Emission Factor Documentation for AP-42,
Section 13.2.1
Paved Roads
Monitoring Policy Group
Office of Air Quality Planning and Standards
U.S. Environmental Protection Agency
June 2010
-------
Emission Factor Documentation for AP-42,
Section 13.2.1
Paved Roads
Monitoring Policy Group
Office of Air Quality Planning and Standards
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
June 2010
-------
NOTICE
This document is a draft product. However, it should not be construed to represent Agency
policy. It is has been circulated for comments on its technical merit.
-------
CONTENTS
Preface iii
List of Figures vii
List of Tables vii
1. Introduction 1-1
2. Source Description 2-1
2.1 Public and industrial roads 2-1
2.2 Review of current paved road emission factors 2-2
3. General Data Review and Analysis 3-1
3.1 Literature search and screening 3-1
3.2 Emission data quality rating system 3-2
3.3 Emission factor quality rating system 3-4
3.4 Methods of emission factor determination 3-5
3.5 Emission factor quality rating scheme used in
this study 3-8
4. AP-42 Section Development 4-1
4.1 Revisions to section narrative 4-1
4.2 Pollutant emission factor development 4-2
4.3 Development of other material in AP-42 section 4-23
5. Draft AP-42 Section 5-1
6. References 6-1
IV
-------
LIST OF FIGURES
Number
4-1
4-2
4-3
4-4
Final data set
Validation data from Test Report I
Correlation and regression results for the data set....
Cumulative frequency distribution obtained during
cross-validation study
Page
.4-17
.4-19
.4-21
.4-24
LIST OF TABLES
Number
3-1 Quality rating scheme for single-valued emission factors
3 -2 Quality rating scheme for emission factors equations
4-1 Applicable test reports
4-2 Summary information for Test Report I
4-3 Summary information for Test Report II
4-4 Summary information for Test Report III
4-5 Recommended emission factor models
4-6 Results of cross-validation study
4-7 Results from independent application of the PM-10 model.
4-8 Decision rule for paved road emission estimates
4-9 Ratio of predicted to measured PM-10 emission factors
Page
.3-12
.3-13
...4-3
...4-5
...4-8
.4-11
.4-22
.4-23
.4-25
.4-25
.4-27
-------
-------
SECTION 1
INTRODUCTION
The document "Compilation of Air Pollutant Emissions Factors" (AP-42) has been
published by the U.S. Environmental Protection Agency (EPA) since 1968. Supplements to AP-
42 have been routinely published to add new emission source categories and to update existing
emission factors. AP-42 is periodically updated by EPA to respond to new emission factor needs
of EPA, State, and local air pollution control programs and industry.
An emission factor relates the quantity (weight) of pollutants emitted to a unit of activity
of the source. The uses for the emission factors reported in AP-42 include:
1. Estimates of area-wide emissions.
2. Estimates of emissions for a specific facility.
3. Evaluation of emissions relative to ambient air quality.
The purpose of this report is to compile the existing background report and supplements
into a single report, provide an update of the background information from test reports and other
information to support preparation of a revised AP-42 section to replace existing Section 13.2.1,
"Paved Roads," dated November 2006.
The principal pollutant of interest in this report is "paniculate matter" (PM), with special
emphasis placed on "PMio"—paniculate matter no greater than 10 umA (micrometers in
1-1
-------
aerodynamic diameter) and PM2 5. PMi0 and PM2 5 form the basis for the current National
Ambient Air Quality Standards (NAAQSs) for paniculate matter.
PMio and PM25 thus represent the two size ranges of paniculate matter that are of
greatest regulatory interest. Nevertheless, formal establishment of PMio and PM2 5 as the
standard basis is relatively recent, and many emission tests have referenced other particle size
ranges. Other size ranges employed in this report are:
TSP Total Suspended Paniculate, as measured by the standard high-volume
(hi-vol) air sampler. TSP was the basis for the previous NAAQSs for paniculate matter.
TSP consists of a relatively coarse particle size fraction. While the particle capture
characteristics of the hi-vol sampler are dependent upon approach wind velocity, the
effective D50 (i.e., 50% of the particles are captured and 50% are not) varies roughly
from 25 to 50 umA.
SP Suspended Paniculate, which is used as a surrogate for TSP. Defined as
PM no greater than 30 umA. SP also may be denoted as "PM30."
IP Inhalable Paniculate, defined as PM no greater than 15 umA. Throughout
the late 1970s and the early 1980s, it was clear that EPA intended to revise the NAAQSs
to reflect a particle size range finer than TSP. What was not clear was the size fraction
that would be eventually used, with values between 7 and 15 umA frequently mentioned.
Thus, many field studies were conducted using IP emission measurements because it
was believed that IP would be the basis for the new NAAQS. IP may also be represented
by "PM15."
FP Fine Paniculate, defined as PM no greater than 2.5 umA. FP also may be
denoted as "PM2.5."
1-2
-------
This background report consists of five sections. Section 1 provides an introduction to
the report. Section 2 presents descriptions of the paved road source types and emissions from
those sources as well as a brief history of the current AP-42 emission factors. Section 3 is a
review of emissions data collection and analysis procedures; it describes the literature search, the
screening of emission test reports, and the quality rating system for both emission data and
emission factors. Section 4 details the development of paved road emission factors for the draft
AP-42 section; it includes the review of specific data sets and the results of data analysis.
Section 5 presents the AP-42 section for paved roads.
1-3
-------
-------
SECTION 2
SOURCE DESCRIPTION
Paniculate emissions occur whenever vehicles travel over a paved surface, such as public
and industrial roads and parking lots. These emissions may originate from material previously
deposited on the travel surface, resuspension of material carried by the vehicle, deposits from
undercarriages, engine exhaust gases or tire and brake wear. Depending on the road surface
characteristics, vehicle mix, the most significant emissions may arise from the surface material
loading (measured as mass of material per unit area), or a combination of engine exhaust, brake
and tire emissions. Surface loading is in turn replenished by other sources (e.g., pavement wear,
deposition of material from vehicles, deposition from other nearby sources, carryout from
surrounding unpaved areas, and litter). Because of the importance of the surface loading,
available control techniques either attempt to prevent material from being deposited on the
surface or to remove (from the travel lanes) any material that has been deposited.
2.1 PUBLIC AND INDUSTRIAL ROADS
While the mechanisms of particle deposition and resuspension are largely the same for
public and industrial roads, there can be major differences in surface loading characteristics,
emission levels, traffic characteristics, and viable control options. For the purpose of estimating
paniculate emissions and determining control programs, the distinction between public and
industrial roads is not a question of ownership but rather a question of surface loading and traffic
characteristics.
Although public roads generally tend to have lower surface loadings than industrial
roads, the fact that these roads have far greater traffic volumes may result in a substantial
contribution to the measured air quality in certain areas. In addition, public roads in industrial
areas can be often heavily loaded and traveled by heavy vehicles. In that instance, better
emission estimates might be obtained by treating these roads as industrial roads through the use
2-1
-------
of a silt loading and average vehicle weight appropriate for the road segment. In extreme cases,
public roads, industrial road, or parking lots may have such a high surface loadings that the
paved surface is covered with loose material and in extreme cases is mistaken for an unpaved
surface. In that event, use of a paved road emission factor may actually result in a higher
estimate than that obtained from the unpaved road emission factor, and the road is better
characterized as unpaved in nature rather than paved.
2.2 REVIEW OF PAST AND CURRENT PAVED ROAD EMISSION FACTORS
2.2.1 September 1985 through January 1995.
From September 1985 through January 1995, AP-42 currently contained two sections
concerning paved road fugitive emissions. The first, Section 11.2.5, is entitled "Urban Paved
Roads" and was first drafted in 1984 using test results from public paved roads.2 Emission
factors are given in the form of the following equation:
E = k (sL/0.5)p (2-1)
where: E = paniculate emission factor (g/VKT)
s = surface material content silt, defined as particles < 75 um in
diameter (%)
L = surface material loading, defined as mass of particles per unit area
of the travel surface (g/m2)
k = base emission factor (g/VKT)
p = exponent (dimensionless)
The factors k and p are given by
E
0.9
0.8
0.8
0.6
Particle
size fraction
TSP
PM-15
PM-10
PM-2.5
k (g/VKT)
5.87
2.54
2.28
1.02
2-2
-------
The form of the emission factor model is reasonably consistent throughout all particle size
fractions of interest.
