United States EPA-600/7-85-051
Environmental Protection
A9ency October 1985
4>EPA Research and
Development
SIZE SPECIFIC PARTICULATE
EMISSION FACTORS FOR
INDUSTRIAL AND RURAL ROADS
Source Category Report
Prepared for
Office of Air Quality Planning and Standards
Prepared by
Air and Energy Engineering Research
Laboratory
Research Triangle Park NC 27711
-------
RESEARCH REPORTING SERIES
Research reports of the Qfftee of Research and Development. U.S. Environmental
Protection Agency, have been grouped into nine series. These nine broad cate-
gories were established to facilitate further development and application of en-
vironmental technology. Elimination of traditional grouping was consciously
planned to foster technology transfer and a maximum interface in related fields.
The nine series are:
1. Environmental Health Effects Research
2- Environmental Protection Technology
3. Ecological Research
4. Environmental Monitoring
5. Socioeconomic Environmental Studies
6. Scientific and Technical Assessment Reports (STAR)
7. Interagency Energy-Environment Research and Development
8. "Special" Reports
9. Miscellaneous Reports
This report has been assigned to the INTERAGENCY ENERGY-ENVIRONMENT
RESEARCH AND DEVELOPMENT series. Reports in this series result from the
effort funded under the 17-agency Federal Energy/Environment Research and
Development Program. These studies relate to EPA's mission to protect the public
health and welfare from adverse effects of pollutants associated with energy sys-
tems. The goal of the Program is to assure the rapid development of domestic
energy supplies in an environmentally-compatible manner by providing the nec-
essary environmental data and control technology. Investigations include analy-
ses of the transport of energy-related pollutants and their health and ecological
effects; assessments of, and development of, control technologies for energy
systems; and integrated assessments of a wide range of energy-related environ-
mental issues.
EPA REVIEW NOTICE
This report has been reviewed by the participating Federal Agencies, and approved
for publication. Approval does not signify that the contents necessarily reflect
the views and policies of the Government, nor does mention of trade names or
commercial products constitute endorsement or recommendation for use.
This document is available to the public through the National Technical Informa-
tion Service, Springfield, Virginia 22161.
-------
EPA-600/7-85-051
October 1985
SIZE SPECIFIC PARTICULATE EMISSION FACTORS FOR
INDUSTRIAL AND RURAL ROADS
Source Category Report
by
Chatten Cowherd, Jr. and Phillip J. Englehart
Midwest Research Institute
425 Volker Boulevard
Kansas City, Missouri 64110
EPA Contract No. 68-02-3158
Technical Directive No.~ 12
EPA Project Officer: Dale L. Harmon
Air and Energy Engineering Research Laboratory
Research Triangle Park, North Carolina 27711
Prepared for:
U. S. Environmental Protection Agency
Office of Research and Development
Washington, D.C. 20460
-------
ABSTRACT
Over the past few years traffic-generated dust emissions from unpaved
and paved industrial roads have become recognized as a significant source
of atmospheric particulate emissions, especially within those industries
involved in the mining and processing of mineral aggregates. Although a
considerable amount of field testing of industrial roads has been performed,
most studies have focused on total suspended particulate (TSP) emissions,
because the current air quality standards for particulate matter are based
on TSP. Only recently, in anticipation of an air quality standard for par-
ticulate matter based on particle size, has the emphasis shifted to the de-
velopment of size-specific emission factors.
This study was undertaken to derive size-specific particulate emission
factors for industrial paved and unpaved roads and for rural unpaved roads
from the existing field testing data base. Regression analysis is used to
develop predictive emission factor equations which relate emission quanti-
ties to road and traffic parameters. Separate equations are developed for
each road type and for the following aerodynamic particle size fractions:
^ 15 urn, £ 10 urn, and ^ 2.5 urn. Finally, recommendations are made for in-
clusion of the resulting emission factors into AP-42.
11
-------
CONTENTS
Figures iv
Tables iv
1.0 Introduction 1
2.0 Data Review 3
2.1 Test Report I - Surface Coal Mining 5
2.2 Test Report 2 - Iron and Steel Production ... 10
2.3 Test Report 3 - Construction Aggregate
Industries, Copper Smelting, and Rural
Roads 14
3.0 Multiple Regression Analysis 21
3.1 Unpaved roads 22
3.2 Paved roads 31
4.0 Proposed AP-42 Sections 43
5.0 References 44
Appendices
A. Test Data Used in Regression Analysis A-l
B. Recommended Update of AP-42 Section 11.2.1 B-l
C. Recommended Update of AP-42 Section 11.2.6 C-l
11
-------
FIGURES
Page
IP emission factor versus silt loading for industrial
paved roads
TABLES
Number
1 List of Primary Test Reports by AP-42 Section Number. . . 4
2 Unpaved Road Test Matrix for Surface Coal Mines (Test
Report 1) • • • 5
3 Equipment Deployment for Unpaved Road Testing at Surface
Coal Mines (Test Report 1) 6
4 Range of Test Conditions for Unpaved Roads in Surface
Coal Mines (Test Report 1) 7
5 Emission Factors for Unpaved Roads in Surface Coal Mines
(Test Report 1) 8
6 Emission Factor Equations for Unpaved Roads in Surface
Coal Mines (Test Report 1) 9
7 Road Test Matrix for Iron, and Steel Plants (Test
Report 2) 10
8 Equipment Development for Road Testing at Iron and Steel
Plants (Test Report 2) 11
9 Range of Road Test Conditions in Iron and Steel Plants
(Test Report 2) 12
10 Emission Factors for Roads in Iron and Steel Plants
(Test Report 2) 13
11 Field Test Matrix for Industrial and Rural Unpaved
Roads (Test Report 3) 15
12 Field Test Matrix for Industrial Paved Roads (Test
Report 3) 15
13 Equipment Deployment for Industrial Road Testing (Test
Report 3) 16
14 Range of Test Conditions for Unpaved Industrial and
Rural Roads (Test Report 3) 17
15 Range of Test Conditions for Paved Industrial Roads
(Test Report 3) 18
16 Emission Factors for Industrial and Rural Unpaved Roads . 19
17 Emission Factors for Industrial Paved Roads 20
18 Geometric Mean Source Parameters for Unpaved Industrial
Roads 24
IV
-------
TABLES (concluded)
Number Page
19 Correlation Matrix for "Final" Unpaved Road Data Base . . 25
20 Predicted Versus Actual IP and PM-10 Emission Factors
for Unpaved Roads 27
21 Comparison of Unpaved Road Model Performance for IP and
PM-10 Emission Factors 28
22 Comparison of Unpaved Road Model Performance for FP
Emission Factors 30
23 Correlation Matrix for "Initial" Paved Road Data Set. . . 33
24 Paved Roads—Comparison of Emission Factors and Source
Characterization Parameters by Data Subset 35
25 Correlation Matrix for "Final" Paved Road Data Set. ... 36
26 Predicted Versus Actual IP and PM-10 Emission Factors
for Paved Roads 38
27 Comparison of Paved Road Model Performance for IP and
PM-10 Emission Factors 39
28 Paved Roads—Comparison of Source Characteristics by
Data Set 41
29 Comparison of Paved Road Model Performance for FP
Emission Factors 41
30 Paved Roads—Comparison of Single Value Emission
Factors 42
A-l Input Data for Development of Size-Specific Emission
Factor Equations for Unpaved Industrial and Rural
Roads A-2
A-2 Input Data for Development of Size-Specific Emission
Factor Equations for Paved Industrial Roads A-3
-------
1.0 INTRODUCTION
Over the past few years traffic-generated dust emissions from unpaved
and paved industrial roads have become recognized as a significant source
of atmospheric particulate emissions, especially within those industries
involved in the mining and processing of mineral aggregates. Typically,
road dust emissions exceed emissions from other open dust sources associ-
ated with the transfer and storage of aggregate materials. For example, in
western surface coal mines, dust emissions from uncontrolled unpaved roads
usually account for more than three-fourths of the total particulate emis-
sions, including typically controlled process sources such as crushing
operations.1 Therefore, the quantification of industrial road dust emis-
sions is necessary to the development of effective strategies for the at-
tainment and maintenance of the national ambient air quality standards for
particulate matter.
Although a considerable amount of field testing of industrial roads
has been performed, most studies have focused on total suspended particulate
(TSP) emissions, because the current air quality standards for particulate
matter are based on TSP. Those studies have produced emission factors that
are poorly defined with regard to particle size. Although the high-volume
sampler, which is the reference device for measurement of TSP concentration,
has a very broad capture efficiency curve,4 TSP is generally recognized as
consisting of particles smaller than 30 urn in aerodynamic diameter.
Only recently, in anticipation of an air quality standard for partic-
ulate matter based on particle size, has the emphasis shifted to the de-
velopment of size-specific emission factors in the particle size range
related to adverse health effects. The following particle size fractions
have been of interest in these recent studies:
IP = Inhalable particulate matter consisting of particles equal to or
smaller than 15 urn in aerodynamic diameter
PM-10 = Particulate matter consisting of particles equal to or smaller
than 10 urn in aerodynamic diameter
FP = Fine particulate matter consisting of particles equal to or
smaller than 2.5 urn in aerodynamic diameter
In practice, these particle size fractions have been determined in the field
using inertial sizing devices characterized by calibrated values of 50%
cutoff diameter (D50). The symbol "^" will be used in this report to define
particle size fractions determined in this manner.
-------
emisslo1" fi^o" far Yin! t^T 1s t0 derive size-specific particulate
field testinn riata, K 1ndusrtrial P^ed and unpaved roads from the existing
derived in this rpno^6" ,!mi!?ion factoi"s for rural unpaved roads are also
the result inn « '.eP?rT-- rinaily, recommendations are made for inclusion of
tne resulting emission factors into AP-42.
-------
2.0 DATA REVIEW
This section presents a review of field studies directed to the devel-
opment of uncontrolled particulate emission factors for industrial paved
and unpaved roads and for rural unpaved roads. The particular studies
selected for the derivation of size-specific emission factors are identi-
fied along with the criteria used in the selection process.
Although a substantial body of literature is available which deals in
some way with road dust emissions, relatively few documents are appropriate
for development of AP-42 emission factors. These documents meet the fol-
lowing criteria:
1. The information in the reference document must deal with actual
emission factor development. Many documents discuss emission fac-
tors but do not derive them.
2. Source testing must be part of the referenced study. Some re-
ports develop emission factor by applying assumptions to existing
factors.
3. The document must constitute the original source of test data.
For example, a symposium paper would not be included if the
original study were already- contained in a previous document.
4. The document must be readily accessible to the public.
Recently, these criteria were applied in a study5 to identify test re-
ports (published through 1981) which contained emission factor data on open
dust sources but which had not been referenced in AP-42. Ten reports which
met the criteria contained data on road dust emission factors. However,
with the exception of one study to develop emission factors for western sur-
face coal mines,1 those studies were directed primarily to emission factors
for TSP. The standard high-volume sampler was used as the primary sampling
device in five of the nine other studies. Moreover, direct measurement of
aerodynamic particle size was performed (at one downwind sampling height)
in only two of the studies; in three other studies, microscopy was used to
provide estimates of physical particle diameter.
Subsequent to the release of the test report on western surface coal
mines in November 1981, two additional reports directed to size specific
emission factors for road dust emissions were issued. The first report
dated August 1982, dealt with paved and unpaved roads in the iron and steel
industry;2 and the second, dated December 1982, presented size specific emis-
sion factors for paved and unpaved roads in several industries (asphalt and
-------
concrete batching, copper smelting, sand and gravel processing, and stone
quarrying and processing) and for rural unpaved roads.3
Together with the test report on surface coal mining, these additional
reports constitute an extensive data base of size-specific particulate emis-
sion factors for paved and unpaved roads. The reliability of the particle
size data presented in these three reports is substantially better than the
data presented in the earlier reports for the following reasons:
1. Measurement of particle size distribution was an essential part
of the exposure profiling strategies used to quantify emissions
in these studies.
2. Particle size distribution was measured simultaneously at more
than one height in the road dust plume.
