MIDWEST RESEARCH INSTITUTE
INHALABLE PARTICULATE EMISSION FACTOR ASSESSMENT AND DEVELOPMENT
FINAL REPORT
EPA Contract No. 68-02-2609, Assignment No. 11
MRI Project Mo. 4369-1(11)
Date Prepared: May 22, 1979
Prepared for
Industrial Environmental Research Laboratory
Environmental Protection Agency
Research Triangle Park,
North Carolina 27711
Attn: Frank Noonan
MIDWEST RESEARCH INSTITUTE 425 VOLKER BOULEVARD, KANSAS CITY, MISSOURI 64110 816753-7600
-------
MRI-NORTH STAR LABORATORIES 10701 Red Circle Drive, Minnetonka, Minnesota 55343 612933-7880
MRI WASHINGTON, D.C. 20006-Suite 250,1750 K Street, N.W. 202 293-3800
-------
INHALABLE PARTICULATE EMISSION FACTOR ASSESSMENT AND DEVELOPMENT
FINAL REPORT
by
J. Patrick Reider and Dr.. Chatten Cowherd
Midwest Research Institute
425 Volker Boulevard
Kansas City, Missouri 64110
EPA Contract No. 68-02-2609, Assignment No. 11
MRI Project No. 4369-L(ll)
Date Prepared: May 22, 1979
Prepared for
Industrial Environmental Research Laboratory
Environmental Protection Agency
Research Triangle Park,
North Carolina 27711
Attn: Frank Noonan
MIDWEST RESEARCH INSTITUTE 425 VOLKER BOULEVARD, KANSAS CITY, MISSOURI 64110 816753-7600
-------
PREFACE
This report was prepared for the Environmental Protection Agency's In-
dustrial Environmental Research Laboratory under EPA Contract No. 68-02-2609,
Work Assignment No. 11 Mr. Frank Noonan, Office of Air Quality Planning
and Standards, was the requestor of this work. The report was prepared by
Mr. J. Patrick Reider, Task Leader, and Dr. Chatten Cowherd, Task Manager.
Approved for:
MIDWEST RESEARCH INSTITUTE
M. P\yj Schrag, Deputy Director
Environmental and Materials
Sciences Division
May 22, 1979
iii
-------
CONTENTS
Preface* ................. «.. iii
Figures* ................... ...... vi
Tables . vii
1*0 Introduction . . ................ 1
2.0 Data Requirements* *......... .... 2
3*0 Data Acquisition *.*«................... 5
3.1 Source test data (SOTDAT) system ..*. 11
3.2 Regional air pollution study (RAPS) data system. 11
3..3 Fine particle emissions information system (FPEIS) 11
3*4 NSPS support documents and published data. ..... 11
4.0 Data Analysis and Results. .................. 13
4.1 Determination of IP mass fractions 13
4.2 Determination of IP control efficiencies ...... 18
4.3 Determination of controlled IP emission factors* * . 18
5.0 Key Issues 28
5*1 Issue 1........ 28
5.2 Issue 2. 28
5.3 Issue 3 29
5.4 Issue 4 31
5.5 Issue 5 31
5.6 Issue 6 31
6*0 Conclusions and Recommendations. ............... 34
References ...».........««. 36
Appendices
A. IP mass determination procedure ................ 38
B. IP calculator program 41
C. Development of IP emission estimates for gray iron foundries. . 46
D. Particle size measurement methods 54
-------
FIGURES
Number
1 Procedure for calculating inhalable particulate emission
rate 3
2 Goal-fired power plant flow chart ...» . 6
3 Gray iron foundry - cupola flow chart 7
4 Ferroalloy production plant flow chart. ........... 8
5 Lead battery manufacturing plant flow chart 9
6 Graphical procedure for determining inhalable particles . 14
7 Procedure for calculating inhalable particulate control
efficiencies. ....................... 21
8 Penetration curves for cyclones and scrubbers ........ 26
9 Penetration curves for baghouses and ESPs 27
10 Fitting a measured distribution with two log-normal
distributions 32
vi
-------
TABLES
Number Page
1 IF Data Sources and Number of Tests Compiled. ........ 10
2 Inhalable Particulate Cascade Impactor Data for Four
Industries. ........................ 16
3 Summary of SASS Train Particle Size Distribution Data .... 17
4 Summary of Condensible Particulate Emission Rates .. 19
5 High-Volume area sampling results .............. 19
6 Summary of Inhalable Particulate Data for Fugitive Emissions. 20
7 Summary of Control Device Inhalable Particulate Collection
Efficiencies 22
8 Summary of Inhalable Particulate Mass Fractions for Four
Industries. 24
9 Inhalable Particulate Emission Factors for Four Industries. . 25
10 Comparison of Different Particulate Collection and Sizing
Methods .*... ....... 30
A-l Constants Used in the Log-Normal Data Analysis. ....... 39
C-l Size Distribution for Three Electric Arc Installations. ... 49
C-2 Particle Size Distributions of Green Sand Emissions for
4-In. Cube Pattern. ..............«.. 49
vii
-------
SECTION 1.0
INTRODUCTION
The Environmental Protection Agency (EPA) is considering the establish-
ment of a National Ambient Air Quality Standard (NAAQS) for Inhalable Par-
ticulate (IP). The Health Effects Research Laboratory recommends that a
standard for inhalable particulate should include particles smaller than
15 /im in aerodynamic equivalent diameter. This recommendation is based on
the worst case situation (mouth breathers) and represents that fraction of
particulate matter which can deposit in the conducting air-ways, and the gas
exchange areas of the human respiratory system; particles larger than 15 pirn
are restricted to the nasopharyngeal region*::'
Establishment of an IP standard will necessitate that states develop and
submit revisions to their State Implementation Plans (SIPs) to demonstrate
attainment and maintenance of the standard. These revisions will entail the
development and utilization of emission inventories, air quality simulation
models, and air quality baseline data for purposes of control strategy de-
velopment. IP emission inventories, in turn, presuppose the availability of
emission factors to quantitatively estimate emissions from significant IP
sources.
The overall objective of the work reported herein was to determine the
data required and the suitability of existing data for development of SIP
revisions in response to the establishment of a NAAQS for IP. In particular,
the anticipated needs of the states in carrying out the emission inventory
efforts required for SIP preparation are of special concern.
In order to assess the status of the current IP data base, four example
industries were studied: ferroalloy production facilities, coal burning
power plants, lead battery manufacturing plants, and grey iron foundries.
Available particulate mass emissions data and particle size data pertaining
to significant sources within the four industries were collected and assessed
as to suitability for development of IP emission inventories. The goal of
this study was to demonstrate the feasibility of using existing IP emission
factor data and additional data under development in a presentation format
that would meet the needs of the states in preparing SIP revisions.
-------
SECTION 2.0
DATA REQUIREMENTS
The Glean Air Act requires that a SIP revision be prepared and sub-
mitted for approval within 9 months after the promulgation of a NAAQS. The
revision must contain a control strategy that provides for the attainment
and maintenance of the NAAQS. Emission limitations and other measures to
assure attainment and maintenance must be adopted as rules and regulations
enforceable by the state agency. In addition to the regulations for control
of existing sources, the SIP revision must set forth legally enforceable
procedures for review of new sources and modification of existing sources to
determine control requirements.
The formulation of an IP control strategy would begin with the develop-
ment of a baseline emission inventory of existing IP sources, both ducted
and fugitive, within a study area. Projected IP emission inventories would
also be required for 5-year forecast periods, taking into account source
growth, more stringent controls, etc. The baseline inventory would be input
into appropriate air quality simulation models which would be calibrated
against observed ambient IP concentrations. The calibrated models would then
be used in conjunction with baseline and projected emission inventories to
determine the source-specific emission reductions needed to attain and main-
tain the IP standard.
Development of emission inventories requires the calculation of IP emis-
sion rates for various sources. Two alternative procedures may be used to
calculate a controlled IP emission rate, as depicted in Figure 1. In the first
approach (the upper path in Figure 1), the mass fraction of particles smaller
than 15 /jm in the uncontrolled emission on stream is multiplied by the uncon-
trolled total suspended particulate (TSP) emission factor for the specific
source yielding an uncontrolled IP emission factor. Next, the controlled IP
emission factor is calculated by multiplying the uncontrolled IP emission fac-
tor by the IP collection efficiency for the specific control device applied
to that source. Finally, the controlled IP emission factor is multiplied by
the source extent (process rate) data to give the controlled IP emission rate.
Determination of the uncontrolled fraction of particles less than 15 Am in
this procedure will be discussed later in this report.
-------
Uncontrolled
Mass Fraction
Less Than 15/i.m
Uncontrolled TSP
Emission Factor
(AP-42 or Other
Source )
Uncontrolled
Particle (Mass)
Size Distribution.
Penetration Curve
for Control Device
(Source Independent)
Control Device
I.P. Efficiency
(Source Dependent)
| | Input Data Which Must Be Fully Developed
Figure ! Procedure for calculating inhalable particulate emission rate.