The urban paved road emission factors represented by Equation 2-1 did not change since
their inclusion in the 4th Edition (September 1985) and the January 1995 revision. It should be
noted that these emission factors were not quality rated "A" through "E." (See Section 3 for an
overview of the AP-42 quality rating scheme.)
Section 1 1.2.6, "Industrial Paved Roads," was first published in 19833 and was slightly
modified in Supplement B (1988) to the 4th Edition. Section 1 1.2.6 contained three distinct sets
of emission factor models as described below.
°-7 (2.2)
For TSP, the following equation is recommended:
where: E = emission factor (kg/VKT)
I = industrial augmentation factor (dimensionless)
n = number of traffic lanes (dimensionless)
s = surface material silt content (%)
L = surface material loading across all traffic lanes (kg/km)
W = average vehicle weight (Mg)
The basic form of Equation 2-2 dates from a 1979 report4 and was originally included in
Supplement 14 to AP-42 (May 1983). The version used in AP-42 was slightly revised in that the
leading term (i.e., 0.022 in Eq. [2-2]) was reduced by 14%. The industrial road augmentation
factor (I) was included to take into account for higher emissions from industrial roads than from
urban roads; it varied from 1 to 7. The emission factor equation was rated "B" for cases with I =
1 and "D" otherwise.
For smaller particle size ranges, models somewhat similar to those in Eq. (2-1) were
recommended:
E = k (sL/12)0'3 (2-3)
where: E = emission factor (kg/VKT)
2-3
-------
k = base emission factor (kg/VKT), see below
sL = road surface silt loading (g/m2)
The base emission factor (k) above varied with aerodynamic size range as follows:
Particle
size fraction
PM-15
PM-10
PM-2.5
k (g/VKT)
0.28
0.22
0.081
These models represented by Equation 2-3 were first developed in 19843 from 15 emission tests
of uncontrolled paved roads and they were rated "A."
During the development of Eq. (2-3), tests of light-duty traffic on heavily loaded road
surfaces were identified as a separate subset, for which separate single-valued emission factors
were developed. Section 11.2.6 recommended the following for light-duty (less than 4 tons)
vehicles traveling over roads where the surface material was dry and the road was heavily loaded
(silt loading greater than 15 g/m2):
E = k (2-4)
where: E = emission factor (kg/VKT)
k = single-valued factor depending on particle size range of interest
(see below)
Particle
size fraction
PM-15
PM-10
k (g/VKT)
0.12
0.093
The single-valued emission factors was quality rated "C."
2-4
-------
During the time that AP-42 had four methods for estimating emissions from paved roads
(Sections 11.2.5 and 11.2.6), users of AP-42 noted difficulty selecting the appropriate emission
factor model to use in their applications.5'6'7 For example, inventories of industrial facilities
(particularly of iron and steel plants) conducted throughout the 1980s yielded measured silt
loading values substantially lower than those in the Section 11.2.6 data base. In extreme cases
when the models were used with silt loading values outside the range for which they were
developed, estimated PM-10 emission factors were larger than the corresponding TSP emission
factors.
Furthermore, the distinction between "urban" and "industrial" paved roads was blurred.
For the purpose of estimating emissions, it was gradually realized that source emission levels are
not a question of ownership but rather a question of surface loading and traffic characteristics.
Confirmatory evidence was obtained in a 1989 field program5 which found that paved roads at
an iron and steel facility far more closely resembled "urban" roads rather than "industrial" roads
in terms of emission characteristics.
Finally, it was unknown how well the emission factors of that time performed for cases
of increased surface loading on public roads, such as after application of antiskid materials or
within areas of trackout from unpaved areas.6 These situations were of considerable interest to
several state and local regulatory agencies, most notably in the western United States.
2.2.2 January 1995 through October 2002
The January 1995 update attempted to correct as many of the shortcomings of the
previous versions as possible. To that end, the update employed an approach slightly different
than that used in the past. In addition to reviewing test data obtained since the September 1988
update,8 the test data used for both of the 1988 sections were also included for reexamination in
the final data set. In assembling the data base, no distinction was made between public and
industrial roads or between controlled and uncontrolled tests, with the anticipation that the
reformulated emission factor will be applicable over a far greater range of source conditions.
2-5
-------
The inclusion of controlled tests represented a break with EPA previous guidelines for
preparing AP-42 sections9. Those guidelines presented a clear preference that only uncontrolled
tests be used to develop an emission factor. However, the principal control measures for paved
roads seek to reduce the value of an independent variable in the emission factor equation, i.e., the
silt loading.
The revised emissions factor equation published in the January 1995 update of the paved
road section included silt loading, average vehicle weight and a particle size multiplier as
independent variables. The resulting equation was:
. .0.65 . .1.5
E = k (SL / 2} (w / 3)
where: E = paniculate emission factor (having units matching the units of k),
k = particle size multiplier for particle size range and units of interest (see
below),
sL = road surface silt loading (grams per square meter) (g/m2), and
W = average weight (tons) of the vehicles traveling the road.
The selection of the value for the independent variable for the particle size multiplier was based
upon the units of the emissions factor desired and the size range for the emissions.
Particle Size Multipliers for Paved Road Equation
Size Range
PM25
PM10
PM15
PM30
Multiplier k
g/VKT
2.1
4.6
5.5
24
g/VMT
3.3
7.3
9.0
38
Ib/VMT
0.0073
0.016
0.020
0.082
2.2.3 October 2002 through December 2003
2-6
-------
Prior to October 2002, the basis of the particle sizing information for paved roads
emissions factors was high volume sampler impactors data. While the initial particle sizing was
performed by cyclones, subsequent particle sizing was performed by slotted impactors. The
impactor data had biases created by particle bounce and reintrainment. As such particle sizing
below 10 um was questioned. In October 2002, a three city paved and unpaved road emissions
study was completed that evaluated particle sizing at 10 and 2.5 um and assessed the default
values for silt loading. The results of the three city study formed the basis for revising the PM2 5
particle size multiplier k from 2.1 g/VKT (3.3 g/VMT or 0.0073 Ib/VMT) to 1.1 g/VKT(1.8
g/VMT or 0.0040 Ib/VMT). The form of the predictive equation and the exponents for silt
loading and average vehicle weight were unchanged. The changes in the October 2002 revision
provided recommended default silt loading data for normal and worst case public paved roads
based upon the updated silt loading values for public paved roads. The remaining numerical
revisions that were made in the emissions factor for paved roads included an adjustment for the
normal mitigation effects due to rain events. For long term average conditions, a 25% reduction
in the paniculate emissions was included for every day that there was measureable rain for that
day. A similar adjustment was included that used hourly time intervals rather that a daily time
interval.
2.2.4 December 2003 through November 2006
The December 2003 revision of the AP-42 Section for paved roads incorporated a
constant in the predictive equation for paniculate emissions factors. The AP-42 equations prior
to December 2003 estimated PM emissions from re-entrained road dust, and vehicle exhaust,
brakewear and tirewear emissions. In the December 2003 revision of the section, the component
of emissions due to exhaust, brakewear and tirewear were separated from the composite fugitive
dust emission factor equation. The first stated reason for the separation was to eliminate the
possibility of double counting emissions. With the introduction of EPA'sMobile6.2 model,
estimates of PM emissions from exhaust, brakewear and tirewear were calculated based upon the
vehicle mix, vehicle speed and road class. The double counting of emissions was a possibility
when both the fugitive dust emission factors from AP-42 and Mobile6.2 were used to estimate
emissions from vehicle traffic on paved roads. The second stated reason was to incorporate
decreases in paniculate matter emissions from the exhaust of newer vehicle models and fuel
2-7
-------
sources. Since the majority of data supporting the paved road emission factor equation was
developed at the time prior to when the vehicles in the fleet incorporated significant reductions
of paniculate matter emissions. A technical memorandum provided the basis for estimating PM
emissions due to exhaust, break wear and tire wear. The technical memorandum used estimated
emissions from a 1980's model year vehicle fleet since the emissions tests supporting the
emissions factors equation were performed in the early 1980's to early 1990's. It was believed
that since 1980, there have been and will continue to be improvements in vehicles and fuel that
will result in a decrease in PM emissions from engine exhaust. Depending on the emissions
factors units desired, the constant that was included in the emissions factor equation had values
of 0.2119 g/VKT, 0.1317 g/VMT or 0.00047 Ib/VMT forPM30, PM15 andPM10 emissions. For
PM2.5 emissions, depending on the required emissions factors units, the constant used in the
equation had values of 0.1617 g/VKT, 0.1005 g/VMT or 0.00036 Ib/VMT.
