficiency
vs. particle diameter.
When preparing to generate monodisperse particles using the
vibrating orifice method, it is necessary to know the solution
flowrate and frequency of vibration in order to calculate the
concentration of solute required to yield the desired particle
diameter after drying. In practice, the flowrate can be selected
and the solution prepared precisely, but the frequency at which
the generator is finally found to yield maximum stability is
111
-------
SAMPLING NOZZLE
CYCLONE
050 = 15 ± 2 urn
CASCADE IMPACTOR
PROBE
4181-76
Figure 61. Schematic of a cascade impactor-precollector cyclone system.
SAMPLING NOZZLE
PROBE
CYCLONE SRI- X
D5Q = 15+2 /Urn
CYCLONE SRI III
D50 = 2.5 ± 0.6 jUm
4181-120
Figure 62. Schematic of two-cyclone system.
112
-------
T
Din
'cup"
DIMENSIONS (CENTIMETERS)
CYCLONE
SRI-X
SRI-m
SRI-1X
D
6.14
3.11
5.12
Din
1.83
0.75
1.53
De
2.17
0.83
1.81
B
2.92
0.76
2.43
H
8.47
4.91
7.06
h
2.82
1.40
2.35
Z
5.65
3.51
4.71
S
2.40
1.08
2.00
HCUp
2.635
2.22
2.26
Dcup
6.14
3.10
5.12
Figure 63. Summary of cyclone dimensions.
4181-22
113
-------
AEROSOL STREAM
FROM VIBRATING
ORIFICE AEROSOL
GENERATOR
\BSOLUTE FILTER
OVEN
I KEPT AT
AEROSOL TEMPERATURE
SAMPLING
PORT
MERCURY WATER
MANOMETER MANOMETER
4181-92
Figure 64. Calibration system for heated aerosols.
-------
unpredictable to some extent. Therefore, as indicated in the
figure captions, the actual particle diameters differed slightly
from 15 ym. In these tests the particles had aerodynamic diame-
ters of 15.0 + 0.6 ym.
Results
Figures 65-68 show the calibration data for the three
cyclones. In Figure 65, the particle collection efficiency is
plotted vs. flowrate for cyclone SRI-IX.
As indicated in the figure, data were taken at 23°C, 93°C,
and 150°C. Similar data are shown in Figure 66 for cyclone SRI-X.
The settling velocity in air of 15 ym particles is 0 . 7 cm/sec
and there was some concern that settling might influence the
collection efficiency of the larger cyclones by making them
sensitive to orientation. In Figure 66, which contains calibra-
tion data for cyclone SRI-X, data are shown taken with the cyclone
in both vertical and horizontal orientations. It can be seen
that there is little, if any, effect due to particle settling.
The data reported in Figures 65 and 66 will be used to select
the flowrates of the new sampling trains. Since there is no
adequate theory for calculating cyclone efficiency, cyclone
SRI-III was calibrated at the same flowrates required for proper
operation of cyclone SRI-X. These calibration data are shown in
Figure 67. D5o's and flow conditions for the three cyclones, as
derived from the graphs, are:
Cyclone SRI-III
3.1 ym (aerodynamic) D50 at 23°C, 11 H/min
2.6 ym (aerodynamic) Dso at 93°C, 20 A/min
2.3 ym (aerodynamic) D50 at 150°C, 23 £/min
Cyclone SRI-IX
15 ym (aerodynamic) D50 at 23°C, 6.8 £/min
15 ym (aerodynamic) D5o at 93°C, 12 H/min
15 ym (aerodynamic) D50 at 150°C, 14 H/min
Cyclone SRI-X
15 ym (aerodynamic) Dso at 23°C, 11 A/n»in
15 ym (aerodynamic) D50 at 93°C, 20 Jl/min
15 ym (aerodynamic) D5o at 150°C, 23 5,/min
Also, limited data were taken with particles with diameters
other than 15 ym using cyclone IX. These are shown plotted with
the inhalable particulate performance specifications in Figure 68.
115
-------
" 4 6 8 10 20 40
FLOW RATE, liters/min 4181-78
Figure 65. Collection efficiency of Cyclone IX versus flowrate for particles o
15±0.6 urn aerodynamic diameter.
COLLECTION EFFICIENCY, %
.8 § . g g i
•<,
^.
m*
"^^
:-i
"""--^
KJ
^^L
, *"«!
V^
t""
^
Jx
1
-------
100
80
39
*
O
UJ
iZ 60
O
O
UJ
8
40
20
II III
O VERTICAL, OUTLET UP
A VERTICAL. OUTLET DOWN
• A H
KDRI2
!ON1
•AL
/
(
/
f
/23°C
/
i 1
1
1
t
/
jL
/
93°C /
1}
; i
II
I
I
f
#
150°C
4 6 8 10 20 40
FLOW RATE, liters/min 4181-7S
Figure 66. Collection efficiency of Cyclone X versus flowrate for particles of
15±0.6 nm aerodynamic diameter.
117
-------
00
I
N
^- Q)
v 5
$3
<3£
°^l
Q> §
5 3
^o-
?S^
co5
*=,
m
i
o'
-------
COLLECTION EFFICIENCY, %
Q>
CJ.
O
3
"
Q)
-*
•fc
S
>
J3
H
O
m
O
>
m
m
•R
m
O
O ».
r> z
_> M
01
g 3
(O (O
00 25
CO
00
%
£
B
§
& _.
o o
o
-------
The data reported above define the operating points of
cyclones SRI-IX and SRI-X at three discrete values of temperature.
When the data are plotted on semi-log coordinates and a smooth
line fitted to the points, the following equations of flowrate
vs. viscosity result:
For cyclone SRI-X, Q = (105 logu - 225) 5L/mLn
and for cyclone SRI-IX, Q = (69.5 logy - 150)£/min.
It is not known at present how accurately these equations
can be extrapolated to temperatures greater than 150°C.
In order to gain a better qualitative understanding of the
mechanisms governing the performance of cyclones, Table 15 was
prepared summarizing some of the dimensions and operating para-
meters of cyclones evaluated at Southern Research Institute.
The parameters of greatest interest are the Stokes number St of
15 ym particles in the inlet and the ratio of the settling veloc-
ity to the centripetal velocity of 15 ym particles in the cyclone
body. The latter is calculated assuming that the air flow in the
cyclone body is a jet having the same velocity and diameter as in
the inlet. This is obviously a crude approximation, but probably
accurate enough to allow a reasonable estimate of the desired
ratio.
The settling velocity V of unit density spheres is given
by the expression (equivalent to Equation 19):
Vo = mgg ' (25)
S
where m is the particle mass,
$ is the particle mobility, and
g is the acceleration due to gravity.
The centripetal velocity Vc is given by:
mV?
Vc = -fi S (26)
where V. is the gas velocity in the inlet and
R is the cyclone radius.
The desired ratio, then, is:
in
120
-------
TABLE 15 . OPERATING PARAMETERS OP SRI CYCLONES
to
Cyclone
Number
V
IV
III
II
I
BRINK
IX
X
cm3/sec.
472
472
472
472
472
28
233 (150° C)
190
D,
cm
1.52
2.54
3.11
3.66
4.47
1.27
5.11
6.14
Din'
cm
0.30
0.57
0.75
1.01
1.27
0.51
1.53
1.83
Vin<
cm/sec
6670
2300
1100
600
370
130
130
70
AP,
mm H2O
1800
330
70
40
5
-
-
-
Dso
ym
0.32
0.65
1.4
2.1
5.4
13
15
15
/St
0.1
0.1
0.1
0.1
0.2
0.6
0.3
0.2
gR/vin2
1.6 x 10"s
2.3 x 10~"
1.3 x 10"3
5.0 x 10~3
1.6 x 10"2
4.0 x 10~2
0.2
0.6
Q is the sample flow rate
D is the diameter of the cyclone body
D. is the diameter of the inlet
V. is the gas velocity in the inlet
AP is the pressure drop across the cyclone
St is the Stokes number of the 15-ym particles in the inlet
gR/V? is the ratio of the settling velocity to the centripetal velocity in the cyclone
body"
-------
The values of /St and V /V can be used to estimate whether or
s c
not impaction onto the wall opposite the inlet or particle
settling is likely to play_a large part in the collection effi-
ciency of a cyclone. If /St is less than about 0.4, then impac-
tion is not likely to occur.29 if V_/V is very small, then
S G
settling is not likely to occur. Prom Table 15, it can be seen
that impaction is probably very significant in the cyclone used
with the Brink impactor, while settling probably contributes
somewhat to the collection efficiency of cyclones SRI-IX and SRI-X.
SIZE-SELECTIVE INLET FOR THE ANDERSEN HI-VOL SAMPLER
The Andersen Size Selective Inlet was tested using mono-
disperse aerosols of ammonium fluorescein generated with the
vibrating orifice aerosol generator. This device is designed
to be a 15 jam aerodynamic diameter inhalable particulate pre-
collector for hi-vol samplers operating at 40 ft3/min. The tests
are only of the second or pure impaction part of the inlet; i.e.,
the inlet was not tested for the effects of wind speed.
The size selective inlet (SSI) was mounted on top of a
standard 40 cfm hi-vol in the laboratory. A flat 8 in. x 10 in.
filter collected all particles passing through the SSI. Three
tests were performed. Particles generated by the VOAG were
directed toward the SSI inlet. The SSI was rotated during the
sampling procedure to insure uniform exposure of all portions of
the SSI.
Figure 69 shows an exploded view of the SSI and the various
surfaces which wera washed after each run. These surfaces were:
1. Inlet entrance rim
2. Underside of top cover
3. Top of first plate and bowl sides
4. Bottom of first plate
5. Top of second plate and bowl sides
6. Bottom of second plate and bowl sides
7. Filter
Table 16 shows the test data. The cumulative efficiency
shown in the last column does not include the contribution from
surface (1) because of probable inconsistencies in the procedure
for directing the particles towards the SSI inlet. Results from
the tests lay within the proposed IP specification limits.
A STACK DILUTION SAMPLING SYSTEM
A significant problem which arises in the sampling of par-
ticulate pollutants in a stack environment is the measurement of
122
-------
1. INLET ENTRANCE RIM
2. UNDERSIDE OF TOP COVER
3. TOP OF FIRST PLATE AND BOWL SIDES .
4. BOTTOM OF FIRST PLATE
5. TOP OF SECOND PLATE AND BOWL SIDES
6. BOTTOM OF SECOND PLATE AND BOWL SIDES
7. FILTER
4181-105
Figure 69. Schematic of size-selective inlet.
123
-------
TABLE 16. SIZE-SELECTIVE INLET CALIBRATION DATA
% of Total Mass on Specific Surface
Efficiency
Run Particle Surface Cumulative
Time(hr) Size (ym) 12345 6 Filter 2,3,4,5,6
2.5
2.5
5.0
16.
16.
11.
8
7
8
1.98
2.9
3.1
0.34
1.0
0.02
10.8
12.4
7.6
0.09
1.8
0.14
23.9
27.1
15.4
7.5
10.2
6.0
55.4
44.6
67.7
43
54
30
.5
.1
.2
the particles in the form that they will have after they are
emitted into the atmosphere. In particular, it is difficult to
characterize primary aerosols formed from condensable vapors in
the stack gas as the plume is cooled and diluted by the ambient
atmosphere. For example, the original EPA Method 5 attempted to
measure the condensable fraction collected at 0°C, with a high
probability of condensing water as well. A recently published
technique attempts to simulate plume mixing using dilution with
clean dry air in the turbulent jet of a dilution ejector pump.30
This approach, while clearly a great improvement over previous
methods, may be criticized in that it uses rapid (0.5 sec), highly
turbulent mixing quite different from that in a typical plume.
Since the particle size distribution and mass loading of a conden-
sation aerosol are affected by the competing rates of transport,
condensation, and adsorption of vapor, the collection conditions
of this sampler can be a liability.
In order to provide an independent measure of particles
formed from condensable stack vapors, a new sampling system with
dilution was developed. The goal of the Southern Research
Institute design was to simulate the actual plume formation as
closely as possible in a portable instrument suitable for field
use while allowing some flexibility in choosing the mixing para-
meters. In this system, stack gas is extracted and diluted with
a standardized "ambient" air. The flow patterns are designed to
qualitatively approximate those which occur in an actual plume,
with the mixing time of the gases allowed to be as long as
practical within reasonable constraints of instrument size and
flowrates. A laboratory prototype was constructed and subjected
to various diagnostic tests. As a result of the tests on the
prototype system, a field diluter instrument was also designed
and constructed.
124
-------
Design and Testing of Laboratory Prototype Sampler
Prototype Design—
A block diagram of the laboratory-scale prototype dilution
sampling system is illustrated in Figure 70. The heart of the
sampler consists of a Plexiglas cylinder in which simulated stack
gas is mixed with filtered ambient air and the resulting aerosol
analyzed. The simulated stack gas is introduced through a tube
oriented along the axis of the cylinder. Dilution air is intro-
duced through four radially oriented tubes, is diverted along the
cylindrical dilution chamber, and passes through a series of fine
screens which serve as flow straighteners. The resulting uniform
flow of dilution air enters the mixing region in an annulus
surrounding the simulated stack sample. In tests using smoke in
the sample or dilution air streams, the streams appeared uniform
at the point of introduction and provided a good visual represen-
tation of the mixing of a plume. Flowrates of both sample and
dilution air can be varied; typical conditions are 10 cfm for
dilution air and 0.4 cfm for sample air, giving a dilution ratio
of 25 and a residence time of about 4 sec.
The diluter is adaptable to a variety of sampling devices
for measuring condensed and diluted aerosols. The total flow can
be directed into a hi-vol impactor or filter, the particle loading
can be examined in situ by optical methods, or portions of the
diluted stream can be extracted, further diluted if necessary,
and sampled using an electrical aerosol analyzer, an optical
counter, or diffusion batteries and a condensation nuclei counter.
Tests to characterize the flow, mixing, and condensation in the
instrument have used a Climet optical particle counter and a
Thermo-Systerns Electrical Aerosol Analyzer.
For laboratory experiments, dilution air was taken from the
laboratory compressed air supply, which was filtered through an
absolute membrane filter, and introduced at ambient temperature
and humidity. For some experiments, the air was cooled in an ice
bath condenser.
Sample stack gas was simulated using heated, humidity-con-
trolled particle-laden air. Filtered compressed air was passed
through a Phoenix SG-20 smoke generator (Phoenix Precision Instru-
ments, Gardiner, NY) to form an aerosol which served as a source
of condensation nuclei. A metered portion of this stream was
heated and steam was injected, if desired, to adjust the humidity
of the sample stream. Experiments have also used either a nebu-
lized salt (sodium chloride) aerosol with or without added mois-
ture, or glycerol smoke, which served as a prototype condensable
vapor.
125
-------
a\
FLOW
CONTROL
VALVE
AIR
DILUTION AIR STREAM
"ZED PRESSURE
REGULATOR
SAMPLE AIR STREAM
l-ILItM tx"S/ ROTAMETER
**~J^
DILI
AEROSOL
i-'LiEfi GENERATOR HUMIDIFIER
DETECTORS
(IMPACTOR, EAA,
ETC.)
DILUTION INLET (1 OF 4)
DILUTION-MIXING
CHAMBER
FLOW
.,-STRAIGHTENER
SAMPLE INLET
(VARIABLE NOZZLE
AREA)
4181-69
Figure 70. Block diagram of dilution sampling system.
-------
Performance Characterization of Laboratory Prototype Sampler—
In order to assess the suitability of the sampler design for
simulating the conditions of real plume formation, several tests
were made to investigate the flow and mixing behavior in the
diluter. These tests on the laboratory dilution sampler were
designed to investigate two characteristics desired in the field
version. (1) The sampler must mix the sample gas and dilution
air in a reproducible and controlled manner similar to the mixing
in actual full-scale plumes. (2) The dilution and cooling of the
simulated plume must provide a local environment for condensation
which is equivalent to that in a plume.
In an actual plume, the stack gases exhaust into still air
with a definite upward momentum and a further tendency to rise
due to buoyancy. As the plume rises, transfer of its momentum
leads to mixing by entrainment of ambient air and formation of
turbulent eddies. By the time the plume momentum has been dissi-
pated and the vapors have been cooled to the local temperature,
the stack gases will be diluted by a large but finite factor.
Further dilution will occur due to diffusion and meteorological
factors, but it may be safely assumed that all condensation has
occurred prior to this point.
The dilution sampler attempts to simulate this mixing by
injecting the sample gases into a moving stream of dilution air.
A small amount of turbulent mixing will occur in the absence of a
velocity mismatch. Nevertheless, the most important mixing
mechanisms should be the same as in the actual plume. The studies
were designed to give a quantitative measure of the degree of
mixing in the sampler. A test aerosol was injected in the sam-
ple gas stream. At the top of the sampler, a probe was positioned
to withdraw a sample from an arbitrary point in the cross section
of the diluter tube. The aerosol concentration at this point was
measured using an electrical aerosol analyzer or condensation
nuclei counter. In this manner a profile of the particle distri-
bution across the diluter could be obtained.
Figure 71 shows the concentration of a salt aerosol across
a diameter of the diluter for various sample temperatures. The
degree of mixing increases sharply as the temperature of the
sample is increased. This effect is expected, since the velocity
mismatch is designed to be zero for ambient sample temperatures
and proportional to the temperature above the ambient value. As
the sample temperature is increased, the transfer of momentum to
the dilution air becomes greater and the mixing more thorough.
As can be seen in Figure 71, for sample temperatures typical of
stack conditions substantial mixing has occurred over the entire
diameter of the diluter tube. These results indicate that dilu-
tion ratios, local temperatures, and vapor saturation values
calculated on the assumption of total mixing will be meaningful.
127
-------
• 1 - 60°F
• 2- 125°F
O 3•175°F
Q 4 - 250°F
POSITION ALONG DIAMETER
Figure 71. Concentration profile of salt aerosol across diluter.
Profiling tests were also performed using glycerol smoke
aerosol. At sample temperatures over about 150 F, a substantial
fraction of the original aerosol evaporates in the sample line
and recondenses as it is cooled by dilution. Between 200 and
300°F, evaporation is complete and a detached plume of glycerol
condensation smoke is observed in the diluter. In this system,
as in field samples containing condensable vapors, molecular
diffusion could conceivably aid in the mixing process. Results
of these tests are illustrated in Figure 72. The dependent vari-
able in this figure is the half-width of the concentration profile
relative to the diameter of the diluter tube. As was evident in
Figure 71, the half-width of the profile increases with tempera-
ture above ambient. Data from the salt aerosol and the glycerol
smoke fall along the same curve, indicating that the main mixing
mechanism is probably turbulent mixing in macroscopic eddies.
This behavior corresponds to what is expected for real plumes.
Attempts were also made to monitor the particle-size distri-
bution of the condensing glycerol aerosol at different sample
128
-------
1.00
oc
LU
3
u.
O
-------
temperatures. For these studies the single point probe was
replaced by an averaging probe system using a tube with several
holes. After an impactor removed particles larger than 1 pm,
particle-size distributions were measured with an electrical
aerosol analyzer. Unfortunately, several factors combined to
limit the accuracy and reproducibility of these results. First,
the signal had an excessive amount of short-term noise due to
fluctuations of aerosol concentration reaching the holes in the
averaging probe. These fluctuations, caused by the randomness of
the turbulent eddies, were partially damped by using a two liter
"averaging chamber" in the line to the detector, but longer term
fluctuations remained and proved troublesome. Probably most of
these residual fluctuations were due to instabilities in the
aerosol generator. Due to the nonlinear response of the electri-
cal aerosol analyzer, relative errors are not the same for all
channels. Only the data in the 0.05-1 urn size range is reliable.
Within this range, semiquantitative comparisons of particle-size
distributions under different conditions are feasible.
Particle-size distribution data on glycerol smoke at three
temperatures are shown in Figure 73. Figure 73A shows the size
distribution of the aerosol as formed by the smoke generator.
At higher temperatures the size distribution becomes more nearly
monodisperse and the median size becomes greater. This behavior
can be explained for the higher temperature data (Figure 73C,
195°F) in terms of condensation of the vapor on residual nuclei.
For diffusion-controlled growth under more uniform temperatures
and flowrates a monodisperse aerosol would be expected. The
residual width of the size distribution may be attributed to the
non-uniformity of dilution and cooling rates in the sampler. The
125°F data (Figure 73B) probably reflects a differential evapora-
tion-condensation process in the heating line, as less than 1% of
the aerosol mass should be required to saturate air at the sample
temperature. In the hot region near the heating line walls,
however, evaporation would decrease the size of all particles,
followed by condensation in the cooler region near the center of
the tube. The behavior at this temperature is strikingly differ-
ent from that at ambient temperatures and is of some interest for
that reason. It has no counterpart in the stack environment,
however, and must be considered an artifact of the glycerol smoke
test.
The size distribution studies confirm that the dilutcr is
suitable for the study of condensation behavior in laboratory-
produced samples and presumably in extracted stack gases. For
further evaluation of the prototype dilution sampling system, it
was used in a test of a domestic oil-fired furnace at the EPA
IERL/RTP facilities. The participation in the test served to
demonstrate the sampling system and to obtain data for comparison
with the other sampling devices used in the furnace test. In
addition, the dilution sampling system provided real-time particle
size measurements unavailable from the other devices in the test.
130
-------
105
»
103
A. 70°F
.042 .075 .133 .237 .422 .750
105 _
io4 —
—
—
—
•^••MM
B.125°F
.042 .025 .133 .237 .422 .750
105
Q
1 104
103
C.195°F
-
•
—
: r
••••••
.042 .075 .133 .237 .422 .750
DIAMETER, urn 4181-104
Figure 73. Particle-size distribution of glycerol smoke aerosol.
131
-------
Domestic Furnace Test—
A modified commercial home furnace burning No. 2 fuel oil
was operated under controlled combustion conditions. Between the
outlet of the furnace and the barometric damper, probes were
inserted for a SASS train, a battery of instruments for flue gas
analysis, a smoke tester, and the dilution sampling system. Above
the damper were a second SASS train and duplicate flue gas ana-
lyzers. Finally, the remaining flow of flue gas plus indraft from
the barometric damper was directed into an automotive dilution
tunnel and the particulate matter in the diluted stream was
collected on an absolute filter. Among the objectives of the test
were the characterization of the particulate and organic vapor
emissions and the recovery of samples for bioassay.
The dilution sampling system was used in a modified configu-
ration for the oil-fired furnace test. The stack sampling probe
was replaced by a simple probe consisting of a short heat-traced
stainless steel tube, and the 2-um cyclone precollector normally
on the probe was left off. The sample gas was directed via a
10-ft flexible heated hose through a metering orifice and then
into the body of the dilution chamber. The dilution air consisted
of filtered room air; although provisions exist on the dilution
sampler for humidity and temperature control, these capabilities
were not used in the experiment, consistent with the operating
conditions of the dilution tunnel. In retrospect, a controlled
temperature would have been preferable for consistency of the
data. Dilution ratios of approximately 10, 20, and 40 were used
during the test.
Approximately 0.5 cfm of the diluted stream was withdrawn
into an ultrafine particle sizing system, developed by Southern
Research Institute. In this system, the gas stream was further
diluted by a factor of about 23 and analyzed with ambient par-
ticle-sizing instrumentation. The instruments included a Climet
208A optical particle counter, a Thermo-Systerns 3030 Electrical
Aerosol Size Analyzer, and an Environment One Rich 100 Condensa-
tion Nuclei Counter with or without a parallel-plate diffusion
battery. All three instruments are capable of providing near
real-time data. Unfortunately, the condensation nuclei counter
developed mechanical problems and gave data which may not be
reliable, and the optical counter was not suitable for the small
particle sizes encountered, so the bulk of useful data was
obtained by the Electrical Aerosol Size Analyzer.
The remainder of the diluted sample stream was passed through
a Sierra 230 hi-vol impactor modified to give stage DSQ'S of 0.5,
1, and 2 ym at 15 cfm. Unfortunately, no quantitative inertial
sizing could be done since the cellulose substrates could not be
weighed accurately with the equipment at hand. Water uptake by
the substrates proved to be an insurmountable problem, so finally
the stages were removed and the backup filter used to obtain
total mass loadings.
132
-------
The dilution sampling system described above was set up and
detailed real-time size data were taken using the Electrical
Aerosol Size Analyzer, mostly with the standard dilution ratio
of about 20 in the dilution sampler, but also at high (about 40)
and low (about 10) dilution ratios. All samples were taken at a
constant sample flowrate of 0.78 scfm. The dilution ratio was
changed by changing the flowrate of dilution air. For each
condition, total particulate matter from four cycles was collected,
and a side stream of the filtered sample gas was drawn through
an XAD-2 cartridge to measure organic vapors in the diluted stream.
In addition, time-resolved measurements using the Ultrafine
Sampling System were taken during all combustion cycles.
Results of Dilution Sampling Measurements
Gravimetric Measurements—
As mentioned above, attempts to measure the time-integrated
size distribution using the hi-vol impactor failed due to prob-
lems with the impactor. Unfortunately, this also prevented
measurements of total mass loading, since an unknown but visible
fraction of the mass was deposited on the impactor stages. Only
on the final test day were the impactor stages omitted and reli-
able total mass loadings obtained for the three dilutions studied.
Some information can be derived from the earlier runs, how-
ever. Measurements taken during 20 cycles indicate a loading of
5.8 mg/nm3 in the particle-size range less than 0.5 pm, the D5„
of the last impactor stage. There was visible darkening of the
substrates of all three stages used, indicating the presence of
particles larger than 2 ym.
On the three runs without the impactor stages, the total
mass loadings were measured over four combustion cycles, or 44
minutes of sample time at 0.0275 nm3/min. In addition, the low
dilution series had one 11-minute cycle and 9 minutes sampling
after shutdown. The total filter weights were: 9.1 mg, 10.3 mg,
and 15.8 mg for standard, high, and low dilution ratios, respec-
tively. After correction for the 0.5 cfm extracted for the
ultrafine sizing system, the weights give mass loadings of
7.8 mg/nm3, 8.7 mg/nm3, and 11.7 mg/nm3 for standard, high, and
low dilutions, respectively. (Note that these figures refer to
the volume of undiluted flue gas). The standard and high dilution
values are identical within experimental error. The low dilution
value is measurably higher, but is probably corrupted by the
inclusion of 9 minutes of sampling of the aerosol from the stag-
nant flue following the emissions burst at turnoff. Therefore it
was tentatively concluded that for the total time-integrated
emissions, the total mass loading measured is not demonstrably
dependent on dilution ratio in the range from 10 to 40.
There are obvious indications, however, that the dilution
and cooling process does affect the aerosol measured. The
133
-------
loadings from the dilution sampler are 25% higher than the 6.6
mg/nm3 measured by the hot filter of the first SASS train. More
striking is the obvious presence of agglomerated particles in
the 1-10 urn range, in contrast to the SASS train measurements,
which showed nothing in the cyclones that are designed to catch
particles over 1 ym.
Temporal Behavior—
Figures 74 and 75 show the time variation of emissions in
several size ranges during a typical combustion cycle. The data
in these figures were taken during run 23, which was a standard
dilution run with the furnace adjusted for a no. 1 smoke spot;
however, the general features of the data were the same for both
furnace conditions and all dilutions. Although the concentra-
tion is measured after dilution, the data are presented in terms
of number or mass per unit of the original flue gas, as was done
with the time-integrated loadings. This normalization was chosen
because the total mass of original particulate matter and con-
densable vapor per unit volume of flue gas is presumably constant,
and dilution should only affect the distribution of this mass
among the final aerosol particle sizes. Figure 74 shows the
temporal variation of the particle number distribution as measured
by the Electrical Aerosol Analyzer. All sizes show a steady
increase with burn time except the smallest range measured, which
decreases in number. The total number rises steadily to a near
maximum and may actually decrease slightly before the end of the
burn. Since the fraction of particles in the larger size ranges
increases rapidly towards the end of the cycle, the total mass
shows a steady,.but much more rapid rise, and typically increases
by a factor of 2.5 or more in the last five minutes of a combus-
tion cycle while the total particle number increases only about
20%. This can be seen in Figure 75.
Immediately after heater shutdown for the "off" period, one
or more large bursts of aerosol were noted as current spikes
roughly 10-15 seconds in duration on all channels of the EAA.
The appearance of these spikes is exactly coincident with spikes
in the carbon monoxide concentration, indicating an origin under
conditions of incomplete combustion. The aerosol bursts were
too short and .insufficiently reproducible for particle-size dis-
tribution measurements; however, since measurable currents were
seen in the highest channels, it was apparent that sizes of at
least lum were included, and that concentrations were many times
larger than those measured during the "on" cycle. Unaccountably,
no similar burst of emissions was seen at the beginning of the
combustion cycle, even though similar levels of soot were noted
on the smoke tester, and the gas analyzers suggested similar
transient conditions of incomplete combustion. Perhaps the turn-
off of the combustion air fan was involved in the higher level
of emissions at the end.