3. Inertial sizing devices were used to obtain direct measurements
of aerodynamic particle size distribution.
Table 1 identifies the AP-42 source categories covered by the three test
reports.
TABLE 1. LIST OF PRIMARY TEST REPORTS BY AP-42 SECTION NUMBER
AP-42
Section No.
7.3
7.5
8.1
8.10
8.19
8.20
8.24
11.2.1
11.2.6
Industrial source category
Copper smelting
Iron and steel production
Asphaltic concrete plants
Concrete batching
Sand and gravel processing
Stone quarrying and processing
Western surface coal mining
Unpaved roads
Paved roads
Test
3
2
3
3
3
3
1
1, 2
2, 3
report
, 3
In the following sections, each of the three reports are discussed.
For each report the method of field sampling is described, including sam-
pling equipment and the number and location of test sites. Also, test data
are summarized including the ranges of road and traffic conditions tested.
-------
2.1 TEST REPORT 1 - SURFACE COAL MINING
This field study was directed to development of size-specific emission
factors for western surface coal mines. Field testing was conducted at
three mines, each representing a major western coal field: Powder River
Basin (Mine 1); North Dakota (Mine 2); and Four Corners (Mine 3). The study
included testing of unpaved haul roads and unpaved access roads in the ab-
sence of dust control measures. Table 2 lists the source testing informa-
tion for this study.
TABLE 2. UNPAVED ROAD TEST MATRIX FOR SURFACE COAL MINES
(TEST REPORT 1)
Vehicle type
Haul truck
Test
method
Uw-Dwa
Profiling
Site
(mine)
1
1, 2, 3
Test period
8/79, 12/79
7/79, 8/79,
12/79
No. of
tests
11
21
Light-medium Profiling 1, 2, 3
duty
8/79, 10/79,
8/80
10
Uw-Dw = Upwind-downwind method.
The primary sampling method for road testing was exposure profiling,
using the equipment deployment given in Table 3. Particle size distribu-
tions were determined at two or more heights in the plume by use of dichot-
omous samplers and high-volume cascade impactors with cyclone preseparators.
Other equipment utilized were: (a) high volume samplers for determining
TSP concentrations; (b) dustfall buckets for determining dust particle
deposition; and (c) recording wind instruments to determine mean wind speed
and direction for adjusting the exposure profiler to isokinetic sampling
conditions.
Road dust emission factors in the form of predictive equations were
developed for the TSP, IP, and FP size fractions. This was accomplished by
regression analysis of emission factors and the corresponding values of the
road and traffic parameters measured for each test.
Table 4 presents the ranges of test conditions from Test Report 1, and
Table 5 presents the average emission factors. These single-valued factors
were obtained by substituting geometric means of the test conditions into
the respective emission factor equations developed in this study. The equa-
tions are listed in Table 6.
-------
TABLE 3. EQUIPMENT DEPLOYMENT FOR UNPAVED ROAD TESTING AT
SURFACE COAL MINES (TEST REPORT 1)
Distance
from
source
Location (m)
Upwi nd 5
Downwind 5-10
Downwi nd 20
Downwi nd 50
1
1
2
1
1
1
1
2
2
2
2
2
Equipment
Dichotomous sampler
Hi-vol with standard inlet
Dustfall buckets
Continuous wind monitor
MRI exposure profiler with 4
sampling heads
Hi-vol with standard inlet
Hi-vol with cyclone/cascade
impactor
Dichotomous samplers
Dustfall buckets
Warm wire anemometers
Dustfall buckets
Dustfall buckets
Intake
height
(m)a
2.5
2.5
0.75
4.0
1.5 (1.0)
3.0 (2.0)
4.5 (3.0)
6.0 (4.0)
2.5 (2.0)
2.5 (2.0)
1.5
4.5 (3.0)
0.75
1.5 (1.0)
4.5 (3.0)
0.75
0.75
Alternative heights for sources generating lower plume heights are
given in parentheses.
-------
TABLE 4. RANGE OF TEST CONDITIONS FOR UNPAVED ROADS IN SURFACE COAL MINES (TEST REPORT 1)
Road surface properties Mean vehicle properties
Moisture Silt Silt Wind
No. of content content loading Speed Weight No. of speed
Vehicle type tests (%, w/w) (%, w/w) (g/m2) (mph) (tons) wheels (mph)
Light-medium 10 0.9-1.7 4.9-10.1 5.9-48.2 24.8-42.9 2.0-2.6 4.0-4.1 6.5-13.0
duty
Haul truck 27 0.3-8.5 2.8-18.0 3.8-254 14.9-36.0 24-138 4.9-10.0 1.8-15.4
-------
TABLE 5. EMISSION FACTORS FOR UNPAVED ROADS IN SURFACE
COAL MINES (TEST REPORT 1)
No. of
Vehicle type Tests
Parti cul ate emission factor by
aerodynamic size range
S 30 urn
15 |jm ^2.5 urn
Haul truck
27
17.4
8.2
0.30
Units
Light-medium 10 2.9
duty
1.8 0.12 Ib/VMT
Ib/VMT
TSP and ^ 15 urn emission factors were determined by applying the mean
correction correlation parameters in Table 13-9 (page 13-14) of test
report to the equation in Table 15-1 (page 15-2) of test report. The
^ 2.5 urn emission factors were determined by applying the appropriate
fraction found in Table 15-1 (page 15-2) of test report to the ^ 30 urn
emission factors.
-------
TABLE 6. EMISSION FACTOR EQUATIONS FOR UNPAVED ROADS IN SURFACE COAL MINES
(TEST REPORT 1)
Particulate emission factor equation3 K
Vehicle type TSP g 15 urn
Light-medium duty 5.79 3.22
(M)4-0 (M)4'3
Haul trucks 0.0067 (w)3'4 (L)0'2 0.0051 (w)3'5
S 2.5 um/TSP~ Units
0.040 Ib/VMT
0.017 Ib/VMT
10
Note: The range of test conditions are as stated in Table 4. Particle diameters are aerodynamic.
a From page 15-2, Table 15-1 of test report.
Multiply this fraction by the TSP predictive equation to determine emissions in the ^ 2.5 pm size
range.
M = moisture content of road surface material (%, w/w)
w = number of wheels
L = silt loading of road surface material (g/m2)
-------
2.2 TEST REPORT 2 - IRON AND STEEL PRODUCTION
In a second study directed to evaluation of open dust source controls
in the iron and steel industry, emissions from paved and unpaved roads were
tested prior to application of control measures. The testing was performed
at two steel plants, in Ohio (Plant F) and Texas (Plant B). This work has
been supplemented by testing in Illinois (Plant AG) and Missouri (Plant AJ).
Table 7 lists the source testing information for this study.
TABLE 7. ROAD TEST MATRIX FOR IRON AND STEEL PLANTS
(TEST REPORT 2)
Road type
Unpaved
Paved
Test site
location
Ohio
Indiana
Missouri
Ohio
Vehicle type
Light duty
Medium duty
Heavy duty
Medium duty
Heavy duty
Average mix
Test period
July 1980
November 1980
November 1980
June 1982
September 1982
July 1980 )
No. of tests
4
1
2
3
3
Texas
Average mix
October 1980
November 1981
July 1981
Exposure profiling was the primary test method using the equipment de-
ployments given in Table 8. Particle size distributions were determined at
two heights in the plume by use of high-volume cascade impactors with
cyclone preseparators. Other equipment utilized were: (a) high-volume
samplers with standard or size-selective inlets for measurement of TSP and
IP concentrations, respectively; (b) high-volume samplers with cyclone pre-
collectors and 37 mm filters for collection of samples to be analyzed for
trace metals and for largest particle size (by microscopy); and (c) record-
ing wind instruments to determine mean wind speed and direction for adjust-
ing the exposure profiler to isokinetic conditions.
Tables 9 and 10 present the ranges of test conditions and the average
emission factors, respectively, from test Report 2.
10
-------
TABLE 8. EQUIPMENT DEPLOYMENT FOR ROAD TESTING AT IRON AND
STEEL PLANTS (TEST REPORT 2)
Distance from source (m)
Plant F
Sampler
MRI exposure
profiler
Hi-vol with
cyclone/
impactor
Hi-vol with
size selec-
tive inlet
Hi-vol with
standard
inlet
Hi-vol with
cyclone
37 mm cassette
Location
Downwind
Downwi nd
Upwi nd
Downwi nd
Upwi nd
Downwi nd
Upwi nd
Downwi nd
Downwi nd
Upwi nd
Deploy-
ment 1
1.0
2.0
3.0
4.0
-
1.0
3.0
2.0
1.0
3.0
2.0
2.0
_.
-
-
—
Deployr
ment 2
1.0
2.0
3.0
4.0
5.0
1.0
3.0
_
1.0
3.0
-2.0
2.0
_
-
-
~
Plant B
1.0
2.0
3.0
4.0
5.0
1.0
3.0
2.0
2.0
2.0
2.0
_
-
-
™
Plant AG/AJ
1.5
3.0
4.5
6.0
—
1.5
4.5
3.0
-
-
-
1.5
1.5
4.5
3.0
Runs F27 to F32, F34, and F35.
Runs F61, F62, and F68 to F70.
11
-------
TABLE 9. RANGE OF ROAD TEST CONDITIONS IN IRON AND STEEL PLANTS (TEST REPORT 2)
Road
surface properties
Road type/vehicle type
Unpaved roads/light-duty
vehicles
Unpaved roads/medium- duty
vehicles
Unpaved roads/heavy-duty
vehicles
Paved roads/average
No. of
tests
4
4
5
11
Silt
content
(%, w/w)
5.8-14.0
6.3-16
6.4-35.7
Total
loading
(g/m*)
10,200-14,200
1,370-2,150
-
Mean vehicle properties
Speed
(mph)
15
20-27
20-24
-
Weight
(Mg)
2.7
20-25
45-49
10-36
No. of
wheels
4.0
5.9-9.8
6.0-10
_
Wind
speed
(m/s)
0.72-2.8
1.9-3.3
0.91-3.7
1.6-5.'
vehicle mix
-------
TABLE 10. EMISSION FACTORS FOR ROADS IN IRON AND STEEL PLANTS (TEST REPORT 2)
Road type/vehicle type
Unpaved roads/light-duty
vehicles3
Unpaved roads/medium-duty
vehicles
Unpaved roads/heavy-duty
vehicles
Paved roads/average
No. of
tests
4
4
5
11
Mean emission factors by aerodynamic size range
TP ^ 15 urn ^ 10 urn ^2.5 urn
3.33 0.860 - 0.227
13.1 3.36 1.01e 0.658
17.8 4.01 0.841f 1.10
0.728 - - 0.0607
Units
kg/VKT
kg/VKT
kg/VKT
kg/VKT
Emission factors are arithmetic means of test runs F28, F29, F30, and F31 from page 60, Table 3-19 of
test report.
Emission factors are arithmetic means of test runs AG1, AG2, and AG3 from supplementary testing and
F68 from page 49, Table 3-12 of test report.
c Emission factors are arithmetic means of test runs F69, and F70 from page 49, Table 3-12 of test report
and test runs AJ1, AJ2, and AJ3 from supplementary testing.
d Emission factors are arithmetic means of test runs F27, F32, F34, F35, F61, F62, B57, B58, B59, and B60
from page 73, Table 3-26 of test report.
e Emission factors are arithmetic mean of test runs AG1, AG2, and AG3 from supplementary testing.
Emission factors are arithmetic mean of test runs AJ1, AJ2, and AJ3 from supplementary testing.
-------
2.3 TEST REPORT 3 - CONSTRUCTION AGGREGATE INDUSTRIES, COPPER SMELTING,
AND RURAL ROADS
The objective of the third study was to expand the road dust emission
factor data base by conducting field testing in other industries with sig-
nificant road dust emissions. It was anticipated that the combined data
base would include ranges of road and traffic conditions that encompass most
industrial settings where road dust emissions are significant.