2.2.5 November 2006 through May 2010
In November 2006, the particle size multiplier k was lowered to 0.66 g/VKT, 1.1 g/VMT
or 0.0024 depending on the needed units for the emissions factor. The revision was based upon a
broad based assessment of the biases associated with the cyclone/impactor method for paniculate
sizes less than 10 um in aerodynamic diameter. While the December 2003 update revised the
particle size multiplier, the update was based upon limited test data. In addition, the impact of
biased emissions factor ratios for PM2 5 impacted fugitive sources other than paved roads. The
impact was due to particle bounce from the cascade impactor stages to the backup filter
potentially inflating PM2 5 concentrations. The impact was possible even though steps were
taken to minimize particle bounce in the earlier studies. The assessment study was sponsored by
the Western Regional Air Partnership and conducted by the Midwest Research Institute (MRI).
The testing was conducted at MRFs Aerosol Test Facility (ATF) in Deramus Field Station in
Grandview, Missouri using surface dust collected from seven locations in five western states.
The tests provided the basis for comparing the average PM2 5 concentration and the collocated
PMio concentration. The study compared the fine fraction ratios derived from FRM samplers to
those derived from the cyclone/impactor method. The cyclone/impactor samplers and operating
2-8
-------
method used in the study were the same as those that generated the original AP-42 emission
factors and associated PM2 5 /PMio ratios. The study consisted of 100 test runs covering PMi
concentration from approximately 0.3 mg/m3 to 7 mg/m3.
2.2.6 May 2010
This update recommends an updated equation for paved roads that is based upon
additional test data that was conducted on roads with slow moving traffic and stop and go traffic.
The emissions tests were performed for the Corn Refiners Association by Midwest Research
Institute (MRI). The testing focused on PMIO emissions at four corn processing facilities.
2-9
-------
-------
SECTION 3
GENERAL DATA REVIEW AND ANALYSIS
To reduce the amount of literature collected to a final group of references from which
emission factors could be developed, the following general criteria were used:
1. Emissions data must be from a primary reference:
a. Source testing must be from a referenced study that does not reiterate information
from previous studies.
b. The document must constitute the original source of test data. For example, a
technical paper was not included if the original study was contained in the previous document.
If the exact source of the data could not be determined, the document was eliminated.
2. The referenced study must contain test results based on more than one test run.
3. The report must contain sufficient data to evaluate the testing procedures and
source operating conditions.
A final set of reference materials was compiled after a thorough review of the pertinent
reports, documents, and information according to these criteria.
3-1
-------
3.1 LITERATURE SEARCH AND SCREENING
Review of available literature identified three paved road testing programs (presented
later as Table 4-1) since the time of the last Section 11.2 update.8 The individual programs are
discussed in detail in the next section. In addition, as discussed at the end of Section 2, earlier
controlled industrial road test data were reexamined. The previous update8 noted that Eq. (2-4)
yielded quite good estimates for emissions from vacuum swept and water flushed roads.
Furthermore, it became apparent that previous distinctions between "industrial" and "urban"
roads had become blurred as interest focused on heavily loaded urban roads (e.g., after snow/ice
controls) and on cleaner industrial roads (as the result of plant-wide control programs).
3.2 EMISSION DATA QUALITY RATING SYSTEM
As part of the analysis of the emission data, the quantity and quality of the information
contained in the final set of reference documents were evaluated. The following data are to be
excluded from consideration:
1. Test series averages reported in units cannot be converted to the selected
reporting units.
2. Test series representing incompatible test methods (i.e., comparison of EPA
Method 5 front-half with EPA Method 5 front- and back-half).
3. Test series of controlled emissions for which the control device is not specified.
4. Test series in which the source process is not clearly identified and described.
5. Test series in which it is not clear whether the emissions were measured before or
after the control device.
Test data sets that were not excluded were assigned a quality rating. The rating system
used was that specified by EPA for preparing AP-42 sections.9 The data were rated as follows:
A Multiple tests that were performed on the same source using sound methodology
and reported in enough detail for adequate validation. These tests do not
3-2
-------
necessarily conform to the methodology specified in EPA reference test methods,
although these methods were used as a guide for the methodology actually used.
B Tests that were performed by a generally sound methodology, but lack enough
detail for adequate validation.
C Tests that were based on an untested or new methodology or that lacked a
significant amount of background data.
D Tests that were based on a generally unacceptable method but may provide an
order-of-magnitude value for the source.
The following criteria were used to evaluate source test reports for sound methodology
and adequate detail:
1. Source operation. The manner in which the source was operated is well
documented in the report. The source was operating within typical parameters
during the test.
2. Sampling procedures. The sampling procedures conformed to a generally
acceptable methodology. If actual procedures deviated from accepted methods,
the deviations are well documented. When this occurred, an evaluation was made
of the extent such alternative procedures could influence the test results.
3. Sampling and process data. Adequate sampling and process data are documented
in the report, and any variations in the sampling and process operation are noted.
If a large spread between test results cannot be explained by information
contained in the test report, the data are suspect and were given a lower rating.
4. Analysis and calculations. The test reports contain original raw data sheets. The
nomenclature and equations used were compared to those (if any) specified by
EPA to establish equivalency. The depth of review of the calculations was
dictated by the reviewer's confidence in the ability and conscientiousness of the
tester, which in turn was based on factors such as consistency of results and
completeness of other areas of the test report.
3-3
-------
3.3 EMISSION FACTOR QUALITY RATING SYSTEM
The quality of the emission factors developed from analysis of the test data was rated
utilizing the following general criteria:
A—Excellent: Developed only from A-rated test data taken from many randomly chosen
facilities in the industry population. The source category is specific enough so that
variability within the source category population may be minimized.
B—Above average: Developed only from A-rated test data from a reasonable number of
facilities. Although no specific bias is evident, it is not clear if the facilities tested
represent a random sample of the industries. The source category is specific enough so
that variability within the source category population may be minimized.
C—Average: Developed only from A- and B-rated test data from a reasonable number of
facilities. Although no specific bias is evident, it is not clear if the facilities tested
represent a random sample of the industry. In addition, the source category is specific
enough so that variability within the source category population may be minimized.
D—Below average: The emission factor was developed only from A- and B-rated test
data from a small number of facilities, and there is reason to suspect that these facilities
do not represent a random sample of the industry. There also may be evidence of
variability within the source category population. Limitations on the use of the emission
factor are noted in the emission factor table.
E—Poor: The emission factor was developed from C- and D-rated test data, and there is
reason to suspect that the facilities tested do not represent a random sample of the
industry. There also may be evidence of variability within the source category
population. Limitations on the use of these factors are always noted.
The use of these criteria is somewhat subjective and depends to an extent on the
individual reviewer.
3.4 METHODS OF EMISSION FACTOR DETERMINATION
3-4
-------
Fugitive dust emission rates and particle size distributions are difficult to quantify
because of the diffuse and variable nature of such sources and the wide range of particle size
involved including particles which deposit immediately adjacent to the source. Standard source
testing methods, which are designed for application to confined flows under steady state,
forced-flow conditions, are not suitable for measurement of fugitive emissions unless the plume
can be draw into a forced-flow system. The following presents a brief overview of applicable
measurement techniques. More detail can be found in earlier AP-42 updates.8'10
3.4.1 Mass Emission Measurements
Because it is usually impractical to enclose open dust sources or to capture the entire
emissions plume, only the upwind-downwind and exposure profiling methods are suitable for
measurement of paniculate emissions from most open dust sources.10 These two methods are
discussed separately below.