134
-------
w
O
c*
PARTICLE DIAMETER
• 0.010-0.018 Mm
A 0.018-0.032 Aim
O 0.032-0.056 pim
O 0.056-0.100 Mm
TOTAL NUMBER
TIME, min
Figure 74. Time variation of particle number concentration per unit flue gas as measured
after dilution. Dilution ratio = 19.6 (standard).
135
-------
m
c
• 0.010-0.018
A 0.018-0.032 fim
O 0.032-0.056
O 0.056-0.100 urn
— TOTAL MASS (xO.5)
4 6
TIME, min
Figure 75. Time variation of particle mass concentration per unit flue gas as measured
after dilution. Dilution ratio - 19.6 (standard).
136
-------
In principle, the particle-size distributions obtained by
the electrical mobility measurements could be independently
obtained using a series of diffusion battery/condensation nuclei
counter measurements. Due to the temporal variation in the
furnace test, such measurements would have been difficult to
implement or interpret, so the condensation nuclei counter mea-
surements were limited to measurements of the total particle
number and the fraction lost in a small parallel-plate diffusion
battery. The qualitative behavior of the CN data tends to confirm
the EAA number distribution measurements. The measured CN number
plotted against time during a combustion cycle looks much like
the total number plot in Figure 74 as determined by the aerosol
mobility measurements. Furthermore, insertion of a small par-
allel-plate diffusion battery before the CN counter causes
attenuation of the CN number by approximately the fraction
predicted from the EAA size distribution. The only disturbing
factor is that the CN counter reading is more than an order of
magnitude lower than the EAA total number, which is beyond the
reasonable calibration error of both instruments. The CN counter
appears to have been malfunctioning during the test. The vari-
ation of measured concentration with flow was anomalous. There
also appeared to be drifts in sensitivity. It was noted that on
subsequent cycles with the same operating conditions, the CN
number measured at the end of the combustion cycle would occa-
sionally vary by a factor of two without any corresponding
variation in the distribution as seen by the EAA. For these
reasons, the CN/diffusion battery data have been viewed with
some caution.
Likewise, the optical counter was not useful in the con-
figuration used. As is evident from Figure 74, essentially all
of the aerosol particles measured by the EAA are smaller than
the 0.3 ym detection limit of the optical counter. Larger par-
ticles present in lower concentrations were attenuated beyond
detectability by the dilution factor necessary for the EAA
measurements. Moreover, the 2-pm cyclone precollector prevented
measurement of larger particles such as were noted on the impactor
substrates.
Effect of Dilution Ratio on Particle Size Distribution—
In order to quantify the effect of dilution ratio on the
particle-size distribution of the aerosol in the diluted sample
stream, it is necessary to define a reference condition for the
time-varying aerosol. This reference point was chosen to be the
population which exists eight minutes into the combustion cycle.
After eight minutes, many of the combustion conditions (e.g.,
flue gas composition and temperature, smoke test emissions, and
total particle number) have begun to stabilize, yet channel
measurements on the EAA may be interpolated without the complica-
tion of the transient at the end of the combustion. Once size
distributions at eight minutes were determined for each cycle,
they were averaged for the cycles at the same dilution conditions.
137
-------
A histogram showing average size distributions at eight minutes
is shown in Figure 76. As before, the mass measurements were
normalized to a constant volume of undiluted flue gas. Several
features of the data are noteworthy. First, the particles are
very small. Considered on a mass basis, the mean particle size
is less than 0.075 ym for all dilution conditions. On a number
basis, the mean size is less than 0.03 ym. Second, there are
signs of bimodality in the data. Under all dilution conditions,
the data can be represented as the sum of two modes with mass
median diameters of about 0.025 and 0.075 ym. A third observation
is that there is a trend toward smaller mean size of the aerosol
as the dilution ratio is increased. Finally, the total aerosol
mass at eight minutes is the same within the 30-40% experimental
error for the three dilution conditions used. This finding is
consistent with the time-integrated mass measurements, and is
what would be expected if the final temperature is low enough to
deplete the bulk of condensable vapors for all dilutions studied.
Conclusions—
The picture that results from these data is that within the
dilution limits studied, a constant mass of organic vapors con-
denses into solid particulate in the diluter. Higher dilution
conditions favor a larger number of small particles; less dilu-
tion gives fewer but larger particles. This result, along with
the higher mass loadings in the diluted stream, as opposed to
the SASS measurements, indicates that condensation is indeed
taking place in the dilution sampling system.
A condensation mechanism also explains the discrepancy
between the SASS train, which saw no particles over 1 ym, the
dilution sampler, which indicated particles of that size range in
the final emission burst at the end of a combustion cycle, and
the smoke tester, which found large soot particles during a
combustion cycle as well as at both ends. In the SASS train,
operated at 200°C, presumably no condensation could occur. In
the dilution sampler, the process of dilution with cooling causes
rapid supersaturation and consequent nucleation condensation of
some of the less volatile organics (GRAV fraction), resulting in
a large number of microscopic particles which rapidly coagulate
into particles of the 0.01-0.10 ym size range. It is conceivable
that the difference in size distribution with dilution is due
entirely to the decreased concentration-residence time product
as the dilution air flowrate is increased. In the burst of
emissions at the end of a combustion cycle, the rates of conden-
sation and coagulation are both increased sharply, giving rise
to large aggregates (>1 ym) in the diluted stream. The difference
between the smoke tester and the diluter may possibly be explained
as follows. If the bulk of the condensation aerosol consists of
unburned hydrocarbons, this fraction may not be as visible in a
filter paper spot as the black soot actually measured. In addi-
tion, this fraction might condense directly on the walls of the
138
-------
water-cooled smoke tester probe rather than forming an aerosol in
the diluted stream. On the other hand, the large soot particles
seen on the smoke tester filter paper would not pass through the
sampling cyclone of the ultrafine particle sizing system.
LOW DILUTION (9.1)
_ I0///////J STANDARD DILUTION (20.6)
I I HIGH DILUTION (39.6)
I
V
0.0100-0.0178 0.0178-0.0316 0.0316-0.0562 0.0562-0.100 0.100-0.178
PARTICLE DIAMETER, Aim
4181-229A
Figure 76. Particle mass concentration per unit flue gas as measured at different dilution
ratios. All concentrations measured eight minutes after beginning of
combustion cycle.
139
-------
SECTION 4
OPERATION AND SAMPLING PARAMETERS FOR THE
INHALABLE PARTICULATE SAMPLING SYSTEM
This section contains information on (1) modified procedures
for the calibration and operation of impactors used for collection
in the inhalable particulate size range, (2) the dual-cyclone
sampling train designed for collection in this range, and (3)
the dilution stack sampling system designed to simulate the condi-
tions of stack plume formation.
CASCADE IMPACTOR CALIBRATION AND OPERATION
In order to measure IP concentrations with cascade impactors,
current operating procedures must be modified. The document
"Procedures for Cascade Impactor Calibration and Operation in
Process Streams" (EPA-6QO/2-77-004)3* was modified to incorporate
the following new procedures:
(1) Operation with a precollector cyclone having an effi-
ciency curve corresponding to that required of IP sam-
plers
(2) Some refinement of the calibration procedures
(3) A new method for selecting and treating substrates
(4) A new method for selecting the sampling rate and
locations
(5) The new data analysis method described above in Section
2.
The revised' calibration and operation manual is being published
as a separate report.
THE INHALABLE PARTICULATE SAMPLING SYSTEM
An IP Cyclone Sampler has been developed to measure particu-
late mass with size cuts at 15 ym and 2.5 ym, and a Dilution
Stack Sampling System was developed to measure the condensable
fraction of stack emissions. The IP Cyclone Sampler is to be
used as a precollector for the dilution system, allowing stack
gas containing the fine particulate fraction and condensable
140
-------
vapors to be mixed with filtered ambient air in a manner that
simulates the dilution/cooling process at the stack/ambient
interface. The simulated plume is passed through a hi-vol impac-
tor and/or filter for measurement of condensable mass concentra-
tion in the diluted gas stream. This section contains a descrip-
tion of the sampling system and instructions for its preparation
and use including sampling, data analysis, and reporting.
The Dual-Cyclone Train
The IP Cyclone Sampler as shown in Figure 77 was designed
to have two stages with D5p's of 15 and 2.5 ym at process stream
conditions in accordance with the IP specifications. It is
operated at a nominal flowrate of 23 Vmin (0.8 ft3/min) at 150°C
(300°F). The sampler consists of two cyclones and a backup filter
in series, which can be connected directly to an EPA Method 5 or
Method 17 sampling train. The backup filter can be a flat filter
or, in cases where there is a large percentage of fine particulate
matter, a thimble filter. Cyclone SRI-X separates the large-
particle fraction from the IP fraction and has a range of nozzles
allowing it to sample isokinetically in flues of different gas
velocities. Cyclone SRI-III separates the fine particle portion
of the IP sample and passes it on to the filter. Thus the fine
particulate matter is collected in the filter and the outlet tube
of cyclone SRI-III and the IP sample is collected in the outlet
tube of cyclone SRI-X and in cyclone SRI-III.
Each IP Cyclone Sampler consists of one each of the following
items:
Fabricated Parts;
Drawing Number Item Name
4181-31-C-05 501 Cyclone III Shell Inlet
4181-31-C-05 502 Cyclone III Inlet Insert
4181-31-C-06 601 Cyclone III Shell Cap
4181-31-C-06 602 Cyclone III Shell
4181-31-C-07 701 Cyclone III Modified Cap
4181-31-C-09 901 Cyclone III Modified Collection Cup
4181-37-C-l'O 1001 Cyclone X (Mk2) Collection Cup
4181-37-C-10 1002 Cyclone X (Mk2) Vortex Tube
4181-37-C-ll 1101 Cyclone X (Mk2) Body
4181-37-C-12 1201 Cyclone X (Mk2) Nozzle (Set of 11)
4181-37-C-14 1401 Cyclone III Body
Supplier: Dean Tool and Machine Shop
Box 3347
Oxford, Alabama 36203
Phone: (205) 831-4430
141
-------
to
15/Jm D50 CYC LONE
2.5/Llrn D50 CYCLONE
4181-77
Figure 77. Schematic drawing of inhalable particulate sampler.
-------
Seals;
Used On
Cyclone X Body
Cyclone III Body
Cyclone X-
Cyclone III
Connection
Cyclone III Shell
Quantity Description
2 2.75 inches O.D. x
0.062 inch diameter
Inconel X-750 Metal
0-Ring
2 1.375 inches O.D. x
0.062 inch diameter
Inconel X-750 Metal
0-Ring
1 1.375 inches O.D. x
0.094 inch diameter
Inconel X-750 Metal
0-Ring
1 3.000 inches O.D. x
0.062 inch diameter
Inconel X-750 Metal
O-Ring
Part Number
E-O-N-002750
-02-07-1
E-O-N-001375
-02-07-1
E-O-N-001375
-04-07-.1
E-O-N-003000
-02-07-1
Supplier:
Fittings;
Advanced Products Company
Defco Park Road
North Haven, Connecticut 06473
Phone: (203) 239-3341
One-half inch NPT Hex Nipple, Stainless Steel (316) , Quantity:
1, part number SS-8-HN. Connects filter holder to Cyclone
III.
Fractional tube adapter to male pipe (NPT), 5/8 inch tube
to 1/2-inch pipe, Stainless Steel (316), Quantity: 1, part
number SS-10-TA-1-8
Supplier: Cajon Company
32550 Old South Miles Road
Solon, Ohio 44139
Final Filters:
Thimble; 43 mm x 123 mm Stainless Steel Filter Holder
Part number: TH-S/2
Silicon "O" rings - package of six (for TH-S/2)
Part number: TH-1/2
143
-------
Filter Clamp Band for TH-S/2
Part number: TH-2
Toyo Roshi Co. Fiberglass Thimbles, 43 mm x 123 mm
Part number: 86R (25 thimbles/box)
Supplier: BGI, Incorporated
58 Guinar Street
Waltham, Massachusetts 02154
Phone: (671) 891-9380
Flat Filter: 2.5 inches In-Stack Filter Holder, 316 Stainless
Steel with 1/2-inch female NPT fittings, 3.0 inches
O.D.
Part number: 272
Supplier: Sierra Instruments, Inc.
P.O. Box 909
Carmel Valley, California 93924
2.4 inches Filter Disc/ Reeve Angel 923-AH (100/box)
Supplier: Arthur H. Thomas Company
P.O. Box 779
Philadelphia, Pennsylvania 19105
The standard material is 316 stainless steel. Other mate-
rials such as titanium (alloy 6A1-4V) or Hastelloy (X) may be
used for special applications. Some special applications may
require the use of flanges or some other device for connecting
the parts other than threads. For those applications it is
necessary to know that the internal dimensions are critical.
Outside dimensions can vary without affecting the collection
characteristics provided that flow into the nozzle is not inter-
rupted. For the same reason, short nozzles should not be used.
The critical (internal) dimensions of a cyclone are illus-
trated in Figure 78. These are considered to be the critical
dimensions because a change in any one of them may affect the
operational characteristics of the cyclone (with the exception
of HC and DCUP for some designs). The wall thickness of the
gas exit tube is not critical as long as it is small compared to
the diameter of the tube. The ratio of exit tube wall thickness
to I.D. is 0.11 for cyclone SRI-III and 0.08 for cyclone SRI-X.
Since there is no cyclone theory that accurately describes how a
cyclone's performance changes as the critical dimensions change,
it is necessary to keep all the dimensions within close tolerances
when manufacturing the sampler. Likewise, it is not known what
effect the surface finish has on cyclone performance, so it is
advisable to maintain the finish specified in the design drawings.
144
-------
Din
JL
'cup-
B
H
h
Z
s
"'cup
-*cup
DIAMETER OF THE CYLINDER PORTION OF THE CYCLONE,
ALSO CALLED THE DIAMETER OF THE CYCLONE
DIAMETER OF THE INLET
DIAMETER OF THE GAS EXIT
DIAMETER OF THE DUST EXIT
HEIGHT OF THE CYCLONE
HEIGHT OF THE CYLINDER
HEIGHT OF THE CONE
INSIDE LENGTH O* THE GAS EXIT TUBE
HEIGHT OF THE COLLECTION CUP
DIAMETER OF THE COLLECTION CUP 4181-27
Figure 78. The critical internal dimensions of a cyclone.
145
-------
The cyclones are designed to provide a leak-free system
when used with proper O-rings. The 0-rings recommended are made
of Inconel X-750; however, other materials may be used. Care
should be taken when ordering O-rings. The wrong thickness of
0-ring can cause leaks if the O-ring thickness is too small, and
can lead to improper alignment if the 0-ring is too thick. Thick-
ness of the O-rings for which the sampler was designed is 0.062
inch except for the 0-ring in the connection between the cyclone
SRI-X vortex tube and the cyclone SRI-III shell which is 0.094
inch.
The final stage of the sampler is an absolute filter. A
Method 17 thimble or a flat filter holder may be used. Criteria
to be considered in choosing a filter are discussed further in
the next section.
The IP Cyclone Sampler was designed to operate with straight
nozzles, which increases the total diameter of the system to just
under six inches. Use of nozzles shorter than those in the design
drawings is not advised.
Cleaning, Inspection, and Assembly—
Before using the sampler, it should be cleaned with a mild
detergent, rinsed with water, and dried. The interior surfaces
should be inspected for quality of finish and for defects due to
corrosion, scouring, or deformation. The 0-ring grooves and
mating surfaces should be flat and smooth and have no transverse
scratches. Certain concentric scratches from machining tools are
allowable if they do not affect the leak-seal integrity of the
cyclone. The O-rings should also be inspected for dents, cracks,
scratches, and deformation. Nicked or bent O-rings should be
replaced. The metal O-rings should last several tests, depending
on the test conditions, but the pliable O-rings in the thimble
holder may need to be replaced after each run. All filter material
should be removed from the filter holder between runs. The nozzles
should be cleaned and inspected for dents especially on the rim
of the nozzle. Large dents that cannot be straightened by hand
tools may require re-machining the nozzle rim.
Cyclone SRI-III is encased in a shell which must be disassem-
bled so that all the internal parts can be cleaned. Any part
which comes in contact with the sample gas is considered an inter-
nal part. For example, the entire Cyclone SRI-III collection cup
and body is considered an internal part or surface. The inlet
insert for Cyclone SRI-III need not be removed each time if no
particles are captured between the inlet and the cyclone body.
The sampler should be assembled as shown in the assembly
drawing shown in Figure 77. The proper nozzle and filter holder
should be selected and connected to the sampler at the appro-
priate places. The threads should not be coated with anything,
146
-------
such as thread lubricant, that might contaminate the sample.
Before the sampler is used in the field, the threads can be sil-
ver plated to prevent galling of the threads. No sealant is
needed for the threads on the nozzles the metal-to-metal seal
is sufficient. Teflon pipe tape should be placed on tapered pipe
threads such as those on the adapter connecting the final filter
to cyclone SRI-III.
Sampling Parameters—
Most of the sampling parameters should be measured by approved
EPA reference methods. In general, many of the parameters are
the same as for a Method 5 test. Values other than these should
be measured accurately and recorded carefully. In particular,
the results of a velocity traverse, stack gas temperature measure-
ment, and gas analysis are used to determine the testing para-
meters such as flowrate and nozzle size.
Gas velocity and analysis—The velocity traverse should be
performed in accordance with EPA Method 2. The velocities and
temperature in the flue as well as the static pressure in the
flue should be measured. An extra effort, if necessary, should
be made to obtain the gas velocity at the sampling ports. If
possible, velocity fluctuations at those points should be noted
and the percent isokineticity estimated. If the velocity fluctu-
ations are so great that + 20% isokineticity cannot be maintained,
then another sampling position should be found, if possible, where
the velocity is more stable. A constant flowrate through the
sampler must be maintained.
A gas analysis to determine the composition (molecular weight)
of the gas, including the amount of water vapor, should be per-
formed using the appropriate EPA reference methods (3, 4, and
others as necessary). The molecular weight of the gas can then
be accurately determined for viscosity and flowrate calculations
and a decision can be made regarding supplemental heating of the
sampler. An estimation of the flue gas composition based on
previous tests may be acceptable, specifically if the same type
of source has been often tested. In such a situation, an estimate
of the gas composition for the purpose of determining the gas
viscosity is appropriate.
Sampling time—The length of time required to collect an
adequate sample is dependent upon the mass loading of the gas, the
size distribution of the particles, and the flowrate of the sam-
pler. If the results of a mass test are available, the mass
loading can be obtained from them. If not, an estimate should
be made based on the pre-test survey or other information. Given
the mass concentration, an estimate of the sampling time for
initial tests can be obtained from Figure 79. Results from the
initial tests can then be used to more accurately establish the
optimum sampling time.
147
-------
00
FLUE GAS MASS LOADING
0.044 gr/acf 0.44
0.1 mg/fi 1.0
= I It I II
READ DOWN FROM MASS LOADING TO SAMPLE
RATE. READ LEFT TO TIME REQUIRED TO
COLLECT A 1 GRAM SAMPLE AT THAT SAMPLE
RATE.
, I l
SAMPLE
LOADING x RATE x TIME
I I I Mill
1 0.035
C/min 100 50
acfm 3.53
30 20
0.35
4181-29
Figure 79, Nomograph for selecting proper sampling duration.
-------
Back-up filter—The selection of the back-up filter must be
made according to the size distribution of the dust, the mass
loading, the flowrate, the amount of sample required, and the
type of analysis of the dust which will be made. If a large
fraction of the dust is fine particulate, a flat filter will load
up before the cyclones have collected an adequate amount of dust.
Therefore, a thimble, which has more filtering area, should be
used. Similarly, if the mass loading is high, a thimble should
be used to allow longer run times so well integrated samples can
be obtained. If the mass or the amount of fine particulate is
low, a flat filter should be used so that weighing errors will be
minimized.
Under some conditions, the material of which the filter
fibers are composed will react with constituents in the gas
stream (especially sulfur oxides SO ) and the filter will lose
A
or gain weight. This weight change due to flue gas exposure may
be comparable to the weight change due to dust loading. If so,
steps must be taken to reduce the weight changes due to flue gas
exposure. This can be accomplished in three ways: (1) a filter
material which has a small weight change upon flue gas exposure
should be used (such as Reeve Angel 934AH), (2) the filter material
may be preconditioned by exposing it to the flue gas prior to its
initial weighing and use in a sampling test, and (3) the filter
material may be preconditioned by exposing it to SO in the labora-
a
tory prior to its use in a sampling test. Sometimes a combination
of (1) and (2) or (1) and (3) will be necessary to sufficiently
stabilize the filters against weight changes. Method (3) is
carried out by washing the filter with sulfuric acid; the
following procedure has been found to be suitable.
Procedure for acid washing filters--
1. Submerge the glass fiber filters to be conditioned in
a 1:1 mixture (by volume) of distilled water and reagent
grade concentrated sulfuric acid at 100-115°C (212-239°F)
and keep them at this temperature for 2 hours. This opera-
tion should be carried out in a hood with clean glassware.
Any controllable laboratory hotplate is suitable.
If the filters need to be weighted down to keep them submerged,
place Teflon disks on the top and bottom of the filter stack.
The top disk can be held down with a suitable glass or Teflon
weight.
2. After removal of the filters from the acid bath, allow
them to cool to room temperature. Then place them in dis-
tilled water and rinse them with a continuous water flow of
10-20 ml/min. The filters should be rinsed until the pH of the
rinse water, after a few minutes in contact with the filters,
149
-------
is nearly the same as that of the distilled water, as
measured with a pH meter. The importance of thorough
washing cannot be overemphasized.
3. After rinsing the filters in distilled water, rinse
them in reagent grade 2-propanol (isopropyl alcohol, isopro-
panol). Submerge them and allow them to remain in it for
several minutes. Repeat this step four or five times, each
time using fresh 2-propanol.
4. Allow the filters to drain and dry. Spread them out
in a clean dry place after they are dry enough to handle.
5. When the filters are quite dry to the touch, bake them
in a laboratory oven to vaporize residual moisture or
2-propanol. Bake them at 50°C (122°F) for about two hours,
at 200 °C (392 °F) for about two hours, and finally at 370 °C
(700°F) for about three hours, to vaporize any residual
sulfuric acid. (The filters are now ready for in-situ
conditioning.)
As a final check that the acid has been removed, measure the
pH of a water extract of the filters. Tear two filters into
small pieces, immerse them in about 50 ml of distilled water,
stir the water for about 10 minutes, and measure the pH with
a pH meter. If the pH is significantly lower than that of
the distilled water, then the filters must be baked at 370°C
(700 °F) for several hours more to remove residual sulfuric
acid. A high temperature must be used since the boiling
point of sulfuric acid is 340°C (640°F).
Procedure for in-situ conditioning of filters—Even after
acid washing, unwanted weight gains have been observed in some
process streams, particularly those at extremely high tempera-
tures and containing relatively large concentrations of sulfur
oxides. If such a problem exists, further conditioning is
required. Place the filters, loosely packed, in a suitable
container preceded by a filter, insert the container in the
process stream, and draw filtered gas through the container for
6-24 hours, followed by desiccation and weighing of the filter
before use.
Sampling points—Figure 80 shows the locations of the sam-
pling points that are recommended. After the velocity profile
has been measured, IP samples should be taken at the locations
shown. Four single-point samples should be taken with the IP
trains and the mass trains, or the tests repeated with the IP
train. Any measurement of total mass concentration that differs
from the mean by more than 50% should be considered suspect. The
suspect measurement should be compared with the value found by
the other train used at the same point. If these values disagree
by less than 50%, the deviations are probably due to stratification
150
-------
a/4
b/4
4181-28
Figure 80. Recommended samp/ing points for circular and square or
rectangular ducts.
151
-------
of the dust, and all of the data should be retained. In sampling
the effluents from variable or cyclical processes, the entire
test sequence must be repeated until a good average value of the
emissions is obtained. In any event, the sequence should be
repeated at least once at each site.
Regardless of the velocity distribution in the flue, the
flowrate through the sampler must remain constant. It is desir-
able to maintain the isokinetic velocity + 20% or better, through-
out the sampling period. Minimum sampling efforts, which are
appropriate only for flues with well-developed flow profiles and
with no significant concentration stratification, are these: at
least two points within a duct should be sampled in each measure-
ment plane, and at least two samples taken at each point. In the
event of non-uniform velocity or gradients in mass concentration,
the number of samples may need to be increased for reliability.
Selecting the sample flowrate and nozzle—Since the D50 of
Cyclone SRI-X must be 15 |im aerodynamic but depends on the vis-
cosity of the gas and the flowrate through the cyclone, the sample
flowrate will be determined by the viscosity of the gas. Figure
81 illustrates the relationship between the flowrate and viscosity
for Cyclone SRI-X to obtain a D5 0 of 15 jam. Using the results of
the gas analysis, the viscosity can be calculated; otherwise the
following equation can be used: ja = (174.4 + 0.406T(°C)) x 10~6
poise. The sampling flowrate is then determined from Figure 81.
The flowrate can be used with the gas velocity to select the
appropriate nozzle using Figure 82. Only straight nozzles should
be used. The D50 of Cyclone SRI-III can be estimated from previous
calibration curves. (See Figure 60.)
Leak testing—Before use, the sampler should be leak tested.
Immediately after assembly, the operator should perform a leak
test on the sampler. This can be performed in the lab with a
minimum amount of equipment. Leak tests should also be made at
operating temperature - after the Inconel O-rings have thermally
expanded. This expansion will generally help to seal the cyclones.
It is suggested that the cyclones be checked for leaks at
the same time the mass trains are checked.
Sampling—
There can be no substitute for trained and experienced per-
sonnel to operate the IP Cyclone Sampler. The same care should
be taken with the sampler during testing as is required for Method
5 and cascade impactor operation. Essentially, the sampler should
be operated the same way as in Method 5 except for the sampling
points and the need to maintain a constant flowrate.
The IP Cyclone Sampler is still in the prototype stage of
development and has not been used extensively in field applica-
tions. It is expected that better, and more detailed, operating
152
-------
30
25
20
ui
15
10
-L
150 200 250
VISCOSITY (77), micropoise
300
41S1-23A
Figure 81. Gas flowrate versus viscosity at
diameter for IP cyclone SRI-X.
15 (Jtm aerodynamic
153
-------
Ul
100
90
30 40 50 60 80 100
4181-25
Figure 82. Nomograph for selecting nozzles for isokinetic sampling, using cyclone samplers.
3 4 5 6 8 10 20
GAS VELOCITY, m/sec
-------
procedures will be developed as this experience is gained.
Potential problem areas include possible reentrainment from
overloading Cyclone SRI-III at high mass loadings and long run
times, extremely long run times at flues with low mass loadings
and loss of particles during the unloading process.
Flowrate—During the run, the predetermined flowrate must
be maintained to ensure stable cut points. Any attempt to modu-
late flow to provide isokinetic sampling will destroy the utility
of the data by changing the cut points of the cyclones. As the
final filter collects particulate, the pressure drop across it
will increase, lowering the sample flowrate. The flowrate should
be monitored and adjusted when necessary.
Traversing—During traversing (moving to a new point or new
port) all motion should be smooth and brief to avoid bumping or
vibrating the sampler. When removing or inserting the sampler,
care must be taken not to scrape the nozzle on the port wall.
Also, the sampler should not be allowed to bump against the far
inside wall of the flue.
Orientation—The IP Cyclone Sampler is designed to operate
in any orientation with equally accurate results. However, when
the sampler is operated in the upright position (the backup
filter above Cyclone SRI-III which is above Cyclone SRI-X), the
flow should not be terminated until the sampler is in the HORI-
ZONTAL position. Otherwise some dust might fall from one stage
of the sampler into another and thus be measured where it is not
collected. After the flow has been terminated, the sampler can
be transported to the lab in the horizontal position with the
nozzle plugged or covered to avoid contamination or loss of the
samples.
Data logging—In addition to the sampling parameters and
process information usually recorded in a Method 5 test, the
operator of the IP Cyclone Sampler should also record the sampler
identification, sampler orientation, filter identification, gas
viscosity, and the sampler flowrate.
Sample Retrieval and Weighing—
Unloading the sampler—After the sampler has cooled down to
nearly ambient temperature and brought into the lab, it is
"unloaded" by removing the catch of particulate matter. Great
care is needed to insure that all of the particulate matter is
recovered and transferred to the proper sample containers. The
following procedure has been found to be suitable.