-------
The second approach for calculating a controlled IP emission rate (the
lower path outlined in Figure 1) requires the particle size distribution in
the uncontrolled emission stream, the uncontrolled TSP emission factor, the
penetration curve for the specific control device, and the source extent
(process rate). This procedure requires a much more sophisticated calculation
program to determine the uncontrolled emission factor, by combining the un-
controlled emission factor and the corresponding particle size distribution
with the control device penetration curve. In both of the above procedures,
particle size distribution information constitutes the most important data re-
quired for developing an inhalable particulate emission rate. Particle size
distribution data are obtained using cascade impactors, cyclones, and di-
chotomous samplers in conjunction with various sampling trains. Each of these
measurement techniques have limitations which need to be considered when using
particle size data. A large portion of the inhalable particulate emissions
data presented in this report were measured using cascade impactors. Further
comments on the limitations of the various particle size measuring techniques
are presented in Appendix D.
-------
SECTION 3.0
DATA ACQUISITION
As stated earlier, the four industries identified as test cases for IP
emission factor development were coal-fired power plants, gray iron foundries,
ferroalloy production plants, and lead-acid battery manufacturing plants.
These four industries represent a range of available source testing informa-
tion for IP analysis. General process descriptions for each industry can be
found in Refs. 4 and 5.
Significant emission points (particulate emission rates greater than 100
tons/year) were identified for each industry, based on Ref. 6. Figures 2 to
5 present typical flow diagrams for each industry with all emission points for
which data have been compiled also indicated in each figure.
The computerized data base systems searched for mass emission data and
particle size distribution data were the Source Test Data System (SOTDAT),
Regional Air Pollution Study (RAPS), and Fine Particle Emission Information
System (FPEIS). NSPS support documents and published data were used as sup-
plemental data acquisition sources.
Data procurement activities were conducted during this study to provide
particulate emissions data (controlled and uncontrolled), particle mass size
distributions, and condensible particulate emissions data for the specific
industrial sources. Particle-size distribution data found in the existing
data systems were considerable for coal-fired power plants and gray-iron
foundries, but rather limited for ferroalloy plants and nearly nonexistent
for lead battery manufacturing plants. The data systems were searched for
pertinent data detailing the source operating conditions, sampling param-
eters and test results. Data pertaining to specific control devices applied
to sources other than power plants stacks were meager in most cases.
Acquisition of additional particle-size distribution data determined
with the SASS train was an integral phase of the program, and efforts were
focused on gathering Environmental Assessment Level I test data for the
specific industries under investigation. Also included in this phase was
the procurement of condensible particulate and fugitive emissions data
pertinent to the subject industrial sources. The IP data are summarized in
Table 1 which presents type of IP data found and data sources included for
each industry.
-------
To Atmosphere
Coal
Preparation
* Major Emission Point (> 100 Tons/ Yr)
.; Test Data Available
Figure 2. Coal-fired power plant flow chart,
-------
To Atmosphere
Flux Receiving
(Rail. Truck)
Coke Receiving
(Rail)
Scrap Iron
Receiving
(Rail)
Sand Receiving
(Rail, Truck,
Bagged )
Pneumatic or
Mechanical
Mechanical
Mechanical
1
Pneumatic or
Mechanical
Flux Storage
(Bins)
Coke Storage
(Bins or Piles)
Scrap Storage
(Piles)
Sand Storage
(Bins)
* Major Emission Point (> 100 Tons/Yr)
Test Data Available
1
1
Control
Device
t
Cupola *
Muller
Core
Preparation
Particulate
Disposal
Pouring *
Moid
Machine
Foundry Returns
Sand Return J3/ i
S/ Sand handling
by scrubber or
Cooling *
1
Shakeout *
i
Cleaning &
Finishing ~Produ
Spent
Sand
i '
Waste Sand
*" (Landfill)
systems normally controlled
fabric filter.
Figure 3. Gray iron foundry - cupola flow chart.
-------
00
To Atmosphere
Crushing &
Screening
Storage
Shipment
* Major Emission Point (> 100 Tons/Yr)
..: Test Data Available
Figure 4. Ferroalloy production plant flow chart.
-------
To Atmosphere
Leod-
To Atmosphere To Atmosphere
. Dry Battery Line
Acid
* Major Emisiion Point (> 100 Tora/Yr)
' Test Data Available
Figure 5« Lead battery manufacturing plant flow chart.
-------
TABLE 1. IP DATA SOURCES AND NUMBER OF RUNS COMPILED
Cascade impactor data
Published
RAPS FPEIS data
SASS train
particle size
distribution
Vaporous
partic-
ulate
Fugitive
emis-
sions
Electric generation
plant
Ferroalloy production
plant
Gray-iron foundry
Lead battery manu-
facturing
104
36
95
27,
15
39
59
19
-------
3.1 SOURCE TEST DATA (SOTDAT) SYSTEM
SOTDAT is a component of the Aerometric and Emissions Reporting Systems
(AEROS), a comprehensive set of air pollution data systems* SOTDAT provides
for the storage, retrieval, and analysis of source test data. The National
Emissions Data System (NEDS) point source identification scheme was adopted by
SOTDAT for universal source identification.
The existing SOTDAT data were not used because the limited available data
for the specific sources of interest lacked detailed information pertaining to
sampling characteristics, particle sizing (typically confined to the 0.2 to
2.0 fjim range), and source operating conditions.
3.2 REGIONAL AIR POLLUTION STUDY (RAPS) DATA SYSTEM
The RAPS data system comprises detailed emissions data gathered during
1975 and 1976 from stationary point sources in the St. Louis Interstate Air
Quality Control Region (AQCR). The accuracy of the RAPS emission inventory
was based on an emission factor verification program. The program was accom-
plished by source testing of typical sources using standard EPA methods.
Coal burning power plant and gray iron foundry particle-size data were
obtained from the RAPS data system*!' The limited amount of available par-
ticle size data from RAPS were for controlled particulate emissions only.
3.3 FINE PARTICLE EMISSIONS INFORMATION SYSTEM (FPEIS)
FPEIS contains data on primary particle emissions from stationary point
sources and includes data on control device performance. Also included are
process descriptions of the sources and descriptions of the sampling equip-
ment and measurement techniques used. FPEIS has the capabilities to include
source test data including particle size distributions and chemical, radio-
nuclide, physical and bioassay testing results from subsequent analyses of
collected particulate samples.
FPEIS was found to contain the best available and most useful particle
size distribution data in any of the data systems. Most of the particle-
size data presented in this report for coal-fired power plants, ferroalloy
plants, and gray iron foundries were procured from FPEIS.
3.4 NSPS SUPPORT DOCUMENTS AND PUBLISHED DATA
Because of the lack of adequate particle-size data on specific process
sources within the computerized data base systems, NSPS support documents
and published data were relied upon for additional information. The supple-
mental data sources provided all of the lead-acid battery manufacturing
11
-------
particle-size data, measured with conventional cascade impactors. Most of the
SASS train data were obtained from reports on Environmental Assessment Level
I testing programs. Unfortunately, not all of the SASS train particle-size
data were accessible for inhalable particulate emission factor development at
this time.
Three EPA test reports for lead-acid battery manufacturing plants present
the full extent of the available source test data. The facilities at which
emission sampling was performed included: ESB Canada, Ltd., Mississauga,
Ontario;^' ESB, Inc., Buffalo, New York;,2' and Globe Union, Inc., Ganby,
Oregon.IP/
None of the test data from the three studies were sufficiently complete
to enable the derivation of inhalable particulate emission factors. Total
particulate emission rates were not presented in any of the reports. Only
emission rates for lead particulate were presented and the proportions of
lead in the total emitted particulate were not indicated. Therefore, neither
total uncontrolled nor total controlled particulate emissions could be
calculated.
Particle size data for both controlled and uncontrolled emissions were
presented in the three studies. However, the controlled particle size data
(using Andersen impactors) were summarized in a nonusable format. Weight
percents for the individual Andersen stages were listed, but were not trans-
formed into exhaust gas concentrations. Thus, controlled emission rates and
emission factors based on particle size could not be derived. Uncontrolled
size data, obtained with Brink impactors, were correctly presented in a com-
puter printout format. These data were given in terms of milligrams per
actual cubic feet and milligrams per actual cubic meters for each impactor
stage.
12
-------
SECTION 4.0
DATA ANALYSIS AND RESULTS
4.1 DETERMINATION OF IP MASS FRACTIONS
In order to use the existing mass size distribution data to extract in-
formation on inhalable particulates, the distributions were assumed to fol-
low a certain form. Log-normal, weighted log-normal, Wiebull, and weighted
Wiebull size distributions were checked to determine the distribution with
the best fit to the data points. Several different measured particle size
distributions were used to test the four different distribution methods. In
almost every case, the log-normal distribution had the highest linear cor-
relation coefficient. Approximately 90% of the test data analyzed using the
log-normal had a correlation coefficient greater than 0.95. Therefore, all
size distribution data were assumed to have the log-normal form.
A log-normal distribution function is represented by a straight line on
the log-probability plot (percentage cumulative quantities versus log D).
Therefore, the mass mean diameter (50%) and the geometric standard deviation
of the distribution can be easily determined graphically from this kind of
plot, as shown in Figure 6. It is also easy to interpolate or extrapolate
data graphically from these plots. However, to draw the best-fit line
through several data points involves subjective judgment and therefore poses
a source of errors.