The basic procedure of the upwind-downwind method involves the measurement of
paniculate concentrations both upwind and downwind of the pollutant source. The number of
upwind sampling instruments depends on the degree of isolation of the source operation of
concern (i.e., the absence of interference from other sources upwind). Increasing the number of
downwind instruments improves the reliability in determining the emission rate by providing
better plume definition. In order to reasonably define the plume emanating from a point source,
instruments need to be located at two downwind distances and three crosswind distances, at a
minimum. The same sampling requirements pertain to line sources except that measurement
need not be made at multiple crosswind distances.
Net downwind (i.e., downwind minus upwind) concentrations are used as input to
dispersion equations (normally of the Gaussian type) to backcalculate the paniculate emission
rate (i.e., source strength) required to generate the pollutant concentration measured. Emission
factors are obtained by dividing the calculated emission rate by a source activity rate (e.g.,
number of vehicles, or weight of material transferred per unit time). A number of meteorological
3-5
-------
parameters must be concurrently recorded for input to this dispersion equation. At a minimum
the wind direction and speed must be recorded on-site.
While the upwind-downwind method is applicable to virtually all types of sources, it has
significant limitations with regard to development of source-specific emission factors. The
major limitations are as follows:
1. In attempting to quantify a large area source, overlapping of plumes from upwind
(background) sources may preclude the determination of the specific contribution
of the area source.
2. Because of the impracticality of adjusting the locations of the sampling array for
shifts in wind direction during sampling, it cannot be assumed that plume position
is fixed in the application of the dispersion model.
3. The usual assumption that an area source is uniformly emitting does not allow for
realistic representation of spatial variation in source activity.
4. The typical use of uncalibrated atmospheric dispersion models introduces the
possibility of substantial error (a factor of three according to Reference 11) in the
calculated emission rate, even if the stringent requirement of unobstructed
dispersion from a simplified (e.g., constant emission rate from a single point)
source configuration is met.
The other measurement technique, exposure profiling, offers distinct advantages for
source-specific quantification of fugitive emissions from open dust sources. The method uses
the isokinetic profiling concept that is the basis for conventional (ducted) source testing. The
passage of airborne pollutant immediately downwind of the source is measured directly by
means of simultaneous multipoint sampling over the effective cross section of the fugitive
emissions plume. This technique uses a mass-balance calculation scheme similar to EPA
Method 5 stack testing rather than requiring indirect calculation through the application of a
generalized atmospheric dispersion model.
3-6
-------
For measurement of nonbuoyant fugitive emissions, profiling sampling heads are
distributed over a vertical network positioned just downwind (usually about 5 m) from the
source. If total paniculate emissions are to be measured, sampling intakes are pointed into the
wind and sampling velocity is adjusted to match the local mean wind speed, as monitored by
anemometers distributed over height above ground level.
The size of the sampling grid needed for exposure profiling of a particular source may be
estimated by observation of the visible size of the plume or by calculation of plume dispersion.
Grid size adjustments may be required based on the results of preliminary testing. Paniculate
sampling heads should be symmetrically distributed over the concentrated portion of the plume
containing about 90% of the total mass flux (exposure). For example, assuming that the
exposure from a point source is normally distributed, the exposure values measured by the
samplers at the edge of the grid should be about 25% of the centerline exposure.
To calculate emission rates using the exposure profiling technique, a conservation of
mass approach is used. The passage of airborne paniculate (i.e., the quantity of emissions per
unit of source activity) is obtained by spatial integration of distributed measurements of exposure
(mass/area) over the effective cross section of the plume. The exposure is the point value of the
flux (mass/area/time) of airborne paniculate integrated over the time of measurement.
3.4.2 Emission Factor Derivation
Emissions factors are typically derived from the ratio of the emissions to an activity
level. It is assumed that the emissions are linearly proportional to the selected activity level.
Usually the final emission factor for a given source operation, is the arithmetic average of the
individual emission factors calculated from each test of that source type. In rare instances, the
range of individual emission factor values is also presented.
3-7
-------
As an improvement over the presentation of a final emission factor as a single-valued
arithmetic mean, an emission factor may be presented in the form of a predictive equation
derived by regression analysis of test data. The use of a predictive equation with a relatively
good correlation coefficient (R2) provides a means for improving the accuracy of the emissions
factor in estimating the actual emissions when the independent variables are known. Such an
equation mathematically relates emissions to parameters when characterize source conditions.
These parameters may be grouped into three categories:
1. Measures of source activity or energy expended (e.g., the speed and weight of a
vehicle traveling on an unpaved road).
2. Properties of the material being disturbed (e.g., the content of suspendable fines
in the surface material on an unpaved road).
3. Climatic parameters (e.g., number of precipitation-free days per year on which
emissions tend to be at a maximum).
An emission factor equation is useful if it is successful in "explaining" much of the observed
variance in emission factor values on the basis of corresponding variance sin specific source
parameters. This enables more reliable estimates of source emissions on a site-specific basis.
A generic emission factor equation is one that is developed for a source operation defined
on the basis of a single dust generation mechanism which crosses industry lines. An example
would be vehicular traffic on unpaved roads. To establish its applicability, a generic equation
should be developed from test data obtained in different industries.
3.5 EMISSION FACTOR QUALITY RATING SCHEME USED IN THIS STUDY
The uncontrolled emission factor quality rating scheme used in this study is somewhat
different than was used in earlier updates8'11 of this section and represents a refinement of the
rating system developed by EPA for AP-42 emission factors, as described in Section 3.3. The
scheme entails the use of the same rating assessment of source test data quality followed by an
3-8
-------
initial rating assessment of the emission factor(s) based on the number and quality of the
underlying source test data.
Test data that were developed from well documented, sound methodologies were
assigned an A rating. Data generated by a methodology that was generally sound but either did
not meet a minimum test system requirements or lacked enough detail for adequate validation
received a B rating.
In evaluating whether an upwind-downwind sampling strategy qualified as a sound
methodology, the following minimum test system requirements were used. At least five
paniculate measuring devices must be operated during a test, with one device located upwind
and the other located at two downwind and three crosswind distances. The requirement of
measurements at crosswind distances is waived for the case of line sources. Also wind direction
and speed must be monitored concurrently on-site.
The minimum requirements for a sound exposure profiling program were the following.
A one-dimensional, vertical grid of at least three samplers is sufficient for measurement of
emissions from line or moving point sources while a two-dimensional array of at least five
samplers is required for quantification of fixed virtual point source missions. At least one
upwind sampler must be operated to measure background concentration, and wind speed must be
measured on-site.
Neither the upwind-downwind nor the exposure profiling method can be expected to
produce A-rated emissions data when applied to large, poorly defined area sources, or under very
light and variable wind flow conditions. In these situations, data ratings based on degree of
compliance with minimum test system requirements were reduced one letter.
Following the assignment of the individual source test quality ratings, the factor quality
rating of the single-valued emission factor will be evaluated. Recently approximately 20 "A"
and "B" rated source test reports have been required to justify a factor quality rating of "A".
Each halving of the number of source test reports results in a one letter grade reduction in the
3-9
-------
final factor quality rating. Several of the source test reports used as the basis for the emissions
factor development include measurements conducted at different locations. To the extent that
there are more than two tests at the different locations and that the different locations within a
given reference represent differences in source conditions, each of the different source
conditions will be counted as an independent test. The development of the paved road emissions
factor differs from typical in that it includes the use of stepwise multiple non linear regression.
Following the initial factor quality rating, the adjusted correlation coefficient will be used to
increase the emissions factor quality rating. Only correlation coefficients above 0.4 will be used
to increase the emissions factor quality rating.
3-10
-------
SECTION 4
AP-42 SECTION DEVELOPMENT
4.1 REVISIONS TO SECTION NARRATIVE
The draft AP-42 presented later in this background document is intended to replace the
current version of Section 13.2.1 "Paved Roads" in AP-42. The last update of this section is
dated November 2006. The general form of the emissions factor equation presented in the paved
road section has been consistent since the January 1995 major revision. Since this date revisions
have been made addressing the influence of rain events, estimating default silt loading levels for
various classes of roads, separating paniculate emissions associated with the roads verses those
associated with the vehicles and addressing biases in the measurement of PM2.5 with devices
that use impactors to perform paniculate sizing.