As indicated in the test matrix for unpaved roads (Table 11) and the
matrix for paved roads (Table 12), testing was performed in five different
industry categories; and testing also included rural (nonindustrial) un-
paved roads. Field tests were conducted in three different geographical
regions: Rocky Mountain region (sand and gravel processing, gravel rural
road), Great Plains region (stone crushing, asphalt and concrete batching,
and rural roads), and the southwestern region of the United States (copper
smelter).
Exposure profiling was the primary test method using the equipment de-
ployment given in Table 13. Particle size distributions were determined at
two heights in the plume using high-volume cascade impactors with cyclone
preseparators. Other equipment utilized were: (a) high-volume samplers
with standard inlets and size-selective inlets for measurement of TSP and
IP concentrations, respectively; and (b) recording wind instruments to de-
termine mean wind speed and direction for adjusting the exposure profiler
to isokinetic conditions.
Tables 14 and 15 give the ranges of test conditions for unpaved and
paved industrial roads, respectively. The average emission factors are
given in Tables 16 (unpaved roads) and 17 (paved roads). These statistics
are based only on those tests meeting the quality assurance criteria stated
on page 33 of Test Report 3.
14
-------
TABLE 11. FIELD TEST MATRIX FOR INDUSTRIAL AND RURAL UNPAVED ROADS
(TEST REPORT 3)
Industrial category
Stone crushing
Sand and gravel
Test site
location
Kansas
Kansas
Sampling
Vehicle type period
Medium duty December 1981
Heavy duty July 1982
No. of
tests
5
3
processing
Copper smelting
Rural roads
Crushed limestone
road
Dirt road
Gravel road
Arizona
Kansas
Missouri
Colorado
Light duty
Light duty
Light duty
Light duty
April 1982
August 1981
September 1981
March 1982
April 1982
3
6
4
2
TABLE 12. FIELD TEST MATRIX FOR INDUSTRIAL PAVED ROADS
(TEST REPORT 3)
Industrial category
Test site
location
Vehicle type
Sampling
period
No. of
tests
Sand and gravel
processing
Asphalt batching
Concrete batching
Copper smelting
Colorado Heavy duty
April 1982
Missouri Medium duty October 1981
Missouri Medium duty November 1981
Arizona Medium duty April 1982
4
3
3
15
-------
TABLE 13. EQUIPMENT DEPLOYMENT FOR INDUSTRIAL ROAD TESTING
(TEST REPORT 3)
Location
Distance
from
source
(m)
Equipment
Intake
height
(m)
Upwind 5-10 Hi-vol with standard inlet 2.0
Hi-vol with cyclone impactor 2.0
Hi-vol with size-selective inlet 2.0
Downwind 5 MRI exposure profiler with 5 1.0
sampling heads 2.0
3.0
4.0
5.0
Hi-vol with cyclone impactor 1.0
3.0
Hi-vol with standard inlet 2.0
Hi-vol with sire selective inlet 2.0
16
-------
TABLE 14. RANGE OF TEST CONDITIONS FOR UNPAVED INDUSTRIAL AND RURAL ROADS (TEST REPORT 3)
Road
surface properties
Industrial category
Stone crushing
Sand and gravel
processing
Copper smelting
Rural roads
Crushed limestone
road
Dirt road
Gravel road
No. of
tests
5
3
3
6
4
2
Silt
content
(%, w/w)
10.5-15.6
4.1-6.0
15.9-19.1
7.7-9.5
5.8-35.1
5.0
Total
loading
(g/m2)
3,360-7,190
13,000-15,200
2,300-3,490
2,140-4,890
2,290-7,820
1,200
Mean vehicle properties
Speed
(mph)
10-15
5
10
25-35
25
35-40
Weight
(Mg)
10-14
27-29
2.1-2.4
1.9-2.3
2.3
1.8-2.1
No. of
wheels
4.4-5.6
12.5-16.6
4.3-4.8
4.0
4.0
4.0
Wind
speed
(m/s)
1.1-4.2
1.0-2.1
1.9-3.1
1.1-5.9
2.9-5.9
4.3-5.0
-------
TABLE 15. RANGE OF TEST CONDITIONS FOR PAVED INDUSTRIAL ROADS (TEST REPORT 3)
oo
Road
surface properties
Industrial category
Sand and gravel
processing
Asphalt batching
Concrete batching
Copper smelting
No. of
tests
3
4
3
3
Silt
content
(%, w/w)
6.4-7.9
2.6-4.6
5.2-6.0
15.4-21.7
Total
loading
(g/m2)
755-1,480
2,820-4,200
189-239
1,220-1,840
Mean vehicle properties
Speed
(mph)
23
10
10-15
10-20
Weight
(Mg)
39-42
3.6-3.8
8.0
3.1-7.0
No. of
wheels
11-17
6-7
10.0
4.2-7.4
Wind
speed
(m/s)
1.4-3.4
2.1-2.7
3.0-4.4
2.2-3.9
-------
TABLE 16. EMISSION FACTORS FOR INDUSTRIAL AND RURAL UNPAVED ROADS
(TEST REPORT 3)
No. of
Mean emission factors by
aerodynamic size range
Industrial category
Stone crushing3
Sand and gravel
tests
3
3
TP
7,050
4,010
^ 15 urn
2,000
1,580
^ 10 urn
1,180
1,120
^ 2.5 urn
114
296
Units
g/VKT
g/VKT
processing
Copper smelting0
Rural roads
2,540 725
471
89.4
g/VKT
Crushed limestone
road
Dirt road f
Gravel road
4
4
2
6,180
14,000
1,890
1,080
2,220
352
612
1,190
235
94.2
236
103
g/VKT
g/VKT
g/VKT
Emission factors are arithmetic means of test runs AA1, AA4, and AA5 from
page 37, Table 16 of test report.
Emission factors are arithmetic means of test runs AF1, AF2, and AF3 from
page 37, Table 16 of test report. v
c Emission factors are arithmetic means of test runs AC1, AC2, and AC3 from
page 37, Table 16 of test report.
Emission factors are arithmetic means of test runs U2, U3, U4, and U5
from page 37, Table 16 of test report.
e Emission factors are arithmetic means of test runs AB1, AB2, AB3, and
AB4 from page 37, Table 16 of test report.
f Emission factors are arithmetic means of test runs AE1 and AE2 from page
37, Table 16 of test report.
19
-------
TABLE 17. EMISSION FACTORS FOR INDUSTRIAL PAVED ROADS
(TEST REPORT 3)
No. of
Mean emission factors by
aerodynamic size range
Industrial category
Sand and gravel3
processing
Asphalt batching
Concrete batching0
Copper smelting
tests
2
4
2
3
TP
1,550
516
1,340
3,160
g 15 urn
287
136
468
1,130
^ 10 |jm
178
83.1
330
784
S 2.5 |jm
57.2
36.6
107
171
Units
g/VKT
g/VKT
g/VKT
g/VKT
Emission factors are arithmetic means of test runs AD2 and AD3 from
page 38, Table 17 of test report.
Emission factors are arithmetic means of test runs Yl, Y2, Y3, and Y4
from page 38, Table 17 of test report.
Emission factors are arithmetic means of test runs Zl and Z2 from
page 38, Table 17 of test report.
Emission factors are arithmetic means of test runs AC4, ACS, and AC6
from page 38, Table 17 of test report.
20
-------
3.0 MULTIPLE REGRESSION ANALYSIS
In deriving recommended AP-42 participate emission factors for indus-
trial paved and unpaved roads, the first step was to investigate whether
size-specific emission factors correlated with source parameters and whether
these correlations crossed industry lines. Such correlations would lead to
predictive emission factor equations of greater reliability than single-
valued mean emission factors.
Stepwise Multiple Linear Regression (MLR) was the basic method used to
evaluate source parameters for possible use as correction factors in a pre-
dictive emission factor equation for a specific particle size fraction.
Various stepwise routines are available as part of the Statistical Analysis
System (SAS) computer package.6 The MaxR2 routine was employed in this
study.
In essence, the MaxR2 routine begins by selecting from the predictor
pool the source parameter that is the best predictor of emission factors.
In other words it selects the predictor that accounts for the highest per-
centage of the variation in emission factors. It changes the dependent
variable values to reflect the impact of this variable. Then another vari-
able, the one that would yield the largest increase in R2, is added. Once
this two-variable model is obtained, the variables in the model are compared
to each variable not in the model. The MaxR2 routine determines if replac-
ing one of the variables in the model by another from the predictor pool,
would increase the R2- After comparing all possible switches, the one that
produces the greatest increase in R2 is made. The resulting model is con-
sidered the "best" two-variable model that the technique can find. The same
process is repeated to find the three-variable model that yields maximum R2,
and so forth.
The steps followed in developing predictive emission factors are listed
below:
1. Create a data array of all monitored independent variables with
corresponding emissions measurements.
2. Input these data into the MLR program using appropriate code to
transform both independent and dependent variables to their
natural logarithms.
3. Evaluate the MLR output, using the classical significance tests
(F-ratio, partial F-ratios) as guidelines in assessing the
importance of potential correction parameters.
21
-------
4. Determine the form of the emission factor equation, exclusive of
the coefficient (base emission factor).
5. Assume typical values for the correction parameters.
6. Calculate adjusted emission factors at the average conditions for
all the correction parameters, using the relationships established
in the emission factor equation.
7. Determine the geometric mean for the adjusted data set. This mean
is the base emission factor or coefficient in the emission factor
equation.
8. Finalize the emission factor equation as the base emission factor
times each correction parameter normalized to average conditions.
9. Determine the precision factor for the emission factor equation.
The independent variables evaluated initially as possible correction
factors were silt content (%, w/w) silt loading (g/m2), total loading (g/m2),
average vehicle speed (kph), average vehicle weight (Mg), and average number
of vehicle wheels. Silt denotes that portion of loose surface dust that
passes a 200 mesh screen during standard dry sieving. The notation used to
define the various independent and dependent variables considered in the
analysis is presented below.
eTp = IP emission factor, expressed in kilograms per vehicle
kilometer traveled (kg/VKT).
ePM in = PM-10 emission factor, expressed in kilograms per vehicle
kilometer traveled (kg/VKT).
s = Silt content of roadway surface particulate matter (%, w/w).
sL = Silt loading of roadway surface particulate matter, ex-
pressed in grams per square meter (g/m2).
W = Mean vehicle weight (Mg).
w = Mean number of vehicle wheels.
S = Mean vehicle speed expressed in kilometers per hour (kph).
3.1 UNPAVED ROADS
3.1.1 Analysis and Results
Based on the criteria discussed in Section 2, it was determined that
three data sets (Test Reports 1 to 3) were available for the development of
IP and PM-10 emission factor equations. These data sets are tabulated in
Appendix A. It should be noted that each data set contains only those tests
which met the quality assurance guidelines outlined in the respective test
22
-------
reports. A summary by industry of pertinent source characteristics is pre-
sented in Table 18.
The correlation matrix associated with MLR analysis of the entire data
base (n = 49), indicated relationships that are consistent with those ob-
tained in an earlier nonparametric analysis of the data.7 For example,
silt content and vehicle weight both exhibited reasonably strong relation-
ships with IP and PM-10 emissions. However, the results of the MLR analysis
for the entire data set were disappointing in the sense that the "best" equa-
tions accounted for only about 40% and 38% of the variation in the respective
IP and PM-10 emission factors. The equations output from the SAS MaxR2
routine, were as follows:
e 0.85 0.32
eIP = 0.098 (s) (W) (1)
e 0.81 0.34
ePM-10 = 0.064 (s) (W) (2)
Analysis of the residuals from regression indicated that these
models performed reasonably well for much of the data base, but that they
did not adequately account for emissions variability in the surface mining
industry. The models tended to significantly overpredict emissions from
mine roads. This was thought to be due to the high degree of compaction of
mine roads which are designed to handle heavy mine vehicles. In support of
this reasoning, the silt loadings on the test mine roads were much lower
than the loadings found in other industries (see Table 18).
Based on the above considerations, the decision was made to exclude
the surface mining data set from the data base. The correlation matrix as-
sociated with the resultant final data set (n = 26) is presented in Table 19.