An analytical technique utilizing the TI-59 programmable calculator was
developed as part of this study to fit particle size data to a log-normal
distribution. The program transforms both coordinates into a linear format
and then performs a standard linear regression analysis to find the slope
and intercept of the least squares line of best fit to the data. The mass
mean diameter is the anti-log of the y-intercept, as shown in Figure 6. The
linear correlation coefficient is also calculated.
To find the mass fraction of particles smaller than 15 /im, the log of
15 (y-coordinate) is entered and the corresponding value of the x-coordinates
is computed using the least squares line previously determined. (This pro-
gram can be modified very easily if the mass fraction for another particle
13
-------
WEIGHT % GREATER THAN STATED SIZE
99.8 99 98 95 90 80 70 60 50 40 30 20 10 5 2 I 0.5 0.2 O.I
1-
0-
100
50
20
1 1
PARTICLE DIAMETER (MICRONS)
o P P
fO U _ N WC
j
SAa
Dia
- 'Y
3S M
mett
Inte
/
ean
sr
;rce
/X
Pt
/j
A
/
/
/
/
r
x
S
/
/
/
Fra(
^ if
^ l~.
/
:tion
)/xm
Slope = Log
s (Geometric Standard
Deviation)
0.1 02 03 1 2 5 10 20 30 4O 5O 60 70 8O 90 95 98 99 99fl 9<
WEIGHT % LESS THAN STATED SIZE
* 1 I 1 1 1 1 1
-3 -2-10 1 2 3
I0(
50
20
10
5
2
1
0.5
0.2
0.1
J.9
LINEARIZED COORDINATES
Figure 6. Graphical procedure for determining inhalable particles,
14
-------
cut-size is desired.) The computed x-coordinate value is then converted back
to a mass fraction. For a detailed analysis procedure, Appendix A outlines
the equations used to approximate the probability scale. A copy of the pro-
grammable calculator program is presented in Appendix B.
4.1.1 Cascade Impactor Test Data
Table 2 summarizes the conventional cascade impactor particle size data
compiled in this study, including the calculated IP mass fractions. The
majority of the accessible and useful particle size distribution data were
acquired from the FPEIS. As indicated in Table 2, the test data were avail-
able from FPEIS for electric power generators (104 tests), gray iron foundries
(95 tests), and ferroalloy production plants (36 tests). The remainder of the
data were available through RAPS, NSPS support documents, and other published
data. The analysis of the test data provided several applications of installed
air pollution control equipment for the electric power generators, ferroalloy
plants, and gray iron foundries.
An indication of the reliability of the particle size data in Table 2 is
given in the last column. The evaluation considered the amount of data, source
of the data, sampling parameters, and operating conditions of the process that
were included with the test results. The ratings range from 1 to 5 with 1
being the most reliable.
An example of data rated No. 1 would be the first set of power plant data
in Table 2. The extracted IP emissions data contained sufficient process and
sampling information to derive emissions factors and rates based on the par-
ticulate sampling information supplied by the data source. A good example of
a No. 5 rating is the lead-acid battery data. These data were insufficient to
enable the derivation of an IP emission factor for that source.
4.1.2 SASS Train Particle Size Data
The environmental assessment testing programs with Levels I or II analysis
require the use of the SASS train sampling system. It was felt necessary to
include all available SASS train particle size data in this analysis of in-
halable particulate data since the use of the SASS train will be increasing.
Many EPA personnel and EPA contractors were contacted to locate SASS train and
condensible particulate data pertaining to the four specific industries of
interest.
Table 3 summarizes the available SASS train data including the calculated
IP mass fractions. Unfortunately, not all of the test results are accessible
at this time. Additional results will be available in the near future from
environmental assessment testing programs. The second set of power plant data
was from the same source as the first set of data except that the burners were
modified in the second test sequence.
15
-------
TABLE 2. INHALABLE PARTIGULATE CASCADE IMPACTOR DATA FOR FOUR INDUSTRIES
Source/Control Device
Electric Power Generation (Dry
Conventional E.S.P.
(Source rate « 112-160 MW)
Conventional E.S.P.
(Source rate 3 100 MW)
Prototype Wet Scrubber
(Source rate - 90-150 MW)
Electrostatic Spray Scrubber
(Source rate 665 MW)
Venturi Scrubber
(Source rate 620-760 MW)
Ferroalloy Plants
Preformed Spray Wet Scrubbe
Gray Iron Foundries
Impingement Wet Scrubber
Atomized Spray Wet Scrubber
Atomized Spray Wet Scrubber
Preformed Spray Wet Scrubber
Not Specified
Lead-Acid Battery Plant^'
Baghouse
Baghouse
Scrubber
Data Source
Testing
Location
So.
of
Tests
Average
Fraction
<15 urn
Standard
Deviation
Sizing Instrument Used
Approx. Range
of Particle
Size Data (urn)
Reli-
ability
Rating
Pulverized Coal)
FPEIS
RAPS-'
FPEIS
Published^
Data
Published^/
Data
FPEIS
FPEIS
FPEIS
FPEIS
FPEIS
RAP&i'
Published
Data
(ESB Canada,
Ltd.)
Published
Data
(Globe Union,
Inc.)
Published
Data
(ESB, Inc.,
Buffalo)
Inlet
Outlet
Outlet
Inlet
Outlet
Inlet
Outlets/
Outlet
Outlet
Inlet
Outlet
Inlet
OutletS/
Inlet
Outlet
Inlet
Outlet
Inlet
Outlet
Outlet
Inlet
Outlet
(single
baghouse)
Outlet
(four
baghouses)
Inlet
Outlet
Inlet
Outlet
22
16
2
33
33
13
6
7
1
29
7
6
6
9
9
15
15
17
18
1
3
3
3
10
14
3
3
0.2452
0.8091
0.955
0.2218
0.7607
0.3500
0.8600
0.9800
0.9672
0.7035
0.9551
0.9900
0.9953
0.9897
0.9998
0.7876
0.7117
0.9967
0.9885
0.466
0.8665
0.3265
0.6804
0.4028
0.7494
0.5124
0.7466
0.0652
0.0790
_
0.0401
0.1689
0.0921
0.1200
0.0100
_
0.2101
0.0334
0.0098
0.0019
0.0123
0.0003
0.1688
0.2874
0.0072
0.0091
-
0.1137
0.1023
0.2368
0.1890
0.1756
0.4075
0.2783
Brink BMS-II Impactor
Andersen Model III Impactor
Andersen Model II Impactor
Brink BMS-II Impactor
MRI Model 1502 Impactor
UW Mark V Impactor
UW Mark III Impactor
UW Mark III Impactor
MRI Model 1502 Impactor
Brink Impactor
Andersen Impactor
Andersen Impactor
Andersen Impactor
UW Mark III Impactor
UW Mark III Impactor
UW Mark III Impactor
UW Mark III Impactor
UW Mark III Impactor
UW Mark III Impactor
-
Brink Impactor
Andersen Impactor
Andersen Impactor
Brink Impactor
Andersen Impactor
Brink Impactor
Andersen Impactor
0.6-17
0.6-14
0.4-11
0.7-12.5
0.4-6
-
-
-
0.4-29
0.5-12.6
0.65-16
0.5-13
0.5-13
0.4-27
0.3-10
0.4-27
0.7-14
0.2-40
0.3-25
0.01-7
0.2-1.2v
0.2-5 |
\
I
0.2-5 '
0.1-1
0.2-4
0.07-1
0.1-3.5
1
]
1
2
1
3
1
3
3
4
5
4
6
_a/ Reference 7.
J>/ Reference LI.
£/ Effluent stream passed through only the first of two stages in the electrostatic scrubber.
d/ Reference 12.
&/ Particulate data Is questionable because the excessive hardness in the scrubber spray water generated additional particles
within the scrubber.
_f/ Reference 13,
£/ Particle density Is approximately 9 g/cm3 (a factor of 3 or 4 higher than other sources).
16
-------
TABLE 3. SUMMARY OF SASS TRAIN PARTICLE SIZE DISTRIBUTION DATA
Source/control device
Electric power generation
Conventional ESP
(source rate - 15 - 22.5 Mw)
Conventional ESP-^
(source rate =7-21 Mw)
Data Sampling
source location
-
FPEIS Inlet
FPEIS Inlet
Outlet
No. of
tests
2
7
4
Avg fraction
< 15 Mm
0.605
0.783
0.713
Multiclone
(source rate = 20 Mw)
Ferroalloy plant'
Silicomanganese
Venturi scrubber
FPEIS
EA
Outlet
Outlet
0.593
0.984
Ferromanganese
EA
Inlet
0.505
_§/ The combustion boiler burners were modified for this series of tests.
_b/ The data were obtained from an Environmental Assessment Level I test
program.-!^.'
17
-------
4.1.3 Condensible Particulate Data
Table 4 presents the condensible particulate data collected for the
specific industries. The emission rates were calculated on the basis of the
related source operating conditions and sampling parameters accompanying the
test results. The power plant and gray iron foundry emission rates are ex-
pressed in pounds of condensible particulates per ton of either coal burned
or material produced. The ferroalloy plant emission factors are presented
in pounds of condensible particulates per megawatt of heat input to the
furnacej the supporting data for the test results did not include material
process values.