4.2 POLLUTANT EMISSION FACTOR DEVELOPMENT
This update to Sections 13.2.1 is planned to address the application of the emissions
factor equation addressing only the component associated with paved road surface materials and
at speeds lower than 10 miles per hour. In order to achieve this goal, the following general
approach was taken
1. Assemble the available test data for paved roads in a single data base, making no
distinction between public and industrial roads or between controlled and
uncontrolled roads.
2. Develop PMio and PM2 5 engine, tire wear and brake ware emissions estimates for
each of the available data sets. For each of the available data sets, estimate the
emissions associated with the road surface material by subtracting the engine, tire
wear and brake wear from the measured PMIO emissions.
4-1
-------
2. Conduct a series of stepwise linear regression analyses of the revised and adjusted
data base to assess the most critical parameters and to develop an emission factor
model with:
• silt loading,
• mean vehicle weight, and,
• mean travel speeds
as potential correction parameters.
3. Conduct an appropriate validation study of the reformulated model.
4.2.1 Review of Specific Data Sets
4.2.1.1 Street Sanding Emissions And Control Study, PEI Associates, Inc., Cincinnati, OH,
October 1989. (Reference 15)
This test program was undertaken to characterize PM-10 emissions from six streets that
were periodically sanded for anti-skid control within the Denver area. The primary objective
was given as development of a predictive algorithm for clean and sanded streets, with a
secondary objective stated as defining the effectiveness of control measures. Summary
information is given in Table 4-1.
Sampling employed six to eight 8 PM-10 samplers equipped with volumetric flow
control. Samplers were arranged in two upwind/downwind configurations. The "basic"
configuration consisted of six samplers arranged in identical patterns upwind and downwind of
the test road, with one sampler and one pair of samplers at nominal distances of 20 and 5 m,
respectively, from the road.
The second configuration was used for tests of control measure effectiveness. The road
segment was divided into two halves, corresponding to the treated and experimental control
(untreated) portions. Identical sampling arrays were again used upwind and downwind on both
halves, at nominal distances of 20 and 5 m. Because this array employed all eight samplers
available, no collocation was possible for the second configuration.
4-2
-------
-------
TABLE 4-1. SUMMARY INFORMATION FOR TEST REPORT I
Operation
Vehicle traffic
Vehicle traffic
Vehicle traffic
Vehicle traffic
Vehicle traffic
Vehicle traffic
Location
Colfax
York St.
Belleview
1-225
Evans
Louisiana
State
Colorado
Colorado
Colorado
Colorado
Colorado
Colorado
Test dates
3-4/89
4/89
4/89
4/89
5-6/89
6/89
No. of tests
17
1
4
9
29
7
PMio emission
Geom. mean
1.33
1.07
1.62
0.31
1.06
0.96
factor (g/VKT)
Range
0.53-9.01
1.07
1.10-4.77
0.17-0.51
0.21-7.83
0.42-1.73
4-0
-------
In addition to the PM-10 concentration measurements, several other types of samples
were collected:
• Wind speed/direction and incoming solar radiation were collected on-site, and the
results were combined to estimate atmospheric stability class needed to calculate
emission factors.
• Colorado Air Pollution Control Division (APCD) representatives collected traffic data,
including traffic counts, travel speeds, and percentage of heavy-duty vehicles.
• Vacuums with disposable paper bags were used to collect the loose material from the
road surface. In addition to samples taken from the travel lanes, the field crew took
daily samples of material adjacent to curbs and periodic duplicate samples.
The study collected PM-10 concentration data on 24 different days and calculated a total
of 69 different emission rates for baseline, sanded and controlled paved road surfaces. Emission
factors were obtained by back-calculation from the CALINES dispersion model12 together with a
series of assumptions involving mixing widths and heights and an effective release height.
Although data collected at the 20 m distance were used to evaluate results, the test report did not
describe any sensitivity analysis to determine how dependent the emission rates were on the
underlying assumptions.
The testing program found difficulty in defining "upwind" concentrations for several of
the runs, including cases with wind reversals or winds nearly parallel to the roadway orientation.
A total of eight of the 69 tests required that either an average concentration from other test days
or a downwind concentration be used to define "upwind" conditions. In addition, the test report
described another seven runs as invalid for reasons such as wet road surfaces, nearby dust
sources or concentrations increasing with downwind distance.
A series of stepwise regression analyses were conducted, with different predictive
equations presented for (a) baseline conditions, (b) sanded roads, and (c) roads swept to remove
the sand applied, and (d) all conditions combined. In each case, only one independent variable
4-1
-------
was included in the predictive equation: silt loading, for cases (a) and (d); and time since
treatment, for (b) and (c).
In general, Reference 15 is reasonably well documented in terms of describing test
conditions, sampling methodology, data reduction and analysis. A chief limitation lies in the
fact that neither sampling configuration fully met minimum requirements for the upwind-
downwind method presented in Section 3.4. Specifically, only two or three samplers were used
downwind rather than the minimum of four.
Furthermore, a later report6 drawing upon the results from Reference 15 and 17
effectively eliminated 24% of the combined baseline tests because of wind directions. In
addition, the later report6 noted that the baseline data should be considered as "conservatively
high" because roughly 70% of the data were calculated assuming the most unstable atmospheric
class (which results in the highest backcalculated emission factor). Because of these limitations,
the emission data have been given an overall rating of "D."
4.2.1.2 RTF Environmental Associates 1990. Street Sanding Emissions and Control Study,
prepared for the Colorado Department of'Health. July 1990. (Reference 17)
This test program was quite similar to that described in Reference 15 cited in paragraph
4.2.1.1 and used an essentially identical methodology. In fact, the two test reports are very
similar in outline, and many passages in the two reports are identical. The primary objective was
given as expanding the data base in Reference 15 to further develop predictive algorithms for
clean and sanded streets. Summary information is given in Table 4-2.
The test program employed the same two basic PMio sampling arrays as did Reference
15. A third configuration was used for "profile" tests, in which additional samplers were placed
at 10 and 20 ft heights. (Analysis of results from elevated samplers is not presented in Reference
17.)
4-2
-------
TABLE 4-2. SUMMARY INFORMATION FOR REFERENCE 17
PM-10 emission factor (g/VKT)
Operation
Vehicle traffic
Vehicle traffic
Vehicle traffic
Vehicle traffic
Vehicle traffic
Vehicle traffic
Vehicle traffic
Vehicle traffic
Location
Mexico
State Hwy 36
Colfax
Park Rd.
Evans
Louisiana
Jewell
Bryon
State
Colorado
Colorado
Colorado
Colorado
Colorado
Colorado
Colorado
Colorado
Test dates
2/90
1-3/90
2-4/90
4/90
2-3/90
1,3/90
1/90
4/90
No. of test
3
13
41
11
11
9
1
3
Geom. mean
2.75
1.31
1.32
1.26
2.10
3.24
6.36
8.38
Range
1.08-6.45
0.14-4.18
0.27-5.04
0.69-3.33
0.87-7.27
1.40-5.66
6.36
5.53-14.72
4-3
-------
As was the case in Reference 15, additional samples were collected including:
• Wind speed/direction were collected on-site, and the results used in estimating
atmospheric stability class needed to calculate emissions factors. (Unlike
Reference 15, solar radiation measurements were not collected.)
• Traffic data, including traffic counts, travel speeds, and percentages of heavy-
duty vehicles were collected.
• Vacuums with disposable paper bags were used to collect the loose material from
the road surface. The program developed an extensive set of collocated samples
of material along the edges of the roadway.
The study collected PMio concentration data on 33 days and calculated a total of 131
different emission rates for baseline, sanded and controlled paved road surfaces. Emission
factors were obtained by back-calculation from the CALINES dispersion model12 together with
essentially the same assumptions as those in Reference 15. This report also noted the same
difficulty as Reference 15 in defining "upwind" concentrations in cases with wind reversals or
winds nearly parallel to the roadway orientation. Unlike Reference 15, however, this report does
not provide readily available information on how many tests used either an average
concentration from other test days or a downwind concentration to define "upwind" conditions.