Perhaps the most significant feature of the matrix is the fact that the silt
loading parameter exhibits stronger correlation with the IP and PM-10 emis-
sion factors than does silt content, the road surface parameter used in MRI's
suspended particulate (SP) emission factor equation.7 Examination of the
matrix also suggests that the influence of vehicle type on emissions, as
parameterized by mean weight and wheels, increases when considering smaller
particle size fractions.
It should also be noted that the weak simple correlations between
vehicle speed and emissions imply that speed is not a primary influence on
emissions variability. However, as indicated below, speed and emissions do
exhibit substantial partial correlation after taking into account the in-
fluences of roadway surface loading and vehicle type.
The "best" MLR equations, as determined from the SAS MaxR2 output, were
as follows:
p 0.68 0.34 0.84
eIP = 0.00097 (sL) (W) (S) (3)
p 0.65 0.44 0.75
ePM-10 = 0.00059 (sL) (W) (S) (4)
These equations explain 72% and 73% of the variation in IP and PM-10 emis-
sions, respectively.
23
-------
TABLE 18. GEOMETRIC MEAN SOURCE PARAMETERS FOR UNPAVED INDUSTRIAL ROADS
ro
Road
surface parameters
Industry
Surface coal mining1
Haul trucks
• Light/medium
duty vehicles
Iron and steel produc-
tion2
Various industries3
Industrial roads
Rural roads
No. of
tests
14
9
7
9
10
Silt
Content
(%, w/w)
7.9
6.3
8.6
10.6
9.8
Silt
Loading
(g/m2)
27.4
12.9
395
607
255
Vehicle parameters
Weight
(Mg)
70
2.0
34
9.2
2.1
Speed
(kph)
41
54
27
13
46
No. of
wheels
7.9
4.0
6.9
6.9
4.0
49
-------
TABLE 19. CORRELATION MATRIX FOR "FINAL" UNPAVED ROAD
DATA BASE (n = 26)
IP
emission
factor
IP Emission factor 1.0
PM-10 Emission factor
Silt
££ Silt loading
Total loading
Vehicle weight
Vehicle wheels
Vehicle speed
PM-10
emission Silt Total
factor Silt loading loading
0.97 0.42 0.67 0.51
1.0 0.33 0.64 0.54
1.0 0.51 -0.11
1.0 0.80
1.0
Vehicle
weight
0.41
0.52
-0.29
0.20
0.52
1.0
Vehicle
wheels
0.24
0.33
-0.42
0.31
0.66
0.78
1.0
Vehicle
speed
-0.05
-0.12
0.11
-0.40
-0.55
-0.54
-0.73
1.0
-------
Equations 5 and 6 present the comparable predictive emission factor
equations normalized to typical values of the correction parameters.
eIP = 1.22 £L_ °-7 W °-4 _S °-8 (5)
400 7 24
ePM-10 = 0.766 sL_ °-7 W °-4 _S °-8 (6)
400 7 24
The normalization procedure consists of steps 5-9 as outlined at the begin-
ning of this section. It should be noted that previous MRI research indi-
cates that very little predictive accuracy is lost by rounding the exponents
to one figure. The validity of this procedure was verified for the above
equations.
Table 20 presents the predicted versus observed IP and PM-10 emission
factors, and provides a comparative statistic--the ratio of predicted to ac-
tual emission factors for each test. As indicated, the equations generally
provide very acceptable estimates of the actual emission factors. In the
case of the IP equation, all 26 predictions are within a factor of 2.5 of
the actual emissions; for the PM-10 equation, 25 of the 26 predictions are
within a factor of 2.5.
It should also be noted that a nonparametric analysis of the residuals
from the MLR indicated that the equations do not show any systematic pre-
dictive bias with respect to industry category.
3.1.2 Comparative Evaluation
Equations 5 and 6 predict the data set with precision factors of 1.60
and 1.64 for the IP and PM-10 emissions, respectively. The precision factor
is defined such that the 68% confidence interval for a predicted value (P)
extends from P/f to Pf. The precision factor is determined by exponentiating
the standard deviation of the differences (standard error of the estimate)
between the natural logarithms of the predicted and actual emission factors.
The precision factor may be interpreted as a measure of the "average" error
in predicting emissions from the regression equation. The effective outer
bounds of predictability are determined by exponentiating twice the standard
error of the estimate. The resultant estimates of predictive accuracy, in
this case 2.55 and 2.68 for IP and PM-10 emission, respectively, then en-
compass approximately 95% of the predictions.
As a basis for evaluating the emission factor equations developed in
this study, Table 21 presents three alternative models that may be used to
represent the IP and PM-10 emission factor data base. The first alterna-
tive (Model 2) consists of MRI's suspended particulate (SP) emission factor
equation (based on particles less than 30 urn Stokes diameter) with adjust-
ments to the coefficient to approximate IP and PM-10 emission factors.5 It
should be noted that the modified coefficients were developed from limited
particle sizing information that is not of the same quality as the measure-
ments comprising the study data base analyzed in this section. Limitations
26
-------
TABLE 20. PREDICTED VERSUS ACTUAL IP AND PM-10 EMISSION
FACTORS FOR UNPAVED ROADS
IP emission factor
Industry
category
Copper smelting
Iron and steel
production
Stone quarrying
and processing
Sand and gravel
processing
Rural roads
Run
ID
AC-1
AC-2
AC- 3
F-68
F-70
AG-2
AG-3
AJ-1
AJ-2
AJ-3
AA-1
AA-4
AA-5
AF-1
AF-2
AF-3
U-2
U-3
U-4
U-5
AE-1
AE-2
AB-1
AB-2
AB-3
AB-4
predicted
0.667
0.608
0.806
4.27
3.91
3.71
3.52
0.906
0.825
1.15
1.59
1.93
1.88
0.960
1.35
2.11
1.75
1.21
0.719
0.956
0.495
0.425
5.06
1.27
0.607
1.34
(kg/VKT)
actual
0.716
0.623
0.838
9.45
9.25
2.11
1.49
1.45
1.01
0.829
0.902
2.38
2.73
1.12
0.945
2.67
1.41
0.894
1.00
1.02
0.310
0.392
5.98
0.919
1.18
0.798
ratio3
0.93
0.98
0.96
0.45
0.42
1.76
2.36
0.62
0.84
1.39
1.76
0.81
0.69
0.86
1.42
0.79
1.24
1.35
0.72
0.94
1.60
1.08
0.85
1.38
0.51
1.68
PM-10 emission factor
(kg/VKT)
predicted actual
0.373
0.338
0.445
2.98
2.98
2.61
2.51
0.693
0.626
0.866
1.05
1.30
1.26
0.694
0.965
1.52
0.965
0.667
0.397
0.537
0.276
0.233
2.84
0.716
0.341
0.751
0.460
0.412
0.539
7.28
7.01
1.56
1.08
1.18
0.739
0.603
0.606
1.27
1.64
0.733
0.660
1.96
0.871
0.494
0.527
0.556
0.201
0.270
3.41
0.268
0.561
0.524
ratio3
0.81
0.82
0.84
0.41
0.42
1.67
2.32
0.59
0.85
1.44
1.73
1.02
0.77
0.95
1.46
0.78
1.11
1.35
0.75
0.96
1.37
0.86
0.83
2.67
0.61
1.43
Predicted divided by actual.
27
-------
TABLE 21. COMPARISON OF UNPAVED ROAD MODEL PERFORMANCE FOR IP AND PM-10 EMISSION FACTORS
Model
No.
Model origin
Model
Precision factor
PM-10 IP
This study
(
Modification to MRI I
SP equation (co- \
efficient only) I
Modification to MRI
SP equation (co-
efficient and
exponents)
Single-valued
emission factor
( eIP = 1.22 /sL \ °-7 / W\ °-4 /_S\ °-
) Uoo) \~7l
TM-10 = 0.766 / sL\ °-7 /W\ °-4 /_S\ °-!
UOO/ \7/ V24/
eIP = 0.95 /_s\ /_S\ /W\ °-7 /w\ °-5
\12J \48/ \3/ U/
JPM-10 = 0.745 / s\ / S\ /W\ °-7 /w\ °-5
Il2/ 1487 U/ U/
6IP = 1.34 / s\ /W\ °-3 /w\ L2 / <:\ °-8
PM-10 = 0.847 / s\ /W
(24)
/_i\ /W\°'3 w *•* I_S\ °-
\10) \7) 6 \24/
'IP = x = 1.29
-PM-10 = x =0.814
1.64
1.81
2.41
1-60
2.04
2.28
Represents the interval encompassing 68% of the predicted values.
IP = IP emissions
ePM-10 = PM-10 emissions
sL = Road surface silt loading
S = Average vehicle speed
s = Silt content of road surface
material
W = Average vehicle weight
w = Average number of wheels per vehicle
kg/VKT
Kg/VKT
g/m2
kph
%, w/w
Mg
-------
notwithstanding, these equations still predict the IP and PM-10 emissions
more accurately than do single-value emission factors.
The second alternative (Model 3) retains the same form as the MRI SP
equation but with adjustments to both the coefficient and the exponents of
the correction terms based on regression analysis against the study data
base. The fact that these equations provide reasonably accurate predictions
suggests that the original SP model formulation is also valid for smaller
particle size fractions. It reduces the uncertainty in estimating emissions
considerably over the use of single-value emission factors. In addition to
indicating slightly different relationships between correction parameters
and particulate emissions, the changes in exponents for the vehicle param-
eters in the MRI SP equation may partially reflect the greater diversity in
traffic characteristics present in the study data base.
Based on a comparison of precision factors, it is apparent that the
model incorporating silt loading to characterize the amount of surface ma-
terial available for entrainment, provides better estimates of IP and PM-10
emission factors than do the alternative models that use silt percent.
3.1.3 Extension of the Predictive Equations to FP Emissions
The FP emission factor equations were developed using a slightly dif-
ferent approach than that used in the case of the IP and PM-10 equation.
Rather than develop estimates of the model exponents and coefficients di-
rectly through MLR, constant multipliers—the geometric mean ratios of
FP/IP and FP/PM-10 emission factors—were -computed. These were then ap-
plied to the various IP and PM-10 emission factor equations. The resultant
models are presented in Table 22.
For the unpaved road situation, Model Nos. I/IP and 3/IP represent the
IP silt loading and silt models scaled by the geometric mean ratio of FP/IP
emission factors to approximate FP emissions. Similarly Model Nos. l/PM-10
and 3/PM-10 are scaled by the geometric mean FP/PM-10 ratio. As indicated,
the models scaled by the FP/PM-10 ratio provide better estimates of FP emis-
sion factors than do the corresponding models scaled by the FP/IP ratio;
all are considerably better than the single-value factor. Perhaps more im-
portantly, the silt model (3/PM-10) provides better estimates of FP emis-
sions than does the silt loading model (l/PM-10).
3.1.4 Applicability
Recommendations for incorporation of unpaved road emission factors into
AP-42 must balance the reliability of each candidate factor against the rela-
tive ease with which the factor can be used for emission inventory purposes.
Although the emission factor equations presented above have greater precision
than the single-valued counterparts, the equations require determination of
suitable input parameters. An important consideration in deciding which set
of size-specific emission factor equations for unpaved roads is most appro-
priate, centers on the reproducibility of the surface characterization
parameters for situations in which a potential user intends to collect in-
dependent observations to apply in the predictive equation.
29
-------
TABLE 22. COMPARISON OF UNPAVED ROAD MODEL PERFORMANCE FOR
FP EMISSION FACTORS
Model Precision
origin3 Model9 factor
/ cl \ 0 7 / W\ 0 4 / <;\ ° 8
I/TP o - n i=in I SLi • (-1 • I —I • 2 27
I/IP eFP - 0-150 UQO/ \7/ \24/ ^'
°4
l/PM-10 epp= 0.161 ' ' - 2.21
/ «: \ /W\ ° 3 /w\ ! 2 / ^\ ° 8
3/IP epp = 0.165 (^) (5) ' (I ' (2!) ' 2.14
3/PM-10 ecn = 0.176 (^)(5)--- (=)'•" (^}''8 2.05
Single- ePC> = x = 0.159 2.71
valued rK g
emission
factor
See Table 21 for starting models and definition of symbols.