4.1.4 Fugitive Emissions Data
The only process fugitive particulate emissions data pertaining to any
of the four industries are presented in Table 5. These high-volume sampler
data were measured at a distance of 100 m from a lead battery smelter tested
by an EPA contractor in Denmark«i;L' The typical feed composition for the
smelter furnace charge contained approximately 13% PVG scrap batteries. This
large fraction of polyvinyl chloride battery plate separators lead to the
formation of lead chloride which is more volatile than other materials in the
furnace.
Table 6 presents calculated IP mass for open dust sources based on data
from MRI source testing programs. Particle size data were obtained with a
parallel-slot cascade impactor and cyclone precollector operated at 20 cfm.
The sizing apparatus was positioned 5 m downwind of the source near the plume
centerline. The sources tested are considered typical of materials handling
and transportation activities found at most industrial sites.
4.2 DETERMINATION OF IP CONTROL EFFICIENCIES
Figure 7 depicts the procedure used for calculating the IP collection
efficiency. This procedure requires mass emissions data and IP fractions at
control device inlet and outlet locations. IP mass fractions were determined
by fitting particle size data to log-normal distributions as described above.
Table 7 presents the calculated IP control efficiencies for installed
air pollution control equipment within the four industries. The average col-
lection efficiency for inhalable particulates ranged from 76% for a wet
scrubber to 99% for an ESP.
4.3 DETERMINATION OF CONTROLLED IP EMISSION FACTORS
A controlled IP emission factor for a given source may be calculated by
chain multiplication of three quantities as shown in Figure 1 (upper path).
18
-------
TABLE 4. SUMMARY OF CONDENSIBLE PARTICULATE EMISSION RATES
Source
Electric power genera tionS/
Gray iron foundry
Baghouse
Ferroalloy plant
Open EAF
Sealed EAF
Partially enclosed EAF
Sampling
location
Outlet
Inlet
Outlet
Outlet
Outlet
Outlet
No. of
tests
6
6
13
33
24
2
Average condensible
particulate emission rate
0.096 A Ib/ton5-/
1.406 Ib/ton
0.3462 Ib/ton
0.1137 Ib/Mw^
0.9977 Ib/Mw
0.0151 Ib/Mw
_§/ Boiler rated at 12,500 Ib/hr of steam.
_b/ "A" is the ash content of coal burned, expressed as weight percent.
_c/ Emission rate given as pounds of particulate per megawatt of heat input.
TABLE 5. HIGH-VOLUME AREA SAMPLING RESULTS
Date
9/26/78
9/26/78
Sampling
rate
(m3/hr)
88.04
91.16
Sampling
period
(hr:min)
23:00
25:56
Total
particulate
collected
(M,S)
197,700
201,900
Total
particulate
concentration
(MS An3)
97.64
85.42
Ambient air
volume
sampled
(m3)
2,025
2,364
19
-------
TABLE 6. SUMMARY OF INHALABLE PARTICULATE DATA FOR FUGITIVE
a/
EMISSIONS-
Source/control method
No. of
tests
Avg fracti on
Standard
deviation
Unpaved roads
Uncontrolled
Dirt 12
Dirt and slag 2
Crushed slag 6
Crushed rock 2
Crushed rock/glacial till 6
Controlled
Dirt and slag treated with Coherex 1
Crushed rock treated with Trex 3
0.42
0.58
0.41
0.21
0.28
0.67
0.44
0.14
0.03
0.05
0.08
0.06
0.19
Paved roads ^uncontrolled)
Lightly loaded
Moderately loaded
Aggregate storage pile stacking
Iron ore pellets
Coal
Lump iron ore
Fine slag
7
3
4
1
5
7
0.79
0.48
0.21
0.22
0.20
0.076
0.14
0.17
0.25
0.18
0.06
aj Data source is MRI source testing, a Sierra Cascade Impactor was used
for particle size determination.
20
-------
Inlet
Outlet
Figure 7. Procedure for calculating inhalable particulate control efficiencies.
-------
TABLE 7. SUMMARY OF CONTROL DEVICE INHALABLE PARTICULATE COLLECTION EFFICIENCIES
N>
Inlet AveraRes Outlet Averages AVK. Total
Source/Control Device Race (kR/hr) (Ib/hr) (kR/hr) (Ib/hr) (kg/hr) (Ib/hr) (kit/hr) (Ib/hr) Efficiency
Electric Power Generation
Conventional ESP
Normal current density 112-160 MW 1,807 (3.984) 445.5 (982.2) 5.17 (11.40) 4.16 (9.17) 99.71
Reduced current density 112-160 MW 1,836 (4,048) 459.3 (1,012.6) 15.29 (33.69) 12.92 (28.48) 99.16
Prototype Wet Scrubber 90-150 MW 304.5 (671.2) 65.7 (144.9) 4.68 (10.33) 3.39 (7.47) 98.34
Electrostatic Spray Scrubber-S/ 665 MW 84.9 (187.2) 28.1 (62.0) 0.15 (0.33) 0.14 (0.32) 99.80
Ferroalloy Plants
Cray Iron Foundries
Atomized spray wet scrubber No Data 120.7 (266.0) 119.7 (264.0) 1.49 (3.28) 1.49 (3.28) 98.53
Atomized spray wet scrubber No Data 93.8 (206.8) 93.5 (206.1) 1.55 (3.42) 1.53 (3.38) 98.35
Preformed spray wet scrubber 48.500 kg metal/hr 31.21 (68.81) 22.75 (50.16) 7.60 (16.75) 5.02 (11.07) 69.98
Lead-Acid Battery Plant*!/
a/ Data source is Ref. 11.
Inhalable I'itrt Iculate Control Efficiency (%>
Avery^e Deviation Uunp.e
99.06 0.23 98.74-99.24
97.18 0.08 97.09-97.25
94.62 1.67 91.23-98.54
99.50
98.25 0.60 97.40-98.68
98.50 1.16 96.45-99.57
98.36 0.52 97.21-98.95
76.55 14.51 40.97-90.07
b/ Source operating data and sampling parameters were insufficient Cor analysis.
-------
1. The uncontrolled TSP emission factor obtained from AP-42 or other
data source*
2. The uncontrolled IP mass fraction taken from Table 8.
3. The IP control efficiency taken from Table 7.
It would also be necessary to add in the quantity of IP formed as condensible
particulate.
Table 9 presents the results of these calculations for the source/
control device combinations for which all data elements were available, ex-
cluding the condensible particulate contribution. Included are the uncon-
trolled and controlled emission factors for specific control device applica-
tions in the four industries.
The alternate scheme for calculation of controlled IP emission factors
(lower path in Figure 1) entails the use of a control device penetration
curve. Figures 8 and 9 present penetration curves from the RAPSi^/ data sys-
tem. As shown, these curves require extrapolation to the 15 /Ltm IP cut-point.
The concept of a control device penetration curve assumes that the per-
formance of any specific control device is predictable based on its principles
of operation and the properties of the inlet stream. EPA has undertaken to
develop performance models which simulate the operation of major control de-
vice types: electrostatic precipitators (EPA 600/7-78-111 a/b), Venturi
scrubbers (EPA 600/2-77-112), and fabric filters (EPA 600/7-77-084). These
models indicate that the development of control device penetration curves
which are dependent only on inlet particle size distribution may be unattain-
able.
Appendix C presents an example analysis of one of the four subject in-
dustries (iron foundries) from the standpoint of the IP emission factor de-
velopment effort. The calculation strategy entailing control device penetra-
tion curves is used to develop IP emission estimates for a typical foundry.
23
-------
TABLE 8. SUMMARY OF INHALABLE PARTICULATE MASS
FRACTIONS FOR FOUR INDUSTRIES
Fraction < 15 /im Fraction < 15 /^m
Source/Control Device Cascade Impactors SASS-Train Cyclones
Electric Power Generation
Uncontrolled Emissions 0.254 0.743
Conventional ESP 0.825 0.713
Prototype Scrubber 0.761
Electrostatic Scrubber 0.925
Ferroalloy Plants
Uncontrolled Emissions 0.704 0.505
Preformed Spray Scrubber 0.955 0.984
Gray Iron Foundry
Uncontrolled Emissions 0.928 -
Atomized Spray Scrubber 0.820
Preformed Spray Scrubber 0.989
Lead Battery Plants
Uncontrolled Emissions 0.510
Baghouse 0.675 -
Scrubber 0.747
24
-------
TABLE 9. INHALABLE PARTICULATE EMISSION FACTORS FOR FOUR INDUSTRIES
Source/control device
Electric power generation
Bituminous coal
Pulverized dry bottom
Conventional ESP
Scrubber
Uncontrolled
TSP
emission
factor£'
(Ib/ton)
n. b/
A
..
Uncontrolled
Uncontrolled IP emission
fraction factor
< 15 ptm (Ib/ton)
0.254 4.3 A
-
tm
Average
IP collection
efficiency
-
99
97
Controlled
IP emission
factor
(Ib/ton)
«
0.04 A
0.13 A
Ferroalloy production plant
Semi-covered furnace
N> Ferrochrome
Scrubber
Gray iron foundry
Cupola
Atomized spray scrubber
Preformed spray scrubber
Lead battery manufacturing!:/
17
0.704
0.928
31.7
15.8
98
98
76
0.55
0.25
3.70
aj Uncontrolled TSP emission factors are found in AP-42, Ref. 4.
b/ "A" constitutes the ash fraction in the coal burned, expressed as weight percent.
cj The uncontrolled TSP emission factor for a ferromanganese process was used in deriving the IP emis-
sion factor. Since the boiling points for chromium and manganese are relatively the same, it is
assumed the particulate emissions are qualitatively similar.
dj Data are insufficient to develop an IP emission factor.