First, remove sample collected in Cyclone SRI-X. With the
nozzle plugged, turn the sampler from a horizontal position to a
position 45° from the horizontal. Cyclone SRI-X should be lower
155
-------
than the backup filter. Holding the sampler in this position,
carefully unscrew the body of Cyclone SRI-X from the cap. Holding
the body underneath the cap, brush the particles adhering to the
bottom of the cap and the outside of the gas exit tube into the
cyclone body. A no. 7 camels hair artists' brush (or the small
nylon brush made for cleaning electric shavers) is suggested.
Rotate the cyclone so that the nozzle is pointing upward at
30-45° from the horizontal and brush the particles caught inside
the nozzle down into the cyclone. Rotate the cyclone back to an
upright position and use a downward, pushing motion to brush the
particles on the inside walls of the body of the cyclone down
into the collection cup. Carefully detach the collection cup
from the body and, holding the body over the cup, brush the
particles adhering to the underside of the body into the cup.
At this point, all of the particles caught by Cyclone SRI-X should
be in the cup. If a cup insert is used, remove it with a pair of
tweezers for desiccation and weighing. If not, transfer the
sample by pouring and brushing to a pre-weighed container.
Cyclone SRI-III should still be in the 45°-from-horizontal
position, with the backup filter higher than the cyclone. Very
carefully, unscrew the shell of Cyclone SRI-III and, holding it
over a sample container, brush the dust down into the container
from the shell, the cyclone, and the inside of the exit tube and
connecting pieces of Cyclone SRI-X. Then, holding the container
underneath, brush off the dust clinging to the outside of the
Cyclone SRI-III body and collection cup and the now exposed part
of the cap. After all this dust has been removed, slowly unscrew
the Cyclone SRI-III body from the' cap. Hold the body upright and
brush all the particles down into the collection cup. Remember
to brush the dust out of the inlet. Then slowly remove the
collection cup from the body and transfer its contents to the
sample container. Finally, hold the sample container underneath
the vortex tube and brush the dust on the outside of the exit
tube and the underside of the vortex tube into the container.
Now turn the connecting tubing and filter so that the
Cyclone SRI-III cap is directly above the filter. Brush the dust
collected on the inside of the gas exit tube and the connecting
tubing down onto the filter. When all the internal surfaces
directly ups.tream of the filter have been brushed clean, remove
the filter and place it in a proper sample container for desicca-
tion and weighing.
In some cases it may also be necessary to wash the internal
nozzle and cyclone surfaces with a solvent, such as methylene
chloride, into a preweighed bottle or aluminum cup. If the dust
is hard and dry, the particles can be brushed off into the weigh-
ing container; if the particles are sticky or wet, some type of
washdown procedure should be used. Use a solvent that is consi-
derably more volatile than the particulate matter. Cover the
156
-------
sample container loosely and allow the solvent to evaporate
completely before desiccation and weighing. Desiccation is
complete when consecutive weighings (two hours apart) of the
sample yield the same values. In a sealed container of a
desiccant like silica gel, desiccation takes about 24 hours.
Drying and weighing — Each of the particulate containers must
be dried to a constant weight, with two hour checks used to
establish the uniformity of the weights. Samples of hard, non-
volatile particles are often dried in a convection oven at a
temperature of 212 °F, kept in a desiccator until cooled to room
temperature, weighed, and then check-weighed two hours later.
Volatile particles present special problems which have to be
dealt with according to the particulate characteristics and
sampling goals. One technique for particles which are volatile
at elevated temperatures is to desiccate them 24 hours before
weighing. Whatever the technique used, constant weight of the
samples after further drying is the criterion to be met. Record
the results of the weighings and any notes in a notebook and keep
it with the run sheets.
Data Analysis and Reports —
Data taken in various inhalable particulate studies should
be readily comparable and independent of the sampling team. The
data should therefore be presented in identical format,
Calculation of mass concentration-- The total mass loading
or total concentration is found by dividing the total mass
collected, M, by the volume of gas sampled (product of the sample
flowrate, Q, and the sampling duration, t) . Similarly, the
inhalable particulate concentration is calculated by dividing
the mass collected in Cyclone SRI-III and the filter by the
volume of gas sampled, and the fine particulate concentration
is found by dividing the mass on the filter by the gas volume.
The product of the sample flowrate at stack conditions and the
sampling duration gives the total actual volume of gas that was
sampled. The equation below is used to calculate the equivalent
volume of gas at standard (normal) temperature (20°C) and pressure
(760 torr) and at dry conditions (water vapor diluted),
\T = rvH Stack Pressure x 293°K n -F ^
vdry yr Stack Temperature x 760 torr x ( H2Cr
where fv _ is the fraction of water vapor present in the gas.
** 2 ^
The mass concentration at "normal" conditions is thus found by
dividing the mass by V, . The results should be summarized in
a format like that below:
157
-------
PARTICIPATE MASS CONCENTRATION
(Example)
Cumulative %
Actual Conditions Normal Conditions Cumulative Including
mg/m3 gr/ft3 mg/nm3 gr/ft3 % Condensables
Total 1.6xl05 0.30 2.5xl05 0.46 100.0 60
IP (<15 jim) 1.5xlOs 0.27 2.2xl05 0.41 52.5 54.5
FP (<2.5 pin) 2.7x10" 0.05 4.1x10" 0.08 8.0
FP (<2.5 pm, 4.2x10" 0.08 6.4x10" 0.12 — 11.9
including
condensables)
IP is inhalable particulate, FP is fine participate
Cyclone calibration data—In each report, calibration curves
showing the collection efficiency and D5 0 cut point for the type
of cyclone that was used should be included. Any extrapolation
from these curves to the actual field test conditions should be
recorded with explanatory notes. Likewise, the filter material
that was used should be identified fully and any conditioning or
acid treatment should be described. The particle retention
characteristics, in the form of calibration curves or manufac-
turer's data, should also be reported. If a chemical analysis
of the filter is performed, then the background elemental analyses
should also be reported. This information can be obtained from
the manufacturer or, better, determined by the laboratory on
clean filters.
Reporting data—Each report should include all the parameters
of each test.In general, the data required are those required
by Method 5. Specifically, the following items should be reported:
- Gas viscosity
- Sample flowrate at actual and normal conditions
- Mass collected in each stage of the sampler
- D5 o cut point of each cyclone for each test
- Particle density
- Sampling duration
- Percent isokinetic gas flow at each point sampled.
A simple report outline might follow the scheme of the one
below:
I. Conclusions
II. Recommendations
III. Introduction
IV. Plant/Process Description
V. Sampling Strategy
VI. Mass Train Data (if applicable)
VII. Cascade Impactor Data (if applicable)
VIII. IP Sampler Data
IX. Appendices
X. References
158
-------
The Dilution Stack Sampling System
Design and Operating Procedures—
A diagram of the major components of the Stack Dilution
Sampling System is shown in Figure 83. In operation, gases from
the process stream are drawn through the IP Cyclone Sampler, in
which particles with aerodynamic diameter greater than 15 ym and
those in the range 2.5-15 ym are removed in two stages. The
stack gas containing the fine particle fraction (<2.5 ym) and
condensable vapors passes through the heat-traced probe and
flexible sample line and is introduced axially into the bottom
of the cylindrical dilution chamber. At this point the stack gas
is mixed with dilution air to form a simulated plume which flows
upward through the dilution chamber, through a standard 20 cm x
25 cm (8 in. x 10 in.) hi-vol filter which collects the fine
particulate and any new particulate formed by condensation. The
diluted stream is exhausted by a 1 HP blower or optionally by a
standard hi-vol blower.
Dilution air is drawn from the ambient through a blower and
forced through an ice bath condenser. In this condenser, the air
is cooled to 5-8°C, depending on the flow and ambient temperature.
More significantly, the dilution humidity is reduced to about
0.57% by volume, corresponding to saturated air at the ice point.
After the condenser, the air is reheated as required to reach 21°C
(70°F) at the dilution chamber inlet, filtered through an HEPA-
type absolute filter, and introduced into the dilution chamber.
The dilution air enters through a single tangential inlet at the
base of the dilution chamber and passes through a set of flow-
straightening screens into the annular region surrounding the
sample gas inlet. The ratio of the areas of the two inlets is
such that for sample gas at room temperature the velocities of the
sample and dilution streams are equal. Sample gas at stack
temperature will be injected at a higher velocity proportional to
the thermal expansion of the heated gas stream. This was judged
the best simulation of a buoyant plume injected into stagnant air.
System Description—
The geometric and flow specifications have been set by
several constraints. The sample flowrate is set by the flow
requirements of the IP Cyclone Sampler. Ideally, to approximate
the conditions found in actual plumes, the dilution ratio should
be high (approaching 103-101*) and the mixing times long (tens of
seconds). The actual dilution conditions represent a compromise
dictated by limitations on the size of a portable field instru-
ment. Geometric and flow specifications are as follows:
159
-------
HI-VOL IMPACTOR
AND/OR FILTER
PROCESS STREAM
o
/\
TO HEATERS, BLOWERS
TEMPERATURE SENSORS
MAIN CONTROL
EXHAUST BLOWER
. DILUTION
CHAMBER
TO ULTRAFINE
PARTICLE SIZING
SYSTEM (OPTIONAL)
DILUTION AIR
HEATER
• DILUTION AIR
BLOWER
ICE BATH
TO ORIFICE
PRESSURE TAPS
_EL
FLOW, PRESSURE
MONITORS
4181-266C
Figure 83. Diagram of Stack Dilution Sampling System.
-------
Active length of dilution tube: 1.22 m (4 ft)
Total height of sampler: 1.8-2.1 m (6-7 ft)
I.D. of dilution tube: 21.3 cm (8.4 in.)
I.D. of sample inlet tube: 4.27 cm (1.68 in.)
Active dilution volume: 43.6 I (1.54 ft3)
Sample flow: 17 n£/min (0.6 ft3/min-determined by cyclone
cut point)
Dilution flow: 425 njl/min (15 ft3/min)
Dilution factor: 25
Residence time: 6.2 sec
Sample velocity: 25 cm/sec at 150°C
Dilution air velocity: 19.8 cm/sec at 21°C
Since the effect of varying dilution air temperature and
humidity cannot be easily predicted for all typical process
streams, standard conditions of 0.57% moisture by volume at 21°C
(corresponding to about 24% RH at 70°F) were chosen. This
relatively dry dilution air should not be subject to water
condensation for normal stack samples, yet is more realistic
than totally dry air.
Sample probe and heated hose—The sample probe and heated
hose assemblies are shown in Figure 84. The probe is a heat-
traced stainless steel assembly consisting of a core of 5/8-in.
tubing wrapped with high-temperature heating tape, glass-fiber
insulation, and a stainless steel outer jacket. The heating tape
covers only the final 4 ft of the probe which extends outside the
stack. Thermocouples monitor the temperature of the process
stream, of the inner tube walls, and of the gas sampled between
the probe and the hose. The heated hose is a commercially
available, heat-traced, flexible sample hose with a 1/2-in. O.D.
Teflon core, a nichrome wire heating element, specially installed
RTD and thermocouple temperature sensors, glass-fiber insulation,
and a weather-resistant outer coating. The hose is rated for
operation below 200°C.
Dilution air line—The dilution air line, pictured in Figure
85, consists of a portable blower, a high-flow condenser, air
heater (consisting of a coil heating element), an HEPA-type
absolute filter, and a flow metering orifice, all joined by
flexible tubing. The purpose of the line is to provide a high
flow of filtered air with controlled temperature and humidity.
Dilution air temperature is monitored downstream of the condenser
and in the base of the dilution chamber. The cooling-reheating
process provides a means of reducing the humidity to the specified
value, and also allows temperature control over a wide range of
ambient conditions. Very dry ambient air may have to be pre-
humidified to reach the proper level, but such conditions should
not commonly occur except in sub-freezing weather.
Dilution chamber inlet—The dilution chamber inlet assembly
shown in Figure 86 determines the mixing conditions of the sample
161
-------
620-186
84. Sample probe and heated hose for the Dilution Stack Sampling System.
F/^t/re 55. 7?7e dilution air line of the Dilution Stack Sampling System.
162
620-187
-------
620-188
Figure 1. Inlet assembly of the dilution chamber of the stack dilution sampling system.
163
-------
gas and dilution air streams. The sample inlet is heated and
offers no flow restriction. The temperature of the inlet tube is
monitored at two points. The dilution air section is thermally
isolated from the sample gas inlet by a spacer of glass-reinforced
Teflon in order to minimize preheating of the dilution air or
cooling of the top of the sample inlet tube, which might lead to
condensation on the tube walls. The dilution air is injected
tangentially; the shape of the dilution chamber inlet and the
fine mesh flow straightening screens combine to minimize the
vortex effect and provide a nearly-uniform dilution air velocity
profile.
Dilution chamber body—The dilution chamber itself is simply
a long tube in which the mixing and cooling process occurs.
Provisions have been made to extract a small sample (10-30 £/min)
for sizing with optical, electrical mobility, or diffusional
means. The top of the dilution chamber body adapts to a square
flange for mounting the hi-vol filter or impactor assembly.
Hi-vol filter and impactor assembly—The hi-vol filter and
impactor assembly are shown in Figure 87. In the figure, the
impactor base plate and one impaction stage are mounted on the
hi-vol filter holder. For standard operation, measurements would
include only filter measurements of total particulate mass loadings
in the diluted stream.
620-189
Figure 87. Hi-vol filter and impactor assembly of the Stack Dilution Sampling System.
164
-------
Instrument packages—Figure 88 shows the Pressure and Flow
Measurement Module, which is the smaller of the two suitcase-
mounted instrument packages. In this module, magnahelic pressure
gages monitor the pressure differential in the dilution chamber
with respect to ambient, the pressure drop across the impactor
and/or filter, and the flow through orifices in the sample,
dilution air, and exhaust lines.
620-190
Figure 88. Pressure and flow measurement module.
The Heater and Flow Control Module contains all electrical
connections to the Stack Dilution Sampling System (Figure 89).
Powered by two independent 110V, 20A power lines, the package
contains proportional temperature controls for the probe, heated
hose, sample inlet, and dilution air reheat. A digital thermo-
couple readout monitors temperatures at various points in the
system, and variable auto transformers supply power to the
dilution and exhaust blowers.
165
-------
S 20-1 91
Figure 89. Heater and flow control module.
Assembly— , , ,
The dilution chamber, dilution air line, probe, and hose
should be assembled as shown in Figures 83-87. Since the inter-
connections are made with standard tube fittings or with 1.5-in.
I D flexible tubing, assembly is relatively straightforward, and
only details which may not be obvious are presented below. The
assembly of the IP Cyclone Sampler is described in the preceding
section of this report. Electrical and thermocouple connections
to the control module are straightforward, since all connectors
are either numbered, color-coded, or unique to the mating compo-
nent. Likewise, pressure taps leading the pressure/flow monitor
module are clearly labeled.
Setup at a typical site would proceed as follows. The dilu-
tion chamber body is set up in the optimum location near the
sampling port. The dilution chamber base is assembled .as shown
in Figure 86, using the special tool shown to screw in the Teflon
insulating spacer. The flow straightening screens and retaining
gasket are positioned in the base, the O-ring inserted, and the
base bolted to the mating flange in the dilution chamber. Since
the orientation of the dilution air inlet is not important, it
166
-------
should be positioned in a direction convenient for setup of the
dilution air line. The heated hose and probe should be attached
to the sample gas inlet, and the dilution air line set up in the
order shown in Figure 85. While the interconnection tubing
lengths for the dilution air line are not specified, it is pref-
erable that they be kept short in order to minimize pressure drop
and warming of the chilled dilution air. If the heating tape and
insulation have not been installed on the sample -gas inlet tube,
they should be installed at this time, taking care that the
insulation extends over the connector to the heated hose. The
IP cyclone sampler should be prepared according to the cyclone
sampler procedures manual and attached to the probe with the
appropriate fittings. For normal operation, the extraction tube
for fine particle sampling is removed and fitting capped.
Since the hi-vol filter assembly is operated in an upsidedown
configuration, the filter must be mounted with some care. For
systems which have the optional hi-vol impactor, it is most
convenient to use the impactor base plate as a retainer even
though no impactor stages are used. Alternatively, a hi-vol
filter hold-down frame, such as the Sierra 305-2017, may be used
as a support for the filter. The filter holder is removed from
the apparatus, the filter mounted, and the impactor base plate
(or hold-down frame) secured above the filters as shown in Figure
87. The assembly can then be mounted on the dilution chamber
using a second rubber gasket. Finally, the exhaust orifice tube
and exhaust line can be connected.
Sampling Parameters—
Most of the sampling parameters should be measured by approved
EPA reference methods. Many of the parameters are the same as
for a Method 5 test. Values other than these should be measured
accurately and recorded carefully. In particular, the results of
a velocity traverse, stack gas temperature measurement, and gas
analysis are used to determine testing parameters such as flow-
rate and nozzle size for the cyclone train. Stack temperature
and moisture content will determine the final moisture content of
the diluted stream and may in some cases preclude normal operation
due to water condensation in the dilution chamber. Ambient tem-
perature and humidity may likewise require adjustment in operation.
Temperature and moisture content of stack gas and ambient
air—While the standard dilution air humidity was deliberately
chosen to be rather low (corresponding to a dew point of 0°C)
situations might arise which make testing at the standard humidity
difficult or unfeasible. One such situation might arise if the
process stream has such a high moisture content that condensation
of water occurs in the dilution chamber. When process streams of
typical compositions are diluted by air of the composition used
in the Stack Dilution Sampling System, the relative humidity of
the diluted stream reaches a maximum at relatively low dilution
factors of 2-5. Thus, transient condensation of water plume
167
-------
might occur in the dilution chamber followed by evaporation prior
to the hi-vol filter. This effect would be detrimental to system
operation if the water fog persists at the full dilution, causing
the filter to blind rapidly. Such blinding would occur in the
rather unlikely case of a stack gas consisting of over 33% mois-
ture by volume. If such a stream is encountered measurements
with the Stack Dilution Sampling System must be modified by
completely drying the dilution air with desiccant columns and/or
increasing the dilution air flow and thus the dilution ratio.
A more likely complication might occur if the dew point of
the ambient air is lower than 0°C. This would certainly be the
case in subfreezing weather conditions; it could also occur under
warmer but extremely dry conditions. In such circumstances it
will not be possible to maintain the standard dilution air humi-
dity without some preliminary humidification. If such low
humidity conditions are encountered, values of ambient tempera-
ture and humidity should be recorded every hour during the
sampling cycle and notation made on the data sheets that the
dilution air humidity is expected to be low. While the effects
of such lower humidity on the condensation process are not
completely understood, the differences in the data obtained are
expected to be minimized.
Sampling time—Even in the absence of condensable matter,
the proportion of particulate in the fine particle fraction
(<2.5 um) passed by the cyclone samplers is expected to be high
enough that the filter is adequately loaded before sufficient
sample is collected in the cyclones. In general, measurement of
the cyclone contents is important, so sampling time will be
determined by the need for an adequate sample in the cyclone
train. Some guidelines for estimating the sampling time are
given in the previous section on operation of the IP Sampler.
If only the condensable and fine particle fractions are of inter-
est, the sample run can be terminated when a substantial increase
in pressure drop across the filter is noted.
Gas velocity, temperature, and analysis—Since the flowrate
of the cyclone sample is dependent upon the viscosity of the
stack gas, and since the nozzle selected for this sampler depends
on the stack velocity, these parameters must be-measured carefully.
The procedures for measurement of stack parameters and selection
of sample nozzle and flowrate are outlined in detail in the
previous section of this report on operation of the IP Sampler.
Filter and substrate selection—Under some conditions, the
material of which the hi-vol filters (or impactor substrates) are
composed will react with constituents in the gas stream (espe-
cially SO ) and the filter will gain or lose weight due primarily
X
to sulfate formation. This problem is common in in-stack sampling
and has been noted in ambient hi-vol sampling as well. The
168
-------
diluted stack gas represents an intermediate situation more
similar to ambient than in-stack sampling. Difficulties due to
sulfate formation are expected to be avoided by a suitable choice
of filter material. Candidate glass fiber materials are Reeve
Angel 934AH or Schleicher and Schuell 30, both of which exhibit
low weight gain at stack conditions. Alternatively, some glass
fiber filters, such as Gelman Spectrograde, exhibit low weight
gain at ambient temperatures even though they are unsuitable for
in-stack use. Other filter materials, such as fibrous quartz,
Teflon, or cellulose, may be suitable for specialized applications,
especially when chemical analysis of the particulate matter is
planned. However, for accurate gravimetric determinations these
materials are generally inferior to glass fiber filters and are
not recommended for general use.
Desiccation and weighing of filters and substrates should
be performed with the care customary for Method 5 and cascade
impactor tests.
Sampling—
As for a Method 5 or cascade impactor operation, the Stack Di-
lution Sampling System operation requires great care in all phases.
Operation of the system will be similar to the operation of the
IP Cyclone Sampler without dilution; the previous section of this
report contains information regarding operation of the IP Sampler.
The Stack Dilution Sampling System is still in the prototype
stage of development and has not been used extensively in field
applications. It is expected that better and more detailed
operating procedures will be developed as this experience is
gained. Potential problems include difficulty in flow measurement
and control, difficulties with port access and portability due to
the size of the instrument, loss of particulate to the walls or
by re-evaporation from the hi-vol filter, and unwanted interaction
between the collected particulate matter and the gas stream.
Flowrate—As mentioned above, the sample flowrate is pre-
determined by the requirements of the cyclone sampler. The
diluted flowrate is fixed at its standard value of 425 normal
liters per minute in order to maintain a reproducible aerodynamic
pattern in the dilution chamber. In addition, when the optional
hi-vol impactor is used, the diluted flowrate must remain con-
stant to insure stable cut points for the impactor stages. Both
sample and diluted flowrates may tend to change due to drift in
the two blower system or loading of the hi-vol filter. Both
flowrates should be monitored and adjusted as necessary.
Sampling points—Because of the size of the Stack Dilution
Sampling System, it may not be possible to sample at several
points in a duct. It is recommended that a port be chosen at a
spot in the duct where stratification of particulate matter is
not expected, and that this lack of stratification be confirmed
169
-------
by measurements with mass trains or with the IP train without
dilution. Measurements with the Stack Dilution Sampling System
should be made at two representative points in the duct where
undiluted IP or mass train measurements have been made. At least
two samples should be taken at each port. In the event of non-
uniform velocity or gradients in mass concentration, the number
of samples may have to be increased to obtain reliable results.
In sampling the effluents from variable or cyclical processes,
the test sequence must be repeated until a good average of the
emissions is obtained.
Data logging—The operator should record the sampling para-
meters and process information appropriate for the IP Cyclone
Sampler. In addition, the exhaust, sample, and dilution air
flowrates and pressure drops in the system should be logged, along
with readings of the temperature at all monitoring points in the
system, and the temperature and relative humidity of the ambient
air.
Size analysis of diluted aerosol—While the standard operating
procedure for the Stack Dilution Sampling System calls only for
measurement of total particulate mass in the diluted sample, it
may be desirable for some studies to know the particle size
distribution in the diluted gas stream. For this purpose, the
dilution chamber was designed to accept a Sierra Model 230 High
Volume Cascade Impactor through which the entire diluted flow
passes prior to the hi-vol filter. Operation of the Stack Dilu-
tion Sampling System in this configuration requires only a few
modifications to the normal operating conditions. The impactor
stages 3-6 will be mounted on the impactor base plate prior to
loading the filter. The stages and substrates are secured to
the mounting posts on the base plate by the knurled mounting
screws. While some attention must be paid to proper alignment
of the substrates and slotted stages, the procedure is straight-
forward. As mentioned above, it is important that the diluted
flowrate should be maintained at its specified value in order to
maintain constant stage D50's. At the diluted flowrate of 425
£/min, the nominal D5Q values for stages 3-6 are 2.4, 1.6, 0.84,
and 0.47 ym, respectively.
In order to allow study of submicron particles formed in
the dilution/cooling process, provision has been made to extract
a portion of the diluted stream for subsequent size analysis
using electrical mobility, diffusional, or optical size analysis.
The extraction probe can be inserted in the bored-through tube
fitting at the top of the dilution chamber. Approximately 10-30
i/iciin can be withdrawn without excessive losses for particles
below 1 urn. The extracted stream may have to be further diluted
for use with the sensitive sizing instruments mentioned above.
170
-------
Sample Retrieval and Weighing—
After the end of a sample run, the hi-vol filter (and sub-
strates, if applicable) and the cyclone catches should be removed
with care, dried, and weighed. The hi-vol filter assembly should
be removed as a unit from the dilution chamber body. The filter
and substrates should then be removed, taking care to avoid
tearing or loss of filter material. Detailed proceedings for
unloading the IP Cyclone Sampler are found in the previous section
on IP Sampler operation.
Data Analysis and Reports—
The data taken with the Stack Dilution Sampling System should
be reported with that taken using the IP Cyclone Sampler. The
reporting procedures are like those for the Dual Cyclone Operation
Section of this report.
REFERENCES
1. Johnson, J. W., G. I. Clinard, L. G. Felix, and J. D. McCain.
A Computer-Based Cascade Impactor Data Reduction System.
U.S. Environmental Protection Agency Report EPA-600/7-78-042
(NTIS PB 285433), Southern Research Institute, Birmingham, AL,
March 1978. 601 pp.
2. McCain, J. D., G. I. Clinard, L. G. Felix, and J. W. Johnson.
A Data Reduction System for Cascade Impactors. U.S. Environ-
mental Protection Agency Report EPA-600/7-78-132a (NTIS PB
283173), Southern Research Institute, Birmingham, AL, July
1978. 44 pp.
3. Scheid, Francis. Numerical Analysis. McGraw-Hill, New York,
1968. p. 65.
4. Smith, F. H. The Effects of Nozzle Design and Sampling Tech-
niques on Aerosol Measurements. EPA-650/2-74-070, U.S. Envi-
ronmental Protection Agency, Research Triangle Park, North
Carolina, 1974. 89 pp.
5. U.S. Environmental Protection Agency. Standards of Perfor-
mance for New Stationary Sources. Federal Register 42 (160):
41776-41782, 1977.
6. Smith, W. B., P. R. Cavanaugh, and R. R. Wilson, Jr. Tech-
nical Manual: A Survey of Equipment and Methods for Particu-
late Sampling in Industrial Process Streams. U.S. Environ-
mental Protection Agency Report EPA-600/7-78-043 (NTIS PB
282501), Southern Research Institute, Birmingham, AL, March
1978. 280 pp.
171
-------
7. Blake, D. E. Source Assessment Sampling System: Design and
1 Development. EPA-600/7-78-018, U.S. Environmental Protection
; Agency, Research Triangle Park, North Carolina, 1978. 221 pp.
8. Elder, J. C., L. G. Littlefield, M. I. Tillery, and H. J.
Ettinger. Preliminary Design of a Prototype Particulate
Stack Sampler. Los Alamos Scientific Laboratory Informal
Report LA-7286-MS, 1978. 17 pp.
9. Ringwall, C. G. Compact Sampling System for Collection of
Particulates from Stationary Sources. EPA-650/2-74-029,
U.S. Environmental Protection Agency, Research Triangle Park,
North Carolina, 1974. 108 pp.
10. Smith, W. S., and E. W. Stewart. Field Evaluation of an
Autoisokinetic Stack Particulate Sampling System. EPA-600/
2-77-035, U.S. Environmental Protection Agency, Research
Triangle Park, North Carolina, 1977. 58 pp.
11. Miller, F. J., D. E. Gardner, J. A. Graham, R. E. Lee, Jr.,
W. E. Wilson and J. D. Bachmann. Size Considerations for
Establishing a Standard For Inhalable Particles. J. Air
Pollut. Control Assoc. 29(6):610-615, 1979.
12. Corn, M. and F. Stein. Reentrainment of Particles from a
Plume Surface. Amer. Ind. Hyg. Assoc. J. 26:325, 1965.
13. Rao, A. K., and K. T. Whitby. Non-ideal Collection Char-
acteristics of Single Stage Cascade Impactors. Amer. Ind.
Hyg. Assoc. J. 38:174, 1977.
14. Gushing, J. M., J. D. McCain, and W. B. Smith. Experimental
Determination of Sizing Parameters and Wall Losses of Five
Source-Test Cascade Impactors. Environ. Sci. Technol. 13:
726, 1979.
15. Smith, W. B., R. R. Wilson, Jr., and D. B. Harris. A Five-
Stage Cyclone System for In-Situ Sampling. Environ. Sci.
Technol. 13:1387, 1979.
16. Yule, T. J., and C. G. Broniarek. An Experimental Study of
Virtual Impactors. In: Proceedings, Advances in Particle
Sampling and Measurement (Daytona Beach, FL, Oct. 1979),
Wallace B. Smith, Ed. U.S. Environmental Protection Agency
Report EPA-600/9-80-004, Southern Research Institute, Birm-
ingham, AL, Jan. 1980. pp. 1-21.