Reference 6 does, however, describe seven tests as invalid because of filter problems or because
upwind concentrations were higher than downwind values.
As with the Reference 15 program, a series of stepwise regression analyses were
conducted. This test program combined data from Reference 15 and 17 and considered
predictive equations for (a) baseline conditions, (b) sanded roads, and (c) roads swept to remove
the sand applied, and (d) all conditions combined.
4-4
-------
Unlike Reference 15, however, Reference 17 appears to present silt loading values that
are based on wet sieving (see page 8 of the test report) rather than the dry sieving technique (as
described in Appendix E to AP-42) routinely used in fugitive dust tests. (MRI could not obtain
any clarifying information during telephone calls to the testing organization and the laboratory
that analyzed the samples.) Wet sieving disaggregates composite particles and results from the
two types of sieving are not comparable.
There is additional confusion over the silt loading values given in Reference 17 for
cleaning tests. Specifically, the same silt loading value is associated with both the treatment and
the experimental control. This point could not be clarified during telephone conversation with
the testing organization. Attempts to clarify using test report appendices were unsuccessful.
Two appendices appear to interchange silt loading with silt percentage. More importantly, it
could not be determined whether the surface sample results reported in Appendix D to Reference
17 pertain to treated or the experimental control segment, and with which emission rate a silt
loading should be associated.
Reference 17 contains substantial amounts of information, but is not particularly well
documented in terms of describing test conditions, sampling methodology, data reduction and
analysis. In addition, the same limitations mentioned in connection with Reference 15 are
equally applicable to Reference 17, as follows:
• not meeting the minimum number of samplers.
• numerous tests conducted under variable wind conditions.
• frequent use (70% to 80% of the tests) of the most unstable atmospheric stability class in
the CALINE 3 model which will result in the highest calculated emission rate.
Because of these limitations, emission rate data have been given an overall rating of "D."
Furthermore, the silt loading data in this report are considered suspect for reasons noted above.
4-5
-------
4.2.1.3. T. Cuscino, Jr., et al., Iron And Steel Plant Open Source Fugitive Emission Control
Evaluation, EPA 600/2 83 110, U. S. Environmental Protection Agency, Cincinnati,
OH, October 1983. (Reference 6)
This study evaluated paved road control techniques at two different iron and steel plants.
(See Tables 9 and 10 in Reference ]8J_ Data were quality rated as"A," and uncontrolled test
results were incorporated into the data base for Section 11.2.6 published in 1983. The only use
of the controlled test results, however, was the following addition to Section 11.2.6.4 in 1988:
"Although there are relatively few quantitative data on emissions from
controlled paved roads, those that are available indicate that adequate estimates
generally may be obtained by substituting controlled loading values into ..
[Equations (2-2) and (2-3)].... The major exception to this is water flushing
combined with broom sweeping. In that case, the equations tend to overestimate
emissions substantially (by an average factor of 4 or more)."
In the current update, the controlled emission factors have been used as part of the overall
data base to develop predictive models. Although PM-10 emission data are not specifically
presented in the report, appropriate values were previously developed by log-normal
interpolation of the PMi5 and PM2 5 factors.8
4.2.1.4 G. E. Muleski, Measurement of Fugitive Dust Emissions from Prilled Sulfur
Handling, Final Report, MRI Project No. 7995-L, Prepared for Gardinier, Inc., June
1984 (Reference 30)
This was first report identified to suggest that heavily loaded paved roads may be better
considered as unpaved in terms of emission estimates. The program produced three tests of
emissions from end-loader travel over paved surfaces. Two of the three tests were conducted on
very heavily loaded surface, while the third was on a cleaned paved surface. (See Tables 20 and
21 of the 1987 update.)8
Comment [RM1]: It is unclear which
reference this means and what the
tables state.
4-6
-------
No PM-10 emission factors were reported; results were presented for total paniculate
(TP) and suspended paniculate (SP, or PM-30). Data were quality rated "A" in the 1987 report.
Because no PM-10 data were given, Test Report 5 data were most directly useful as
independent data against which the TSP emission factor model (Eq. (2-2)) could be assessed.
This comparison showed generally good agreement between predicted and observed with
agreement becoming better as source conditions approached those in the underlying data base.
The 1987 update8 developed PM-10 emission factors based on information contained in
the test report. When compared to the single valued factors (Equation [2-4]), agreement for the
first two tests was within a factor of approximately two. The third test—that of the cleaned
surface—could not be used to assess the performance of either Eq. (2-1) or Eq. (2-3) because the
surface loading value could not be converted to the necessary units with information presented in
the report.
4.2.1.5 T. F. Eckle and D. L. Trozzo, Verification of the Efficiency of a Road-Dust Emission-
Reduction Program by Exposure Profile Measurement, Presented at EPA/AISI
Symposium on Iron and Steel Pollution Abatement, Cleveland, Ohio, October 1984.
(Reference 31)
This paper discussed the development of an exposure profiling system as well as an
evaluation of the effectiveness of a paved road vacuum sweeping program. Because no
reference is made to an earlier test report, this paper is considered to be the original source of the
test data. Although ten uncontrolled and five controlled tests are mentioned, test data are
reported only in terms of averages. (See Tables 24 and 25 in Reference 8.) Only TSP emission
factors are presented. Although data were obtained using a sound methodology, data were rated
"C" because of inadequate detail in the paper.
4-7
-------
Averaged data from Test Report 8 were used in an independent assessment of Eq. (2-2).
Although only average emission levels could be compared, the data suggested that TSP
emissions could be estimated within very acceptable limits.
4.2.1.6 Roadway Emissions Field Tests at U.S. Steel's Fairless Works, U.S. Steel
Corporation, Fairless Hills, PA, USX Purchase Order No. 146-0001191-0068, May
1990. (bref01_13s0201Janl995.pdf) (Reference 1)
This 1989 field program used exposure profiling to characterize emissions from paved
roads at an integrated iron and steel plant near Philadelphia, Pennsylvania, in November 1989.
In many respects, this program arose because of uncertainties with paved road emission factor
models used outside their range of applicability. During the preparation of an alternative
emission reduction ("bubble") plan for the plant, questions arose about the use of AP-42
equations and other EPA guidance13 in estimating roadway emissions involved in the emissions
trade. This program provided site-specific data to support the bubble plan. This testing program
also represented the first exposure profiling data to supplement the AP-42 paved road data base
since the 1984 revision. Site "C" was located along the main access route and had a mix of
light- and medium-duty vehicles. Site "E" was located near the southwest corner of the plant
and the traffic consisted mostly of plant equipment. Table 4-3 provides summary information
and Table 4-4 provides detailed information.
The program involved two paved road test sites. The first (site "C") was along the four-
lane main access route to the plant. Average daily traffic (ADT) had been estimated as more
than 4,000 vehicle passes per day, with most vehicles representative of "foreign" equipment (i.e.,
cars, pickups, and semi-trailers rather than plant haul trucks and other equipment). Site "E," on
the other hand, was located near the iron- and steel-making facilities and had both lower ADT
and heavier vehicles than site "C." The plant regularly vacuum swept paved roads, and two
cleaning frequencies (two times and five times per week) were considered during the test
program.
4-8
-------
Eight tests were conducted at Site C-l and four tests were conducted at Site E-2. The
paved road test sites were considered uncontrolled. The road width, moisture content, and mean
number of wheels were not reported. The test data are assigned an "A" rating. Table 4-3
presents summary information and Table 4-4 presents detailed test information. Warm wire
anemometers at two heights measured wind speed.