Represents the interval encompassing 68% of the predicted values.
30
-------
Recognizing that any surface characterization measurement requires some
judgment on the part of the sampling personnel, it is felt that silt percent
is more easily quantified than silt loading, primarily because it is not as
sensitive to variations in sampling procedure. In this context, it should
be noted that reproducibility comparisons performed as part of an internal
MRI QA program indicate that co-located silt measurements are on the average
within 15%, while silt loading measurements are normally within approximately
20%. Though based on limited data, these comparisons are generally con-
sistent with previous experience which indicates that the collection of a
silt loading measurement does not typically pose any additional problems
(beyond that for silt content) except for instances in which there is no
well developed hard pan underlying the loose road surface material.
The models incorporating silt percent may also be preferable to those
using silt loading for some applications in which no independent measurements
of the parameters will be made. This follows from the fact that there is
more information currently available on the expected range of percent silt
for industrial roads. For example, some industrial facilities may already
have developed emission inventories based on measured silt content values.
To provide a comparable amount of information for the silt loading parameter,
it would be necessary to perform additional road surface characterization
work.
For these reasons the better of the two models incorporating silt per-
cent (Model 3) is recommended over the model incorporating silt loading
(Model 1) for the IP and PM-10 particle size fractions. For the FP size
fraction, the recommended model is 3/PM-10.which incorporates silt content
and is also the most accurate model.
3.2 PAVED ROADS
3.2.1 Analysis and Results
Applying the criteria of Section 2, it was determined that two data
sets (Test Reports 2 and 3) were available for the development of paved
road IP and PM-10 emission factor equations. These included test data col-
lected within the following industry categories: iron and steel production;
copper smelting; concrete batching; and sand and gravel processing. One
test was deleted from the former data set due to incomplete collection of
the source characterization parameters; two tests were dropped from the
latter set because they did not meet the QA guidelines for acceptable wind
direction. After deletion of these tests, the data base consisted of 21
tests, as tabulated in Appendix A. The independent variables considered
initially as possible correction factors were the same as those in the un-
paved roads analyses.
Prior to the analysis, it was recognized that the measured correction
factors would probably not account for a substantial portion of the variabil-
ity in IP and PM-10 emissions. One of the major reasons for this is that
any direct contribution of particulate from vehicle underbodies, exposed
haulage loads (i.e., aggregate materials), or vehicle exhaust is not para-
meterized by the available correction factors. Similarly, the influence of
31
-------
emissions from unpaved shoulders generated by the wakes of large vehicles
is not considered in the correction parameters. Because paved road emis-
sions are generally much lower than those from unpaved roads, the influence
of these sources is potentially greater in paved road emission factors. It
should be noted that previously published equations for paved road emissions
have used augmentation or judgment factors in an attempt to partially account
for the influence of these sources.7'8
The initial correlation matrix for the paved road data base is presented
in Table 23. As indicated, the correlations between emissions and the cor-
rection parameters are generally low, with the road surface characterization
parameters—silt and silt loading—exhibiting the strongest relationships
(r = 0.30) with emissions.
Plots of the simple linear relationships between road surface loading
parameters and emission factors were constructed to determine whether the
low correlations could be attributed to a particular test series or set of
source conditions. One plot, IP emissions versus silt loading, is shown in
Figure 1. It suggests that much of the "scatter" in the relationship is
associated with four tests conducted at an asphalt batching facility. The
roads at this facility had relatively heavy silt loadings yet produced rela-
tively low emissions. This situation is not consistent with either previous
MRI research or the remainder of the tests in the data base, which tend to
indicate a positive relationship between loading and emissions. This ap-
parently incongruous result may be linked to the fact that the roads were
traveled by predominantly light-duty vehicles; the mean vehicle weight for
each test was less than 4 Mg. One of the tests from the iron and steel
control efficiency program also deviated considerably from the positive re-
lationship between loading and emissions. In this case, the vehicle mix
included predominantly light-duty traffic (~ 80% pick-up trucks and cars),
with a mean vehicle speed considerably, higher than that of the other tests
in the data base.
Based on these considerations, the decision was made to partition the
data base into two subsets. Comparative statistics for each subset are pro-
vided in Table 24. As indicated, Subset 1 includes tests for relatively
heavily loaded roads traveled by predominantly light-duty vehicles (i.e.,
mean vehicle weight < 4 Mg). In contrast, Subset 2 includes tests for
roads with generally moderate surface loadings and vehicle mixes that can
be considered more typical of industrial facilities (i.e., mean vehicle
weight ~ 16 Mg). It is also important to note that the mean emission fac-
tors (IP and PM-10) for Subset 1 are less than 50% of those of Subset 2.
The correlation matrix based on Subset 2 is presented in Table 25. It
shows that for this data set the relationship between roadway surface load-
ings and emissions is reasonably strong. The inverse relationships between
emissions, and vehicle weight and speed cannot be explained by current
knowledge of the physical mechanisms responsible for the generation of fugi-
tive emissions. For this reason, the latter correlations should not be
construed as being applicable to the population of industrial paved roads.
Rather, they should be interpreted as products of this specific sample.
32
-------
TABLE 23. CORRELATION MATRIX FOR "INITIAL" PAVED ROAD
DATA SET (n = 21)
IP
emission
factor
IP Emission factor 1.0
PM-10 Emission factor
Silt
to Silt loading
Total loading
Vehicle weight
Vehicle wheels
Vehicle speed
PM-10
emission Silt Total
factor Silt loading loading
0.99 0.32 0.28 0.12
1.0 0.37 0.19 0.02
1.0 -0.22 -0.56
1.0 0.93
1.0
Vehicle
weight
-0.02
0.01
0.42
-0.51
-0.59
1.0
Vehicle
wheels
0.18
0.12
-0.36
0.11
0.22
0.45
1.0
Vehicle
speed
0.16
0.19
0.12
-0.24
-0.35
0.62
0.04
1.0
-------
2.00 -
1.00
-* 0.50
en
o
o
^ 0.10
o
0.05
0.01
O
A Asphalt Batching
• Concrete Batching
* Copper Smelting
• Iron and Steel
O Sand and Gravel Processing
i i i i i i
i i i
i i i
10
50 100
500
Silt Loading
Figure 1. IP emission factor versus silt loading
for industrial paved roads.
34
-------
TABLE 24. PAVED ROADS—COMPARISON OF EMISSION FACTORS AND SOURCE
CHARACTERIZATION PARAMETERS BY DATA SUBSET
CO
ui
Data subset description n
Subset 1:. Light-duty 6
traffic0
Subset 2: Medium- and 15
heavy-duty traffic
IP emission PM-10 emis- Silt Vehicle Vehicle
factor sion factor loading weight speed
(kg/VKT) (kg/VKT) (g/m2) (Mg) (kph)
xa xa xax oxa
gg gg ggg ggg
0.158 2.03 0.110 1.97 108 3.08 3.9 1.26 22 1.67
0.336 2.00 0.247 1.95 12.5 5.09 15.6 1.96 27 1.52
All statistics are geometric means and standard geometric deviations.
Includes four tests at asphalt batching facility, one test copper smelter, one test iron and steel
plant.
Includes nine tests at iron and steel plants, two tests each at copper smelter, concrete batching
plant, and sand and gravel processing plant.
-------
TABLE 25. CORRELATION MATRIX FOR "FINAL" PAVED ROAD
DATA SET (n - 15)
IP
emission
factor
IP Emission factor 1.0
PM-10 Emission factor
Silt
Silt loading
Total loading
Vehicle weight
Vehicle wheels
Vehicle speed
PM-10
emission Silt Total
factor Silt loading loading
0.99 0.09 0.77 0.71
1.0 0.14 0.70 0.63
1.0 0.08 -0.26
1.0 0.94
1.0
Vehicle
weight
-0.53
-0.58
0.09
-0.09
-0.12
1.0
Vehicle
wheels
0.10
-0.01
-0.59
0.49
0.67
0.38
1.0
Vehicle
speed
-0.47
-0.50
0.38
0.03
-0.17
0.71
0.01
1.0
-------
The "best" MLR equations as determined from the SAS MaxR2 output, were
as follows:
eIP = 0.148 (sL)0'32 (7)
ePM-10 = 0.120 (sL)0'29 (8)
These equations explain 59% and 49% of the variation in IP and PM-10 emis-
sion factors, respectively. It should be noted that though a greater per-
centage of the emissions variability could be accounted for by including
either vehicle weight or speed as a second correction factor, the resulting
equation would probably not provide stable emission factor estimates in
independent applications.
Equations 9 and 10 present the comparable predictive emission factor
equations normalized to the typical value for silt loading.
6IP = 0.322 (Ife)o-s (9)
ePM-10 = 0.244 (St) °'3 (10)
Table 26 presents (for Equations 9 and 10) the predicted versus actual
IP and PM-10 emission factors as well as the ratio of predicted to actual
emission factors for each test. As indicated, the equations generally pro-
vide very acceptable estimates of the actual emission factors. In the case
of the IP equation, all 15 of the predictions are within a factor of 2.5;
for the PM-10 equation, 14 of the 15 predictions are within a factor of 2.5.
3.2.2 Comparative Evaluation
The emission factor equations, as developed above, predict the data set
with precision factors of 1.59 and 1.64 for IP and PM-10 emissions, respec-
tively. Table 27 presents alternative models that may be used to represent
the emission factor data base. It is clear that the equations developed in
this study predict the IP and PM-10 emissions more accurately than do single-
value emission factors. The remaining alternative consists of MRI's sus-
pended particulate (SP) emission factor equation (< 30 urn Stokes diameter)
with adjustments to the original coefficient to approximate IP and PM-10
emission factors.5 As indicated, this model does not provide acceptable
prediction of the new emission factor data base.
The relatively poor performance of the scaled SP model may be attri-
buted largely to two factors. First, the proportionality constants probably
introduce significant error in the emission factor estimates, because as
noted in connection with the unpaved road equation, these constants are
based on limited particle sizing information. Second and more importantly,
the range of source conditions that provided the basis for the SP equation,
is much smaller than that of the new data base. This is reflected in the
fact that the surface loading term in the SP equation is raised to the first
power, while the newly developed equations indicate a much lower dependence
37
-------
TABLE 26. PREDICTED VERSUS ACTUAL IP AND PM-10 EMISSION
FACTORS FOR PAVED ROADS
Industry
category
Copper smelting
Iron and steel
production
Concrete
batching
Sand and gravel
processing
Run
ID
AC-4
AC- 5
F-34
F-35
F-45
F-61
F-62
B-57
B-58
B-59
B-60
z-i
1-2
AD-2
AD-3
IP emi
predicted
0.860
0.757
0.214
0.194
0.257
0.372
0.372
0.191
0.310
0.198
0.223
0.326
0.335
0.548
0.518
ssion factor
(kg/VKT)
actual
1.57
1.25
0.151
0.239
0.172
0.381
0.262
0.156
0.305
0.280
0.333
0.275
0.660
0.355
0.221
ratio3
0.55
0.60
1.42
0.81
1.49
0.98
1.42
1.22
1.02
0.71
0.67
1.18
0.51
1.54
2.34
PM-10 emission factor
(kg/VkT)
predicted actual
0.632
0.556
0.158
0.142
0.189
0.273
0.273
0.140
0.228
0.145
0.164
0.240
0.246
0.403
0.381
1.09
0.883
0.117
0.184
0.132
0.288
0.197
0.121
0.229
0.233
0.273
0.197
0.460
0.212
0.145
ratio3
0.58
0.63
1.35
0.77
1.43
0.95
1.38
1.16
1.00
0.62
0.60
1.22
0.54
1.90
2.63
Predicted divided by actual.