17. Fuchs, N. A. The Mechanics of Aerosols. Pergamon, New York,
,; 1964. p. 110.
172
-------
18. Stein, F., W. A. Esmen, and M. Corn. The Shape of Atmosphe-
ric Particles in Pittsburgh Air. Atmos. Environ. 3:443,
1969.
19. Rimberg, D., and J. W. Thomas. Comparison of Particle Size
of Latex Aerosols by Optical and Gravity Settling Methods.
J. Colloid Interface Sci. 32:101, 1970.
20. Corn, M., F. Stein, Y. Hammad, S. Manekshaw, W. Bell, S. J.
Penkola, and R. Freedman. Physical and Chemical Character-
istics of 'Respirable' Coal Mine Dust. Ann. N. Y. Acad.
Sci. 200:17, 1972.
21. Medical Research Council Panels, cited by R. J. Hamilton
and W. H. Walton. The Selective Sampling of Respirable
Dust. In: C. N. Davies, ed. Inhaled Particles and Vapors,
Pergamon Press, Oxford, 1961.
22. Mercer, T. T. Aerosol Technology in Hazard Evaluation.
Academic Press, New York, 1973, Chapter 8.
23. Natanson, G., cited in N. A. Fuchs, The Mechanics of Aero-
sols, Pergamon Press, New York, 1964. p. 112.
24. Rusanov, A. A. Determination of the Basic Properties of
Dust and Gases. In: Ochistka Dymovykh Gasov v Promyshlennoy
Energetike (Cleaning Stack Gases in Industrial Power Engi-
neering) , A. A. Rusanov, I. I. Urbakh, and A. P. Anastasiadi,
eds. "Energiya," Moscow, 1969.
25. Hochstrasser, J.M. The Investigation and Development of
Cyclone Dust Collector Theories for Application to Miniature
Cyclone Presamplers. Ph.D. Thesis, University of Cincinnati,
Cincinnati, 1976.
26. Lippmann, M., and T. L. Chan. Cyclone Sampler Performance.
Staub Reinhalt. Luft 39:7-12, 1979.
27. John, W. G., P. Reischl, and J. J. Wesolowski. Size-Selec-
tive Monitoring Techniques for Particulate Matter in Cali-
fornia Air. AIHL/SP-12. Final Report, California Air
Resources Board Contract No. A5-00487, 1978.
28. Chan, T. L., and M. Lippmann. Particle Collection Effi-
ciencies of Air Sampling Cyclones: An Empirical Theory.
Environ. Sci. Technol. 11:377-382, 1977.
29. Marple, V. A. A Fundamental Study of Inertial Impactors.
Ph.D. Thesis, University of Minnesota, Mechanical Engineer-
ing Department, Minneapolis, Particle Technology Laboratory
Publication 144, 1970. 243 pp.
173
-------
30. Heinsohn, R. J./ J. W. Davis, G. W. Anderson, and E. A.
Kopetz, Jr. The Design and Performance of the Stack Sampling
System with Dilution. Presented at the 69th Annual Air
Pollution Control Association Meeting, Portland, Oregon,
June 1976.
31. Harris, D. Bruce. Procedures for Cascade Impactor Calibra-
tion and Operation in Process Streams. U.S. Environmental
Protection Agency Report, EPA-600/2-77-004. January 1977.
174
-------
APPENDIX A
MODIFICATIONS TO CIDRS CUMULATIVE MASS CURVE FITTING
PROGRAMS FOR INHALABLE PARTICULATE DETERMINATION
PROGRAM CONVERSION AND PROGRAM LISTINGS
„ The primary purpose of developing the OSCFIT subroutine in
its proposed form is, of course, to expand the range of accuracy
of the cumulative mass size distribution as calculated by the
CIDRS series of programs. Another consideration in its develop-
ment was the ease of conversion from the original CIDRS programs.
For those who have already converted the CIDRS programs to a form
compatible with their computer systems, the OSCFIT subroutine
minimizes the changes to be made. Only changes in the program
SPLINl are necessary. In the revised SPLINl, calling variables
for the OSCFIT subroutine are defined, OSCFIT is called defining
the cumulative mass distribution beyond the first stage DSO in a
functional form, and the spline fitting technique continues as in
the original SPLINl program.
To simplify the program conversion, Table A-l lists the
specific lines of SPLINl which are to be replaced followed by
a brief comment on the change. Listings of the original SPLINl
version (Table A-2) and of the new SPLINl version (Table A-3)
follow with the changed lines boxed and labeled according to the
index of Table A-l. Also given here is a listing of the OSCFIT
subroutine in Table A-4. Calling arguments are defined in SPLINl
and also in the OSCFIT subroutine. Section 2 of the report con-
tains a description of the program development.
175
-------
TABLE A-l.
CONVERSIONS FROM ORIGINAL SPLINl TO
PROPOSED SPLINl
Conversion
Index
New line(s)
of SPLINl
Replaces
former line(s)
of SPLINl
Comments
227
200
B
267-287
288-289
237-243
244-245
H D
291-302
247-252
E
313-321
262
Correction of error made in comment
cards of original CIDRS report from
."(1+4) point" to "(1+3) point."
Action in program is not changed.
Commend cards describing hyperbola
fit are replaced with description of
osculating polynomial fit.
Variable name HYPL (hyperbola length)
is changed to OSCL (osculating poly-
nominal length). This is the area
over which the osculating polynomial
fits. OSCL=HYPL.
Coefficients Bl and B2 for a hyper-
bola over the fitted region are re-
placed by the variables needed by
subroutine OSCFIT and the call to
OSCFIT. OSCFIT then returns 3rd
order polynomial coefficients BI,
B2, B3/ and B,,, and the zero slope
point ZSPT.
Values of log (mass concentration)
along the hyperbola are replaced by
values of mass concentration along
the oscullating polynomial length
and then converted to log (mass con-
centration) . Log (mass concentra-
tion) is defined as the log of maxi-
mum concentration for diameters
_> zero slope diameter, ZSPT.
-------
TABLE A-l (continued)
Conversion
index
New line(s)
of SPLIN1
Replaces
former line(s)
of SPLIN1
Comments
331
270
In new SPLIN1, the spline fitting
routine is begun two intervals back
from the original SPLIN1 program so
that the SPLIN1 fit over these last
two intervals before the first DSO
will "feel the effect" of the oscu-
lating polynomial points. (This is
an improvement over letting the
extrapolated polynomial points beyond
the first stage D50 guide the fit as
in the original SPLIN1 version.)
-------
•4 <
H
E-"
en
H
I-H
i
t-Z,
13
H
I—I
QJ
a
1^4
H
rn
Pn
o
•
fN]
1
•^
W
Jv? iT
liJ *"**
rij _J
EH a.
V.
£_ .
•a
or
0
o
QC
0.
^
«->
i.
o c
c c
M i^ a IT <
) VS tOGIO(OSO)
RFF POLYNOMIALS.
E CONTINUOUS'.
ti ir a
Z Ul f
•-i e.
O CO
•« O Ul
C 2 >
_JJ (V •-•
t—
CO CO «
CC Z) >
<* o •-.
z z> or
2 Ul
Ul »-< O
> 1^
l-i Z 1—
t- o to
•< o —
1
r • ec
— z u.
t_> 1-1 I
«- a »-
o a
— •« c
o _ z
O^g
*»
-j tt
Ul CO
to > _i
t- C «t
•-* »— 1
U U I.
o a
— z
n to >-
•-« Ui _l
• •_« ^1
a. tt a
to u>
to bJ
Z X
QC
t: i i
^*! V— ^»
a »-i c
QL. X C
) «_' CI CU C,'
0 0 O O
r~«o
*
*
*
*
4(
*
•K
*
4jt
41
*
*
4jr
4J
41
*
41
4J
41
*
4JI
4j
•It
44
4r
*
•K
*
*
4,
*
*
41
*
4c
4)1
41
*
4t
41
41
41
*
«
4C
H
*
J
*
41
4j
*
l_» CJ I
O 0 C
p o ~« rut* •« r- cc o>
c
c <~
z to
o -
« c
•~ IT
0 —
X^ v4
C Ul
z o
o u
Ul •
O «
^* • >O
o <*^ «
«- er- o
o o «
>- _IUt «
• O O <-t
r* Z »J Ul
0 O • O
«H « *•« f^
^* ^> 5J •
Z CO *^ Ul
U. «^ B) •-> C.
a vo « IM i_>
C) X *"* ^* ^^ V
Z U Z) •« Z »«
X Z U — • •« •- Z
««-« . «- Z -H »H
O O *-* *" \t\ IL *^ >~ *
CZCJf^*-* » ho ^-* «
_- ^ n. IT -^ *^ *- » %
0 M a. *^ >• w T* *^ o
* >. « »- cc *- c
3T Z ^^ »-«*-J »XO
•—* »-»30KllfI/5O* X5t
(/j «y> 3t3l-Z2.7J2rZ]ZZ'
aacccccuj.;
tl1 «,' «j iZ 2 Z A. 2 >
itiw 3D CL UJ UJ UJ UJ Uk. *~* *4.
*-~ ^xxrxz:""*-
j
0
fv
** «^*
o o
O 0
» *
o o
» *
o c
•)>-« w-»
»-1 »-»
« »
•X. „_>
•a a
^ 2 in
» -i _j
z *-••*
•-* u. u.
OC * *
_: r *-4 c
% c c.
% * *
QL »-* p*^
IT IT O
«|IM LI
•i. C •-"
• «••>—
X. ^-* *— •
-J U- U*
Or. *zr y
(/> t-4 »— |
-J U. U.
u c c
«f 1 f
»- —1 —
<< «t ^
l\i «\l (V
I»J s»
1-
z z
Q£ tt
j£
Z X _1
*-* o ^
Z •« Z)
t-i ui C.
4 ^4
i- or >
IT O •-
uiu o
a. z
z c
o oc. o
«— o z
•— o •-•
•< Ul Z
X OC t->
a •«
O «VI-
u «- z
z o
^4 CO I.J
z
_:=• or
4 K Ul
a cc
u, u. z
u. z
cr •
o c
QO Z Or
Z II O
•- Z CJ
•*. zs u.
>- D Ot
^ ^
O 1-
t_> • to •
CO •< O
•-• z ' »*•
OZ!
-~ or n ui
c _i ~ •-
Cr _ «- U
0 < U
u en z
U C *— ' •*•*
a »-
(V m rvj IM r\»
0 — «W l«r S» 4
o
c •
c
(O Ul
1—
X9 •-*
Z U
ft~l
^ cr
c
-SO
»-
CO
c^ co
«« 1-
r u
10
•
r z
Z3 •-!
U
O
u •<
0 U.
a
CO
1- 1
u
CO 0
_l N
i
< iV
O'
c z
c.
1- «
<-«Ui
b. a
•>£
•*.
— 1C •
r •< u
ct «i s: _i o
Z * t-> «J CJ Ls
rt»r\iK>»oiotopOf.
INTERVAL <
<
ACM PAIR OF
KC).
Ul t-
•- i-
to z »-
•< U! Z
_> u. a
3E U'
Z t- O.
IM U,
CD -49
a. n
z> a z
1- 0
Ul t- t-
ia ui
. • to to
U! X II
CC <« Ul Z
X CD
C C
»- o •
o ^~ u>
tO Z t£
_l •« CO •«
^ * 1^-
> Ul 4 CO
a cs >
Witf - •*-*
»-> Z 3".
ec >- «-< u M
r? ec ec. c u.
cc — ~ en
CO 0 «-<
u. u kr •
o o u c *-»
C H
a o c x
u1 IT a »- u,=
CC ^1 .ti O
z cc a z x
:? z T r> •-• uj
i U.' ^ C.
n % "" % r c •-
•- II 0 CO Z 3-11
£. U, IT II II «3 Z >•
U t_. t
r^lt^irOK1, ^ 33 9?sr=r;
C O- 0
BE FITTFD. «
STOP'. <
c
fr- C*"
*-
^
i- it:
•« z
o ««
t
a oo
cr
u o
•4 O
»•«
t-
«co
o •«
Z _l
O Ul
a >
o <
O UJ
Ui -1
ct
•1
CO C5
Ui II
a
»-
* o
0 Z
II C.
4
t- UJ
E *
Z iL'
c, u
** M
a . •
x. c
to
u _
1 3t 9
-»f\. KV « IT *
••
u>
a:
•a
c
a
c
u
ua
or
u.
o
u
z
•i
c
^
^» •-
K
C
a
u.
• •-» rr
•^ f^ fr-
C; — U!
Ob, *^ 5" *•' ^
at «- •« —
x c: •-« _J
W
z
o z>
z x
*-* •"!
es x
•« >•
o x
_i
•
co to
to >
4
Z O
z
U, l-l
> o
•-« <
»- c
•< —I
i
z> to
X CO
r; <
o z
U —1
c «t
h-. «
tr o o
u *- u.
^ #—
X QL t-
r> o •-<
r u u
i
II
u.
Z
CJ iJ (-
IT. Ul lf>
0 —
a.
U'
»-
u
X
c
z
X
^v
X
«x
X •
to
C. Ul
z z:
*4 I
•4
CO >
Ul
3 O
-J Z
•« (-4
> O
o c
IT _l
C.
CO
U CO
c •«
X
1-
Ul U.
CO C
• •-,
If— -
UJ
>» to
t-4
u •
3T
* fr—
«»•» cV4
II U.
»—i 2
% *•>
*—» w-t
c-* II
V 1-4
Z »
a..
»—
c ^
i C
X >
u> o
in -o
00
00000000000000000000000000000000000000 c-ococcooooococoooooooo
-------
061 C OTHER VALUES READ HF.RE ARE NOT USED. 72
062 C ™
061 310 »f AOno^IAVnS.NFIT.GRNAM.lO.RHO.TKS.POA.FGHPO.DSMAfnMAX, 7tt
06" iOPC.CHMG,DMOLD.GFOMO.ONOLD,CYCl.MC3,MOO,MS.VVt 75
065 2(XMOPf H(l).I«l.'
07*, r THIS ;-no 100" LOOP FITS A 2ND DEGREE POLYNOMIAL TO i 87
077 r I OGlOfCtlHULATTVF «ASS LOAOINK1 VS*. LOGlO(DbO) POINTS ON FACH 88
07« C TPAVEBSF, IE.J 89
070 C IOG10 OF fXNflPENM).YG{m,fXNnPtN(2).YOC2)),tXNf>PFNO),YOm)| 90
080 c " '" CXNOPEN_f2),Ypt?)),fXNpPEN(;S)fYp(3)),fXMnPEN(a),Yp(«))| 91
081 C f ..I........."..'..'....'....'..'.. -'.'..'.. ..'..'..'.......'.............. I 92
082 C " " rXNDPEM(MFlT2),YO(NFTT2)),(XNOPEN(MFlTi),YO(NFTT1))» 93
081 f. (XNfPFMfNFin.YOCNFITn 9«
08« C IF Tup FITTING POIYNOMIAL HAS NON-NEGATTVE SLOPE AT ROTH 95
085 C LOGinfXMi)PE^(I).Yncn) AND I OG10 (XNDPFNf 1 + 1 ) , YO f I »1) ) . THE 9fc
06fe C FfTTTMG COE^F1CIEMTS ARE LtSFD BETWEEN THESE 2 POINTS TO OFFINE 97
087 C 3 P'TERMEOIATE POTffTS EVENl Y SPACED ON LOG10 SCALE. IF THERE IS A 98
08* C fiFR&TTVr SLOPE AT r.ITHEP- OF THESE 2 "POINTS, A STRAIGHT LINE. FIT 99
OB9 C BFTWfFN THE POINTS IS USER TO DEFINE THF 3 INTERMEDIATE POINTS. 100
0«n C THF VFCTOWS XI AHt) YI REPRKSt'NT LOGlO OF ORIGINAL CUM'. MASS 101
0<»t C L'UnifiG VS. 050 POINTS ANO THF FABRICATED INTF.PMpni AT£ POINTS.
092 C lf>3
091 C THE i-MiM&ER OF POINTS REPRFSFNTF.O BY THE XI, Yl VpCTORS IS 10<»
09U C (HFTT?*<»)*t*2'. THFRF ARE (NFIT2*«1 + 1 POI'JTS BFTXFFN POINTS AT 10«i
095 f LOG10(050) OF LAST STAGE ANO !.OG10(D50) OF 1ST STAGF INCL'-'SIVF. tf>6
09h C ? MORF AKF FXTRAPOt.ATFO BEYONO LOG10(050) OF 1ST STAGF. 107
097 C 108
09B PO JOO I"l,NFIT2 109
099 JJsM-t 110
100 trfHFTT?-n"o,«c.Po in
tOl «»« .IJeM+2 112
10? PO Ms(T-tl*N*l 113
101 Xl(Mlsf)LOG10CXNDPf N(I)) llfl
100 YlfMlsniOG10fYn(T)1 115
105 XrNC=(nLOGtOfXMC:PEN(I + m-0|.OG10fXNDPEN(I)))/P lift
10* C 117
107 r. Sinn SOLVES N SIMULTANEOUS LINEAR EQUATIONS, AX s B. H£RF 118
to* c N s v. Twp HATKIX OF COEFFICIFNTS, A, is DESTROYED in THF IIP
tfl" C COMMUTATION. THF VfCTOR OF ORIGINAL CONSTANTS, B. IS RFPLACFO 1?0
1)0 C BY THf FIWAL SOLUTION VALUFS. VECTOR x'. COFFFl.CIEwT "ATRIX A AND
111 r CON?TANT Vtr.TOf? H A»F DEFINED IN THIS LOOP. 122
112 t 123
ill on 1100 11*1,4 124
lltt ^MsI-14-II 125
115 B(II1snLOG10fYn(MHn 126
lib Ks**ni-l) \27
117 DP 1 ICO Jsl,l 128
11* M!=I-1+J 129
UO linn AtK + .T)sOlOGJf>(XNOPEN(Ml))**ni«l) 110
120 KS=0 131
179
-------
-io ri o r» r» fj or»r»o
»^ «Mk »^
»•» ** a o -\i >~ a
.ji o c 00,0 jt
nx3B3ta«»»jx-3 o- en -. ii 3 -* ii o ii o rna-n o*-»-H3OG'-»*^<-«'-.-«ir-o me »»
>~ M -«<— — ic — w H c ~« -»3 x x »• z rn x f- H -t»- 3?— M aD—i z •-»•-» T r w -• o »v -» 3 z en r~
f~o»-'-»->—>-»>.4.i-. »,— ^._,.^,_/it/>"3 -«f-» — o tr> f r- o >-» + — + + ••«»•- c TJ -^ —«x;*+»--*-«>-»<.»r—we<»"O«— 3 "n r~
ll-.J'lli.-»p-40HIIII — -* — — ~ t*1 II-. — C^^-* — -.>— f-i-n •-«— —C.S-iO«M3»|!— f. —
tn •<-. -< — -> -.x- — -
-* -«^<-4-»l|^.t. -«>-» IIT13 •• 3« -< -»l II I I *• X ~ 31 Z) T — «'i"~— 3TI3J« -n-l >-«
X I i u •- x i ii u --«^- —-nr^ -< -< M M < -t i ;z_,_ix —1_.-«-. r-_iffi m«f~ c«m>-ooor-m*^ n T
B -«'ui\»«-'«^(\»ro —. t:« »i>«\j <->'%uiz-»ii a ^*^n-»>-'ii^M ?n3 i— o « (\» «-»o«-»ivt, 3^ o
*^ >—' • 1«-*» «*-»• <»<£:<-• •!» •* -^«O| <-» »»-J3:»^» t- 3: -»i «-»C3 t— O •— >t O •-•*-' M 3QM »^
>• » tf4 -• » tH IX! ^ .-» •> •»• ^* ^f tit ' » Z»>^Z» • -t J »-« •»• »- • 3 Z — — •«•>— >-« S> »•
• z »z »x <-.i\j«^v^fT»>-eo* M*MW rnaj aj o c_ ^oooiw* ox«
3D — » >^ » -« O • • » -I ^> •" 3 » -« T| |» r^ ^-» C. -->-t *ux om-<«o 2 •
W +>- * n- (I -. TJ urf -« z -»Z -»Z3— *>.— rfj
• *^ -ox z-oifrrt>» i »-»m IA! »o »-««-«^^ •— on !—•«-••
X «- ->O-»H X i— X XH ^^^^.-, ZX
C/» I W O » TI 3 CO «rf • OX-. t— •»« <->4--*>«M»^rT'CO
-* X >-''^Z3"Tl<^ -<'~X-»Z I *- w» • «•» z »«
'^ it X -< O t-i «_» o« 7C>^ OJ ^* «— •-» * Hl^l
*-» X —* —» ^ Z —• TJ-< *^*-« *^'X^V**O !*!»-<
~» « IU«--II -O-H*^ +•>-» IMX-<«-»r~ 3J-*
^^ »A«-»«_«s-*i-4 O C0«^ ^^ZO4>O-<*
v^ • . X O CD M >• T3 »^+ »O--»a >»
< <^ z z-< a MC. o •»!-« »- r-z
»->-^I\>P» -^»-« •-• » r~ rn>_» o • o
iv* r — -«c ai -»z »x o z •-•'-» -«
<-> n! i x x »-x-« x«-« ci^^x
^*XT?« ~ -. a» -• r ^.» •• #^ *-% n o ^^ o o
X3 — O «*-»KX -» -• T» ~~
C^'-'-^TI «i-«w> x^>ni-<
CO -< " I 1-1 -< -« » Z Z
-izn-t ^ x * o « ti
a^in ~» tvi -o -i »-•
xxn>> »»> m «• -<
»• t-l O O> -« v« Z C. «
<:»!•-« «'M «.§
•t r* o -« Z »^»«M
x m ac -^ o *- «-• TI
"o ~n /^ ci z ^ ^*
a •-> — o o o z
co-«»''''i"n -«rn in
«-••—• rn e> o
-4 -4 "O * XJ »•
"• I -I O GO «rq -I
< * I «-• -« rq >-,
rn o ci z <^ <
= -• TI x -o m
M o -< 3 — a
r- x — «-• • f 01
o r» z -< < r-
D-«3"'I-« —Z O
m i o ^o« <-» o TI
-• a; 3 x rn
»» a> a i n •«•-•
-I »-t -«l O »• »
TJ z -t *-*r- ~i
O >• X -« Z
>-• -< r*i —< -«»-• «i
Z "1 CO Z «
-t •-« -•
• «• *• x
30 TO
1*1 TO
-------
1B1
162
183
Ifiu
185
186
187
ISfl
189
190
191
192
193
194
195
196
197
19H
1<»9
201
202
20 i
204
205
206
207 .
30*
209
210
211
?1 ?
213
215
C 1 &
217
?1 H
21"
220
221
222
223
224
225
226
227
228
22"
230
231
23?
233
234
?35
236
J3"
?Ufl
_--— -~
C
c
r.
t
c
c
r.
c
c
c
r
t
c
c
c
c
r
c
p
r
r.
c
r
c
c
c
. — - —
20 roE(i.t)«flfll
ii = riRST INTERVAL FOR WHICH FITTING COEFFICIENTS ARE DEFINED.
JNTSl z LAST INTERVAL fcHERE POLYNOMIAL FITS WERE USED TO FABRI-
CATE TNTERMfnTATE POJNTS. THE UPPER BOUNDARY OF THIS INTERVAL
IS LOGlOfXMOPf MfNFTTn.YO(NFITl)) HERE',
I js]
IftNG ROUTINE **LOOKS
AHEAP" TO LFT COMING POINTS INFLUENCE CURVE DIRECTION,
23 IV) 50 IsII.INTSl
JJ=1
P(l)=0.0
HO 25 Js?,3
H = I-1
If f r.tn.1 )K»I
?5 H(l)a(ifl)+-(J-n*tCOE(K,J))*X(Il**(J«»2)
H(?)s('OE(K» 1 5
nn 30 js?., 3
30 H
-------
CO
to
r» o n o r>r«or>
o Ji
H» 0 111
n» -n o *•% JE "-• o •-<
Z O O-"» X! < ~* Z CD rl O2
o n z o •-• .»•»-• -4 ~n z H
T-4O-4 — r~ II m -4 -4 en
> •-• m ~*i c "*i z z o —
-4 Z t— *•» T iQ II II
^«-tii_«>^ » C 4 rnz wti
-. c,r-i «. o> —5 i
rv •-- J* r TO Z O> "
t_ -• O T)
n z "n o •
• -4 -4 ""
• x r -4 z
>HI *•• (T> *4 ~4
n -» m m
K- M O O 3C
• v O <
*-« » m TJ ^
z -« ~n o r~
— » n -*i >-• oo
» r> — 4 ~*i
z •-> <» a
-4 Z
• 0> X X
«•» m —•
XTI o
•-> OM X
*-• — 4 T|
^ mm -«
« ^ 31 -x
M n < -•
• x >• •->
~ r- z
Z Z CP
Tl — 4 O n
-« m c o
*-* a z m
« -co rn
r~~ v,iMWruM
[^
M
T
1
n
o
T
^^
f TU aj V Mi\i fu V
i4> O> ^1 J1 *J1 fc k*4
r> o o n *"* o
7
°
r>
r- c« M 3D o
C »» -5 3 -1
-n 33 -n Z
O > ^ rn
•n >»
Z -4 "-• «
O O
•-1 r-4 O
•n z 01
-« TI •n o
— 1 33 M *
-4 •< 5D •
Z » C/l
SB f -«*-
- 3
11 i-t O
o •-• z -o
r- z -i
-< -4 rni-i
z c* zi z
o — et
O X»4
m >c xz
-n «-• t» z
•»> -4TTIM
-4 X 3OZ
n mo
1-4 O O
n i r~ *
z »• » z
-4 X« O
OJ **1
o » fn z
< m ci o
m -4 •-•• •
n c z
n cam
-< z m
X »C1
m -4 -4>-«
CO O Z
m o z
>-< o o z
Z t) (0
^ "^
O)
ru ru n< O> ~4 ^ ill I*
TC
M
i
C-.
H
3)
—
a
•\
^
—
0
V
r«h
*x
0
•
o
^
^
X
!••
*^
\J
•t^
IX
1
;
ru i\i A> iv* ru - u O IM r~ -• -4
-.1 II O T T
c, 4- •— z. 4- a ru z
X O 'V -* ~
i— X » X
-» -• 2 3 T
3 M C 3 m
«— * «^ v rn jo
+ » mo
>-> z z o
» W /-« TJ T)
X Z O
*-• -n •-• T
Z -4 Z D
<-*
-4 -41\
-> Tt
V »• C
• -Q M
o c«
r- o
^
-4 >
rn r-
o o
z
< 0
f -4
C X
n in
c*
X
O> X
m -o
X f*t
O TO
Z3>
3 O
r~
^
•^
w
z
z
*
ru
N
0»
IM
in u» in* c c
\> *- o -o *> -.
3- B < X X X
-.-» i
II II ^ *U >rtk *•"
-< •-* U II H II
•JI X O — X —
C £/!<-• — •
•I ^ 3 3 «•• ~
i- CD S> 0 Z c
1 (M »• X. x >.
1 1 O X V
-> x -> z z
ru 01 x O c
* CT^'n r»
X — Z Z Z
C >. M Z Z
CD .-» -4 11 "*
•^ X •«' *-4 •"
Ot ""^ ™4 "
c -• —
3) >-
•M
1
X
CO
c
31
M*
fc^
ru ixi ru TU ixi n,
»<»•»• cr- Ji u
Ij4 IU *^ ^^ A fl
1
O-
«
~
-<
-4
1
*""
••
Z
+
IXI
IU (V
A C
in c
«x
" f~
i 3
x r~
<^ c
39 3
TG *~>
X
2
C
T
T
^
1
**
-
*^
*^
1
C
r~
a
i^
—
a
^
x
z
-
T
n
Z
*^
2
*«
—
**
*-*
*-•
ru t\j
m a
» is
i
XI Hj (\ll
& 0 fcl
...i
•« r»r> 1
1
a> >
** I
«• T!
-",
a> xi
»> CE
V 3
•-* f~
^ *"
33
x r^
• -4
•y
m
z
—4
X
m
CO
m
fu
•a
3
*4
z
-4
CA
O
^
-4
X
PV
"n
o
a
i
f~
o
CT
**
O
^*
X
^*
n
u ru ix»
/im at
B !<• ru
1
c JL
-------
CO CO COCO CO CO
*•• CO
0 .J
in 4, •
O >•* CO
*•* X 3.
0 Q O
e >- z
_J C >-
O •• z
t£ CO O
2 * U!1J
£4 *~ o: u.
rn IS « Of
M Z£"*
•« *-• o
1-7 C CO
•4 O U.