-------
ro
99.99
99.95
99.9
99.8
99.5
99
98
95
90
80
70
60
50
40
30
20
10
5
iin~[
o.oi
Centrifugal Collector
High Efficiency
* Medium Efficiency
Wet Scrubber
Q High Efficiency
A Medium Efficiency
O Low Efficiency
10.0
JJ99.99
100.0
Particle Diameter, Microns
Figure 8. Penetration curves for cyclones and scrubbers.
-------
99.99
0.01
High Efficiency
& Medium Efficiency
O Low Efficiency
O.I
1.0
Particle Diameter, Microns
10.0
.L-m.99
100.0
Figure 9, Penetration curves for baghouses and ESPs,
-------
SECTION 5.0
KEY ISSUES
Several key issues have been identified as pertinent to the development
of IP emission factors. Each issue is sizeable in scope and will require con-
siderable investigation in order to determine a definitive answer. Brief dis-
cussions for each key issue are presented in the following paragraphs.
5.1 ISSUE 1: How reliable are available particle size data, and what are
the limitations and problems with these data in relation to
the objectives of this project?
Available particle size data for emission sources consist mostly of
cascade impactor measurements. Cascade impactor design cut points for par-
ticle classification are generally well below the 15 jum value for IP. Cyclone
precollectors used with impactors have cut points nearer the IP value.
In addition to the performance problems associated with impactors
(notably particle bounce which increases with particle size), the expedi-
tious use of impactors to define particle size in the fine particle range
has introduced additional sources of error in the size range bracketing the
IP cut point. For example, because fine particles have negligible inertial
properties and because the impactor has to be operated at a single sampling
rate, it was often judged unnecessary to sample isokinetically or at more than
one sampling point within the emission stream. However, because larger par-
ticles in the range of the IP cut point have significant inertial properties,
errors may have resulted in defining the upper end of the IP size range.
The available particle size data are confined mostly to ducted sources.
Data on process fugitive emissions and on open dust sources are sparse. More-
over, little data exist on condensible particulate, which would be obtained
from the back half of the Method 5 train.
5.2 ISSUE 2: Do cascade impactors and SASS train cyclones give comparable
particle size distribution results, and also, should impactor
precollector cyclone values be included in the data analysis?
The precision of measuring particulate emissions has been compared for
several total particulate collection and size distribution measurement
28
-------
methods ..22/ Simultaneous tests were run using the SASS train, Joy train with
cyclones, standard Method 5 procedure, and an Andersen cascade impactor. The
total particulate and size distribution data from these tests are given in
Table 10.
Based on the results, the precision of the sampling methods for particulate
mass and size distribution has a relative standard deviation of within + 30%.
The precision of the SASS train and the Joy train cyclones (which had the same
respective cut sizes as the SASS cyclones) was determined for measuring the
aerosol size distribution. The SASS train responded slightly better than the
Joy train.
To utilize an impactor precollector cyclone as a particle sizing device,
it is usually necessary to determine an adjusted particle cut-off diameter,
because cyclone D$Q corrections are needed, when the temperature and/or flow
rates are not maintained at design values. Although isokinetic sampling is
achieved to the closest degree possible by selection of probe nozzle diameter,
adjustments to the sampling rate are also common.
The data shown in Table 8 do not indicate good comparison of IP mass
fractions determined by cascade impactors and SASS trains. However, this is
thought to be due mainly to source differences, since the two methods were not
tested on the same specific sources.
5.3 ISSUE 3: Can particle size data below 10 /^m or less be extrapolated to
15 p.m in a way which is acceptable to the scientific and engi-
neering communities?
It has been demonstrated that specific aerosols generated by a single
formation mechanism tend to be log-normally distributed. However, because
more than one mechanism may be active within a given source type, bimodal
particle size distributions may be encountered. In some cases, the second
mode may be formed by condensation in the ambient air subsequent to plume dis-
charge.
Examination of available particle size data (which are generally confined
to the sub-10 jtim size range) for source emissions indicates that the log-normal
size distribution is a good representation of the data. Other two-parameter
distributions such as the Wiebull distribution are less representative of the
data.
It should be safe to extrapolate data from the sub-10 /jm range to the
IP cut point if there is no reason to suspect that a second aerosol formation
mode is active. This judgment may be made more reliably if emissions sources
are classified as to the potential active aerosol generation mechanisms: corn-
munition, condensation, ashing, etc.
29
-------
TABLE 10. COMPARISON OF DIFFERENT PARTIGULATE COLLECTION AND SIZING METHODS
U)
o
Particulate
emission
Method of
collection
SASS
Joy
Method 5
Andersen
impactor
Mean
Stand dev.
Percent stand
dev.
With imp
0.0510
0.0365
0.0660
-
0.0512
0.0148
29
(gr/dscf)
. Without imp.
0.0229
0.0276
0.0396
M
0.0300
0.0086
29
Less
With imp
89
64
-
-
76
18
23
Percent of particles
than 10 |im
Without imp.
74
50
M
80
68
16
23
Less
With imp.
76
58
tm
-
67
13
19
than 1 [aa
Without imp.
47
38
-
38
41
5
13
-------
However, it should be noted that control devices may transform a unimodal
particle size distribution at the inlet to a bimodal size distribution at the
outlet* This is due to the size-dependent mechanisms of particle collection,
agglomeration, and reentraintnent, Typically, modes in control device penetra-
tion are observed at about 0,5 and 5 jum.
5,4 ISSUE 4: Is there an acceptable method for extrapolating bimodal particle
size distribution data to 15
If two aerosols produced by two different mechanisms are mixed together
to form a composite aerosol, the resulting bimodal distribution will have two
distinct peaks. When plotted on log-probability paper, the bimodal distribu-
tion will have a distribution with two distinct and different slopes,
A composite distribution as shown in Figure 10 can be represented fairly
well by adding together two log-normal distributions weighed so as to represent
their respective fractions of the total mass.23/ jf the larger mode has a MUD
between 10 and 20 jum, it is probably necessary to have size distributions of
both modes to represent the continuous distribution in order to extrapolate
data from 10 ^m or less to 15 fj,m»
5,5 ISSUE 5: How does the near 15 jum particle fraction in control equipment
effluent air streams avoid capture by the control device?
Penetration of particles up to 15 jitm in size may be due to any of the
following effects:
1, To begin with there is a finite statistical probability that par-
ticles of any size will penetrate a well designed and operating control de-
vice,
2, Formation of particle coagulation chains which may behave as fine
particles in the collection process within the control device but which behave
as coarse particles within an inertial sizing instrument. The rate of chain
formation is high in flames and electrical fields,
3, Reentrainment of collected (and possibly agglomerated) particles,
especially during the control device cleaning cycle,
4, Defective sections of the control device (e.g., ruptured bags, dis-
connected electrodes, etc,),
5,6 ISSUE 6: How will IP emission factors be developed for sources where
there are no particle size distribution data available?
The mass fraction of. IP in uncontrolled source emissions may be esti-
mated by generic grouping of source categories according to particulate
31
-------
a.
Q
V
O Test Data Points
a Composite Distribution
0.)
Particle Diameter Dp, fim
Figure 10. Fitting a measured distribution with two log-normal distribution.
-------
generation mechanisms. Process fugitive emissions may be assigned the same
IP mass fraction as the uncontrolled ducted emissions for the same sources.
It has been shown that open dust sources within the same generic source cate-
gory have nearly the same proportion of IP in the emissions plume»
In addition, it should be possible to demonstrate that some source emis-
sions (for example, condensation of metal fume) consist entirely of IP* Also,
it is usually a good approximation to treat controlled particulate as IP since
control devices typically collect most of the particulate mass consisting of
particles larger than 15 jUm»
33
-------
SECTION 6.0
CONCLUSIONS AND RECOMMENDATIONS
This study has focused on two alternative schemes for the calculation of
control (or uncontrolled) IP emission rates as required for IP emission in-
ventories. Both methods utilize existing uncontrolled particulate emission
factors as found in AP-42 or other sources. Both schemes require data on
particle size distribution in the IP size range. In addition, the simpler
of the two methods requires IP control efficiency values for each source/
control device combination, and the more complex method requires control
device penetration curves which must be based on extensive modeling and
verification by testing. Either of the two schemes allows for the develop-
ment of an IP data presentation format which is an adaptation of the format
currently used in AP-42.
There are a limited number of industries whose atmospheric emissions of
particulate matter have been characterized through extensive source testing.
The available data for each significant source must be evaluated for reli-
ability and for suitability in developing IP emission factors. For the in-
dustries investigated, extrapolation of data were necessary to determine IP
mass fractions. In addition, the lack of fugitive emission data and con-
densible particulate data constituted a major information gap for signifi-
cant sources.
More satisfactory measurement techniques for inhalable particulates may
be available through modifications of existing equipment by using cyclone
particle sizing in conjunction with the back-half impingers of the Method 5
train for vaporous particulate collection. A special particulate sampling
system of this type has been used to measure the mass size distribution and
in some tests, the impinger catch has been included»££' However, this
special sampling system was not ideally suited to inhalable particulate sam-
pling because it was still necessary for some extrapolation of particle size
data to 15 /j,m. A device such as the dichotomous sampler which has a 15 /urn
cut point for the capture of airborne particles and direct quantification of
IP mass concentrations, should be employed in stack sampling trains for pur-
poses of IP emission quantification.