38
-------
TABLE 27. COMPARISON OF PAVED ROAD MODEL PERFORMANCE FOR
IP AND PM-10 EMISSION FACTORS
Model
origin
Precision factor3
Model PM-10 IP
This study
(Equations
9 and 10)
eIP = 0.332 (^
,0 3
PM-10 = 0.244 '
0 3
1.64
1.59
Modification to
MRI SP equa-
tion
IP-0.058
(!) °7
2.28
2.02
Single-valued
emission factor
eIP = x = 0.336
TM-10 = x = 0.247
1.95
1.99
Represents the interval encompassing 68% of the predicted values,
IP = IP emissions
PM-10 = PM-10 emissions
I = Industrial road augmentation factor
L = Surface dust loading on traveled
portion of road
n = Number of traffic lanes
sL = Road surface silt loading
s = Silt content of road surface
material
w = Average number of wheels per vehicle
W = Average vehicle weight
kg/VKT
kg/VKT
kg/km
g/m2
%, w/w
Mg
39
-------
of emissions on silt loading (0.3). In this context it should be noted that
the scaled SP equation consistently overpredicts the IP and PM-10 emission
factors of the new data set.
Table 28 provides a summary of selected source characterization param-
eters for a number of paved road data sets. It clearly shows the differences
in source conditions between the SP data set, and that used in developing the
new equation. For example, the mean road surface silt loading for the SP
data set is less than 25% of that for the new data set. Similarly, the mean
vehicle weight for the former ,data set is only about one-third of that of
the latter data set. In addition, it should be noted that the vehicle
weight range for the new data set is about four times greater than that used
in developing the SP equation.
3.2.3 Extension of the Predictive Equations to FP Emissions
Using the same approach described in Section 3.1.3 for unpaved roads,
FP emission factor equations were developed for paved roads. The resultant
models are shown in Table 29.
Examination of the precision factors for the paved road models suggests
that little predictive accuracy would be gained by using the silt loading
model in preference to the single-value factor. However, it should be noted
that the precision factor associated with the single-value indicates that
there is not a great deal of inherent variability in paved road FP emis-
sions.
3.2.4 Applicability
Partitioning the data base into two subsets as explained in Sec-
tion 3.2.2, restricts the applicability of the newly developed equations
to roads traveled by predominantly medium- and heavy-duty vehicles, at
mean speeds less than 48 kph (30 mph). As guidance, it is recommended that
use of the equations be limited to roads for which the mean vehicle weights
(based on all traffic) fall within the range of 6 to 42 Mg.
For roads that are traveled by predominantly light-duty traffic, the
single-value emission factors represented by the geometric mean emission
factors for Subset 1, should provide reasonable upper limits for IP and
PM-10 emissions. The geometric mean emission factors developed from the SP
data set, probably represent reasonable lower limits for industrial paved
road emissions. As indicated in Table 30, these mean emission factors
developed from the SP data set are approximately 50% of the mean emissions
factors for the new data set.
40
-------
TABLE 28. PAVED ROADS—COMPARISON OF SOURCE CHARACTERISTICS
BY DATA SET3
Data base
description n
Data set for 13
SP Equation
New data - 15
Subset 2
New data - 6
Subset 1
Silt loading
x a
g g
2.96 6.00
12.5 5.09
108 3.08
(g/m2)
Range
2.62-124
1.91-287
15.4-400
Vehicle weight (Mg)
x a Range
5.0 1.64 3-12
15.6 1.96 5.7-42
3.9 1.08 3.1-5.1
Reported values are geometric means and standard geometric deviations.
TABLE 29. COMPARISON OF PAVED ROAD MODEL PERFORMANCE FOR
FP EMISSION FACTORS
Model
origin3
Model'
Precision
factor
Equation
Single-
valued
emission
factor
e = 0.0795
epp = xg = 0.0788
1.66
1.75
a See Table 27 for starting models and definition of symbols.
b Represents the interval encompassing 68% of the predicted values.
41
-------
TABLE 30. PAVED ROADS—COMPARISON OF SINGLE
VALUE EMISSION FACTORS
Data Base
Description N
SP Equation
data set 13
New data -
Subset 1 6
Ratio3
Emission factors (kg/VKT)
IP PM-10
0.0781 0.0622
0.158 0.110
0.49 0.56
a Ratio of SP equation data set emission factor to
new data Subset 1 emission factor.
42
-------
4.0 PROPOSED AP-42 SECTIONS
Appendix B and Appendix C present the proposed revisions to the AP-42
sections for unpaved roads (Section 11.2.1) and for industrial paved roads
(Section 11.2.6), respectively. Updates for these sections were recently
developed by MRI5 and are included in Supplement 14 to AP-42. To the extent
possible, the format used in Supplement 14 was retained for the purpose of
incorporating the size-specific particulate emission factors developed in
this document.
Based on the rating quality scheme developed in the earlier study,5
all of the recommended size-specific particulate emission factors are
A-rated based on two criteria. First, the test data were developed from
well documented sound methodologies. Second, a total of at least six tests
were performed at two or more plant sites.
With regard to unpaved road emission factors for western surface coal
mining, it is recommended that the new AP-42 Section 8.24 be used without
modification. That section, which was developed in the earlier study,5
contains predictive emission factor equations for specified particle size
fractions (see Table 6).
Exposure profiling was the primary test method used for collecting the
emission data reported in the source category report. Particle size dis-
tributions were determined using high-volume cascade impactors with cyclone
preseparators. The manufacturer reported a 50% cut point of 11 microns for
the cyclone preseparator at the flow conditions used. However, the cyclone
preseparator was calibrated by Midwest Research Institute, and it was deter-
mined that the 50% cut point was actually 15 microns.
Although most of the test results reported in the source category re-
port were calculated using 11 microns as the cut point, these results were
recalculated for the AP-42 section using a 15 micron cutpoint. This change
has made the particle size multipliers used in the emission factor equations
in the AP-42 section slightly different from those presented with the same
equations in the source category report.
43
-------
5.0 REFERENCES
1. K. Axetell, Jr. and C. Cowherd, Jr., Improved Emission Factors for
Fugitive Dust from Western Surface Coal Mining Sources, Volumes 1
and 2, EPA Contract No. 68-03-2924, U.S. Environmental Protection
Agency, Cincinnati, Ohio, July 1981.
2. T. Cuscino, et al. , Iron and Steel Plant Open Source Fugitive Emission
Control Evaluation, EPA-600/2-83-110, U.S.Environmental Protection
Agency, Research Triangle Park, North Carolina, October 1983.
3. J. Patrick Reider, Size Specific Participate Emission Factors for Un-
controlled Industrial and Rural Roads, Draft Final Report, EPA Contract
No. 68-02-3158, Technical Directive No. 12, Midwest Research Institute,
Kansas City, Missouri, January 1983.
4. Wedding, J. B., "Ambient Aerosol Sampling: History, Present Thinking,
and a Proposed Inlet for Inhalable Particulate Matter," prepared at
the Air Pollution Control Association Specialty Conference, Seattle,
Washington, April 1980.
5. C. Cowherd and B. Peterman, AP-42 Update, Section 11.2, Fugitive Dust
Sources, EPA-450/4-83-010, U.S. Environmental Protection Agency,
Research Triangle Park, North Carolina, March 1983.
6. Statistical Analysis System - SAS User's Guide, SAS Institute, Inc.,
Raleigh, North Carolina, 1979.
7. C. Cowherd, et al., Iron and Steel Plant Open Source Fugitive Emission
Evaluation, EPA-600/2-79-103, U.S. Environmental Protection Agency,
Research Triangle Park, North Carolina, May 1979.
8. J. A. Maser and C. L. Norton, "Uncontrolled and Controlled Emissions
from Nontraditional Sources in a Coke and Iron Plant: A Field Study
Analysis," presented at the Air Pollution Control Association Spe-
cialty Conference on Air Pollution Control in the Iron and Steel
Industry, Chicago, Illinois, April 1981.
44
-------
APPENDIX A
TEST DATA USED IN REGRESSION ANALYSIS
A-l
-------
TABLE A-l.
INPUT DATA FOR DEVELOPMENT OF SIZE-SPECIFIC EMISSION FACTOR EQUATIONS
FOR UNPAVEO INDUSTRIAL AND RURAL ROADS
Source characterization parameters
Emission factors
Industry
category
Copper smelting
Iron and steel
production
Stone quarrying
and processing
Sand and gravel
processing
Rural roads
Run
ID
AC-1
AC- 2
AC- 3
F-68
F-70
AG-2
AG-3
AJ-1
AJ-2
AJ-3
AA-1
A A- 4
AA-5
AF-1
AF-2
AF-3
U-2
U-3
U-4
U-5
AE-1
AE-2
AB-1
AB-2
AB-3
AB-4
IP
(kg/VKT)
0.716
0.623
0.837
9.45
9.25
2.11
1.49
1.45
1.01
0.829
0.902
2.38
2.73
1.12
0.944
2.67
1.41
0.894
1.01
1.02
0.310
0.392
5.98
0.919
1.18
0.798
PM-10
(kg/VKT)
0.460
0.412
0.538
7.28
7.01
1.56
1.08
1.18
0.739
0.603
0.606
1.27
1.64
0.733
0.660
1.96
0.871
0.493
0.527
0.555
0.201
0.270
3.41
0.268
0.561
0.524
FP
(kg/VKT)
0.0798
0.0837
0.104
2.18
2.40
0.280
0.137
0.258
0.206
0.140
0.0809
0.110
0.151
0.176
0.175
0.539
0.115
0.0857
0.0913
0.0846
0.0708
0.136
0.699
0.0253
0.0792
0.143
Silt
(%, w/w)
19.1
15.9
16.0
14.0
16.0
5.8
7.2
6.3
7.4
7.7
13.7
15.6
15.6
4.2
6.0
4.1
9.1
7.7
8.6
9.2
5.0
5.0
35.1
16.7
16.8
5.8
Silt
loading
(g/m2)
440
394
558
1,100
667
1,050
1,020
114
101
165
484
911
911
545
908
583
445
262
184
255
60.0
60.0
2,740
414
384
133
Vehicle
weight
(Mg)
2.2
2.1
2.4
20
48
22
25
49
47
45
11
14
13
29
27
27
1.9
1.9
1.9
2.3
2.1
1.8
2.3
2.3
2.3
2.3
Total
loading
(g/m2)
2,300
2,480
3,490
7,860
4,170
18,100
14,200
1,810
1,370
2,150
3,530
5,840
5,840
13,000
15,100
14,200
4,890
3,400
2,140
2,770
1,210
1,210
7,820
2,480
2,280
2,290
Vehicle
wheels
4.8
4.0
4.3
5.9
10
7.3
6.6
6.0
6.0
7.1
5.0
5.6
5.0
14.5
16.6
12.5
4.0
4.0
4.0
4.0
4.0
4.0
4.0
4.0
4.0
4.0
Vehicle
speed
(kph)
16
16
16
32
32
27
25
24
24
24
24
16
16
8
8
8
56
56
40
40
64
56
40
40
40
40
-------
TABLE A-2. INPUT DATA FOR DEVELOPMENT OF SIZE-SPECIFIC EMISSION FACTOR EQUATIONS
FOR PAVED INDUSTRIAL ROADS
>
CO
Industry
category
Run
ID
Emission factors
IP
(kg/VKT)
PM-10
(kg/VKT)
FP
(kg/VKT)
Source characterization parameters
Silt
(%, w/w)
Silt
loading
(g/m2)
Total
loading
(g/ro2)
Vehicle
weight
(Mg)
Vehicle
wheels
Vehicle
speed
(kph)
Subset 2 - Medium- and Heavy-Duty Vehicles
Copper smelting
Iron and steel
production
Concrete batching
Sand and gravel
processing
Subset 1 - Light-Duty
Asphalt batching
Copper smelting
Iron and steel
production
AC-4
AC- 5
F-34
F-35
F-45
F-61
F-62
B-57
B-58
B-59
B-60
Z-l
Z-2
AD- 2
AD- 3
Vehicles
Y-l
Y-2
Y-3
Y-4
AC- 6
F-27
1.57
1.25
0.151
0.239
0.172
0.381
0.262
0.156
0.305
0.280
0.333
0.275
0.660
0.355
0.221
Traveling on
0.100
0.148
0.0862
0.209
0.570
0.101
1.09
0.882
0.117
0.184
0.132
0.288
0.197
0.121
0.229
0.233
0.273
0.197
0.460
0.212
0.145
Heavily Loaded
0.0725
0.113
0.0226
0.124
0.381
0.0813
0.239
0.202
0.0414
0.0584
0.0488
0.0922
0.0691
0.0417
0.0556
0.0942
0.122
0.0564
0.158
0.0547
0.0595
Roads
0.0392
0.0603
0.0120
0.0350
0.0733
0. 0299
19.8
15.4
16.0
10.4
28.4
21.0
20.3
6.4
17.9
14.0
13.5
6.0
5.2
7.9
7.0
2.6
2.7
4.6
4.6
21.7
35.7
287
188
2.80
2.00
5.10
17.5
17.5
1.91
9.58
2.14
3.21
11.0
12.0
64.0
53.0
91.0
76.0
193
193
400
15.40
1,450
1,220
17.7
19.6
18.0
83.4
83.4
36.0
53.5
14.7
23.8
189
239
805
755
3,490
2,820
4,200
4,200
1,840
43.1
5.7
7.0
25
23
15
36
33
11
16
10
11
8.0
8.0
39
40
3.6
3.7
3.8
3.7
3.1
13. Oa
7.4
6.2
6.1
6.0
5.3
7.6
7.4
6.2
5.9
5.3
6.4
10
10
17
15
6.0
7.0
6.5
6.0
4.2
4.4
16
24
43
42
40
40
40
18
18
18
18
16
24
37
37
16
16
16
16
32
b
a Approximately 80% of the vehicle passes during this test were pickup trucks and cars; the mean value reflects the influence of < 5% of
the passes by very heavy equipment.