H oz >
2 -"T
H CO CO ^t
•4 «o 3 >
Q. 4 C t~«
rn X 3 tt
W z u.
ui »- o
9j > >—
M •- Z 1-
5 »-oio
** 4 CJ «"•
3 • a:
. Z Ow
7 5£*
1 «- a i-
<; o a.
-- 4 e
M £-5
•4 _ ae
n uj U. U. Z
_« o o
a. — z
»c z co >
»-% uJ _)
r _i »-i c
4 a a a
CC US —1
iS CO U.
0 IT
K^ ^ ft—
^ ^ 9"
c. a
C I T
Z — »- t-
»- a »- c
4 Qb "3C ft
a.
*
4i
OCJ OOCJ 0
CO CO
41
•K
41
*
41
«
41
4(
*
41
41
«
41
41
41
41
#
«
41
«
41
*
41
41
41
4l
41
41
41
41
•*
41
«
41
4C
*
Jg
4i
•fc
4t
«
«
•*r
«
4i
*T
*
41
4t
4C
41
4i
«
41
4t
^
4t
*
«
*
(J C"
CO CO CO CO CO CO
»•%
t>
^>
c
_i
c »•«
c •
• c.
•* V
CT *"•*
C U.
X O
O CJ
UI •
o n
o «-> %
— C- €3
%^ *^ if*
C O *•"
V -JUi
• c c
« r cj
o o •
C *
• >- *
*? 5T ^^ * ••"*
C C1 C ^ f*
•— * •—! ac ro IP
OT CO *^ l/^ 1b'
^-1 M O *"* *^l
O O *•* 5C X
> '•*! k&*
> tt ft Z ~Z 2
flL CX C5 C C
!.£«..«<.£ A ^
ftfc* M l£ taJ. 4>Mt U-"
i- r - i 3 r
— •• — — 1* -n-" 1 — P -TT—f -^f rnir1 IHF -Mi* *>^ «J n i • mi
CO4I34O CCCOCOIOfOtOCOCOCOCOCO
e o
0 C
^% % »
•~ o o
ttl » *
C o ~
*i Z^»r
U4 * «•
*^ O X _*
CC CD » %
u *-*>-* * o c;
* **i v- « « v^ »•
41— *-/ % % % 4* •
M CC. *"* C IX *»f ^*
*" CC «* C> CC if C
_I » X O _J f\i IP
t*. *^ • X *•* * *
CO C* X ^£ U.. G *""*
J *— V. % % w-i w-i
>-« «J, ^ ^ ^^ ^«^
Uk ^ li. X .in-1 UJ UJ
tJ < £L .7. .£
2" Z Z 2E CO •-* i^*
(/j yj «^ ii ^ ^ j^t
i 2. .>
Uj *J ' ^* *t *I _L/ -.1 *
-;
S
cc
7
3
t- ~
•4 4
c
c-
Z 1-
ac e
z
•« z
3 *-•
c •«
> a
»- U-
O Q.
z
•t Z
o
C9 •-•
Z 1-
•-• 4
Z X
M Ot
•< o
•- u.
z z
O t-«
DC 4
u a
OC UJ
•c.
=> UI
z o
C. CO
o t_ i_- «_
— i*j_ j_i_ j— j«j.j«j
cococococococococo
CO
(9
z
c
if*"
—
cc
(C
^f
J
u.
»— 1
*—
•*
X
u
z
4
z
o
o
3:'
I
CO
c
a
c
lii
or
u.
O
• .
a. co
Ui t—
c uj
Z CO
3
IT
^ ^
I-i
U.. •
co co
t- >
*
4i
41
4T
*
*
J|
*
41
41
41
4>
41
£
J
*
41
41
41
41
41
41
41
*
4C
«
«
41
4t
«
Z 41
r> 41
a. *
;• ^, 4c
4 «
»•>• fXJUl 41
— 4r a *
o 2 st *
— > 2 ^ *
•• a -* 41
— n C. *
4 U. 4 41
IU U3 ^j • 41
a •- a 41
41
* 41
41 41
cj L; u; «.! «-• «_•
«l — J »J .^ _J
CO CO CO CO CO
u.
c
_i a
2* ^
or a.
»- 3
z o
t1
t-
CO Z
«« Ui
3
Z i-
o.
3 O-
t-
CO IL-
• CO
U. X
er 4 ui
z z
c e
1- C •
O t- U,
CO IT O
_J 4 CC 4
> U! 4 CO
oc ts >
u. 4 o: »-
*- »- UJ It
Z CR •- -•
•-« Z
tr H- •- u.
3 CC £C C.
to •- ~
IO 0
U U. IT
O 0 -. 0
c
£t 0 C
Uk' tO DC. ^~
co a u.
z cc a
3 Z Z —
UJ 2 CO
II 31 »
»- II O
Z U. IT
~ c: .7t c.
_I_I_I
COCOCRCOCOCCCOCO
41
4!
41
41
«
41
*
41
4c
4i
4>
£
41
b
^
*
41
41
41
*
4r
41
41
41
41
41
41
4i
41
41
4C
41
41
41
41
41
* Z
41 •»*
* U.
41 CO
41 M
4t •>*
41 II
* X
41 U-
* CJ
4t Z X
41 •-. UI
* C
41 - C 2
41 Z C t-
* CO 2 9 II
* II II 3 Z >
* .£ CE II II C 4
« z a 2 a c. •-
41
4t r--
* o
41 Kl
U O
.4 —1 •_* — J
(O CO CO CO
•k
41
41
41 -
41 C
* U.'
41 t-
* 1-
41 U.. •
4i OL
41 UJ C
41 X *~
41 CO
4> t- C
41 ^-
41 t v
41 4 Z
4i at «
41 U. O
41 tt
4C 7^ 4
41 4 O
41 »-
* «co
* O 4
41 Z -I
41
41 a uj
* a •>
* 04
41 O U.
41 4U _>
* a
41 »
41 CO
41 U> II
41 t-
41 % n
41 c Z
* M C
* «t
4r o a
* z *
41
41 O U.
4l 4 I i
* U.
« a «
41 X C
41 {£
41 ' U. _
41 I-" 4
41
41
«
41
O CJ O O
_> _l
CO CO
41
41
41
41
41
41
41
4l
41
41
^|
*
•ft
^
f|
*
*
f(
^(
41
4t
*
4t
4t
^
4" fO
* s
* H
*
4T
J
41
41
•ft
4t
»-
It
+r
«!
41
^
.j
«
«
*
^
•«
41
^
^
41
U C.'
o o o o o o o o o «-« v« *^ •-• ^^ •-« *^ *-* *^ «^* f\i f\i ro i _
ooeooooocrcoooooooococoooooooocoocooocce-ococc-ooocooooo e o o o coo
-------
06!
062
063
060
065
066
067
068
069
070
071
072
073
075
07*.
077
078
079
080
081
08?
083
064
065
086
087
088
089
090
091
09?
093
09U
095
096
097
09fi
099
ion
101
102
105
10U
1 OS
Iftfr
107
10R
109
1 JO
111
11?
in
1 1 1
1 15
1 1*
117
118
119
120
2
C*****
C*
c*
c*
c*
c*
c*
c*
c*
r *
310
c
("****<
c*
c*
c*
c*
c*
c*
c****/
313
c****
c*
c*
c*
r*
c*
c*
c*
r*
c*
c*
c*
c*
r*
c*
c*
r*
c*
r*
c*
c*
c*
c*
c*
c****
HO
90
IFfKRf 40)410,310.2
PFApf2.1)IAV
I F ( I A v '. fc" fJ , C ) S T np
****************************************************************
VALUf S TO |$F USED FROM READING OF RECORD ARE|
rjFlT - NUMKFW OF ClJMULATlVF MASS LOADING VS. D5« POINTS ( + 1
FOR TOTAL MASS LOADING VS. MAXIMUM DIAMETER) TO RE
FITTED'.
XNOPf N(l),!5l.NFlT • SET OF 050 VALUES AND MAXIMUM DIAMFTER.
YD(n.l»l,!JFIT - SET OF MASS LOADING VALUES,
OTHFR VALUES REAP HFRE ARE NOT USFO.
: * *^*ft4tft£44t'A^4E &1t1Hi1i4t4tiiil1t1tit'1t1i'k * H A
DE*r>(l"'IAV)IS.NFIT,GRNAM,Ip,RHO,TKS,POA.FGH20.0SMA,r»MAX.
DPC.Ci'MG.DMOLD.GFOMDjDNDLD.CVCI.WCStMOOpMS.MF , VV,
MXK'DPCNfl )»Isi,NFIT),fYOrn,Ist»NFIT)
I****************************************************************
NFITi = NUMBfB Of CUMtlLATTVF MASS LOADING VS. D50 POINTS
(F-XfLUDFS TOTAL LOADING VS. DMAX)'.
MPT = TOTAL DUMBER OF POINTS USED FOR FITTING BETWfE11! (AND
INCI DOING) HAXIMIJM PARTICLE SIZE AND 050 OF LAST STAGE.
*****************************************************************
"FITlsNf IT-1
NPTs((NFITl-l)*N)+NN+l
MFT T?sMFIT-2
it****************************************************************
THJS "DO 100" LD()P FITS A 2ND DEGREE POLYNOMIAL TO 3
LOG1fHCUMUl.ATT.VF MASS LOADING) VS. LOG10CD50) POINTS ON fACH
TRAVFRSF, IF,I
LOG 10 OF fXMDPFNJf 1),YPCl)),fXNDPtNC2),VO(2}),(XNOPEN(3).YO(3)) »
" " (XNOPKN (2) , YOf 2) ) , (XNDPEN(3) , Y0(3) )| (XNRPEN(O) , Y0(t) ) I
CXNDPFN(NFIT).YO(»"FIT)}
IF THF FITTING POLYNOMIAL HAS MflN.NEGATIVF Si OPF AT BOTH
LtU,1orXNOPENfT).YO(I)J AND L OfilO (XNDPEN f I * I ) , YO f 1 + 1 ) ) . THE
FITTiKiG COEFFTCTF.NTS ARE USED BETWEEN THFSf 2 POINTS To DEFINE
3 INThRMEniATE POINTS EVENLY SPACFD ON LOG10 SCALF,IF THERE IS
NFGATTVF SLDPF AT EITHFR OF THESE 2 POINTS, A STRAIGHT LI*£ FTT
RFTWFFN THf. POINTS IS USED TO DEFINE THF 3 INTE"MEDTATF POINTS.
THf VFCTORS X1 AMD Vt REPRESENT LOGIC OF ORIGINAL CD"'. MASS
LDADJNG VS. HSO POINTS AND THF FABRICATED I^TERMFOI ATF POINTS.
TH£ NiiMfltR OF' POINTS RfPPFSFNTED BY THE XI, Yl VECTORS IS
(NFTT?*«)+1+?. THFRF ARE (NFIT2*a)*l POINTS BETWEEN POINTS AT
LnGlnfD50) OF LAST STAGF AMD 1.0010(050) OF 1ST STAGF INCLUSIVE.
? MOf
-------
121
124
125
1?6
127
12*
131
13?
133
134
135
136
137
13B
139
111
la?
143
144
145
146
147
149
150
151
152
153
154
15s-
156
157
15*
161
16?
163
165
166
167
170
171
17?
173
170
175
17*
177
17«
17P
180
("LOG 1 P f XMDPF.N (I+ 1) )-BLOC 1 0 f XNDPfN ( T ) )
c*
c*
c*
c*
r*
c*
SOLVES " SIMULTANEOUS IINEAR EQUATIONS, AX s H. HF«F
•••> s 3'. THE MATRIX OF COEFFICIENTS, A, is DESTROYED IN THF
COMPUTATION. THF VECTOR OF ORIGINAL CONSTANTS, B. IS REPLACED
HY TH FINAL SOLMTIOM VALUES, VECTOR x'. COEFFICIENT MATRIX A AND
CONSTANT VECTOR » ARE OEFINEO IN THIS LOOP.
*SPII
*SPLI
*SPLI
*SPLI
*SPLI
*SPLI
*3PLI
STRAIGHT LINE FIT, NOT POI Y FIT,
OF ORIGINAL INTERVAL,
1104
.
SLOP? sn(?3+?.n*nm*0i.OGiofXNr>PENn+j
IF rsinpf-H 100,1110,1119
•1))
c*
c*
c*
c*
r*
THf FIRST 3
TO Of F I ME
FOUTVALFNT
fX,Y)
SLOPE
*SPLI
NEGATIVE SLOPF AT FITHF.R*SPLI
*SPI.I
SPLI
i
SPLI
SPLI
SPLI
SPLI
SPI i
SPLI
SPLI
SPLI
SPl I
SPLI
SPLI
SPt I
SPLI
SPLI
I
*SPLI
POINTS WITH A 2ND DFGR.FE POIYNOMIAL IN *SPLI
AT rxm.Yd)). fNOTE - tXl.Yl) POINTS ARF*SPLI
iaj,3
Ks3*(I-15
nn no Jsi,3
A(K+J)
"0 US
MO
115 B(T)BVfM)
XfM)**(T-1 J
1 = 1,3
C*
C*
C*
CHECK THf Slf.PE HF THIS CURVE FIT AT THF FIRST POINT. IF IT
NEGATTVF, Ann A POIMT PN THE OTHER SIDE OP POINT 1 FROM THE
IS
SPLI
SPLI
SPLI
SPLI
SPLI
SPLI
SPLI
SPLI
SPLI
SPLI
I
*SPLI
*SPLI
*SPLI
129
130
111
132
133
114
135
136
nr
138
139
140
1"!
14(1
145
146
147
148
149
150
151
15?
153
154
155
156
157
158
199
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
1«1
182
183
184
18S
186
18?
188
185
-------
1P1
18?
183
184
185
186
187
16*
189
190
191
19?
193
194
195
19*
197
19ft
199
200
201
20?
203
204
205
?07
?0B
?09
210
211
21?
?1 3
214
21S
21*
217
?1«
?!"?
?20
?2 1
22?
2?1
??4
225
?.?.h
P«?7
??.P
229
230
f*
C*
C*
C*
c*
c****
a
s
10
15.
19
?0
r* * * A *
1 K » ft W
r*
e*
c*
r*
r*
r*
r****
2>'n POINT A DISTANCE (X(2)-X(m FROM X(U. THE Y COORDINATE
VAIJIF IS SET s Tf) Y(2). THF POLVNOMIAL FIT THROUGH TH|S POINT.
(xrn.Yrm, A^D rxc?).Y(2n MUST HAVF POSITIVE SLOPF AT
fX(1),Yfl)).
******************************************************************
SLOPFs«(2)+2.0*Hf <)*X(1)
IF f SI OPE14, 19,19
no s i»i . i
A m s r.
A(fl)«Xfl )-fX(N+11-X(l))
A(7)«fA(4) )*»2
no 10 i si,?
Ks3*T
nn to 1=2,3
i"s» + f ( J.?)*N)
A(K + J)5X(*4)**I
P- M 1 = Y ( 1 )
no t«; Ts?.,3
*sl + ( (T-2)*^)
R(I)=YfM)
KS = O
CAt L STMn(A,8,3,KS)
!>0 20 Tsl,3
COF(1.T)=a(I)
*****************************'****4r'ft***********A****A*******A******
11 s FIRST iNTfcPVAl FOR WHICH FITTING COEFFICIENTS ARF OEFINfD,
IMTSl = LAST INTERVAL WHFRg POL^NOMTAl FIT3 WFRE USED TO FABRI-
CATE TK'Tf-.RMEDIATF POINTS. THE UPPER BOUNDARY OF THIS INTERVAL
1 Si n<; 1 otXNpPENC-^ ITl),YO(KiFTTl )) HERE'.
******************************************************************
T 1 = 1
TNTSt-NPT-NN-1
*SPLI 189
•SPLI 190
*SPLI 191
•SPLI 192
*SPLI 193
SPLI 195
SPLI 196
SPLI 197
SPII !9fl
SPLI 199
SPLI 200
SPLT 201
SPLI ?02
SPLI 203
SPLI ?04
SPLI 205
SPII 206
SPLI 207
SPLI 20fl
SPLI 209
SPLI 210
SPLI 2tl
SPLI 212
SPLI ?13
. Q n i v D 4 /i
*3 KU- 1 f. 1 tf
*SPII ?15
*SPLI 216
*SPLI 217
*SPLI ?1B
•SPLI 219
*SPLI 2?0
*SPLI 2?1
SPt I ?2?
SPLI 223
r* ***************************************************************** *****SPL I 224
r*
r*
c*
r*
r*
r*
r*
r »
c*
c*
c*
r*
C*
r*
THIS 1 "HP FINDS THf FITTING COEFFICIFNTS FOR EACH INTERVAL. THE
S frOUATJQNS ARf. SOIVFO FOR 3 UNKNOWN COEFFICIENT VALUFS FOR THE
MfTTNG 2»IO OFGPFF POLYNOMIAL". THE EQUATIONS EXPRESS 3
COuniTinNS FOR THF FjTi
i'. THE FITTING POLYNOMIALS OF THE 2 JOINING INTERVALS ARE
CONTINUOUS AT THE MUTUAI BOUNDARY POINT.
f'. THE FIRST OFRIVATIVFS SAME ARE CONTINUOUS &T THF
•HjTllAL ROMNOARY POINT'.
i. THI FITTING POLYNOMIAL OF THIS I TH INTERVAL fFITTING
HFTWEEN PniMTS T AND 1+1) GOES THROUGH THE (I+"i1RH POINT,
I'.E. A POTMT nilTSTDF THF INTERVAL. PITTING ftdUTlNF. "I OOkS
AHEAD" TO. IET CODING POINTS INFLUENCE CURVF OIRECTTON.
?3i r **********************************************************************
?3?
?33
234
23.5
33*
337
238
23<>
?40
?3
2-5
RO Sn TsII.INTSl
JJ=I
H C 1 ) = 0 t 0
no 25 .»a?,3
^sT-1
IFfT'.f'O.UKsI
H(1)«Bfp + (J"l 1*(CnE(K.,I))*X(I1**(J«2)
P (?}=COt f K, i)
no 30 j=2,3
•SPII ??5
*SPI. I 226
*SPLI 227
*SPLI 228
*SPLI 229
•SPLI 230
*SPLI 231
*SPLI 232
*SPLI 233
*SPLI 234
*SPLI 23SK (
*5PL1 ??6
*RPLI 237
*SPLI 238
*SPLI ?39
SPLI 240
SPLI 241
SPLI ?«2
SPLI ?43
SPLI 244
SPLI 245
SPII ?46
SPLI 247
SPLI 248
186
-------
241
242
243
244
245
246
247
24A
250
251
252
253
254
255
256
257
258
260
261.
?62
263
264
265
30
8(3}*YfI+?l
no 3 •» J=t,3
35 A(l )a(J,i)*xm**(J-2)
HO 40 Js1,3
KsJ-1
40 A(KK+2)aX(I)**K
DO 43 J=l,3
KsJ-t
C*
C*
C*
r*
SAVF TMF FITTING COEFFICIENT VECTOR B WHICH FITS OVER INTERVAL
I AS C.CiF,
00 45 J=H
COE(T,.naB(J)
50 CONTINUE.
IFMJ.Frj. (NPT.D)GO TO 55
SPLI
SPLI 250
SPLI 251
SPLI 252
SPLI 253
SPLI 254
SPLI 255
SPLI 256
SPLI 257
SPLI 258
SPLI 259
SPLI 260
SPLI 261
SPLI 262
SPLI 263
264
*SPLI 265
•SPLI 266
*SPLI 267
*3PLI 268
269
SPLI 270
SPLI 27i
SPLI 272
SPLI 273
m —
?6fl
?.69
270
271
37?
?73
274
275
276
277
278
5 7 a
c I "
?80
261
28?
2B3
281
2flS
287
>89
C*
c*
c*
c*
c*
c*
c*
r*
C*
c*
c*
c*
c*
c*
c*
c*
r*
r*
TM» LAST SERIES OF INTERVALS FOR WHICH FITTING COEFFICIENTS ARE
TO fig DfFTNED LIFS BETWEEN LOG10(XNOPEN(NFITi),YnfNFlTi)) ANO
>*5PL! 5T"»
*SPLI ?76
*SPLI 277
*SPLI ?7R
LnGJO(XMr)PEN(NFTT),YO(NFIT}). THE POINTS ARE DEFIED BY AN OSCU-*
I.ATTNG POLYNOMIAL DEFINED IN OSCF'IT SIVEN|
l'. THE t.OG VAI UE OF THF FIRST 050 AND THE VALUF Of CUMULAHVf
MASS < THIS D«;o CALCULATED FROM SPLINF FIT, (XLLD50. YLf>50) .
2. THF OtRIVATIVF OF CUMULATIVE MASS WJTH RESPECT TO LOG
PTAMFTER AT POIMT DESCHlBFO ABOVE, DMOLDO.
i. THE LOG OF THE MAXIMUM PARTICLE DIAMETER AND THE TOTAL
ACCUMULATED "ASS CONCENTRATION, (XLDMAX, YMAX) .
4. THE 'DERIVATIVE OF CUMULATIVE MASS WITH RESPECT TO LOG
DIAMETER AT POINT DESCRIBED ABOVE, 0.0.
OSCFfT RETURNS THF 1 05Ps10'.0**YlfN| . .
f^EL 1 =cnf. (IMTS1.2 5+COE CINTS1, 3) *2'. *XLLD50
PPPsCoFfINTSl.1)
no 5PS 1=2,3
,1 )*XLl.050**(L"l)
YMAXsYOdlF IT)
187
-------
301
302
305
304
305
306
307
30fl
309
310
311
31?
513
SIS
317
519
51"
520
521
S2?
323
3 2 '4
325
326
327
328
329
33fl
[J31
33?
333
335
336
337
33*
33°
340
341
342
343
340
345
546
3 '4 7
34P
349
350
3S1
3S2
CAI 1 CISC* lT(XLLD50,YLr>50.Xt HMAX, YMAX.OMDLOO, 0. 0, 7SPT,
lf(1).l'f2'»l'(3).B(4))
C**********************************************************************
C* THF Mi"F»E.R OF POINTS TO RF U3FO ALONG THt OSCULATING POLYNOMIAL
c* is >-'N*2 = 8+2 THE 2 AOoen POINTS ARE EXTRAPOLATED VALUFS BEYOND
T* LOGl Of XNpPEN(MF TT),Vn(NFIT)),
c*
C*»***»****4*******«***** **********************************************
NisNU*?
DO 1150 Tat,Ni
J=H+I
XI (JlaXl (M)+I*XIMC
IRX1 f.t) .LE.ZSPT1GO TO 550
Gii TO 1150
Yt f J)=Bf 1 )
r>0 114fl K = 2,4
114R YHJ)sY1 fJ)+B(K)*X1 (J)**(K»1)
Yl (JlaALnGIOCYl (,TJ>
1150 CONTINUE
C**********************************************************************
C*
C* RPPFFTHE "00 50" LOOP INRFX BEGINNING AND FNO. BFGINNINR
r* INTfRVAi tl BEGINS TWO (X1,Y11 POINTS BEFORE 050 OF FIRST
c* STAHF'. i AST INTFWVAL, INTSI, FNOS WITH DMAX. RETURN TO TOP OF
c* LOOP TII UNO FITTING POLYNOMIAL COEFFICIENTS OVER Twrsf
C* IWTFPVAI S.
c*
r************«*********************************************************
I I=NPT-NN-2
INTS1aMPT-1
RO Tfl ?J
C*
C* P'T = ii')MBF.R OF INTERVALS FOR WHICH FITTING COEFFICIENTS HAVE
C* HfFN nf f I NED.
C*
r****** **********»***************************#***•**•*****•************
55 InT=NPT-1
c**********************************************************************
c*
r* FILE NIIMHFR OF FITTEO POINTS, THE INTERVAL BOUNDARY POINT
C* VAU'FS, AND FITTING COEFFICIENTS FOR EACH INTERVAL.
C*
r*********************************************************«************
WRITF(H »IAV)MPT. INT, (XI (T 1,1*1 ,NPT), ( Y 1 (H , I«l ,NPT) ,
1 f (COFl f I,J),J=1,3),I = 1 .INT 3
(100 CniMTJNIlf
STnp
1 FOH^AT (il?)
fun
•SPLI 2^3
*
*
•SPLI 2*7
*SPI, I 2^6
•SPLI ?99
SPLI 300
SPLI 301
SPLI 302
SPLI 303
SPLI 30«
•SPLI 306
•SPLI 307
•SPLI 30B
*
*SPLI 310
*SPLI 311
•SPLI 312
•SPLI 313
*SPLI 314
SPLI 316
SPLI 317
*SPL I 31 B
•SPLI 319
•SPLI 320
*SPLI 3?1
*SPL1 322
*SPLI 323
SPLI 3?4
*SPl I 3?5
*SPUI 326
*SPLI 327
•SPLI 32B
•SPLI 32"
•SPLI 330
SPLI 331
SPLI 332
SPLI 333
SPLI 336
SPLT 337
SPLI 345
©
co«'t.
188
-------
TABLE A-4. SUBROUTINE OSCFIT LISTING
001
002
003
004
005
006
007
OOfl
009
010
Oil
012
013
014
015
016
017
01S'
019
020
021
022
023
024
025
026
027
028
029
030
031
032
033
034
035
036
037
038
039
040
041
04?
044
045
046
047
048
049
050
051
052
053
054
055
056
057
05P
059
060
C*
C*
C*
C*
C*
C*
C*
C*
C*
C*
C*
C*
C*
C*
C*
C*
C*
C*
C*
C*
C*
C*
C*
C*
C*
C*
C*
c*
C*
C*i
c*.
c*
c*
c*
c*
c*
c*
c*
c*
c*
f . .
c ;H
SUBROUTINE OSCFlT(XO,YO,Xl,Yl,YPO,VPi,XIN,A4,A3,A?,Al)
OSCFIT01
c*
c*
c*
c*
c*
c*
c*
c*
c*
c*
c*
c*
c*
c*
c*
c*
c*
c*
c*
c*
c*
c*
c*
c*
c*
c*
c*
c*
c****
c****
c*
c*
c*
c*
c*
c*
c*
c*
c*
c****
to
„
THIS SUBROUTINE GENERATES THE COEFFICIENTS OF A FIRST QRDFR
OSCULATING POLYNOMIAL OF DEGREE THREE TO APPROXIMATE Y(X) OVER
THE INTERVAL XO.LF.X.LE.X1 ,
INPUTTED VALUES ARE {
XO THF VALUE OF X AT THE LOK£R LIMIT
YO THE VALUE op Y(X) AT x EQUAL THE LOWER LIMIT
YPO THE VALUE OF THE FIRST DERIVATIVE Dy/DX AT XaXO
XI THE VALUF. OF X AT THE UPPER LIMIT
YI THE VALUE OF Y(X) AT X EQUAL THE UPPER LIMIT
YP1 THE VALUE OF THF FIRST DERIVATIVE DY/DX AT XaXl
OUTPUTTF.D VALUfS ARE :
*
*
X1N j THE NEW UPPER LIMIT (LARGEST VALUE) OF X IN THE INTERVAL
XO.LE.X1N.LF.X1 SUCH THAT THE APPROXIMATING FUNCTION
PCX) is ALWAYS GREATER THAN OR EQUAL TO ZERO (I.F,
NEVER NEGATIVE),
AI i THE COEFFICIENTS OF THE BEST FIT CUBIC EQUATION IN THt
*
*
*
A2 OSCULATING SENSE PASSING THROUGH YO AND Yl AND MATCHING*
A3 THE SLOPES YPO ANn YP1 AT THESE POINTS,. NOTF THAT IF
*4 X1N.LT.X1 THE CALLING PROGRAM SHOULD OEFINK P(X)=Y1
(I.E. A CONSTANT) AND DP(X)/DX a YP1 (ALSO A CONSTANT)
IN THE INTERVAL XIN.LE.X.LE.XI ,
THE GFNERATFD APPROXIMATING POLYNOMIAL IS GIVEN RYl
PCX) = A1*(X**3) + A2*(X**)2 + A3*(X) + A4
***************************************************** ************
******************************************* **********************,
hi IS A COUNTER* OF THE NUMBFR "OF TIMES THRU THE ITERATIVE LOOP
STEP IS THE INITIAL BACKWARD STEP SIZE IN X FOR THF BACKWARD
SUCCESSIVE APPROXIMATION OF X1N THE NEW UPPER LIMIT IN X
STfPl IS fHE FORWARD STEP SIZE USED TO DETERMINE IF THE
POLYNOMIAL HAS A NEGATIVE SLOPE ANYWHERF. WITHIN
THE PAhGE OF X VAUt'FS. STEP1 IS ALWAYS TAKEN AS
CMPPFR LIMIT . LOWER LIMIT)/99,
************************************************************* *****
NsC
STFPsl.OE-1
*X 1=X l
^ELXSX 1 »Xfl
STFPl=OELX/99.