34
-------
The expansion of existing data management systems is needed to establish
a reliable computerized IP data base. A centralized IP data base would sim-
plify locating existing data in the preparation of SIPs. Presently, there
are several data systems which contain IP data. An integral part of an IP
data base should include close adherence to quality assurance protocols.
Conduct of this study has prompted the recommendations outlined in the
paragraphs below.
The existing information base is inadequate in relation to many aspects
of the inhalable particulate problem. Data acquisition programs should be
formulated to fill major data gaps for significant sources for which inhalable
particulate emission factors are needed.
Existing laboratory and field testing methods should be evaluated for
suitability in measuring the particulate concentration (including condensables)
and particle-size distribution up to 15 fj.m» The most suitable IP measurement
technique should be defined for sampling ducted and fugitive sources. Re-
search programs should be undertaken to develop devices for direct measure-
ment of IP mass fractions.
IP source characteristics should be analyzed and classified with respect
to particulate generation mechanisms. This information would aid in the de-
velopment of IP emission factors for sources which lack adequate testing data.
The IP mass fractions in source emissions may be estimated by generic group-
ing of source categories by particulate generation mechanisms.
The AP-42 data presentation format should be adapted in publishing IP
emission factors and control efficiencies. This would constitute an expan-
sion of the conventional versions of the data presentation formats for par-
ticulates and would minimize ambiguities in the preparation of IP emission
inventories as required for SIP revisions. An example of data presentation
format for IP emissions data was given in Table 9.
The impact of changing the IP cut-off diameter should be assessed. For
example, a change to 10 fj,m would allow direct interpretation of EA test data
without the need for an extrapolation procedure.
35
-------
REFERENCES
! Environmental Protection Agency HERL, Health Effects Considerations for
Establishing a Standard for Inhalable Particulate. July 1978.
2. Rodes, C« Inhaled Particles, Environ. Sci, and Tech, 12(13):1353-1355,
December 1978,
3« Environmental Protection Agency, Process Measurements Review. 1(3):8,
Winter Edition, 1979.
4, Environmental Protection Agency, Compilation of the Pollutant Emission
Factors, AP-42, Second Edition, February 1976.
5. Granger, L. S. Preliminary Environmental Assessment of Lead Emissions
from Selected Stationary Sources. Midwest Research Institute, Draft Final
Report, June 1977.
6. Monarch, M, R», et al. Priorities for New Source Performance Standards
Under the Clean Air Act Amendments of 1977. Argonne National Laboratory,
April 1978.
7. Littman, F, E», et al. Regional Air Pollution Study Point Source Emis-
sion Inventory, Rockwell International, March 1977,
8, Martin^ R, Source Testing at a Lead Acid Battery Manufacturing Company.
Environmental Protection Agency, August 1976.
9. Holzschuh, D. P, Emission Study at a Lead Acid Battery Manufacturing
Company. Environmental Protection Agency, August 1976.
10, Martin, R, M, Source Testing of a Lead Acid Battery Manufacturing Plant.
Environmental Protection Agency.
11. Pilat, M. J», and G. A, Raemhild, University of Washington Electrostatic
Scrubber Tests at a Coal-Fired Power Plant, University of Washington,
December 1978,
12. Leavitt, C. A., et al. Utility Conventional Combustion Comparative En-
vironmental Assessment - Coal and Oil,
36
-------
13. Littman, F» E., R. W» Griscom, and H. Wang. Regional Air Pollution Study -
Sulfur Compounds and Particulate Si2:e Distribution Inventory. April 1977.
14. Rudolph, J. L«, et al» Ferroalloy Process Emissions Measurement. Draft
Report, A. D. Little, Inc., August 1.978.
15. Coleman, R. T., and R. Vandervort. Source Characterization of the SB Bat-
tery Smelting Furnace. Draft Report, December 1978.
16. Abramowitz, M., and I. A. Stegon. Handbook of Mathematical Functions.
National Bureau of Standards, June 1964.
17. Midwest Research Institute. A Study of Fugitive Emissions from Metallurgi-
cal Processes (Iron Foundries). Draft Final Report, August 1, 1978.
18. A. T. Kearney and Company. Systems Analysis of Emissions and Emission
Control in the Iron Foundry Industry, Volume I Text. PB 198 348, February
1971.
19. Bates, C. E., and W. D. Scott. Better Foundry Hygiene Through Permanent
Mold Casting. Southern Research Institute for NIOSH, Contract No. 1 R01
OH 00456-01, January 30, 1976.
20. A. T. Kearney and Company. Systems Analysis of Emissions and Emission
Control in the Iron Foundry Industry, Volume II Exhibits. PB 198 349,
February 1971.
21. Georgieff, N. T., and F. L. Runyard. An Investigation of the Best Systems
of Emission Reduction for Electric Arc Furnaces in the Grey Iron Foundry
Industry. Draft, U.S. Environmental Protection Agency, November 1976.
22. Taback, H. J., et al. Fine Particle Emissions from Stationary and Miscel-
laneous Sources in the South Coast Mr Basin. Final Report prepared for
California Air Resources Board, Sacramento, California, February 1979.
23. Lundgren, D. A., and H. J. Paulus. The Mass Distribution of Large Atmo-
spheric Particles. Presented at 66th Annual APCA Meeting, Chicago,
Illinois,' June 1973.
24. Smith, W. B., and R. R. Wilson, Jr. Development and Laboratory Evalua-
tion of a Five-Stage Cyclone System. EPA-600/7-78-008, January 1978.
37
-------
Particle size data fitting a log-normal distribution yields a straight
line when plotted on log-probability graph paper. To graphically determine
the mass fraction of particles smaller than 15 ^m in diameter, the data
points would have to be plotted. Then, the best-fit line would be drawn
through the data points and the IP fraction determined. Such a graphical
approach is time consuming and requires a subjective judgment in drawing
the best-fit line through the data points.
An analytical technique utilizing the TI-59 programmable calculator
was developed as part of this study. The program transforms both coordi-
nates into a linear format, as shown in Figure 6, and then performs a stan-
dard linear regression analysis to find the slope and intercept of the least
squares line fit to the data. The ordinate is linearized by taking the log-
arithm of the aerodynamic particle diameter. The abscissa or the probabil-
ity function is represented by the integral
-/.
x t:2/2
e -
This integral can not be solved explicitly, but can be approximated by
2
0 < F < 0.5 x = -t + CQ + C]_t + c2t + S(F), where t =Vln(l/F2) and
.2
d3t3
0.5 < F < 1.0 x = t -
CQ + cit + c2t _
+ d]_t + d2t2 + d3t3
The constants needed for the probability function approximation are given
in Table A-l.
167
TABLE A-l. CONSTANTS USED IN THE LOG-NORMAL DATA ANALYSIS
b, = 0.31938153
1
b^ = -0.356563782
bl = 1.781477937
b^ = -1.821255978
bs = 1.330274429
5
. . -8
|e(x)| < 7.5 x 10
o
c, =
1
C2 =
r =
2.515517
0.802853
0.010328
0.2316419
d = 1.432788
d^ = 0.189269
d^ = 0.001308
-4
e(F)j < 4.5 x 10
39
-------
Once the data points are transformed to linear coordinates, the stan-
dard linear regression function of the TI-59 is used to determine the slope
and intercept of the least squares line fit through the data points. The
mass median diameter is the anti-log of the y-intercept, as shown in Fig-
ure 6, and the geometric standard deviation is the anti-log of the slope.
The linear correlation coefficient is also calculated.
To find the mass fraction of particles smaller than 15 u.m, the log of
15 (y-coordinate) is entered and the corresponding value of the x-coordinate
is computed using the least squares line previously determined. This pro-
gram can be modified very easily if the mass fraction for another particle
cut size is desired. The computed x-coordinate value is then converted
back to a mass fraction using the following formulas:
x < 0 F = f(x)[btt + b2t2 + b3t3 + b4t4 + b5t5] + e(x)
x > 0 F = 1 - f(x)[b t + b2t2 + b3t3 + b4t4 + b5t5] + e(x)
1 -x2/2
where f(x) = _ e and t
V 2n 1 + r|x|
The constants for the formula are presented in Table A-l. Appendix B
contains the log-normal distribution program used for analysis of the
particle size test data.
The log-normal method is a useful procedure for interpolating between
points as well as extrapolating beyond the measured range of the particle
size distribution. It is common to find deviations from log-normality at
the extremes of the size distribution. There are limitations of the log-
normal method; however, this procedure facilitates the extrapolation needed
to arrive at a mass fraction less than 15 ^m, from measured particle size
distribution data.
40
-------
APPENDIX B
IP CALCULATOR PROGRAM
41
-------
TITLE
.PAGE.
.OF.
PROGRAMMER
\ \
DATE
Tl Programmable
Coding Form
LOC CODE KEY
COMMENTS
LOC
CODE
KEY
COMMENTS
LOC
CODE
KEY
COMMENTS
\\
jft.l_
4.5L
;A)TO
_AS:.-
23
61
llo
Uftv-
» A'
.O.SL.