No speed data obtained
-------
APPENDIX B
RECOMMENDED UPDATE OF AP-42 SECTION 11.2.1
B-l
-------
LI. 2.1 UNPAVED ROADS
11.2.1.1 General
Dust plumes trailing behind vehicles traveling on unpaved roads are a
familiar sight in rural areas of the United States. When a vehicle travels an
unpaved road, the force of the wheels on the road surface .causes pulverization
of surface material. Particles are lifted and dropped from the rolling wheels,
and the road surface is exposed to strong air currents in turbulent shear with
the surface. The turbulent wake behind the vehicle continues to act on the
road surface after the vehicle has passed.
11.2.1.2 Emissions And Correction Parameters
The quantity of dust emissions from a given segment of unpaved road varies
linearly with the volume of traffic. Also, field investigations have shown
that emissions depend on correction parameters (average vehicle speed, average
vehicle weight, average number of wheels per vehicle, road surface texture and
road surface moisture) that characterize the condition of a particular road and
the associated vehicle
Dust emissions from unpaved roads have been found to vary in direct
proportion to the fraction of silt (particles smaller than 75 micrometers in
diameter) in the road surface materials.^- The silt fraction is determined by
measuring the proportion of loose dry surface dust that passes a 200 mesh
screen, using the ASTM-C-136 method. Table 11.2.1-1 summarizes measured silt
values for industrial and rural unpaved roads.
The silt content of a rural dirt road will vary with location, and it
should be measured. As a conservative approximation, the silt content of the
parent soil in the area can be used. However, tests show that road silt con-
tent is normally lower than in the surrounding parent soil, because the fines
are continually removed by the vehicle traffic, leaving a higher percentage
of coarse particles.
Unpaved roads have a hard nonporous surface that usually dries quickly
after a rainfall. The temporary reduction in emissions because of precipita-
tion may be accounted for by not considering emissions on "wet" days (more than
0.254 millimeters [0.01 inches] of precipitation).
The following empirical expression may be used to estimate the quantity of
size specific particulate emissions from an unpaved road, per vehicle kilometer
traveled (VKT) or vehicle mile traveled (VMT) , with a rating of A:
9/85 Miscellaneous Sources 11.2.1-1
B-2
-------
TABLE 11.2.1-1. TYPICAL SILT CONTENT VALUES OF SURFACE MATERIALS
ON INDUSTRIAL AND RURAL UNPAVED ROADSa
I
to
I
en
to
Industry
Copper smelting
Iron and steel production
Sand and gravel processing
Stone quarrying and processing
Taconite mining and processing
Western surface coal mining
Rural roads
Road Use Or
Surface Material
Plant road
Plant road
Plant road
Plant road
Haul road
Service road
Access road
Haul road
Scraper road
Haul road
(freshly
graded)
Gravel
Dirt
Crushed limestone
Plant
Sites
1
9
1
1
1
1
2
3
3
2
1
2
2
Test
Samples
3
20
3
5
12
8
2
21
10
5
1
5
8
Silt (%, w/w)
Range
[15.9 - 19.1]
4.0 - 16.0
[4.1 - 6.01
[10.5 - 15.6]
[ 3.7 - 9.7]
[ 2.4 - 7.1]
4.9 - 5.3
2.8 - 18
7.2 - 25
18 - 29
NA
5.8-68
7.7 - 13
Mean
[17.0]
8.0
[4.8]
[14.1]
[5.8]
[4.3]
5.1
8.4
17
24
[5.0]
28.5
9.6
oo
Ul
References 4-11.Brackets indicate silt values based on samples from only one plant site.
NA = Not available.
-------
where: E = emission factor
k = particle size multiplier (dimensionless)
s = silt content of road surface material (%)
S = mean vehicle speed, km/hr (mph)
W = mean vehicle weight, Mg (ton)
w = mean number of wheels
p = number of days with at least 0.254 mm
(0.01 in.) of precipitation per year
The particle size multiplier, k, in Equation 1 varies with aerodynamic particle
size range as follows:
Aerodynamic Particle Size Multiplier For Equation 1
<30 ym
0.80
<15 ym
0.50
£10 urn
0.36
<5 ym
0.20
<2.5 ym
0.095
The number of wet days per year, p, for the geographical area of interest
should be determined from local climatic data. Figure 11.2.1-1 gives the geo-
graphical distribution of the mean annual number of wet days per year in the
United States.
Equation 1 retains the assigned quality rating if applied within the ranges
of source conditions that were tested in developing the equation, as follows:
RANGES OF SOURCE CONDITIONS FOR EQUATION 1
Equation
1
Road silt
content
(%, w/w)
4.3 - 20
Mean vehicle weight
Mg
2.7 - 142
ton
3 - 157
Mean vehicle speed
km/hr
21 - 64
mph
13 - 40
Mean no.
of wheels
4-13
Also, to retain the quality rating of the equation applied to a specific unpaved
road, it is necessary that reliable correction parameter values for the specific
road in question be determined. The field and laboratory procedures for deter-
mining road surface silt content are given in Reference 4. In the event that
site specific values for correction parameters cannot be obtained, the appro-
.priate mean values from Table 11.2.1-1 may be used, but the quality rating of
the equation is reduced to B.
Equation 1 was developed for calculation of annual average emissions and
thus, is to be multiplied by annual vehicle distance traveled (VDT). Annual
average values for each of the correction parameters are to be substituted into
9/85
Miscellaneous Sources
B-4
11.2.1-3
-------
110
ISO
CO
CO
CO
0 S0100 ZOO 300 400 500
120
MILES
00
Ul
Figure 11.2.1-1. Mean number of days with 0.01 inch or more of precipitation in United States.
10
-------
the equation. Worst case emissions, corresponding to dry road conditions,
may be calculated by setting p = 0 in the equation (which is equivalent to
dropping the last term from the equation). A separate set of nonclimatic
correction parameters and a higher than normal VDT value may also be justified
for the worst case averaging period (usually 24 hours). Similarly, to calc-.
ulate emissions for a 91 day season of the year using Equation 1, replace the
term (365-p)/365 with the term (91-p)/91, and set p equal to the number of wet
days in the 91 day period. Also, use appropriate seasonal values for the
nonclimatic correction parameters and for VDT.
11.2.1.3 Control Methods
Common control techniques for unpaved roads are paving, surface treating
with penetration chemicals, working into the roadbed of chemical stabiliza-
tion chemicals, watering, and traffic control regulations. Chemical stabilizers
work either by binding the surface material or by enhancing moisture retention.
Paving, as a control technique, is often not economically practical. Surface
chemical treatment and watering can be accomplished with moderate to low costs,
but frequent retreatments are required. Traffic controls, such as speed limits
and traffic volume restrictions, provide moderate emission reductions but may
be difficult to enforce. The control efficiency obtained by speed reduction
can be calculated using the predictive emission factor equation given above.
The control efficiencies achievable by paving can be estimated by com-
paring emission factors for unpaved and paved road conditions, relative to
airborne particle size range of interest. The predictive emission factor
equation for paved roads, given in Section It.2.6, requires estimation of the
silt loading on the traveled portion of the paved surface, which in turn depends
on whether the pavement is periodically cleaned. Unless curbing is to be
installed, the effects of vehicle excursion onto shoulders (berms) also must be
taken into account in estimating control efficiency.
The control efficiencies afforded by the periodic use of road stabili-
zation chemicals are much more difficult to estimate. The application para-
meters which determine control efficiency include dilution ratio, application
intensity (mass of diluted chemical per road area) and application frequency.
Between applications, the control efficiency is usually found to decay at a
rate which is proportional to the traffic count. Therefore, for a specific
chemical application program, the average efficiency is inversely proportional
to the average daily traffic count. Other factors that affect the performance
of chemical stabilizers include vehicle characteristics (e. g., average weight)
and road characteristics (e. g., bearing strength).
Water acts as a road dust suppressant by forming cohesive moisture films
among the discrete grains of road surface material. The average moisture level
in the road surface material depends on the moisture added by watering and
natural precipitation and on the moisture removed by evaporation. The natural
evaporative forces, which vary with geographic location, are enhanced by the
movement of traffic over the road surface. Watering, because of the frequency
of treatments required, is generally not feasible for public roads and is used
effectively only where water and watering equipment are available and where
roads are confined to a single site, such as a construction location.
9/85 Miscellaneous Sources 11.2.1-5
B-6
-------
References for Section 11.2.1
1. C. Cowherd, Jr., et al., Development of Emission Factors for Fugitive
Dust Sources. EPA-450/3-74-037, U. S. Environmental Protection Agency,
Research Triangle Park, NC, June 1974.
2. R. J. Dyck and J. J. Stukel, "Fugitive Dust Emissions from Trucks on
Unpaved Roads", Environmental Science and Technology, 10(10):1046-1048,
October 1976.
3. R. 0. McCaldin and K. J. Heidel, "Particulate Emissions from Vehicle
Travel over Unpaved Roads", Presented at the 71st Annual Meeting of the
Air Pollution Control Association, Houston, TX, June 1978.
4. C. Cowherd, Jr., et al., Iron and Steel Plant Open Dust Source Fugitive
Emission Evaluation, EPA-600/2-79-103, U. S. Environmental Protection
Agency, Research Triangle Park, NC, May 1979.
5. R. Bohn, et al., Fugitive Emissions from Integrated Iron and Steel Plants,
EPA-600/2-78-050, U. S. Environmental Protection Agency, Research Triangle
Park, NC, March 1978.
6. R. Bohn, Evaluation of Open Dust Sources in the Vicinity of Buffalo, New
York, U. S. Environmental Protection Agency, New York, NY, March 1979.
7. C. Cowherd, Jr., and T. Cuscino, Jr., Fugitive Emissions Evaluation,
Equitable Environmental Health, Inc., Elmhurst, IL, February 1977.
8. T. Cuscino, Jr., et al., Taconite Mining Fugitive Emissions Study,
Minnesota Pollution Control Agency, Roseville, MN, June 1979.
9. K. Axetell and C. Cowherd, Jr., Improved Emission Factors for Fugitive
Dust from Western Surface Coal Mining Sources, 2 Volumes, EPA Contract
No. 68-03-2924, PEDCo Environmental, Inc., Kansas City, MO, July 1981.
10. T. Cuscino, Jr., et al., Iron and Steel Plant Open Source Fugitive
Emission Control Evaluation, EPA-600/2-83-110, U. S. Environmental Pro-
tection Agency, Research Triangle Park, NC, October 1983.