Depending on traffic characteristics of the road being tested, a 6 to 7.5 m high profiling
array was used to measure downwind mass flux. This array consisted of four or five total
paniculate sampling heads spaced at 1.5 m heights and was positioned at a nominal 5 m distance
downwind from the road. A high-volume sampler with a parallel-slot cascade impactor and a
cyclone preseparator (cutpoint of 15 umA) was employed to measure the downwind particle size
distribution, and a standard high-volume sampler was utilized to determine the downwind mass
fraction of total suspended paniculate matter (TSP). The height for downwind sizing devices
(2.2 m) was selected after review of prior test results. It approximated the height in a roadway
dust plume at which half the mass emissions pass above and half below. The upwind
(background) particle size distribution was determined with a high-volume cyclone/ impactor
combination. Warm wire anemometers at two heights measured wind speed.
Additional samples included:
• Average wind speeds at two heights and wind direction at one height were
recorded during testing to maintain isokinetic sampling.
• Traffic data, including traffic counts, travel speeds, and vehicle class were
recorded manually.
• Vacuums with disposable paper bags were used to collect the loose material from
the road surface.
The sampling equipment met the requirements of a sound exposure profiling
methodology specified in Section 3.4 so that the emission test data are rated "A." The test report
4-9
-------
presents emission factors for total paniculate (TP), total suspended paniculate (TSP) and PMio,
for the ten paved road emission tests conducted.
Reference 1 found that the emission factors and silt loadings more closely resembled
those in the "urban" rather than the "industrial" data base. That is to say, emissions agreed more
closely with factors estimated by the methods of September 1985 AP-42 Section 11.2.5 than by
methods in Section 11.2.6. Given the traffic rate of 4000 vehicles per day at Site "C," this
4-10
-------
TABLE 4-3. SUMMARY INFORMATION FOR REFERENCE 1
Operation
Vehicle traffic
Vehicle traffic
Vehicle traffic
Location
AU-X
(Unpaved road)
Paved road
Paved road
State
PA
PA
PA
Test
dates
11/89
11/89
11/89
No. of
tests
2
6
4
TSP emissior
Geom. mean
0.61
0.033
0.078
factor, Ib/VMT
Range
0.39-0.96
0.012-0.12
0.033-0.30
PM-10 emiss
Geom. mean
0.16
0.0095
0.022
ion factor, Ib/VMT
Range
0.14-0.18
0.0009-0.036
0.0071-0.036
1 Ib/VMT = 281.9 g/VKT.
TABLE 4-4. DETAILED INFORMATION FROM PAVED ROAD TESTS FOR REFERENCE 1
Test runs
AU-C-3
AU-C-4
AU-C-5
AU-C-6
AU-C-7
AU-C-8
AU-E-1
AU-E-2
AU-E-3
AU-E-4
PM-10
emission factor,
Ib/VMT
0.00497
0.0355
0.0337
0.00816C
0.000887
0.0174
0.00709
0.0234
0.0355
0.0199
Duration,
min
103
147
120
187
96
218
154
89
118
130
Meteorology
Temperature,
°F
50
63
62
39
42
40
43
44
41
41
Mean wind
speed, mph
12
11
14
14
12
15
12
13
9.3
9.3
Vehicle characteristics
No. of vehicle
passes
836
1057
963
685
703
779
210
373
330
364
Mean vehicle
weight, ton
5.5
6.0
3.9
6.2
3.0
2.0
12
5.1
2.6
2.6
Mean
vehicle
speed"
(27)
25
29
(27)
(27)
(27)
15
16
(15)
(15)
Silt
loading,
g/m2
0.42
0.52
0.23
0.23b
0.26b
0.1 5b
4.0
4.0
2.2
1.3
Silt, %
10
12
9.7
8.6
7.7
9.9
17
17
18
15
"Value in parentheses is the average speed measured for test road during the field exercise.
bTest conducted on a paved road surface vacuum-swept five times per week.
cMean TSP/TP or PM10/TP ratio applied.
1 Ib/VMT = 281.9 g/VKT.
1 g/m2 = 1.434 gr/ft2
4-11
-------
finding was not terribly surprising. What was far more surprising was that emissions at Site
"E" were also more "urban" than "industrial." Although the TSP and PMio models in Section
11.2.5 showed a slight tendency to underpredict, the Section 11.2.6 PMio model
overestimated measured emissions by at least an order of magnitude. The performance of the
industrial TSP model, on the other hand, was only slightly poorer than that for the urban TSP
model.
4.2.1.7 Midwest Research Institute, Paved Road Particulate Emissions - Source Category
Report, for U.S. EPA, July 1984. (bref02_13s0201Janl995.pdf)- Reference 2
This document reports the results of testing of paved roads conducted in 1980 at sites in
Kansas City, MO, St. Louis, MO, Tonganoxie, KS, and Granite City, IL. Paved road test sites
included commercial/industrial roads, commercial/residential roads, expressways, and a street in
a rural town. The expanded measurement program reported in this document was used to
develop emission factors for paved roads and focused on the following particle sizes: PMi5
(inhalable paniculate matter [IP]), PMio, and PM2 5.
Total airborne PM emissions were characterized using an exposure profiler containing
four sampling heads. High-volume samplers with size selective inlets (SSI) having a cutpoint of
15 umA were used to characterize upwind and downwind PM-15 concentrations. A high-
volume sampler with a SSI and a cascade impactor was also located downwind to characterize
particle size distribution within the PMi5 component. Upwind and downwind standard high-
volume samplers measured TSP concentrations. Warm wire anemometers at two heights
measured wind speed.
A total of 19 paved road emission tests were conducted in four cities. These included
four tests of commercial/industrial paved roads, ten tests of commercial/residential paved roads,
four expressway tests, and one test of a street in a rural town. Additionally, as part of this study,
81 dust samples were collected in 12 cities. The mean number of vehicle wheels was not
reported. The test data are assigned an A rating. Table 4-5 presents summary test data and
Table 4-6 presents detailed test information.
4-12
-------
TABLE 4-5. SUMMARY INFORMATION FOR REFERENCE 2
Operation
Commercial/
Industrial
Commercial/
Residential
Expressway
Rural Town
State
MO
MO,IL
MO
KS
Test
dates
2/80
2/80
5/80
3/80
No. of
tests
4
10
4
1
PM15 emission factor, Ib/VMT
Geom. mean
0.0078
0.0021
0.0004
0.031
Range
0.0036-0.013
0.0006-0.012
0.0002 - 0.0008
0.031
PM10 emission factor, Ib/VMT
Geom. mean
0.0068
0.0017
0.0004
0.025
Range
0.0034-0.011
0.0004 - 0.0093
0.0002 - 0.0007
0.025
PM2 5 emission factor, Ib/VMT
Geom. mean
0.0045
0.0011
0.0002
0.005
Range
0.0030 - 0.0063
0.0002 - 0.0037
0.0001 - 0.0003
0.005
1 Ib/VMT = 281.9 g/VKT.
4-13
-------
TABLE 4-6. DETAILED INFORMATION FOR PAVED ROAD TESTS FOR REFERENCE 2
Category
Commercial/Industrial
Commercial/Industrial
Commercial/Industrial
Commercial/Industrial
Commercial/Residential
Commercial/Residential
Commercial/Residential
Commercial/Residential
Commercial/Residential
Commercial/Residential
Commercial/Residential
Commercial/Residential
Commercial/Residential
Expressway
Expressway
Expressway
Expressway
Rural Town
Run test
No.
M-l
M-2
M-3
M-9
M-4
M-5
M-6
M-13
M-14
M-15
M-17
M-18
M-19
M-10
M-ll
M-12
M-16
M-8
PM-10
emission
factor,
Ib/VMT
0.0110
0.00340
0.00781
0.00712
0.000400
0.00153
0.00304
0.00680
0.00301
0.00323
0.00582
0.000800
0.000390
0.000390
0.000700
0.000190
0.000530
0.0247
Duration,
min.