.31_
65*
1 J
_U_
II
It
\3
.o!_
.151
33
_Q7_
_/.*.
V2-
ya.
.AtS_(4
X ^^ *t*
1 °
_11L
_VJ
0.3.
16
_5.TJ5L
'0
±3_
cia.
TU
*t_ & L.
6.T6
sro
03
o f .
3i
151
±3_
~Ar~
1SL
-t_
_^c^
_o.8_
x_
_&Ci-_
6-1
JA
JJL_
ff
.OP
Tt> ro
01
JSu
ja_
A3_
J_V_
J2I
Js/.4_.
5TO
Ai.).£k»y_
of
(3
_B/i_
_? "
-3J-.
i3L
SflX.
JA_
3
_JJp_
_13__
_?J_
3JSL
v -* V
X <». T
r/i_
AE5S T/fAW
v
J.5^«jfl_
__ST10_
sro
of
"WV.
_as-.
/r
J.51.
_X*L
3J,_
_6J_
23_
_oy_
j£-
'0
A3_
__J.o_
iACJL_tJ>_
er
Lai__4«CL
MERGED COOES
62 Ed m 72fsro] HI 83[cTO| IB!
63 HM IBl 73 |Rop BIM 84 ! BT1
64fja m 74|sUit| mill 92HNV] (SSR]
33
TEXAS INSTRUMENTS
INCORPORATED
331977 Texas Instruments Incorporated
42
-------
TITLE Uarj - K\W rcvoA
PROGRAMMER TTX TtN
PAGE
. DATE
OF
Tl Programmable
Coding Form
LOG
ODE
KEY
COMMENTS
LOG
CODE
KEY
COMMENTS
LOG
CODE
KEY
COMMENTS
a/
J-ILZL
I I
..(.
1/3
J^u
_js_
.*£
L3
.1L
-*-_
JJ_
-g-
_23_
II
_*'
JL
_1L
_l-3_
.13_
JLl_
J.i.
L3JL
/l
_IJ_
.3_L
_3J_
_^^
ll
J£.
a_
>/=_
AJ/SJ-'y-'J'J-
A3_
h
J3U31
-iii.
.3A.SL
3&IL
A,.
V-
TO
ST3
-6.5:
-^
_A
44L
l/_
J$L
_u_
STS
X-3
_HJU&_
J-^iX-1
zt
AC.
53_
L13_
a
MJ
<5.0/63J.^_
_4.tD__iL_
.^-
MERGED CODES
62 Ed Iffil 72 [sfoj iffii 83 (5rol
63 m JIIBi 73 IRO) IB! 84 MM
64E1BI1 74|sUJM{ m 92|W| |
TEXAS INSTRUMENTS
INCORPORATED
© 1977 Texas Instruments Incorporated
43
-------
TITLE /fl£ -AJdgrtMfrL
PROGRAMMER 710 "TYX
.PAGEJ OF
DATE 12-Z.K
Partitioning (Op 17) I . . T . I Library Module Was fa
Tl Programmable
Program Record
/ fcTil i
_Printer Cards / ( Jip^li
PROGRAM DESCRIPTION
A
,4-AffUg
THE
TOA
common
t3sc JSJA ( duj
) /£S5
(jue.Af.ne. ie>& of
O
C>. of SLOPC ) J^
, R .
USER INSTRUCTIONS
STEP
PROCEDURE
ENTER
PRESS
DISPLAY
OATA P^/IVT:
M (fat
CLi/U
lift-
JD£
51
-&-
J
S,
a?4 £)>JT* POINT*
J£.t±Akl-
c,
flinjl uT
£L.
JL
D
DATA
USER DEFINED KEYS
DATA REGISTERS ([Ml BB )
LABELS (Op 08)
, .
,f>eiTl!iiT V
I?, &SO
11
12
2°
,5
M-
rfs] Tv£1__|sfol fH51_|sDg_rrT1 _
(in rn rn nn IGTOI nn
lafii r=i ....IM! _r*i _[g/si_nn_
[»/-! | = | PB3 IPPI iBi KB
Q| ECT QQ m 2Q| P!ti
QJ Q| IQI Q3 ^D B^
HQ3 IBB Q IB1 £jg liBl
FLAGS
© 1977 Texas instruments Incorporated
44
1014966-1
-------
TITLE i_o* t-tf^
UU^_
fj
K^*- ^
L, 2_ i
h * f^<^^
(_-j_-^
LV^
tb^
, ^
" ai£^
T Ufr-^"
j
PRESS
/li
-------
APPENDIX G
DEVELOPMENT OF IP EMISSION ESTIMATES FOR GRAY IRON FOUNDRIES
46
-------
This appendix outlines the procedure for the development of IP emission
estimates for a typical gray iron foundry. Gray iron foundry was selected
because sufficient amounts of information were found pertaining to plant
operation, emission sources, and particle size data.
For the purpose of this study, a source is considered to be significant
if, when uncontrolled, it has the potential for emitting 100 tons/year of
particulate. In a study directed toward helping EPA set priorities for de-
veloping New Source Performance Standards, Argonne National Laboratory has
identified all major point sources (those sources having the potential to
emit 100 tons/year of any of the criterial pollutants) .' Those point
sources in the foundry identified as having the potential of 100 tons/year
emissions, the emission factor, process capacity, and total annual particu-
late emissions are identified below.
Emission factor Capacity Emissions
Source (lb/ton)fL/ (tons/year) (tons/year)
Cupola 17 9 x 104 711
Reverberatory furnace 10 6 x 10^ 279
Electric arc furnace 7 9 x 104 293
(EAF)
a/ Reference 4.
Additional data on fugitive emissions are available from a previous MRI
study on fugitive emissions from metallurgical processes.il/ Since cupolas
and EAFs are the primary melting furnaces in gray iron foundries, a typical
production rate of 9 x 104 tons/year was utilized in conjunction with the
emission factors determined during the previous study to identify the follow-
ing significant sources.
Emission factor Emissions
Source (Ib/ton) (tons/year)
Iron pouring 4.0 180
Cooling of casting 4.3 194
Shakeout 3.15 142
47
-------
It should be noted that a previous EPA studyi^/ indicates that rever-
beratory furnaces account for only about 2% of the total melt in iron
foundries. Hence, this will be a relatively minor source on a nationwide
basis.
In order to comply with SIP requirements, state agencies must be able
to estimate emissions of IP from each significant source. In the absence
of actual test data, this will generally be accomplished by using an IP
emission factor for uncontrolled sources and an estimate of fractional ef-
ficiency for the applied control system.
In general, emission factors for IP are not available. It is necessary
to utilize total particulate emission factors and analyze particle size data
to determine the average fraction of emissions less than 15 /im in diameter.
The emission factors for the three point sources were given in Ref. 6.
The value of 17 Ib/ton for the cupola is taken from AP-42-L/ and is probably
the best value currently available. The value for reverberatory furnaces is
significantly higher than the AP-42 value of 2 Ib/ton. Data are insuffi-
cient to determine the reliability of either of these values. Reference 21
indicates that emissions from EAFs have been reported in the range of 4 to
40 Ib/ton. In the absence of specific source data, an emission factor of
17 Ib/ton is probably the most reasonable.
The emission factors for the three fugitive sources were developed
from data obtained by Bates and Scott!5L/ in a bench-scale experiment. These
tests are not as reliable as full-scale emissions tests. However, since
full-scale test results are not available, these are the best known emission
factors.
Few particle size data are available for iron foundry sources. A
search of the FPEIS data system produced four series of tests, all for cu-
polas. No data were found for reverberatory furnaces. Particle size data
for two EAFs were reported in Ref. 20. Finally Bates and Scott found par-
ticle size for the bench tests on pouring, cooling, and shakeout. Each of
these results is described below.
Particle size results for two EAFs are presented in Table C-l. It can
be seen that 93 and 94% of the emissions from the two furnaces are less than
15 /im in diameter. Since information is not sufficient to allow greater
than single digit accuracy, 90% of the emissions will be assumed to be less
than 15 jum in diameter.
The particle size data for pouring and shakeout generated by Bates and
Scott are presented in Table C-2. These data indicate that about 99% of the
emissions from both operations are less than 15 jum. This value will also be
assumed for cooling operations.
48
-------
TABLE C-l. SIZE DISTRIBUTION FOR THREE
ELECTRIC ARC INSTALLATIONS^^/
Particle size
distribution (/^m)
Foundry B
Foundry C
< 1
< 2
< 5
< 10
< 15
< 20
< 50
8
54
80
89
93
96
99
18
61
84
91
94
96
99
a/ Reference 20.
b_/ Data from Foundry A were not utilized since
they were not representative due to agglom-
eration.
TABLE C-2. PARTICLE SIZE DISTRIBUTIONS OF GREEN SAND EMISSIONS
FOR 4-IN. CUBE PATTERN*/
Pouring
Shakeout
Size
< 0
0.54 -
0.84 -
1.35 -
2.68 -
4.15 -
6.09 -
9.96 -
> 14
(Mm)
.54
0.83
1.34
2.67
4.14
6.08
8.95
14.36
.36
Mass (g)
3.98
8.35
23.01
16.69
1.86
0.97
0.53
0.40
0.68
% of total
7.0
14.8
40.7
29.5
3.3
1.7
0.9
0.7
1.2
Mass (g)
5.14
2.28
1.36
0.36
0.56
0.24
10.88
0.34
0.28
% of total
24.0
10.6
6.3
1.7
2.6
1.1
50.7
1.6
1.3
Total
56.41
21.44
a/ Reference 19.