11. J. Patrick Reider, Size Specific Emission Factors for Uncontrolled Indus-
trial and Rural Roads, EPA Contract No. 68-02-3158, Midwest Research
Institute, Kansas City, MO, September 1983.
12. C. Cowherd, Jr., and P- Englehart, Size Specific Particulate Emission
Factors for Industrial and Rural Roads, EPA-600/7-85-038, U. S. Environ-
mental Protection Agency, Research Triangle Park, NC, September 1985.
.13. Climatic Atlas of the United States, U. S. Department of Commerce,
Washington, DC, June 1968.
11.2.1-6 EMISSION FACTORS 9/85
B-7
-------
APPENDIX C
RECOMMENDED UPDATE OF AP-42 SECTION 11.2.6
C-l
-------
11.2.6 INDUSTRIAL PAVED ROADS
11.2.6.1 General
Various field studies have indicated that dust emissions from industrial
paved roads are a major component of atmospheric particulate matter in the
vicinity of industrial operations. Industrial traffic dust has been found to
consist primarily of mineral matter, mostly tracked or deposited onto the
roadway by vehicle traffic itself when vehicles enter from an unpaved area or
travel on the shoulder of the road, or when material is spilled onto the paved
surface from haul truck traffic.
11.2.6.2 Emissions And Correction Parameters
The quantity of dust emissions from a given segment of paved road varies
linearly with the volume of traffic. In addition, field investigations have
shown that emissions depend on correction parameters (road surface silt content,
surface dust loading and average vehicle weight) of a particular road and
associated vehicle traffic.^"^
Dust emissions from industrial paved roads have been found to vary in
direct proportion to the fraction of silt (particles £75 ym in diameter) in
the road surface material. ^-"2 The silt fraction is determined by measuring the
proportion of loose dry surface dust that passes a 200 mesh screen, using the
ASTM-C-136 method. In addition, it has also been found that emissions vary in
direct proportion to the surface dust loading.^-"2 The road surface dust loading
is that loose material which can be collected by broom sweeping and vacuuming of
the traveled portion of the paved road. 'Table 11.2.6-1 summarizes measured silt
and loading values for industrial paved roads.
11.2.6.3 Predictive Emission Factor Equations
The quantity of total suspended particulate emissions generated by vehicle
traffic on dry industrial paved roads, per vehicle kilometer traveled (VKT) or
vehicle mile traveled (VMT) may be estimated, with a rating of B or D (see below),
using the following empirical expression^:
n
where: E = emission factor
I = industrial augmentation factor (dimensionless) (see below)
n = number of traffic lanes
s = surface material silt content (%)
L = surface dust loading, kg/km (Ib/mile) (see below)
W = average vehicle weight, Mg (ton)
9/85 Miscellaneous Sources 11.2.6-1
C-2
-------
TABLE 11.2.6-1. TYPICAL SILT CONTENT AND LOADING VALUES FOR PAVED.-ROADS
AT INDUSTRIAL FACILITIESa
No. of
Industry Flint Sices
Copper smelting 1
Iron and steel
production 6
Asphalt batching 1
Concrete batching 1
Sand and gravel
processing I
No. of
No. of Silt (X, w/w) Travel Total loading x
Samples Range Mean Lanes Range
3 [15.4-21.7] [19-0) 2 [12.9-19.5]
[45.8-69.2]
2 0.006-4.77
20 1.1-35.7 12.5 2 0.020-16.9
4 [2.6-4.6] [3.6] 1 [12.1-18.0]
[43.0-64.0]
3 [5.2-6.0] [5.5] 2 [1.4-1.8]
[5.0-6.4]
3 [6.4-7.9] [7.1] 1 [2.8-5.5]
[9.9-19.4]
Mean
[15.9]
[55.4]
0.495
1.75
[15.7]
[55.7]
[1.7]
[5.9]
[3.8]
[13.3]
Silt loading
10 1.0, the rating of the equation drops to D because of the subjectivity
in the guidelines for estimating I.
The quantity of fine particle emissions generated by traffic consisting
predominately of medium and heavy duty vehicles on dry industrial paved roads,
per vehicle unit of travel, mav be estimated, with a rating of A, using the
following empirical expression :
11.2.6-2
EMISSION FACTORS
C-3
9/85
-------
E = k || (kg/VKT)
/ sL\ 0-3
E = k(3.5) ^0.35^ (Ib/VMT)
where: E = emission factor
sL = road surface silt loading, g/m2 (oz/yd2)
The particle size multiplier (k) above varies with aerodynamic size range
as follows:
Aerodynamic Particle Size
Multiplier (k) For Equation 2
(Dimensionless)
£15 urn <10 ym <2.5 ym
0.28 0.22 0.081
To determine particulate emissions for a specific particle size range, use the
appropriate value of k above.
The equation retains the quality rating of A, if applied within the range
of source conditions that were tested in developing the equation as follows:
silt loading, 2 - 240 g/m2 (0.06 - 7.1 oz/yd2)
mean vehicle weight, 6 - 42 Mg (7 - 46 tons)
The following single valued emission factors^ may be used in lieu of
Equation 2 to estimate fine particle emissions generated by light duty vehicles
on dry, heavily loaded industrial roads, with a rating of C:
Emission Factors For Light Duty
Vehicles On Heavily Loaded Roads
£15 ym £10 ym
0.12 kg/VKT 0.093 kg/VKT
(0.41 Ib/VMT) (0.33 Ib/VMT)
These emission factors retain the assigned quality rating, if applied within
the range of source conditions that were tested in developing the factors, as
follows:
silt loading, 15 - 400 g/m2 (0.44 - 12 oz/yd2)
mean vehicle weight, £4 Mg (£4 tons)
Also, to retain the quality ratings of Equations 1 and 2 when applied to a
specific industrial paved road, it is necessary that reliable correction para-
meter values for the specific road in question be determined. The field and
9/85 Miscellaneous Sources 11.2.6-3
C-4
-------
laboratory procedures for determining surface material silt content, and surface
dust loading are given in Reference 2. In the event that site specific values
for correction parameters cannot be obtained, the appropriate mean values from
Table 11.2.6-1 may be used, but the quality ratings of the equations should be
reduced by one level.
11.2.6.4 Control Methods
Common control techniques for industrial paved roads are broom sweeping,
vacuum sweeping and water flushing, used alone or in combination. All of
these techniques work by reducing the silt loading on the traveled portions of
the road. As indicated by a comparison of Equations 1 and 2, fine particle
emissions are less sensitive than total suspended particulate emissions to the
value of silt loading. Consistent with this, control techniques are generally
less effective for the finer particle sizes.^ The exception is water flushing,
which appears preferentially to remove (or agglomerate) fine particles from the
paved road surface. Broom sweeping is generally regarded as the least effec-
tive of the common control techniques, because the mechanical sweeping process
is inefficient in removing silt from the road surface.
To achieve control efficiencies on the order of 50 percent on a paved road
with moderate traffic ( 500 vehicles per day) requires cleaning of the surface
at least twice per week.^ This is because of the characteristically rapid
buildup of road surface material from spillage and the tracking and deposition
of material from adjacent unpaved surfaces, including the shoulders (berms) of
the paved road. Because industrial paved roads usually do^not have curbs, it
is important that the width of the paved road surface be sufficient for vehicles
to pass without excursion onto unpaved shoulders. Equation 1 indicates that
elimination of vehicle travel on unpaved or untreated shoulders would effect a
major reduction in particulate emissions. An even greater effect, by a factor
of 7, would result from preventing travel from unpaved roads or parking lots
onto the paved road of interest.
References for Section 11.2.6
1. R. Bohn, et al., Fugitive Emissions from Integrated Iron and Steel Plants,
EPA-600/2-78-050, U. S. Environmental Protection Agency, Research Triangle
Park, NC, March 1978.
2. C. Cowherd, Jr., et al., Iron and Steel Plant Open Dust Source Fugitive
Emission Evaluation, EPA-600/2-79-103, U. S. Environmental Protection
Agency, Research Triangle Park, NC, May 1979.
3. R. Bohn, Evaluation of Open Dust Sources in the Vicinity of Buffalo,
New York, U. S. Environmental Protection Agency, New York, NY, March 1979.
4. T. Cuscino, Jr., et al., Iron and Steel Plant Open Source Fugitive Emis-
sion Control Evaluation, EPA-600/2-83-110, U. S. Environmental Protection
Agency, Research Triangle Park, NC, October 1983.
5. J. Patrick Reider, Size Specific Particulate Emission Factors for Uncon-
trolled Industrial and Rural Roads, EPA Contract No. 68-02-3158, Midwest
Research Institute, Kansas City, MO, September 1983.
11.2.6-4 EMISSION FACTORS 9/85
C-5
-------
6. C. Cowherd, Jr., and P. Englehart, Size Specific Partlculate Emission
Factors for Industrial and Rural Roads, EPA-600/7-85-038, U. S. Environ-
mental Protection Agency, Research Triangle Park, NC, September 1985.
9/85 Miscellaneous Sources n ? 6-5
C-6
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing}
. REPORT NO^
EPA-600/7-85-051
2.
3. RECIPIENT'S ACCESSION-NO.
4. TITLE AND SUBTITLE
Size Specific Particulate Emission Factors for
Industrial and Rural Roads; Source Category Report
5. REPORT DATE
October 1985
6. PERFORMING ORGANIZATION CODE
, AUTHOR(S)
Chatten Cowherd, Jr. and Phillip J. Englehart
8. PERFORMING ORGANIZATION REPORT NO.
. PERFORMING ORGANIZATION NAME AND ADDRESS
Midwest Research Institute
425 Volker Boulevard
Kansas City, Missouri 64110
10. PROGRAM ELEMENT NO.
11. CONTRACT/GRANT NO.
68-02-3158, Task 12
12. SPONSORING AGENCY NAME AND ADDRESS
EPA, Office of Research and Development
Air and Energy Engineering Research Laboratory
Research Triangle Park, NC 27711
13. TYPE OF REPORT AND PERIOD COVERED
Task Final; 6/81- 6/85
14. SPONSORING AGENCY CODE
EPA/600/13
15. SUPPLEMENTARY NOTES
2429.
project officer is Dale L. Harmon, Mail Drop 61, 919/541-
is. ABSTRACT Tne repOrt gives results of a study to derive size-specific particulate emis-
sion factors for industrial paved and unpaved roads and for rural unpaved roads from
an existing field testing data base. Regression analysis was used to develop predic-
tive emission factor equations which relate emission quantities to road and traffic
parameters. Separate equations were developed for each road type and for three
aerodynamic particle size fractions: < or = 15, < or = 10, and < or = 2. 5 micro-
meters. Recommendations are made for including the resulting emission factors
in EPA document AP-42. Over the past few years, traffic-generated dust emissions
from unpaved and paved industrial roads have become recognized as a significant
source of atmospheric particulate emissions, especially within industries involved in
mining and processing mineral aggregates. Although a considerable amount of field
testing of industrial roads has been performed, most studies have focused on total
suspended particulate (TSP) emissions, because the current national ambient air
quality standards (NAAQS) for particulate matter are based on TSP. Only recently,
in anticipation of a NAAQS for particulate matter based on particle size, has the
emphasis shifted to the development of size-specific emission factors.
7.
KEY WORDS AND DOCUMENT ANALYSIS
a.
DESCRIPTORS
O.IDENTIFIERS/OPEN ENDED TERMS
c. COSATI Field/Group
Pollution
Roads
Industries
Rural Areas
Dust
Emission
Size Separation
Regression Analysis
Pollution Control
Stationary Sources.
Industrial Roads
Rural Roads
Particulate
Emission Factors
13B
05C
05K
11G
14G
07A.13H
12 A
13,
. DISTRIBUTION STATEMENT
Release to Public
19. SECURITY CLASS (This Report)
Unclassified
21. NO. OF PAGES
66
20. SECURITY CLASS (Thispage)
Unclassified
22. PRICE
EPA Form 2220-1 (9-73)
C-7
------- |