N = N + )
l.'ELlaXii.xX!
OR 2s-r>fl.l
^ELJSsOEL i**2
R£l 2SsDEl e>**2
DEUC = DEL1S*OE1.1
^Fl2C=f5EL2S*OEL?
KlcYO/DFLtS
«?.=-?. *YO/OEl 1C
K3rYi/T>EL?S
Kis«2 ,*Y1/DEL2C
K5aYPo/DFLlS
*
*
*
*
*
*
*
*
*
t*
ncr c T T «3
v'OUr i 1 u c
»*
*
*
*
*
*
*
*
*
*
*
nSCF!T03
05CFIT04
OSCf-ITOS
nsCFlT06
nSCF IT(!7
OSCFITOfi
OSCFIT09
OSCFIT10
OSCFITtl
OSCF1T12
OSCFIT13
OSCFITU
OSCFTTlb
HSCFIT17
nSCFITl?
OSCHT19
189
-------
061
062
063
064
065
066
067
068
069
070
071
072
073
07U
075
076
077
078
079
080
081
062
083
K63VPJ/DEL2S
AtsK?tKa+K5+K6
A2=CM+K3-(2.*XXUXG)*(K2tK5)-(2,*XO+XXl)*(K4+K6n
e s
OB6
087
088
089
090
09)
092
093
094
095
096
097
09«
009
100
101
102
105
106
1-2.*K1*XX1-2.*K3*XO
AU3(M*CXX1**2)*K3*CXO**2)-XO*(XX1**2)*
-------
APPENDIX B
FINAL REPORT
SRI Contract No. N0014
(Acct. No. 3858-13)
"Modeling of SRI Impactor Data"
MODELING OF SRI IMPACTOR DATA
AND CALCULATION OF IP CONCENTRATION
Prepared by
A. Kapadia, D. Y. H. Pui, and K. T. Whitby
Submitted by
K. T. Whitby, Professor and Chief
Environmental Division
Mechanical Engineering Department
University of Minnesota
Minneapolis, Minnesota 55455
October, 1979
Particle Technology Laboratory Publication No. 403
191
-------
ABSTRACT
The SRI impactor data for emissions from several different stationary
sources have been analyzed using a modal analysis technique, in which the data
are fitted with multi-component log-normal distributions. The technique is used
because the available data do not contain sufficient size classification infor-
mation in the upper particle size limit of 10 to 100 ym. This information is
needed in order to estimate the inhalable particulate (IP) concentration, i.e.,
all particles below 15 ym aerodynamic diameter, for these source aerosols. By
using a simplex minimization method, the present technique projects the entire
size distribution by fitting the available portion of the size distsribution
with log-normal distributions. It has previously been applied successfully to
atmospheric aerosols (Whitby, 1978). This study shows that the technique can
indeed be extended to the source aerosols and work reasonable well. The condi-
tions under which the technique applies are discussed.
The size distributions thus obtained are integrated with the proposed IP
sampler curve. The resulting IP concentrations are presented, together with
their sensitivity to the assumed modal parameters of the size distributions.
The effects of varying the sampler cut size and the sharpness of cut on the
IP concentrations are included. The calculations indicate that the present
specifications for the performance curve of the IP sampler, 059 = 15 ± 2 ym
and 0gQ = 1.0 - 1.7, are adequate because the resulting IP concentrations
change by less than ±20% within the tolerance of the performance curve.
192
-------
Abstract
Contents
List of Figures
I. INTRODUCTION
A. Background ....
B. Estimation of Inhalab
C. Log-Normal Function f
II. ANALYSIS TECHNIQUE . .
A. Minimization Procedur
B. Estimation of IP Cone
C. Sample Data Analysis
1. Aerosol at the in
2. Aerosol at the ou
D. Sensitivity Analysis
1. Modal parameters
a. Variation of
b. Variation of
2. Effect of IP samp
a. Variation of
b. Variation of
III. RESULTS AND DISCUSSION .
A. Coal-Fired Boiler, Ho
B. Coal-Fired Boiler, Co
C. Copper Smelter, Hot S
ABLE OF CONTENTS
Page
i
ii
.......... iv
1
1
e Particulate (IP) Concentration . . 3
r Aerosol Size Distribution .... 4
7
s 7
ntration 11
11
et of control device 12
let of control device . 12
15
15
15
18
er curve on IP concentration .... 21
gD 21
50 • 21
23
Side ESP Inlet (N6P-13) 23
d Side ESP Inlet (JBI-13) 32
de ESP Inlet (AJOI-3, AJOI-5). ... 41
193
g-
WID2
-------
TABLE OF CONTENTS (cont.)
Page
IV. SUMMARY AND RECOMMENDATIONS 51
A. Summary 51
B. Recommendations 52
References 53
APPENDIX A. COMPUTER PROGRAM FOR IMPACTOR DATA MODELING AND
IP CONCENTRATION CALCULATION 54
194
-------
No.
I.I
1.2
Ilil
II.2
II.3
II.4
II.5
II.6
II.7
II.8
III.l
III.2
III.3
III.4
III.5
.1ST OF FIGURES
Title
Modeling of aerosols at the inlet and outlet of an electro-
static precipitator: (ap inlet aerosol, (b) electrostatic
precipitator penetration, and (c) output aerosol ,
Page
Modeling of aerosols at the inlet and outlet of a scrubber:
(a) input aerosol, (b) scrubber penetration, and (c) output
aerosol 6
Graphical representation
minimization procedure.
Modeling of impactor dat
obtained at the inlet of
Modeling of impactor dat
of the search method in the simplex
a for a coal-fired power plant
an electrostatic precipitator.
a for a coal-fired power plant
9
13
obtained at the outlet of an electrostatic precipitator. . . 14
Effect of assumed a of the inlet aerosol and
g2
sampler curve on the IP
Effect of assumed ag2 of
sampler curve on the IP
Effect of assumed MMD2 o
sampler curve on the IP
Effect of assumed MMD2 o
concentration estimate ....... 16
of the
the outlet aerosol and
of the
:oncentration estimate ....... 17
F the inlet aerosol and 0gQ of the
:oncentration estimate ....... 19
F the outlet aerosol and ago of the
sampler curve on the IP concentration estimate.
Effect of IP sampler cur
concentration for the in
Modal fitting of impacto
hot side ESP inlet (NGP-
Effect of assumed aeroso
concentration estimate (
Effect of assumed aeroso
concentration estimate (
e cut size on the estimated IP
et data ,
data from coal-fired boiler,
3, p = 1.0)
and sampler aqQ on the IP
1GP-13, p = 1.0).
MMD2 and sampler aqp on the IP
K3P-13, p = 1.0).
Effect of sampler
(NGP-13, p = 1.0)
on
the IP concentration estimate
Modal fitting of impactor data from coal-fired boiler,
hot side ESP inlet (NGP-13, p = 2.41)
20
22
24
25
26
27
28
195
-------
LIST OF FIGURES (cont.)
No. Title Page
1 1 1. 6 Effect of assumed aerosol 0^2 and sampler agD on the IP
concentration estimate (NGP-13, p = 2.41). . ........ 29
I II. 7 Effect of assumed aerosol MMD2 and sampler agQ on the IP
concentration estimate (NGP-13, p = 2.41) .......... 30
I II. 8 Effect of sampler DSQ on the IP concentration estimate
(NGP-13, p = 2.41) ..................... 31
II I. 9 Modal fitting of impactor data from coal -fired boiler,
cold side ESP inlet (JBI-13, p = 1.0) ............ 33
I II. 10 Effect of assumed aerosol ag£ and sampler agp on the IP
concentration estimate (JBI-13, p = 1.0). . ....... . 34
I II. 11 Effect of assumed aerosol MMD2 and sampler OgQ on the IP
concentration estimate (JBI-13, p = 1.0) .......... 35
III. 12 Effect of sampler 050 on the IP concentration estimate
(JBI-13, p = 1.0) ...................... 36
1 1 1. 13 Modal fitting of impactor data from coal -fired boiler,
cold side ESP inlet (JBI-13, p = 2.34) ........... 37
1 1 1. 14 Effect of assumed aerosol cg2 and sampler agD on the IP
concentration estimate (JBI-13, p = 2.34) .......... 38
1 1 1. 15 Effect of assumed aerosol MMD2 and sampler aqQ on the IP
concentration estimate (JBI-13, p = 2.34). / ........ 39
III. 16 Effect of sampler DSQ on the IP concentration estimate
(JBI-13, p = 2.34) ..................... 40
III. 17 Modal fitting of impactor data from copper smelter,
hot side ESP inlet (AJOI-3) ............ . ..... 43
1 1 1. 18 Effect of assumed aerosol ag2 and sampler agQ on the IP
concentration estimate (AJOI-3) ............... 44
II I. 19 Effect of assumed aerosol MMD2 and sampler agp on the IP
concentration estimate (AJOI-3) .......... ..... 45
I II. 20 Effect of sampler DSQ on the IP concentration estimate
(AJOI-3) ....................... 46
1 1 1. 21 Modal fitting of impactor data from copper smelter,
hot side ESP inlet (AJOI-5) ......... ' ........ 47
196
-------
LIST OF FIGURES (cont.)
No.
111.22 Effect of assumed aeroslol
concentration estimate
111.23 Effect of assumed aeroslol
concentration estimate
111.24 Effect of sampler
(AJOI-5)
on the IP concentration estimate
Title
Og2 and sampler ago on the IP
ui uq^ a
(AJOI-5)
Page
48
MMD? and sampler aaD on the IP
(AJOI-5) y 49
50
197
-------
I. INTRODUCTION
This is the final report on SRI Contract No. N0014, entitled "Modeling
of SRI Impactor Data". A progress report, PTL No. 391, was submitted in May,
1979. It described the analysis technique used in the impactor data reduction
together with the sample data analysis for two sets of in-stack data. The
present report extends the analysis technique to other SRI data (Section IV).
In addition, a computer program developed for this study is included in Appen-
dix A.
A. Background
Atmospheric aerosols have been found to encompass almost six decades of
size, from clusters on the order of 10 A" to particles on the order of 100 ym
in size. Further study of atmospheric aerosols has shown that the combination
of different generation mechanisms with different classification mechanisms
produces modes or sub-range groupings often having more or less homogeneous
physical and-chemical properties. Condensational mechanisms produce particles
primarily less than 1 or 2 urn in size. On the other hand, mechanical processes
produce particles from about 1 ym up. Aerosol growth dynamics often operate to
limit condensation-sized aerosols to less than 1 ym and to limit the amount of
mass transfer between the smaller fine particles and the coarse particles.
Furthermore, it has been found convenient to characterize these groupings
by additive log-normal distributions called modes. In particular, we have found
that three modes—one with a number geometric size peak around .01 ym, called
the nuclei mode; a second one with a number peak around .07 ym, called the
accumulation mode; and a coarse particle mode peaking in the 5 to 20 ym range—
are sufficient to characterize atmospheric distributions well.
198
-------
the
Since it was known that tf
must operate in stationary sour
concentrations and size limits,
analysis techniques developed
aerosols.
The analysis is based on a
atmospheric aerosol analysis
1) Three additive log-
terization for the
2) Geometric standard dev
tions are usually on
the modes do not overl
by analysis techniques
ajnalysis of this report
3) Classification techni
cleaners or by settli
tions of the modes and
shape of the modes
by a log-normal
The usual method of size d
testing has been the cascade
size distribution information i
generally the aerosol between
not characterized well, nor is
practice, this means that the
filter, is not going to be
e same formation and classification mechanisms
ces that produce particulates, with different
it seems reasonable to apply the same modal
for atmospheric aerosols to the stationary source
number of assumptions mostly derived from the
experience findings and assumption as follows:
normal distributions are usually an adequate charac-
aeriosol.
ling
iations of these additive log-normal distribu-
order of 2. This is small enough so that
ap to the point where they cannot be separated
such as the simplex program used for the data
ques.
, such as those introduced by the gas
, tend to modify the geometric standard devia-
the means of the modes, but do not change the
enough so that they can still be fitted quite well
distribution.
stribution analysis for source particulate
imbactor. The cascade impactor yields fairly good
ih the few tenths of a ym to 10 ym range, but
the cut size of the last stage and the filter is
he size distribution from 10 ym and up. In
niiclei mode, which is essentially all on the
characterized very well by size, nor is the coarse
199
-------
mode if its size is larger than 10 ym. The coarse particle mode also may not
be characterized well because impactor stages which operate above 10 ym have
not been too practical.
However, from work on atmospheric aerosols, when the size distribution is
not known in enough detail over the entire particle size range, good estimates
of the total mass in the nuclei and coarse particle modes can be made by partial
fitting to the available portion of the size distribution. The work done for
this project is an extension of the analysis to stack data.
B. Estimation of Inhalable Particulate (IP) Concentration
Currently EPA has under consideration the specification of an inhalable
particulate sampler which would have a cut size of 15 ± 2 ym and a geometric
standard deviation of the classification efficiency curve in the range from 1.0
to 1.7. The idea is to use this pre-classifier on samplers so that the sample
passing would presumably be that which could be inhaled by a normal human being.
One question is how sensitive the mass passing through such a sampler would be
to the cut size and the slope of the IP classification curve. Therefore, we
investigated this by integrating the IP curve with geometric standard devia-
tions varying from 1 for the perfect classification, to IP curves with a ag
of 1.7. These were integrated with a typical size distribution ahead of the
gas cleaner and a typical one after the gas cleaner. Not surprisingly, these
results show that the mass passing from the IP classifier is not very sensitive
to the slope of the efficiency curve, but is quite sensitive to the cut point.
This indicates that for such a sampler, much greater variations on the slope
could be tolerated than in the cut point.
The investigation of the variation of the assumed ag of the coarse particle
mode showed that it too can be varied over quite a range when fitting coarse
particle mode without seriously affecting the IP results.
200
-------
C. Log-Normal Function for Ae
Atmospheric aerosols have
a three-component or a trimoda
size range of 0.002 urn to 60 yr
is used in the present study t
of a control device. It is re
of the control device can be UK
bution, but no a priori assump
the control device can be made
the outlet aerosol by the mult
lowing calculations were made.
The histogram in Figure I
at the inlet of the control de
electrostatic precipitator (Me
ure I.lc is the calculated aer
by taking the product of Figure
inlet and the outlet aerosols \
As can be seen from the figure
multi-component log-normal dist
values are of the same order o
aerosols. Similar results were
scrubber efficiency (McCain, 19
Figure 1.2.
osol Size Distribution
been modeled successfully by Whitby (1978) using
log-normal distribution for particles in the
. The three-component log-normal distribution
model the aerosol at the inlet and the outlet
sonable to assume that the aerosol at the inlet
deled by a multi-component log-normal distri-
ion regarding the aerosol at the outlet of
To demonstrate the feasibility of modeling
-component log-normal distribution, the fol-
ia shows a typical mass distribution measured
ice. Figure I.lb shows the efficiency of an
ain et al., 1975), and the histogram in Fig-
sol at the outlet of the precipitator obtained
s I.la and I.lb. The histograms of both the
ere fitted by bimodal log-normal distributions.
, the aerosol can be modeled satisfactorily by
•ibution. It should be noted here that the x^
magnitude for both the inlet and the outlet
found when the calculations were done using a
78). The resulting calculations are shown in
201
-------
co
PARTICAL PENETRATION, PERCENT
AM/Alog Dp , mg/m3
to
o
tv)
AM/Alog Dp , I02 mg/m5
O)
o>
"o o.
rfr r*-
Cu CU
ft- <-*•
O -'•
T O
t-« °
O m il '
-g M p> 3
T- u1 o> •g
-------
*
m
02
,
fo
AM/Alog D
55
PERCENT
80r-e
O
PARTICALE PENETRA
O
20-
0
4.0
« 3-2
e
- 2.4
O.
Q
1.6
0.8
0.1
INPUT AEROSOL 1
DG./im
1st Mode 2.084 I.
2nd Mode 31.59 2
Xz»
fc
1 1 1 1 1
Figure 1.2 Modeling of aerosol
input aerosol, (b)
0 CONTROL DEVICE
G M.mg/m3
9 756.6
2 4825
/
(a)
1 1 I | I 1 III 1 1 I | I I I
SCRUBBER PENETRATION
(b)
' ' l""l r ^^r —
OUTPUT AEROSOL FROM SCRUBBER
DG,/zm SG M,mg/m3
1st Mode 0.897 1.46 1.402
2nd Mode 1.92 2.31 3.8(-3)
X2 = 6.84 (-3)
SCRUBBER EFFICIENCY FROM:
McCain (1978) FIG. II
1 t f 1 t 1
(c)
10.0
100
Dp , p.m
s at the inlet and outlet of a scrubber: (a)
scrubber penetration, and (c) output aerosol.
203
-------
II. ANALYSIS TECHNIQUE
Several procedures for fitting functions with non-linear coefficients are
available. Bevington (1969) and Daniel and Wood (1971) describe several numer-
ical methods for non-linear curve fitting. Raabe (1978) has recently applied
a non-linear curve fitting method to obtain the modal parameters of aerosol
size distributions. In the present study, a multi-component (up to three)
log-normal distribution was fitted to impactor data from stack sampling mea-
surements. It should be noted here that no correction for bounce and cross-
sensitivity were made to the raw data before the fitting procedure was carried
out.
A. Minimization Procedures
The fitting procedure used in the present study is based on the simplex
minimization method developed by Nelder and Mead (1965).
In order to use this method, the following three requirements have to be
specified: (a) the form of the solution which can be specified by several
parameters (trimodal log-normal distribution), (b) the minimization procedure
which gives the set of parameters that best fit the data, and (c) the function
to be minimized (hereafter referred to as the objective function).
In this method, a function of m variables is minimized using a direct
search method. For a function of m variables, a general simplex of m+1 ver-
tices, denoted by Pj, j=l,2,....m+l is initially formed and the objective func-
tion evaluated at each of the vertices. The value of the objective function at
PJ is denoted by yj. The vertex which has the highest value of the objective
function (Pn) is then replaced by another point.
To determine the new point, first the centroid (?) of the remaining points
(i.e., PJ, j=l,2....m+l, j^h) is calculated and the location of the next point
204
-------
is determined by one of the thrpe operations: reflection, expansion, or con-
traction.
The coordinates of P* determined
variiibl
f'2
minirmm
where o is a positive constant.
and on the line joining F and P
at P*) is between yn and y] (hi
evaluated at the vertices of th
pi ex formed and the process res
To show the process graphi
to be fitted. Let the two
and ?3 form the initial simplex
lie on the line joining PI and
If P* produces a new
and P** located. The coordinat
P*
and in Figure II.l, Y = (P**F)/
less than y*, P* is replaced by
started. Otherwise, Pn is rep!
On the other hand, a
tion fails. The location of P
P
with the contraction coefficient
ure II.l, (P***P)/(Pn/F) equals
i.e., the contraction point is
traction fails, all the P-j's are
started.
by reflection are given by
P* = (1 + a)P - a Ph (II.l)
Thus, P* lies on the far side of F from Pj,,
. If y* (value of the objective function
jhest and lowest value of the objective function
simplex), Pn is replaced by P*, and a new sim-
arted.
ally, consider a two-variable function which is
es be Xi and X£. In Figure II.l, PI, ?2»
and assume that PS is Pn. The point Fwill then
and P* on the line joining ?3 and F.
, i.e., y* < y], an expansion is attempted
is of P** are determined by
= YP* + (1 + Y)F (II.2)
P*F), where Y is greater than unity. If y** is
P**, a new simplex formed, and the process re-
ced by P*.
is carried out, giving p***, when reflee-
is determined by
= fJPh + (1 - B)F (H.3)
taking values between 0 and 1. In Fig-
0. P*** replaces Pn, unless y*** > min(yn,y*),
vjorse than the better of Pn and P*. When con-
replaced by (Pi + Pi)/2, and the process re-
contrcction
*•**
***
205
-------
CM
X
X,
Figure II.1 Graphical representation of the search.method in the simplex
minimization procedure.
206
-------
In this manner, the simple>
present work, a = 1.0, y = 0.5,
continuously adapts itself to the landscape
and contracts in the neighborhood of a minimum. The search process is halted
when dimensions of the simplex tecome smaller than certain set values. In the
and 3 = 2.0 were chosen as recommended by Nelder
and Mead (1965). It should be roted here that the minimum may not necessarily
be a global minimum, but only a local minima. In order to ensure that it is a
global minima, either the minimization process should be repeated with different
starting simplices, or the solution can be taken as reasonably close to the
global optimum if the value of 1
Three forms of the objectix
square, x^m» given by Equation (
(11.5); and the weighted least <
,2 .
1 m
2 (MT' -
V = '
x MT1
WLS • \ on.. - on^;
MT A UMj1 + AM^/2
(II.4)
AM.1
AM.
(II.5J
AM.
(II.6)
0T
s based on the trimodal log-normal distribution
ass on the itn impactor stage; MT and MT1 are
l mass; and a-j is the uncertainty in the mea-
ends on the instrumental uncertainty and the
stability of the measured aeroscl. As no a priori estimate of a-j is available,
oportional to the normalized concentration.
207
-------
The proportionality constant is assumed to be unity for minimization purposes,
as only the relative values WLS are of interest.
From the analysis of two sets of impactor data, the normal chi-square
appears to give a better fit than the modified chi-square in the lower size
limit of the distribution. Subsequent study by Kapadia (1979) shows that the
WLS is capable of giving better results than the normal chi-square. Therefore,
all the fitting results in Section III were obtained using the WLS as the ob-
jective function.
B. Estimation of IP Concentration
Equation (11.7) gives the estimate of the IP concentration, once the op-
timum set of parameters for the three-component log-normal distribution are
determined,
IP = j SDC(D50, OgD) . fM(Dp) d In Dp (II.7)
where SDC(D5Q, .Og0) is the function representing the sampler curve with DSQ as
the cut size and Ogp as the spread of the sampling characteristic, and fj^Dp)
is the optimized three-component log-normal distribution which best fits the
differential mass data for the impactors.
C. Sample Data Analysis
Two sets of data have been selected from the SRI report to the Electric
Power Research Institute (EPRI FP-792). The data were obtained from a coal-
fired power plant equipped with an electrostatic precipitator. Set A gives
the aerosol mass distribution measured at the inlet of the control device, and
Set B gives the mass distribution at the outlet of the control device. These
sets of impactor data compare favorably with the electrical aerosol analyzer
data in the size range where they overlapped, i.e., between 0.1 ym to 1.0 ym.
208
-------
1. Aerosol at the inlet of control device
The inlet data were obtained by SRI using a modified Brink cascade im-
pactor. The impactor has seven stages (SO-S6), an in-line cyclone precollector,
and a back-up filter. Following the recommendation of J. D. McCain, we have
combined the cyclone mass with SO mass and also neglected the back-up filter
mass. The reason for combining cyclone mass and SO mass is due to the uncer-
tainty in the cyclone cut size. The back-up filter mass was not used because
of the impactor bounce problem. The histogram of the impactor data is shown
in Figure II.2. The EAA data are also included, and they are plotted with
hatched lines. In the present study, only the cascade impactor data were used
in the analysis. The EAA data are shown here for comparison purposes.
The impactor data were fitted with a bimodal log-normal distribution using
the simplex minimization procedures described above. The resulting modal param-
eters and fitted curves are shown in Figure II.2. Good fitting is obtained in
the first mode. In the second mode, the fitted curve covers only two histogram
intervals. Section IID describes the sensitivity analysis for this mode, giving
the IP concentration as a function of
-------
30
25
E
M
O
o.
O
o>
o
15
10
I > i Mil
1st Mode
2nd Mode
I i i i ni
INLET DATA
2.22
48.5
0.02
SG M,mg/m3
2.23 1049
2.0 4578
2
(fe"fc)« 3.19 (-2)
TC
V
ij
i i 111
0.1
1.0 10.0
Dp , fj,m
100.0
1000.0
Figure II.2 Modeling of impactor data for a coal-fired power plant obtained at
the inlet of an electrostatic precipitator.
-------
OUTLET DATA
SG M,mg/m3
1st Mode 0.80 2.23 9.35
2nd Mode 7.72 3.12 5.34
(fe'fc)* 545(-2)
TC
0.02
1000
Figure II.3 Modeling of impactor data for a coal-fired power plant obtained at
the outlet of an electrostatic precipitator.
-------
The impactor data were fitted with a bimodal log-normal distribution using
the simplex minimization procedure. The resulting modal parameters and fitted
curves are shown in Figure II.3. It can be seen that the outlet data can be
fitted satisfactorily with the bimodal log-normal distribution. The sensitivity
analysis for the second mode were also done and are given in the next section.
D. Sensitivity Analysis
One of the main objectives of this study is to develop an analysis tech-
nique for estimating the inhalable particulate (IP) concentration from the
existing in-stack data. The accuracy of the estimated IP concentration depends
on the accuracy of the impactor data modeling and the criteria used for the IP
sampler curve. The analysis below shows the sensitivity of the estimated IP
concentration to the variation of the fitted parameters of the impactor data
and the characteristics of the IP sampler curve.
1. Modal parameters
a. Variation of ag
Figures II.4 and II.5 give the variation of IP concentration as a function
of the geometric standard deviation of the second mode, ag2» f°r tne inlet
data (Figure II.2) and the outlet data (Figure II.3), respectively. The IP
concentration is obtained by keeping the 0g2 constant duri.ng the minimization
process. The x^ of the fitted distribution is also plotted in the same figure
to indicate the goodness of fit. The two curves for the IP concentration are
for ago = 1.0 and crgQ = 1.7 with DSQ = 15.0 ym. The discussion for the IP
sampler curve is given in Section 112.
Figure II.4 shows that, for the inlet data, the total mass below 15 ym for
the sampler curve with ago =1.0, M-15, changes by less than 20% as the geo-
metric standard deviation for the second mode, ag2, varies from 1.0 to 2.0.
212
-------
CIS
IP CONCENTRATION , mg/m3
X s
-------
U3
tn
o m
c -h
-5 -*>
< fD
ro O
O
3 O
O
rn
O
m
_O
IP CONCENTRATION , mg/m3
ro oi 5 „ _ oi
o>
ro
Oi
CO
-o 3
ro
O OL
3 Q
OtQ
n> ro
3
<-*• o
-5 -h
O)
c^ c^
_i. 3-
o ro
3
O
(D C
V) r+
c*- —'
^3 ^*"
0)
(D fD
• -5
O
O>
a.
P
to
o
o
-h
ro
o*
•o
ro
Y2 - V (fe-fc)2
A "^ fe
-------
The IP concentration for a sampler curve with DSQ = 15.0 ym and agp = 1.7
changes by about 40% in the same range of og2. These curves indicate that
for a sampler with a narrow cut-off characteristic and assuming that the
of the inlet aerosol size distribution is below 2.0, the IP concentration can
be determined to an accuracy of 20% for the sample data used. The closed sym-
bols in the figure indicate optimum parameters obtained by setting all six modal
parameters as variables during fitting procedures. The parameters are seen to
locate very close to the minimum x2 value. The geometric standard deviation
for the second mode was found to be 1.7, giving an M-15 value of 1100 mg/m^.
For the outlet data, Figure I I. 5 shows that the M-15 and IP concentration
vary by less than 15% over a wide range of ctg2, from 1.1 to 5.0. Therefore,
the IP concentration can be estimated with good accuracy for the sample outlet
data.
b. Variation of MMD2
Figures II. 6 and II. 7 give the variation of IP concentration as a function
of the mass median diameter of the second mode, MMD2, for the inlet data and
the outlet data, respectively. The IP concentration is obtained by keeping the
MMD2 constant during the minimization process. The x2 of the fitted distribu-
tion is also plotted in the same figure.
Figure II. 6 shows that, for the inlet data, the IP concentration is a
strong function of the MMD2. This figure, together with Figure II. 4, indicates
that additional information concerning the second mode will be needed in order
to obtain a good estimated IP concentration for the inlet data.
For the outlet data, Figure I I. 7 shows that the M-15 and IP concentration
vary by less than 20% over a wide range of MMD2, from 2.0 to 40.0 ym. This
figure, together with Figure II. 5, indicates that for the outlet data, the IP
concentration can be estimated with good accuracy.
215
-------
to
c
-s
CO
IP CONCENTRATION , I02mg/m3
o m
C -h
•V-h
n> r>
<-«•
o
a o
-h
fD in
in
J-'C
-o 3
0)
O Q.