49
-------
Based on the above data, the following best estimates of fine particle
emission factors for gray iron foundry sources are given below.
Particulate IP
emission factor emission factor
Source (Ib/ton) % IP (Ib/ton)
Cupola 17 92.8 15.8
Reverberatory furnace 10 a/
EAF 20 90 18
Pouring 4.0 ~ 99 3.9
Cooling 4.3 ~99 4.2
Shakeout 3.15 ~ 99 3.1
a/ Insufficient data.
The only true point source (one emitting directly to a stack) is the
cupola. Reference 18 indicates that the primary control device used on cu-
polas in 1967 was the wet cap. Medium and high energy scrubbers and fabric
filters were also applied to a number of cupolas. ESPs were applied to only
a few cupolas at this time. The authors indicated that the trend was away
from wet caps due to low collection efficiency. Most newer installations
will probably use high energy scrubbers or fabric filters.
Emissions from the EAF can be associated with three phases of the fur-
nace cycle: charging, melting, and tapping. An effective control system
must have the capability of capturing the emissions during each of these
phases and then removing particulate from the gas stream. The five major
capture systems identified in Ref. 21 are:
1. Canopy hoods
2. Roof hoods
3. Side-draft hoods
4. Direct furnace evacuation
5. Enclosures
These systems may be used individually or in combination to better capture
total emissions. The systems are described in detail in Ref. 21. The pri-
mary removal devices used in conjunction with the above system are fabric
filters, ESPs, and wet scrubbers.
50
-------
Control systems for pouring, cooling, and shakeout are described in
Ref. 17. Frequently no control system is used with pouring and cooling.
Capture systems which may be used are building evacuation, mobile vents, or
the use of a stationary pouring station with a side-draft hood. Building
evacuation or a mold tunnel with exhaust can be used to capture cooling
emissions. It should be noted that building evacuation is a costly method
and probably will not be utilized if feasible alternatives are available.
In general, captured emissions are vented to the atmosphere with no removal.
The primary capture systems associated with shakeout are total enclo-
sure and side-draft hoods. Due to the moist nature of the exhaust gas, wet
scrubbers are generally used as removal devices.
Four major removal devices have been identified for foundries: wet
caps, wet scrubbers, fabric filters, and ESPs. Reference 20 indicates that
wet caps effectively remove .only those particles greater than 44 jum in diam-
eter. Hence, it is likely that the effectiveness of wet caps is near 0%
for inhalable particulate.
The effectiveness of a system in controlling inhalable particulate is
dependent on the ability of the system to capture the emissions and then
upon the efficiency of the removal device in the sub 15 /urn range. Avail-
able data are not sufficient to predict accurately the capture efficiency
of the control systems for fugitive emissions of inhalable particles. It
would probably be necessary in the SIP revision process to estimate this
efficiency based on airflow calculations and judgments of agency personnel
based on visual evidence.
For example as a part of the development of the NSPS for electric arc
furnaces, EPA Method 22 is being used to estimate effectiveness of capture
systems based on visual observations. This method is currently being tested
and further data on its acceptability as a monitoring tool should be avail-
able in the near future. It is also possible to estimate the effectiveness
of hooding systems by measuring actual system parameters and using equations
in the Industrial Ventilation Manual to calculate theoretical capture of the
particulate*
Data on the fractional efficiency of ESPs, fabric filters, and wet
scrubbers were compiled by RAPS. These data are presented in Figures 8
and 9 in the report. In order to utilize the data for specific sources,
agencies would need to estimate inlet particle size distribution. The
curves in Figures 8 and 9 can then be used to estimate collection effi-
ciency in the sub 15 jLtm range.
51
-------
To calculate the inhalable particulate emission rate for a typical
source in a gray iron foundry assume the typical production rate of 9 x
tons/year for a cupola with an atomized spray scrubber applied for particu-
lates control. The generalized formula for a controlled inhalable particu-
late rate is as follows:
IP control Average Uncontrolled Production Control device
emission rate = fraction x TSP emission x rate x IP penetration
(IP Ib/year) <15 jj.m factor (tons/year)
(%) (Ib/ton)
This approach is outlined as the top path in Figure 1.
The IP emission rate for the cupola is:
IP control
1?>0 x 9^Q x 1Q ^ _ 0.984)
emission rate
=2.3 x 10 Ib/year, or 11.4 tons/year
where:
0.928 = average fraction < 15 ju,m, %, taken from Table 8
17.0 = uncontrolled TSP emission factor, Ib/ton
9.0 x 10^ = typical production rate, tons/year
0.984 = average IP collection efficiency, T?IP» taken from
Table 7
The total inhalable particulate emission rate for any facility will in-
clude all significant point sources and fugitive sources. (The condensible
particulate emission rate would be developed as an independent value.) For
a typical gray iron foundry, assume that the fugitive particulate emissions
are 80% inhalable since the calculations listed on page 50 show that an uncon-
trolled cupola generates a large fraction of inhalable particulates (92.870).
The inhalable particulate emission rate for each identified significant source
is as follows.
52
-------
IP emission rate
Source (tons/year)
Cupola 700
Iron pouring 144
Cooling of casting 155
Shakeout 114
Total 1,113
53
-------
APPENDIX D
PARTICLE SIZE MEASUREMENT METHODS
54
-------
Cascade impactors have been used extensively to establish the mass
distribution of aerodynamic particle size of an emission stream by frac-
tionating aerosol samples into several components representing different
size ranges. A mass size distribution function is determined from the
mass collected on each stage and the 50% cutoff particle diameter for each
stage.
Although impactors have been extensively used in the past to determine
mass size distribution, these devices have several limitations:
Frequently there is not enough mass collected on some stages to
be weighed accurately.
. Particle bounce and reentrainment cause an unpredictable, but sig-
nificant, error in the stage and backup filter catches.
. When the mass concentration is high, the sampling time may be un-
desirably short.
Impactors are used with lightweight collection substrates which
are often unstable in mass when exposed to the process stream.^'
A series of cyclones with progressively decreasing cut points will
perform similarly to impactors, but without many of the associated problems.
Cyclones, however, also have limitations to their applicability:
. There is no general theory to describe the performance of small
cyclones under field test conditions.
. Sampling times may be undesirably long at sources where the mass
concentration is low.-=-t'
The fraction of the IP emissions consisting of condensible particulate
may be determined using the back half of the particulate mass train results
measured with the EPA Method 5 sampling procedure or other suitable proce-
dure for accurately measuring the particulate which condenses between stack
temperature and ambient temperature. It is assumed that condensible par-
ticulate is comprised entirely of particles smaller than 15 (jjn in aero-
dynamic diameter.
EPA's SASS train uses a series of three cyclones operating at 10 cfm
to provide large quantities of particulate matter size, and these are clas-
sified into three ranges: 10 [im; 3 to 10 ^m; and 1 to 3 p,m. A backup fil-
ter provides a fourth size cut. The SASS train is prescribed by EPA for
environmental assessment work.
55
-------
The Fugitive Assessment Sampling Train (FAST), currently under develop-
ment by EPA, utilizes an existing cyclone separator design and a glass fiber
filter to collect a 500-mg particulate matter sample from the atmosphere
near an industrial source in an 8-hr sampling period. The cyclone provides
a Den for the respirable (2- to 3-(j,m) fraction of the particulate matter,
with the balance of the sample being collected on a 930-cm (1-ft ) filter.'
However, FAST does not provide the mass size distribution of collected par-
ticulates.
Fugitive emissions are those air pollutants (generated by activities
at industrial sites) that are transmitted into the ambient atmosphere with-
out first passing through some stack, duct, or pipe designed to direct or
control their flow. These types of emissions make up a large part of the
total pollution problem. Their generally diffuse nature and the absence of
any restrictions to their dispersion preclude the use of standard stack
sampling methods in the quantitation of their release into the environment.
As a result, an ambient air sampling strategy must be used to measure fugi-
tive particulate emission rates and particle size distribution.
The most widely used device for mass sizing of airborne particulate
matter is the cascade impactor. The conventional cascade impactor does
not correctly size dry particulate especially in the coarse particle range
(> 10 j^m) because of particle bounce problems. Although use of a cyclone
preseparator for removal of coarse particles partially alleviates the prob-
lem, residual effects are significant, and additional, somewhat tenuous,
corrections must be applied.
2/
The recently developed EPA version of the dichotomous sampler, is
virtually free of particle bounce problems. The dichotomous sampler has a
15-u.m cutpoint for capture of airborne particles and is useful for quantifi-
cation of IP mass concentrations. This device separates the collected IP
mass into size fractions below and above the 2.5-jj,m cutpoint (the minimum
in the typical bimodal size distribution of atmospheric particulate). How-
ever, the dichotomous sampler does not provide direct information on TSP
mass size distribution. Moreover, this device operates at a low flow rate
(1 cu m/hr) yielding only 0.024 mg of sample in 24 hr for each 100 u,g/nr
of TSP concentration; thus, an analytical balance of high precision is re-
quired to determine IP mass concentrations.
56
------- |