O
aro
c+
-» o
ft-1*
fD -••
(/> 3
c* — •
c-f o>
o> ro
• -$
o
cu
Q.
o
o
O>
T3
fD
-J
-------
to
H
IO
15
14
Od
\-
§ 12
z
o
o
10
I—I I I I 1
IP CURVE
Cut-size,/im ergo
15.0 1.0
I
I
1 I i 1 I I
I
0.18
0.16
0.14
M
0.12
<
0.10
w
0.08 „
CM
0.06 X
0.04
0.02
2.0 5.0 10.0 2OO
MASS MEDIAN DIAMETER FOR SECOND MODE ,
50.0
Figure II.7 Effect of assumed MMD2 of the outlet aerosol and 0aD of the sampler
curve on the IP concentration estimate. 9 sampler
-------
2. Effect of IP sampler curve on IP concentration
a. Variation of agQ
Figures II.4 through II.7 give the IP concentration for two IP sampler
curves: (1) DSQ = 15.0 ym and agD = 1.0, or M-15, and (2) DSQ = 15.0 ym and
OgQ =1.7. As seen in the figures, the IP concentration is not strongly af-
fected by the cjgD. The IP concentration varies by less than 5% for the outlet
data and by less than 15% for the inlet data.
b. Variation of DSQ
The variation of IP concentration as a function of cut size of IP sampler
curve is shown in Figure II.8. The ordinate gives the ratio of IP concentra-
tion at various cut sizes to the concentration at 15 ym cut size. As expected,
the ratio depends strongly on the cut size. Such a curve is useful in setting
the permitted variability in the cut size parameter.
218
-------
to
14 16 18
IP (0-g*l.7) CUT-SIZE , /xm
Figure II.8 Effect of IP sampler curve cut size on the estimated IP concen-
tration for the inlet data.
-------
III. RESULTS AND DISCUSSION
Three pairs of SRI impactor data have been selected for the analysis. The
data were taken at the inlets of control devices for two pulverized coal boilers
and a copper smelter. For the pulverized coal boilers, each pair consists of
the same set of impactor data except calculated with two different assumed par-
ticle densities. The density difference results in a change in the impactor
cut size, giving a somewhat different particle size distribution. The analysis
of the data therefore provides some information on the effect of the assumed
particle density on the estimated IP concentration.
The analysis here follows the formats of Section II.C and II.D, which in-
clude the fitted modal parameters and the sensitivity of the IP concentration
to the variation of the coarse particle.size and deviation, as well as the per-
formance curve of the IP sampler.
A. Coal-Fired Boiler, Hot Side ESP Inlet (N6P-13)
The results of the modal fitting and the IP concentration calculation
are shown in Figures III.l through III.4 for an assumed particle density of
1.0 g/cm3, and in Figures III.5 through III.8 for an assumed particle density
of 2.41 g/cm3.
The important findings to be noted are:
1. The impactor data can be fitted satisfactorily with a bimodal log-
normal distribution.
2. Because the mass concentration of the second mode is not substantially
higher than that of the first mode, the estimated IP concentration
consequently is not very sensitive to the variation of the geometrical
standard deviation and the mass median diameter of the second mode,
the coarse particle mode.
220
-------
35
to
to
I st MODE
2nd MODE
0.01 O.I 1.0 10.0 1000 1000.0
Dp, /Am
Figure III.l Modal fitting of impactor data from coal-fired boiler, hot side
ESP inlet (NGP-13, p = 1.0).
-------
zzz
-n
C
n>
ro
-5 -•>
Q> -h
n>
3
01 C
rt- 3
Z Oi
o ro
-^ -5
I O
!-• «/>
CO O
•O Q
to
n ro
o o.
• to
o>
•§
n>
O
O
n>
o
o
3
O
(V
IP CONCENTRATION , I02mg/m3
yi
01
o —
m —
O
m
H
O
2.0
STANDAR
O
m
o
m
o
§g
m o
q
•Q
ro
o
m
k
f
J
o
b
ro
i
§
i
§
P
O
00
p
O
WLS=2-
-------
Id
c
0>
K)
to
CO
CO
n- m
T -*>
o» -*>
O
3
C* D)
-J. C/»
3
QJ c
rt 3
n> rt>
o.
z 01
CD (D
•O -5
I O
(-• W)
C*> O
II O
ro
• Q)
O 3
Ql
•a
«-^
n>
o
o
IP CONCENTRATION , I02mg/m3
rn
o
m
H
rn
s
o ,
§3
o
00
00
4
-1
4
°"
GJ
o
I
II
b
aos
— — w
tn en N'
o
c
m
q _
n-
3T
fD
_ 3
WLS=:
o
o
o
n>
-------
to
10
12 14
SAMPLER D50
16
18
20
Figure III. 4 Effect of sampler
(NGP-13, p - 1.0).
on the IP concentration estimate.
-------
to
U1
100.0
1000.0
D
P|
Figure III.5 Modal fitting of impactor data from coal-fired boiler, hot side
ESP inlet (NGP-13, p = 2.41).
-------
to
to
o\
10
^20
e
w
O
- 15
6
LJ
O
O
u
10
CLOSED 'SYMBOLS'
OPTIMUM PARAMETERS
NGP-13 , />*2.4I
A
O 15/im STEP FUNCTION
D IP CURVE (15/im cut,crgD = l.7)
1
0.08
M
0.07 3
i
M
0.06 (/>
0.05
1,1 15 2.0 3.0 4.0 5.0
GEOMETRICAL STANDARD DEVIATION FOR SECOND MODE,
-------
to
CD
IP CONCENTRATION
I02mg/m3
(O
to
O
O
O
-------
1.4
J.3
T T
T T
i.;
N>
ro
oo
I.I
1.0
to
0.9
0.8
0.7
NGP-13
0.6.
J 1 1 L
J L
10
'2 14 16 18
SAMPLER D50 (
-------
3. The estimated IP concentration is not strongly affected by the slope
of the sampler collection curve, orgD. It varies by less than 7% as
the agD varies from 1.0 for a perfect sampler to 1.7, the worst allowed
by the IP performance curve.
4. More variation in the estimated IP concentration is shown as the cut
size of the IP sampler, DSQ, varies. However, within the specified
limits of the performance curve, 13 to 17 ym, there is only a differ-
ence of ±5% in the IP concentration.
5. The particle size distribution changes when a different assumed par-
ticle density for the impactor data is used. However, the density
change does not appear to affect greatly the estimated IP concentration
for these data. The IP concentrations shown in Figures III.2, III.3,
III.6, and III.7 all have a value of about 1500 mg/m3 in the vicinity
of the minimum WLS value.
B. Coal-Fired Boiler, Cold Side ESP Inlet (JBI-13)
The results of the modal fitting and the IP concentration calculation
are shown in Figures III.9 through III.12 for an assumed particle density of
1.0 g/cm3, and in Figures 111.13 through 111.16 for an assumed particle density
of 2.34 g/cm3. The coarse particle concentration is nearly ten times higher
than the fine particle concentration, in contrast to the data above where the
two concentrations are about equal.
The important findings to be noted are:
1. The impactor data can be fitted satisfactorily with a bimodal log-
normal distribution.
2. The estimated IP concentration is sensitive to the variation of
the geometrical standard deviation and the mass median diameter
229
-------
to
u>
o
bu
55
*
10
^ 20
o>
E
O
. 15
o.
O
o
I I0
5
0
1
JBI-13, />=I.O
: INLET DATA
DG,,xm SG M,
_ 1 st MODE 3.53 2.0 €
2nd MODE 44.0 2.5 <
(f -f )2
Wl C - V ..,,.?. £ - 7 97 f-T
W L.O - Z> - f. 76
3360
/
/
^s
0
f
/
J
/
I
I
^
1 '.
'I
/
\
\
10
1
[ \
f \ *1
I
1
\
^^ 1
.0 100.0
[ -1
—
—
L """
1000.1
p,
Figure III. 9 Modal fitting of impactor data from coal-fired boiler, cold
ESP inlet (JBI-13, p = 1.0).
side
-------
CO
c:
T -t»
O) -h
r+ n>
-i. O
O c+
3
O
ro -+>
3 w
fi) C
n- 3
(T> (D
CL
C_> Oi
oo n>
V o
t—> in
CO O
t> q
ua
n ro
,— I Q,
• =3
O Q.
Ol
•o
_J
tn
O
fD
O
O
m r
O
S
m
H
2
o
01
g
O
> ro
§b
o
m
•rib
m
°*
zb
o
S
is
IP CONCENTRATION
|>G -&> CD
O O O
^ mg/m3
o
o
C\3
O
T
o
o
-J
01
o
o
ro
OI
g
01
o
o
->i
01
WLS=2-
-------
n>
IP CONCENTRATION , I02mg/m3
ro
w
to
r* m
-j -*»
01 -«,
O
3
3
O)
r+
n>
DO a>
•-• -»
i o
i-1
OJ o
•o 3
n o
ro
HJ
* cu
0=3
*— • Q.
Ol
-i
a>
-5
Q
(£1
a
o
o
o
m
3
-------
to
U)
CO
in
i
2
10
\2
14 16
SAMPLER D50 (erg = 1.4),
18
Figure III.12 Effect of sampler 050 on the IP concentration estimate
(JBI-13, P « 1.0).
-------
NJ
CO
*».
60
55
ro
1*20
Q.
O
o»
o
15
10
JBI-13 , ^ = 2.34
INLET DATA
I st MODE
2nd MODE
DG, pm
2.994
42.29
SG
1.95
2.48
M, mg/nrr
1001
9281
WLS=2
(f,-fc)'
f.
= 7.718 (-3)
0.01
\
\
O.I
too
1
100.0
loooo
Dp, fj.m
Figure III. 13 Modal fitting of impactor data from coal-fired boiler, cold side
ESP inlet (JBI-13, p = 2.34).
-------
sez
in
c
-5
-5 -h
tu -h
rt fD
-*. O
O rt
3
O
fD -h
rt Oi
_.. w
3 o>
rti
ro ro
Q.
C_, O)
co ro
HI -5
I O
i—• tn
CO O
•o Q
tQ
H ro
rooj
'10
o>
•a
ro
-5
0
ft)
O
O
o
ro
3
I
en
IP CONCENTRATION , I02mg/m3
O ui
ro
o
01
Ol
O
OJ
Ol
m L.
o
m
0
>
?!»
5 o
o
o
m
<
|
Is
§
CO
m
o .&
i*
o
O Ol
So
•r
i
I
O
o
en
b
o
b
o
09
b
o
b
INJ
-------
ua
I
ro
u>
01
m
o>
o <-«•
o
rt> -h
(/>
rt Oi
_i. (/»
3 U)
Oi C
rt 3
fP
CO (T>
•-• 1
I O
H-« w
CO O
•o
ro
i
ro
o*
CL
o»
-a
to
Q
a
o
O
O
O
n>
IP CONCENTRATION , I02mg/m3
-------
to
12
14 16
SAMPLER D50 {
-------
2. (cont.)
of the coarse particle mode. A factor of three difference in the IP
concentration is observed as the Og2 and the MMD£ vary from 1.1 to 5.0
and from 10 to 100 ym, respectively. This is due to the much higher
concentration for the coarse particle mode compared to the fine par-
ticle mode. The results suggest that, for this aerosol, additional
information is needed for the coarse particle mode in order to accu-
rately determine the IP concentration.
3. The estimated IP concentration is affected by the slope of the sampler
collection curve. However, in the vicinity of the minimum WLS value,
the difference in the IP concentration for a sampler with ergo of 1.0
and 1.7 is not too great, the maximum being about 20%.
4. As the cut size of the IP sampler varies, a larger variation is found
in the estimated IP concentration. Within the specified limits of
the performance curve of 13 to 17 ym, the estimated IP concentration
changes by about ±15%.
5. Although the particle size distribution changes with different assumed
particle densities, the change does not appear to affect greatly the
estimated IP concentration. In the vicinity of the minimum WLS value,
the IP concentration in Figures III.10, III.11, III.14, and III.15
all have a value of about 2000 mg/m^.
C. Copper Smelter, Hot Side ESP Inlet (AJOI-3, AJOI-5)
Two sets of data have been selected for copper smelter emissions. AJOI-3
represents an average mass loading run, and AJOI-5 is the lowest mass loading
run. An assumed particle density of 3.58 g/cnr* is used for both sets of data.
The results of the modal fitting and IP concentration calculation are shown
238
-------
in Figures III.17 through III.20 for Run No. AJOI-3 and in Figures III.21
through III.24 for Run No. AJOI-5.
The important findings to be noted are:
1. The impactor data can be fitted reasonably well with a bimodal log-
normal distribution.
2. The estimated IP concentration is sensitive to the variation of the
geometrical standard deviation and the mass median diameter of the
coarse particle mode for the high mass loading run, AJOI-3, and less
sensitive for the low mass loading run, AJOI-5.
3. The estimated IP concentration is not strongly affected by the slope
of the sampler collection curve, agQ, the maximum concentration dif-
ference for a ago range of 1.0-1.7 being about 10%.
4. A larger variation in IP concentration is found as the cut size of
the IP sampler varies. However, within the specified limits of the
performance curve of 13 to 17 ym, the estimated IP concentration
changes by less than ±10%.
239
-------
to
IU
o
I ' I
INLET DATA
SG M,mg/m3
1ST MODE 0.978 2.67 438.9
2ND MODE 50.02 2.05 2417
WLS'Z-1-——»
4.212 X I0
0.01
D,
100.0
1000.0
Figure III.17 Modal fitting of impactor data from copper smelter, hot side
ESP inlet (AJOI-3).
-------
ro
12
10
o»
" 8
2
O
>•
o 4
2
O
O
A-
•T^-^^-2^
I
AJOI-3
«h CLOSED 'SYMBOLS'
J) OPTIMUM PARAMETERS
O l5/*m STEP FUNCTION
D IP CURVE (15/im cut,
-------
UD
C
to
£»
to
t-t- m
T -h
01 -h
r+ n>
-«. o
O
ro -h
3 W
O) C.
<-*• 3
(V 0)
o.
> o>
C-. fD
O -5
HH o
I (/>
CO O
o
ro
Ol
=i
o.
O)
•o
0>
-i
u?
a
o
ct
n>
o
o
IP CONCENTRATION , I02mg/m3
o
n>
-------
OJ
0.6
12 14 16
SAMPLER D50 (erg = 1.4),
18
20
Figure III.20 Effect of sampler DSQ on the IP concentration estimate (AJOI-3)
-------
10
6
10
E 4
CM
O>
E
0
o
O.
O
O>
— 2
x
2
S3.58
INLET DATA
DG, Mm SG M, mg/m3
I st MODE 0.38 1.28 248
2nd MODE 23.5
WLS'2-1
3.44
461
6.5 (-2)
10.0
100.0
D|
1000.0
Figure III.21 Modal fitting of impactor data from copper smelter, hot side
ESP inlet (AJOI-5).
-------
N)
*.
Ul
IO
o>
O
2 3
oc
h-
z
UJ
O
O
AJOI-5 , ^=3.58
® \CLOSED'SYMBOLS'
Jj OPTIMUM PARAMETERS
O 15 /tm STEP FUNCTION
Q IP CURVE (15/im cut,o-gD = l.7)
J I . J I
i
o.ia
0.10
CVJ~.
0.08 f
. «
0.06
0.04
U 1.5 2.0 3.0 4.0 5.0
GEOMETRICAL STANDARD DEVIATION FOR SECOND MODE, o-g
Figure III. 22 Effect of assumed aerosol
tration estimate (AJOI-5).
sampler aqp on the IP concen-
-------
to
c
to
IP CONCENTRATION , 10 mg/m'
— ro t*i 4*
O
O
3
O
-------
to
to
12
20
14 16 18
SAMPLER D50 (
-------
IV. SUMMARY AND RECOMMENDATIONS
A. Summary
1. Aerosols at the inlet as well as at the outlet of the control device
in a stack can be modeled satisfactorily by using multi-component
log-normal distributions.
2. The SRI impactor data analyzed in this study can be fitted with a
bimodal, log-normal distribution. The modal parameters of the first
mode, the fine particle mode, can be determined with good accuracy.
For the second mode, the coarse particle mode, the modal parameters
are obtained through partial fitting to two or three stages of im-
pactor data. These parameters are consequently less accurate and are
sensitive to the assumed mass median diameter and the geometrical
standard deviation. Sensitivity analysis of these parameters was
therefore performed to examine their effect on the estimated IP
concentration.
3. The present analysis technique gives a good estimate, ±20%, of the
IP concentration when the coarse particle concentration is in the
same order of magnitude as the fine particle concentration. This
situation is encountered in all of the outlet data and about half
of the inlet data analyzed here. When the coarse particle concen-
tration is much higher than the fine particle concentration, the
estimated IP concentration is sensitive to the variation of the
geometrical standard deviation and the mass median diameter. However,
by properly selecting the calculated IP concentration using minimum
objective function value as an indicator, it is possible to estimate
the IP concentration to within a factor of two or better.
248
-------
4. The IP concentration is affected more by the cut size, 050, than the
sharpness of cut, ago, of the sampler curve. The present specifica-
tions for the performance curve of the IP sampler, D$Q - 15 ± 2 um
and cfgp = 1.0 - 1.7, are found to be adequate because the resulting
IP concentration changes by less than ±20% within the tolerance of
the performance curve.
5. For a coal-fired boiler, two sets of impactor data have been analyzed
using two different assumed densities, namely, 1 and 2.41 gm/cm3.
Although the particle size distribution changes with different assumed
densities, the change does not appear to affect greatly the estimated
<
IP concentration for the limited data analyzed here.
B. Recommendations
1. While the present analysis technique gives a reasonable estimate of
the IP concentration, the accuracy can be greatly improved if addi-
tional size classification information is available in the size range
of 10-100 pm. This information is especially needed for sampling
situations in which the coarse particle concentration is significantly
higher than the fine particle concentration. It can be obtained by
adding one or two classification stages in the sampling device or
by the microscopic determination of the
-------
REFERENCES
Bevington, P. R. Data Reduction and Error Analysis for the Physical Sciences.
McGraw-Hill, New York, 1969.
Daniel, C. and F. S. Wood. Fitting Equations to Data. Wiley-Interscience,
New York, 1971.
Kapadia, A. Data Reduction Techniques for Aerosol Size Distribution Measuring
Instruments: Electrical Aerosol Analyzer and Diffusion Battery. Ph.D. Thesis,
Mechanical Engineering Department, University of Minnesota, Minneapolis, Minne-
sota, 1980.
McCain, J. D. CEA Variable-Throat Venturi Scrubber Evaluation. EPA-600/7-78-094,
U, S. Environmental Protection Agency, Research Triangle Park, North Carolina,
1978.
McCain, J. D., J. P. Gooch and W. B. Smith. Results of Field Measurements of
Industrial Particulate Sources and Electrostatic Precipitator Performance.
J. Air PoTTut. Control Assoc. 25(2):117-121. 1975.
Raabe, 0. G. A General Method for Fitting Size Distributions to Multi-Component
Aerosol Data using Weighted Least-Squares. Environ. Sci. Techno!. 12(10):
1162-1167, 1978.
Whitby, K. T. The Physical Characteristics of Sulfur Aerosols. Atmos. Environ.
12:135-159, 1978.
250
-------
APPENDIX A
Computer Program for Impactor Data Modeling
and IP Concentration Calculation
Program MODSIM is developed for impactor data fitting and IP concentration
calculation. The input parameters are the particle boundaries and concentra-
tions, the mask values used to determine whether the desired distribution is a
unimodal, bimodal, or trimodal distribution, and an initial estimate of param-
eters which characterize the distribution. The objective function to be mini-
mized can be chosen to be either normal chi-square, modified chi-square, or
weighted least square, as defined in Section IIA. The program uses a simplex
method to minimize the objective function which is calculated by comparing the
input particle concentration to the calculated particle concentration. In addi-
tion, the fitted distribution is integrated with the sampler classification
curve of an input cut-off size, DSQ, and geometrical standard deviation, agQ-
The output gives a,set of parameters for a log-normal distribution which has a
minimum objective function value, the histogram of the fitted distribution, and
the sampler concentration.
In the following pages, the sample input and output data, as well as the
program listing, are included.
251
-------
SAMPLE INPUTS
RUN MODSIM
ENTER DflTfl DESCRIPTION <68 CHflFLflCTERS MRXM)
-------
SAMPLE OUTPUTS
SAMPLE RUN
OBJECTIVE FUNCTION:WEIGHTED LEAST SQUARES
BEGINNING OBJECTIVE FUNCTION VALUE -
FINAL OBJECTIVE FUNCTION VALUE =
FUNCTION COMPUTATIONS = 145
5. 8S6E+01
2.221E-02
INITIAL ESTIMATES
MODE!
DG1=0. 0050
SGI =2. 00
TOTAL1=0. OOOE-Oi
MODE2
DG2= 2, 200
SG2=1. 70
TOTAL2=1. OOOE+02
MODE3
DG3= 40. 00
SG3=2.50
TOTAL3=7. 500E+01
FINAL VALUES OF THE PARAMETERS
MODE1 MODE2
DG1=0. 0050 DG2= 3. 275
SGI =2. 00 SG2=1.69
TOTAL 1=0. OOOE-01 TOTAL2=5. 514E+00
MODE3
DG3= 42.54
SG3=2. 21
TQTAL3-1. 011E+02
INPUT DISTRIBUTION
FITTED DIBTRIBUTIQIM
0.
0.
0.
1.
2.
6.
15.
30.
DP
ISO
350
750
200
200
QOO
000
OOO
DPG
0.
0.
0.
1.
3.
9.
21.
54.
251
512
949
625
633
487
213 '
772
2.
5.
9.
1.
3.
9.
2.
5.
DATA D(DATA)./D(LOGDP)
510E-01
120E-01
490E-01
625E+00
633E+00
487E+00
121E+01
477E+01
8.
1.
4.
6.
8.
2.
7.
1.
691E-01
547E+00
649E+00
173E+00
338E+00
384E+01
047E+01
048E-K>2
6.
1.
1.
1.
4.
9.
.2.
5.
FDATA D
010E-05
410E-02
422E-01
093E+00
258E-tOO
541E+OO
382E+01
354EH-01
< FDATA )/D( Li
2.
4.
6.
4
c?
2.
7.
i.
081E-04
261E-02
968E-O1
1 54E+GG
772E+OO
398E+01
912E+01
024E+05
100. 00
INHALABLE PARTICULATE MATTER
D50 = 15. 000 SG = 1. 400
TOTAL IPM MAdS= 1. 702E+01
253
-------
PROGRAM LISTING
C
C
C MODSIM. FTN IS A SIMPLEX MINIMIZATION PROGRAM FOR THE SRI
C IMPACTOR DATA
C
C
COMMON X(9), XMAX(9), XMIN(9), DELTAX(9), MASK(9),
1DELMIN(9), DATA(20), CDATH21), CDAT2(21),
2CDAT3(21), BDS(21), DPI (20), NEAR, TOTL,
3ALDAT(2O), ALCDAT(20), BCHISQ> FCHISQ, NF, CTOTL
DIMENSION TITLE(15), CALDAT(40), Xl(9)
DATA Y/'YES'/
CALL ASSIGN (3- 'SRI. INP')
CALL ASSIGM(6, 'LP: ')
C
C INPUT TITLE
C
220 WRITE (5, 1Q>
10 FORMAT (5X, 'ENTER DATA DESCRIPTION (60 CHARACTERS MAXM)',/
1,5X. '(NOTE: THE TITLE SHOULD CONTAIN AMONG OTHER THINGS THE',/
2,5X» 'WEIGHTING OF THE DISTRIBUTION, AMD THE UNITS OF THE',/
3, 5X, 'CONCENTRATION VARIABLE)-'
READ(5, 71)
-------
75 FORMAT(/5X, 'ENTER REST OF PARTICLE BOUNDARIES',/)
1READ<5, SO) (BDS(J), J=11,NBAR1>
80 FORMAT(15F10. 0)
WRITE (5,50)
50 FORMAT (/5X, 'ENTER FIRST TEN DATUM',/)
READ (5,6O> (DATA(J). J=l,10)
60 FORMAT (20F10. O)
WRITE (5, 5i>
51 FORMAT (/5X, 'ENTER REST OF DATUM',/)
READ (5,61) (DATA(J), J=11,NBAR)
61 FORMAT(20F10. 0)
WRITE (5,70)
C
C ENTER MASK
C
89 WRITE (5,90)
9O FORMAT
READ (5, 190) XI
190 FORMAT (9F10. 0)
GO TO 193
191 DO 192 J=l, 9
192 X1(J)=X(J)
193 DO 195 J =1,9
195 X(J) - XKJ)
196 DO 197 J*l,9
197 DELMIN(J)=DELMIN(J)*X
-------
c
c
C CALCULATE DATA/DLOGD, TOTL
C
C
TOTL=0. 0
DO 110 J=l,NEAR
A=ALOG10 (BDS(vJ-*-l)/BDS(J>>
ALDAT(J)=DATA(J)/A
TQTL=TOTL+DATA(J)
110 CONTINUE
C
C
C
GO TO 110O
170 CALL ESTIN
1100 CALL STEPIT(MFLAG)
C
C
C
C CALULATE CDATA/DLOGD,CTOTL
C
DO 1300 J=l,NBAR
AAA=ALOG10(BDS(J-H)/BDS(J) )
CALDAT(J)=ALC DAT < J>/AAA
1300 CONTINUE
C
C
245 CALL FITslLPLT=0. 03
WRITE(5, 200)
200 FORMAT',/, I5>
1, '0—STOP PROGRAMS /, 15X
2, '1—USE FINAL VALUES OF PARAMETERS AS INITIAL ESTIMATES'',/
3, 15X, '2—ENTER NEW INITIAL ESTIMATES ', /15X
3, '3—USE DIFFERENT OBJECTIVE FUNCTION', /, 5X)
READ (5,251) IFLAG
251 FORMAT(12)
GO TO (191,160,193) IFLAG
240 STOP
END
256
-------
c
C OBJCT. FTN IS THE OBJECTIVE FUNCTION USED TO FIT A
C TRIMODAL LOGIMORMAL CURVE TO THE INPUT DATA.
C
SUBROUTINE OBJCT(NV, CHISQ, MFLAG)
COMMON X(9), XMAX(9), XMIN(9), DELTAX(9>, MASKC9),
1DELMIN(9), DATA(20), CDATH21), CDAT2(21),
2CDAT3(21>, BDS(21), DP I (20), NEAR, TOTL,
3ALDAT(20>, ALCDAT(20), BCHISQ, FCHISQ, NF, CTOTL
GO TO (700,800,900) MFLAG
C IF MFLAG=1 THE NORMAL CHI SQUARED FUNCTION IS USED
C IF MFLAG=2 WHITBYS MODIFIED CHI SQUARED FUNCTION IS USED
C IF MFLAQ=3 THE WEIGHTED LEAST SQUARED FUNCTION IS USED
C
C
C
C
C WHITBYS MODIFIED CHI SQUARED
800 DO 10 J=l, NV
IF(MASK(J)) 10,20,10
20 IF (X(J)-XMAX(JM 30,30,40
30 IF (X(J)-XMIN(J)) 40,10,10
10 CONTINUE
CHISQ^O. 0
CALL ALOND(CDAT1, X<1),X(2),X(3>)
CALL ALOMD(CDAT2, X(4>,X(5),X(6))
CALL ALOND(CDAT3, X(7), X(S), X(9))
D WRITE(A, 100) CDAT1, CDAT2, CDAT3, X
D100 FORMAT(10(2X, 1PE10. 3),/)
DO 50 J=i,NBAR
A=ALOG10(BDSCJ-M)/BDS(J))
ALCDAT(J)=(CDAT1(J)+CDAT2(J)+CDAT3(J)>
A=ALCDAT(J)-DATA(J)
B= (ALCDAT (J) +DATA (J) ) /2. 0
CHISQ-=CHISQ+( (A**2)/B> /TOTL
50 CONTINUE
GO TO 60
40 CHISQ=1.QE20
60 NF=NF+i
GO TO 262
C
C
C WEIGHTED LEAST SQUARES FUNCTION
900 DO 110 J=l,NV
IF (MASMJ)) 110,120,110
120 IF (X(J)-XMAX(J)) 130,130,140
130 IF (X(J)-XMIN(J)) 140,110,110
110 CONTINUE
CHISQ=0. 0
CALL ALOND(CDAT1, X(U,X(2),X(3))
CALL ALOND(CDAT2, X(4).- X(5),X(6))
CALL ALOND(CDAT3, X(7), X(8),X(9))
CTOTL=0. 0
DO 170 J=1,NBAR
257
-------
170
150
140
160
C
C
C
700
220
230
210
970
250
240
26O
262
CTOTL=CTOTL-»-CDATi < J>+CDAT2< J>+CDAT3< J)
DO 150 J=1,NBAR
ALCDAT < J > = < CDAT1 < J > +CDAT2 ( J ) +CDAT3 < J > )
A*ALCDAT < J ) /CTOTL-DATA ( J ) /TOIL
CHISQ=CHISQ+/TOTL4-1. OE-20)
CONTINUE
CHISQ=CHISQ-K ( C TOIL -TQTL>**2>/ TOIL
GO TO 160
CHISQ=1. OE20
IMF=NF+1
GO TO 262
240,210, 210
NORMAL CHI SQUARED FUNCTION
DO 210 J=1,NV
IF(MASK(J» 210,220,210
IF(X
CALL ALOND
CALL ALQNDCCDAT3, X<7>
CTOTL=Q. 0
DO 270 J=i, NEAR
CTQTL=CTOTL+CDAT
DO 250 J=i, NBAR
X<2>, X<3»
X<5>, X(6»
X(8),X(9>>
J ) -(-CDAT2 < J ) +CDAT3 ( J )
A=ALCDAT < J ) /CTOTL-DATA < J ) /TOTL
CHISQ=CHISQ-KA**2)/(ALCDATCJ)/CTOTL+1.
CONTINUE
CHI SQ=CHI SQ+ ( ( CTOTL-TOTL ) #*2 ) /CTOTL
GO TO 260
CHISQ=1.0E20
RETURN
END
BLOCK DATA
COMMON X<9>, XMAX(9), XMIN(9), DELTAX<9>, MASKC9),
1DELMIN(9), DATA(20), CDATK21), CDAT2<21),
2CDAT3(2i), BDS(21), DPI(20), NBAR, TOTL,
3ALDAT<20), ALCDAT<20), BCHISQ, FCHISQ,NF
DATA DELTAX/0. 0, 0. 0, 0. 0, 0. 0, 0. 0, 0. 0, 0. 0, 0 . 0, 0. O/
DATA DELMIN/0. 03, 0. 03, 0. 03, 0. 03, 0. 03.- 0. 03, 0. 03, 0. 03, 0. O3/
DATA XMAX/1. 0, 5. 0, 100O. , 10. 0, 5. 0, 10000. , LOGO. 0, 5. 0, 100000. /
DATA XMIN/0, 001, 1. 1,0. 0, . 01, 1. 1,0. 0- . i, 1. 1, 0 O/
END
258
-------
BLOCK DATA
COMMON X<9>,
1DELMIN(9>,
2CDAT3<21>,
3ALDAT(20)i
XMAX(9), XMIN(9>, DELTAX(9), MASK(9>,
DATA(20>, CDATK21), CDAT2(21),
BDSC21), DP I (20), NBAR, TOTL,
ALCDAT(2Q), BCHISQ, FCHISQ, NF
DATA
DATA
DATA
DATA
END
DELTAX/0. 0, 0. 0, 0. 0, 0. 0, 0. 0, 0. 0, 0. Q, 0 . 0, 0. O/
DELMIN/0. 03, 0, 03, 0. 03, 0. 03, 0. 03, 0. 03, 0. 03/0. 03, 0. 03/
XMAX/1. 0, 5. 0, 1000. , 10. 0, 5. 0, 10000. , 1000. 0, 5. 0, 100000. /
XMIN/0. 001, 1. 1, 0. 0, . 01, L. i, 0. 0, . 1, 1. i, 0. O/
C
C
C
10
C
C
C
10
20
30
35
40
45
50
60
70
SUBROUTINE ALOND CALCULATES AREA UNDER LO6NORMAL DISTRIBUTION
SUBROUTINE ALOND , XMAX<9), XMIN<9), DELTAX<9>, MASM9),
1DELMIN(9), DATA(20>, CDATK21), CDAT2(21),
2CDAT3<21), BDS(21>, DPI (20),. NEAR, TQTL,
3ALDAT(20), ALCDAT(20), BCHISQ, FCHISQ, NF, CTOTL
DIMENSION CDAT (20)
XMEAN= ALOG (AMEAN)
XSIGMA = ALOG (SIGMA)
DO 10 J= 1, NEAR
A«(AL06 (BDS(J» .- XMEAN) / XSIGMA
B =( ALOG (BDS(J+i)> - XMEAN) / XSIGMA
CDAT (J) = APROX (A,B)*CONI
CONTINUE
RETURN
END
THIS IS AN APPROXIMATION FOR THE CUMULATIVE
NORMAL DISTRIBUTION FUNCTION. REF. ABRAMOWITZ
AND STEGUN: MATHEMATICAL HANDBOOK.
FUNCTION APROX ( A, B)
X1=ABS(A>
X2=ABS ( B )
IF(X1) 10,20, 1O
PX1=P(X1)
GO TO 30
PX1=0. 5
IF(X2) 35/40,35
PX2=P(X2)
GO TO 45
PX2=0. 5
IF (SIM) 60, 50, 5O
APROX=ABS(PX2-PX1)
GO TO 70
APROX=1. 0-(PX2+PX1)
RETURN
END
259
-------
C THIS FUNCTION IS USED WITH APROX
FUNCTION P(X)
DIMENSION B(6>
DATA B/0. 3193Q1530, -0. 3565*3782, 1. 781477937, -1. 82125597S,
11.330274429, 0. 2316419/
T»l. 0/(1.0+B(6)*X)
P=Z < X > •«• < B (1 ) *T+B < 2) *T**2+B (3) *T**3+B < 4) *T#*4 +B < 5) *T##5>
RETURN
END
C THIS FUNCTION IS USED WITH APPROX
FUNCTION Z(X)
PI=3. 1415927
S2PI=SQRT<2. 0*PI>
ARG=X**2/2. 0
IFCARG-87. 0) 10, 20.- 20
10 A=EXP(ARG)
2-1. 0/(S2PI*A>
RETURN
20 Z=0. 0
RETURN
END
C THIS SUBROUTINE CALCULATES THE FRACTION BETWEEN GIVEN BOUNDS
C
SUBROUTINE CNUMS
DO 10 J=l,40
A-(ALOG(BDl)-XANMD)/XSIGMA
B = 20,30,20
20 XIHP
-------
SUBROUTINE STEPIT(MFLAG)
C
C MINIMIZES A PIECEWISE CONTINUOUS FUNCTION SUBJECT TO ARBITRARY
C CONSTRAINTS OF INEQUALITY.
C J. A. NELDER AND R. MEAD, THE COMPUTER JOURNAL 7 (1965) 308
C
C
C
COMMON X(9), XMAX(9), XMIN<9)* DELTAX(9), MASK(9>,
1DELMIN(9), DATA(2G), CDATH21), CDAT2C2D*
2CDAT3<21)i BDS<21), DPK20), NBAR, TQTL,
3ALDAT(2Q), ALCDAT(20), BCHISQ, FCHISQ, NF,CTOTL
DIMENSION CHI(IO), 2(10, 9), ZBAR<9>, ZSTAR(9)
C
C
WRITE(5, 20)
20 FORMATdOX, 'CHOOSE DESIRED OBJECTIVE FUNCTION (ENTER 1,2, OR 3>
1, 15X, 'I—NORMAL CHI SQUARED FUNCTION', /
2, 15X, '2—WHITBYS MODIFIED CHI SQUARED FUNCTION',/
3,15X,'3—WEIGHTED LEAST SQUARED FUNCTION')
READ (5,25) MFLAG
25 FORMAT (13)
NFLAT=1
CHISQ=0. 0
NV=9
NFMAX=500
40 HUGE=0. 5E 38
ALPHA=1. 0
BETA=D. 5
GAMMA=2. 0
C
50 DO 130 J=l, NV
IF(MASK(J))130, 60, 130
60 IF< (X(J)-HDELTAX(J) >-X< J) )100, 70, 100
70 IF(X=0.10*X(J)
100 IF
-------
c
c
C CALCULATE INITIAL P=X(K)
Z(NVA, J > =Z(NVA, J)+DELTAX(J)
XS=X(J)
X(J)=Z 230,280
280'IF(NVA)290, 290, 310
290 DO 300 J=l, IMV
300 MASK540, 540, 1200
540 JH=1
IF(JL-1)560, 550, 560
550 JH=2
C CALCULATE PBAR.
C
560 DO 600 J"l, NV
IF(MASK
-------
c
C ATTEMPT A REFLECTION.
C FORM P#
C
DO 620 J=1,NV
IF(MASK(J))620, 610, 620
610 X(J) = U. 0+ALPHA)*ZBAR
620 ZSTAR=X(J)
CALL OBJCTCNV, CHISQ, MFLAG)
630 CHISTR=CHISQ
IF(CHISQ-CHI
CHKJH)=CHISQ
C
C ATTEMPT A CONTRACTION.
C FORM P*#
C
830 DO 850 J=1,NV
IF(MASK850, 840, 850
840 X (J) =BETA*Z (JH, J) •+• (1. 0-BETA > *ZBAR (J)
850 CONTINUE
CALL OBJCT 1100, 960, 960
263
-------
c
C THE CONTRACTION FAILED.
C REPLACE ALL P+P >990, 1000, 990
990 X NVP
ZMAX«AMAX1(ZMAX, Z-CHI>1190, 1200, 12OO
1190 JL=JH\
1200 DO 1210 J=1,NV
1210 X(J>=Z
-------
SUBROUTINE ESTIN
COMMON X(9), XMAX(9), XMIN(9), DELTAX(9), MASM9),
1DELMIN(9>, DATA(20), CDATK21), CDAT2(21>,
2CDAT3<2i)i BDS(21>, DPI<20), NBAR, TOTL,
3ALDAT(20), ALCDATC20), BCHISQ, FCHISQ, NF
A=1.0
RETURN
END
C
C FINLPLT. FTN IS FOR QUTPUTING THE DATA
C
SUBROUTINE FINLPLT (XI, MFLAG, TITLE, CALDAT)
COMMON X(9>, XMAX(9), XMIN<9), DELTAXC9), MASK(9),
1DELMIN(9), DATA(20), CDATK21), CDAT2(21),
2CDAT3(21), BDS(21), DP I(20), NBAR, TOTL,
3ALDAT(20), ALCDAT(2Q), BCHISQ, FCHISQ, NF, CTQTL
DIMENSION TITLEC15), ABDS<21),XI(9), XIHP(40),CNUM(40),CALDAT(40)
WRITE<6, 10) TITLE
10 FORMAT (IHli 4Xi 15A4i //)
GO TO (201,202,203) MFLAG
201 WRITEC6, 205)
205 FORMAT(5X, 'OBJECTIVE FUNCTION:NORMAL CHI SQUARED',/)
GO TO 38
202 WRITE(6, 206)
206 FORMAT(5X, 'OBJECTIVE FUNCTION: W.HI TBYS MODIFIED CHI SQUARED',/)
GO TO 38
203 WRITE(6, 207)
207 FORMAT (5X, 'OBJECTIVE FUNCTION:WEIGHTED LEAST SQUARES',//
38 NBAR1=NBAR+1
WRITE (6,40) BCHISQ, FCHISQ, NF
40 FORMAT(5X, 'BEGINNING OBJECTIVE FUNCTION VALUE = ',5X, 1PE10.3, /
1,5X, 'FINAL OBJECTIVE FUNCTION VALUE * ', 5X, 1PE10. 3, /
2, 5X, 'FUNCTION COMPUTATIONS =',14,//)
WRITE(6, 45)
45 FORMAT(/5X, 'INITIAL ESTIMATES',/, 8X, 'MODE!', 14X, 'MODE2', 15X
1,'MODE3')
WRITE(6, 50)X1(1),X1(4),X1(7),X1(2),X1(5),X1(8),X1(3),X1(6),
1X1(9)
50 FORMAT(5X, 'DG1=', F6. 4, 10X, 'DG2=', F6. 3, 10X, 'DG3=', F6. 2, /
13X, 'SG1=', F4. 2, 12X, 'SG2=', F4. 2, 12X, 'SG3=', F4. 2, /
35X, 'TOTALS', 1PE9. 3, 4X, 'TOTAL2=', 1PE9. 3, 4X, 'TOTAL.3=', 1PE9. 3, /)
WRITE(6, 55)
55 FORMAT(/5X, 'FINAL VALUES OF THE PARAMETERS',/, SX, 'MQDE1',14X
1. 'MODE2', 15X, 'MODE3')
WRITE (6, 50) X( 1), X<4)> X(7), X<2), X(5)> X(8), X(3), X(6), X<9>
WRITE(6, 70)
265
-------
70
, 1 IX, 'FITTED DISTRIBUTION
'D(DATA) /D(LOGDP ) '
9'0
100
SO
106
107
109
112
110
103
118
D
D120
D
D140
150
1:30
160
188
190
X(8), X(9) )
+CDAT3 ( 1 )
.
JIUM
FORMAT(//, 27X, 'INPUT DISTRIBUTION'
1, /,8X, 'DP', 5X, 'DPG',7X, 'DATA',3X,
2, 5X, 'FDATA',3X, 'D ')
DO 80 J=1,NBAR
ABDS(J)=SQRT(BDS(J+1)*BDS(J))
WRITE(6, 90) BDS(J)
FORMAT (5X, F6. 3)
WRITE(6, 100)ABDS(J),DATA(J), ALDAT ALCDAT(J),CALDAT(J)
FORMAT( 13X, F6. 3, 4X, 1PE9. 3, 3X, 1PE9. 3, 7X, 1PE9. 3, 4X, 1PE9. 3)
CONTINUE
WRITE(6, 90) BDS(NBARl)
BDS1=BDS(1)
BDS2=BDS(2)
BDS(1)=. 001
BDS(2)=15. 0
CALL ALOND(CDAT1,X(1), X(2),X(3))
CALL ALQND(CDAT2, X(4), X(5),X(6>>
CALL ALOND(CDAT3, X(7)
CDAT1(1) =CDAT1(1)+CDAT2(1
BDS(1)=BDS1
BDS(2)=BDS2
JJ=0
WRITE(5, 107)
1,/,5X, 'ENTER D50 , SG',/,5X, 'OR HIT RETURN TO SKIP')
READ (5,109) AIHPMN,AIHPSG
FORMAT (2F10. 0)
IF (AIHPMN) 190,190,112
IF (AIHPSG) 190, 190, 110
IF (JJ . EQ. 1) GO TO 118
WRITE(6, 103)
FORMAT*//, 5X, 'INHALABLE PARTICULATE MATTER")
TIHP=0. 0
CALL AIHP)
DO 130 K=l, 7, 3
CALL CNUMSiCNUM, X.
-------
N)
cn
vo
RELIEVE TO
ROOT DM OF THD.
r i n
•r_4rd
.050
I £H_I y/'TURN TO ROOT DM
TWJ_OF THREAD
JLJx\3
• LMO
/ n
I.r
5HELL CAP .
.100
.200
ffl
US 5 DIA
HOLE TO BE
DRILLED DURING
FINAL ASSEMBLY;
(£ COINCIDENT WITH
I.S45* OIA HOLE IN
ITEM C0|.
£01 I SHEU CAP ]rS*"OM».55,VtaNP . ALUMINUM
SHELL . 3it"Pid » 5.3" lOfrC7 .tLU'-l>\. "
p>ALL FILLETS 4 ROUNDS &" R
UNLESS OTHERIMSE NOTED
„. p>Q-RING GROOVE MUST MEET
5:2 TOLERANCES -SPEC'D. TOOLING
.MARKS MOST LAY CONCENTRIC
TO GROOVE.
SHEI L CYQONF H
TOLE«*NCES UNLESS
SOUTHERN RESEARCH INSTITUTE
»I»MINOHAM, AtAtAMA 35205
SRI CYCLONE JZ
CYCLONE JJJ SHELL
-------
to
«j
o
TURN JO ROOT DIA.
OF THREAD
i_
DIA BORE *£ DEEP
fNPSC STRAIGHT
BOTTOW TAP
7QI I I IMOPIFIED CAP, Ijj-'DIA ». y.^-o"
p>ALL FILLETS 4 ROUNDS T^ R UNLESS OJHERWlf.f.
NOTED
THRU
MODIFIED CAP.CYCLONE: in
K BASSON
SOUTHERN RESEARCH INSTITUTE
tlUMINOHAM, AlABAMA 35205
TITLE
SRI CYCLONE T£
CYCLONE ffl MODIFIED CAP
a.-;
-3f -C-Q7
-------
101
103
305-SIS
COLLECTION cup. 1.5*0 t>tA>a.<75<-ouc,
.8WPM "i.its.LONG ^ ALUMI\ if
ALL FILLETS * ROUNDS rk R UNLESS
OTHERWISE NOTTD
ro
•TURN TO ROOT
DIA OF THREAD
TURN TO ROOT
DIA OF THREAD
ACROSS
FLATS
FOR SRX.
USt ONl-Y
R MAX,TVP A PLS
CY^LONF E NOZZLES
ffl COLLECTION CUP
ITEM
90S
9O6
307
908
909
910
911
912.
913
914
916
ID
.208"
!4-.5/>\m
TOLiRANCES UNLESS
OTHERWISE NOTED
K BASSOIV
0-19.75 topep XTtuva 9oS-9i5
SOUTHERN RESEARCH INSTITUTE
•WMINOHAM, ALAtAMA 35305
SRI CYCLONE E
CO.LECTION CUP 4 NOZZLES
FULL
H-17-7?
1181-31 -C-0?
-------
ar
or
to
-j
to
C
3.1 Z 5 OD.
1125 OD.
O.856 DIA.
THRU
2-16 UN
2.419 ID.
RELIEVE TO ROOT
DIA. OF THREAD
2.540 DIA-
.-ACROSS
HEX FLATS
'/2 DIA. BORE
/I6 DEEP
COLLECTION CUP CjooT)
1002
:OLLECTION CUP.3.I25'QD. « I.765*LON&
VORTEX TUBE. 3.1 25 "QD. x4.094" LONG
0-RING GROOVE: 1382 O.D.*
1.135 ID. x 0.079 DEEP
±S«.
MACHINE OUT l%2 1.0.x
2% O.D.X7/I6 DEEP
NOTES:-
0.100
MATERIALS: 316 STAINLESS STEE.L,
6AL-4V TITANIUM ALLOY^ CR ".'/,'-
ALUMINUM; TO BE SPECIHLD SY
WORK ORDER.
ALL FILLETS AND ROUNOo '/I5 R.
UNLESS NOTED.
VORTEX TUBE (1002)
TOLEHANCei UNLESS
OTHE ROUSE MOTED
SOUTHERN RESEARCH INSTITUTE
URMINGHAM. AIA8AMA 3520S
SRI CYCLONE Z:MARK 2
COLLECTION CUP & VORTEX TUBE
°*re8-l7-79
4I8I-37-C-IO
-------
N)
1.148 DIA.-
60 CHAMFER
BEGIN EVEtj
V,"TH CREST
OK THREAD
RELIEVE TO ROOT
DIA.OF THREAD
Q248
O-RING GROOVE: - _.
2.757 *g* O.D. * 2.580^"*
LD. xa045 DEEP
. HIM mi.
J'lO'l ' I
4'o.g
SMOOTH TRANSITION
OT CRITICAL
-%-)4 STRAIGHT PIPE THREADS
10.848 |«— (NPSC INTERNAL THREADS)*0-35'DP.
'. '22 DiA.TAN:
1.-FN- TO BORE
I.8OO ACROSS
HEX-FLATS
L326
2.226
0-350 TQ25O
/
/
-JL— -.
!
4 12:
'^
asoo 1
t
DETAIL *B'
RELIEVE TO ROOT
DIA.OF THREAD
MATERIALS: 316 STAINLESS STEEL,
P6AL-4V TITANIUM ALLOY, OR 7075
ALUMINUM; TO BE SPECIFIED BY
WORK ORDER
[^ ALL ROUNDS AND FILLETS '/IS R,
V^ UNLESS NOTED.
I** tool
SOUTHEIM RESEARCH INSTITUTE
gUMINGHAM. AIA»AMA 35205
SRI CYCLONE X". MARK 2
BODY (CYCLONE
•""FULL
°*™ 8-20-^9
4I8I-37-C-I I
-------
BST. | ITCH
201
KKUfTIOI
NOZZLE^DIA.VARIES* 2.32'LONG
/3/4-l4 STRAIGHT PIPE THREADS TO
/ FIT NPSC INTERNAL THREADS
<-n
to
TURN TO ROOT
DIA,OF THREAD
TABULATION
*D*
01 A.)
mm
3.0
3.5
4.0
4.5
5.0
6.O
7.0
8.5
10.0
12.0
14.5
*A"°
DEGREES
8°I6'
8°02'
7°46'
7°30'
7° 14'
6°42'
6° 09'
5° 21'
5Q09'
3° 55'
2° 23'
->B»0
DEGREES
I8°28'
I8°05'
I7°42'
17° 19'
16° 55'
I6°08'
I5°ZI'
I4909'
I2°56'
ii°ie'
9° 14'
*L"
INCHES
0.25
0.25
0.25
0.25
0.25
0.25
0.25
0.25
0.50
0.50
0.50
MATERIAL TO BE SPECIFIED
BY WORK ORDER.
SMARP EDGE
ZONE NO.
SOUTHERN RESEARCH INSTITUTE
BIRMINGHAM, ALABAMA 35205
SRI CYCLONE X: MARK 2
•NOZZLES
SCALE 2X
OATEQ_2|_79
4I8I-37-B-I2
-------
to
^J
(Jl
.7lft8*/i«.K.
_ .£.90 R.
TRAWS\TiOW
0.295 OIA, TA»l6mT TO
0.20O DP O. «Z D/A C'BOEE,
0.01.6 DEEP
SCALE)
•'•91-
t
ALk. FIL1-E.TS 4.ROOWDS 'A(e"R.
OTHLRW1&C. NOTEO.
BEiPECIHED
THD.
DE.TAIL*B"
(DOURLE SCALE)
TOLERANCES UNLESS
OTH£ RWISENOTED
if*
ICH9-2 C.LIPP£D Cgg 51 cf THDS
SOUTHERN RiSEARCH INSTITUTE
5IRMINOHAM, ALABAMA 3S20S
TITLE ^^
CYCLOME: HL
FULL,
_ 17-79
^•181-37 -C-I4-
-------
APPENDIX D
\
DESIGN DRAWINGS OF THE INHALABLE PARTICULATE
PRECUTTER CYCLONE
Figure D-l. Cyclone IX Collection Cup.
Figure D-2. Cyclone IX Vortex Tube.
Figure D-3. Cyclone IX Body.
Figure D-4. Cyclone IX Nozzles.
276
-------
nun amr me
6011 I ICO'-lrXTION CUP 2.50(70.0. » t.635'LONG
MATERIALS'- 316 STAINLESS STEEL,
6AL-4V TITANIUM ALLOY, OR 7075
ALUMINUM ; TO BE SPECIFIED BY
WORK ORDER.
ALL FILLETS AND
UNLESS NOTED.
ROUNDS '/I6" R.
2.50O O.D.
0.100-
RELIEVE TO ROOT
/'DIA. OF THREAD
•2 - 16 UN
Ki
^1
-J
2.016 DIA.
v'/2 DIA. BORE,
/I6DEEP
FRACTIONS
W-IMALS
ANGLES
TINtSH
SOUTKE8N KSIAKH INSTITUTE
1IRMIHGHAM. AlABAMA 35J03
SRI C'y CLONE IX: MARK 2
AWRJVED
COLLLCTICN CUP
SCALF 2 / ! DWG NO
Te-ir "'-, 1 4I8I-37-C-06
-------
ro
^j
oo
NOTES:
f. MATERIALS:316 STAINLESS STEEL.
K> 6AL-4V TITANIUM ALLOY , OR 7O75
f ALUMINUM ; TO BE SPECIFIED BY
1 WORK ORDER.
ALL FILLETS AND ROUNDS '/I6 R.
UNLESS NOTED.
2.006
DIA.
-.1
. 1
0.0
!
|.827
3IA.
J
•row
•- 0.787"°--
\
^
ooes-*
037
501 1
/YY
J . f
\
5 *-
«-
Z22
«
m
~ fc-*
L, •> I/a
^-5/8 -.
r/8 ACROSS
HEX FLATS
f
~1L_
2%"I6
MACHINE OUT 2'O.D.x
"l'/32*I.D. * Vl6" DEEP
JN
THRU %,
i500 0.0.
•y-
SOUTHEM KSIAKH MSTITUn
URMINOHAM, AiAIAMA 33105
SRI CYCLONE IX: MARK 2
VORTE.X TUBE
4I8I-37-C-C7
-------
to
-J
vo
fcYCLONE BODY. 3
ftOD.*.3 530 LONG.
60 CHAMFER, BEGIN EVEN
WITH CREST OF THREAD
O.602 DIA. TAN-
GENT TO BORE
I li AC ROSS
HEX FLATS
v x s\:
i i /
i I ' /
SMOOTH TRANS"N
CRITICAL
Z.SOO DIA.
RELIEVE TO ROOT DIA.OF THO.
-16 UN
NOTES:
MATERIALS: 3ie STAINLESS STEEL,
1>6AL-4V TITANIUM ALLOY, OR TOTS
ALUMINUM; TO BE SPECIFIED BY
WORK ORDER.
OflING GROOVE
3.260*^0.0 »
2.090*8- ID.'x
C.O45 DEEP
|>>ALL FILLETS AND ROUNDS '/I6 R
r UN LESS NOTED.
SOUTHERN RESEARCH INSTITUTE
BIRMINGHAM, ALAtAMA 34203
SRI CYCLONE 3£: MARK 2
BODY (CYCLONE EE)
4181 57-C-Q8
-------
00
o
SHARP EDGE
'/2-I4 NPS
1 1
1
1
t
I
\
\
\ '
1
1
1
1
1
I I
SHARP EDGE
3.25 MAX.
TABULATION
(D?A)
mm
3
4
5
-------
APPENDIX E
METRIC SYSTEM CONVERSION FACTORS
Non-metric Multiplied by Yields Metric
acfm 28.317 liters/min
°F 5/9 (°F-32) °C
in. 2.54 cm
gr/acf 0.0023 g/liter
ft 30.48 cm
gr/acf 2.29 g/m3
281
-------
IERL-RTP-1283
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
1. REPORT NO.
EPA-600/7-82-036
2.
3. RECIPIENT'S ACCESSION-NO.
4. TITLE ANOSU8"ITLE
Sampling and Data Handling Methods for Inhalable
Particulate Sampling
S. REPORT DATE
May 1982
6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
W.B.Smith,K.M.Gushing,J.W.Johnson, C.T. Par-
sons, A.D.Williamson, and R.R.Wilson, Jr.
8. PERFORMING ORGANIZATION REPORT NO,
SoRI-EAS-81-245R
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Southern Research Institute
2000 Ninth Avenue, South
Birmingham, Alabama 35225
10. PROGRAM ELEMENT NO.
11. CONTRACT/GRANT NO.
68-02-3118
12. SPONSORING AGENCY NAME AND ADDRESS
EPA, Office of Research and Development
Industrial Environmental Research Laboratory
Research Triangle Park, NC 27711
ERIOD COVERED
14. SPONSORING AGENCY CODE
EPA/600/13
15. SUPPLEMENTARY NOTES T£RL-RTP project officer is D. Bruce Harris, Mail Drop 62,
919/541-7807.
16. ABSTRACT
The report reviews the objectives of a research program on sampling and
measuring particles in the inhalable particulate (IP) size range in emissions from
stationary sources, and describes methods and equipment required. A computer
technique was developed to analyze data on particle-size distributions of samples
taken with cascade impactors from industrial process streams. Research in sam-
pling systems for IP matter included concepts for maintaining isokinetic sampling
conditions, necessary for representative sampling of the larger particles, while
flowrates in the particle-sizing device were constant. Laboratory studies were con-
ducted to develop suitable IP sampling systems with overall cut diameters of 15 mi-
crometers and conforming to a specified collection efficiency curve. Collection
efficiencies were similarly measured for a horizontal elutriator. Design parameters
were calculated for horizontal elutriators to be used with impactors, the EPA SASS
train, and the EPA FAS train. Two cyclone systems were designed and evaluated.
Tests on an Andersen Size Selective Inlet, a 15-micrometer precollector for high-
volume samplers, showed its performance to be with the proposed limits for IP
samplers. A stack sampling system was designed in which the aerosol is diluted in
flow patterns and with mixing times simulating those in stack plumes.
17.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.lDENTIFIERS/OPEN ENDED TERMS
c. COSATI Field/Group
Pollution
Flue Gases
Sampling
Dust
Aerosols
Data Processing
Measurement
Cyclone Separators
Pollution Control
Stationary Sources
Particulate
Inhalable Particulate
13 B
2 IB
14B
11G
07D
09B
07A,13I
13. DISTRIBUTION STATEMENT
Release to Public
19. SECURITY CLASS (This Report)
Unclassified
21. NO. OF PAGES
297
20. SECURITY CLASS (Thispage)
Unclassified
22. PRICE
EPA Form 2220-1 (9-73)
,283
------- |