120
86
120
136
240
226
281
194
178
135
150
172
488
182
181
150
254
345
Temp., °F
28
27
28
50
38
53
35
60
55
77
75
75
70
60
56
65
70
50
Mean
wind
speed,
mph
7.4
6.5
7.8
7.4
7.8
2.2
5.6
2.7
9.2
11.4
4.0
5.1
2.7
2.9
8.7
4.7
4.0
4.7
Road
width,
ft
44
44
44
44
36
36
36
22
22
22
40
40
20
96
96
96
96
30
No. of
vehicle
passes
2,627
2,166
2,144
3,248
2,763
2,473
3,204
5,190
3,940
4,040
3,390
3,670
5,800
11,148
11,099
9,812
15,430
1,975
Mean
vehicle
speed,
mph
30
30
30
30
35
35
30
35
35
35
30
30
30
55
55
55
55
20
Mean
vehicle
weight,
tons
5.6
3.8
4.5
4.1
2.1
2.2
2.1
2.7
2.7
2.7
2.0
2.0
2.4
4.5
4.8
3.8
4.3
2.2
Silt
loading,
g/m2
0.46
0.26
0.15
0.29
0.43
1.00
0.68
0.11
0.079
0.047
0.83
0.73
0.93
0.022
0.022
0.022
0.022
2.50
Silt (%)
10.7
6.2
3.5
12.2
18.8
21.4
21.7
13.7
-
8.1
5.7
7.1
8.6
-
-
-
-
14.5
1 lb/VMT = 281.9g/VKT.
1 g/m2= 1.434gr/ft2
4-14
-------
4.2.1.8 Midwest Research Institute, Size Specific Particulate Emission Factors for
Uncontrolled Industrial and Rural Roads, for U. S. EPA, January 1983. Reference 4
(AP-42Ref5)
This document reports the results of testing conducted in 1981 and 1982 at industrial
unpaved and paved roads and at rural unpaved roads. Unpaved industrial roads were tested at a
sand and gravel processing facility in Kansas, a copper smelting facility in Arizona, and both a
concrete batch and asphalt batch plant in Missouri. The study was conducted to increase the
existing data base for size-specific PM emissions. The following particle sizes were of specific
interest for the study: PM-15, PM-10, and PM-2.5.
Exposure profiling was utilized to characterize total PM emissions. Five sampling heads,
located at heights of up to 5 m, were deployed on the profiler. A standard high-volume sampler
and a high-volume sampler with an SSI (cutpoint of 15 umA) were also deployed downwind. In
addition, two high-volume cyclone/impactors were operated to measure particle size distribution.
A standard high-volume sampler, a high-volume sampler with an SSI, and a high-volume
cyclone/impactor were utilized to characterize the upwind TSP and PM-15 concentrations and
the particle size distribution within the PM-15 fraction. Wind speed was monitored with warm
wire anemometers.
A total of 18 paved road tests and 21 unpaved road tests are completed. The test data are
assigned an A rating. Industrial paved road tests were conducted as follows: three unpaved road
tests at the sand and gravel processing plant, three paved road tests at the copper smelting plant,
four paved road tests at the asphalt batch facility, and three paved road tests at the concrete batch
facility. The industrial road tests were considered uncontrolled and were conducted with heavy
duty vehicles at the sand and gravel processing plant and with medium duty vehicles at the
asphalt batch, concrete batch, and copper smelting plants. Table 4-7 presents summary test data
and Table 4-8 presents detailed test information.
4-15
-------
TABLE 4-7. SUMMARY OF PAVED ROAD EMISSION FACTORS FOR REFERENCE 3
Industrial
category
Asphalt Batching
Concrete
Batching
Copper Smelting
Sand and Gravel
Processing
Type
Medium
duty
Medium
duty
Medium
duty
Medium
Duty
TP, Ib/VMT
Geo.
mean
1.83
4.74
11.2
5.50
Range
0.750-3.65
2.25-7.23
7.07-15.7
4.35-6.64
PM-1 5, Ib/VMT
Geo. mean
0.437
1.66
4.01
1.02
Range
0.124-
0.741
0.976-2.34
2.02-5.56
0.783-1.26
PM-10, Ib/VMT
Geo.
mean
0.295
1.17
2.78
0.633
Range
0.0801-
0.441
0.699-1.63
1.35-3.86
0.513-0.753
PM-2.5, Ib/VMT
Geo.
mean
0.130
0.381
0.607
0.203
Range
0.0427-0.214
0.200-0.562
0.260-0.846
0.194-0.211
1 Ib/VMT = 281.9 g/VKT.
4-16
-------
TABLE 4-8. DETAILED INFORMATION FOR PAVED ROAD TESTS FOR REFERENCE 3
Run
No.
Y-l
Y-2
Y-3
Y-4
Z-l
Z-2
Z-3
AC-4
AC-5
AC-6
AD-1
AD-2
AD-3
Industrial
category
Asphalt
Batching
Asphalt
Batching
Asphalt
Batching
Asphalt
Batching
Concrete
Batching
Concrete
Batching
Concrete
Batching
Copper Smelting
Copper Smelting
Copper Smelting
Sand and Gravel
Sand and Gravel
Sand and Gravel
Traffic
Medium
Duty
Medium
Duty
Medium
Duty
Medium
Duty
Medium
Duty
Medium
Duty
Medium
Duty
Medium
Duty
Medium
Duty
Medium
Duty
Heavy
Duty
Heavy
Duty
Heavy
Duty
PM-10
emission
factor,
Ib/VMT
0.257
0.401
0.0801
0.441
0.699
1.63
4.01
3.86
3.13
1.35
3.27
0.753
0.513
Duration,
min.
274
344
95
102
170
143
109
38
36
33
110
69
76
Mean
wind
speed,
mph
5.37
4.70
6.04
5.59
6.71
9.84
9.62
8.72
9.62
4.92
7.61
5.15
3.13
Road
width,
ft
13.8
14.1
14.1
14.1
24.3
24.9
24.9
34.8
34.8
34.8
12.1
12.1
12.1
No. of
vehicle
passes
47
76
100
150
149
161
62
45
36
42
11
16
20
Vehicle characteristics
Mean
vehicle
weight,
tons
3.6
3.7
3.8
3.7
8.0
8.0
8.0
5.7
7.0
3.1
42
39
40
No. of
wheels
6
7
6.5
6
10
10
10
7.4
6.2
4.2
11
17
15
Mean
vehicle
speed,
mph
10
10
10
10
10
15
15
10
15
20
23
23
23
Moisture
content,
%
0.22
0.51
0.32
0.32
a
a
a
0.43
0.43
0.53
a
a
a
Silt
loading,
g/m
91
76
193
193
11.3
12.4
12.4
287
188
400
94.8
63.6
52.6
Silt, %
2.6
2.7
4.6
4.6
6.0
5.2
5.2
19.8
15.4
21.7
6.4
7.9
7.0
llb/VMT = 281.9 g/VKT.
Ig/m2=1.434gr/ft2
a Not measured
4-17
-------
4.2.1.9. Midwest Research Institute, Iron and Steel Plant Open Source Fugitive Emission
Control Evaluation, for U. S. EPA, August 1983, Reference 3 - (AP-42 Ref 3)
This test report centered on the measurement of the effectiveness of different control
techniques for PM emissions from fugitive dust sources in the iron and steel industry. The test
program was performed at two integrated iron and steel plants, one located in Houston, Texas,
and the other in Middletown, Ohio. Control techniques to reduce emissions from paved roads,
unpaved roads, and coal storage piles were evaluated. For paved roads, control techniques
included vacuum sweeping, water flushing, and flushing with broom sweeping. Particle
emission sizes of interest in this study were total PM, PMi5, and PM2 5.
The exposure profiling method was used to measure paved road paniculate emissions at
the Iron and Steel plants. For this study, a profiler with four or five sampling heads located at
heights of 1 to 5 m was deployed. Two high-volume cascade impactors with cyclone
preseparators (cutpoint of 15 umA), one at 1 m and the other at 3 m, measured the downwind
particle size distribution. A standard high-volume sampler and an additional high-volume
sampler fitted with a SSI (cutpoint of 15 umA) were located downwind at a height 2 m. One
standard high-volume sampler and two high-volume samplers with SSIs were located upwind for
measurement of background concentrations of TSP and PMi5.
Twenty-three paved road tests of controlled and uncontrolled emissions were performed.
These included 11 uncontrolled tests, 4 vacuum sweeping tests, 4 water flushing tests, and
4 flushing and broom sweeping tests. For paved roads, this test report does not present vehicle
speeds, mean number of wheels, or moisture contents. Because vehicle speeds above 15 MPH
and moisture content are not expected to influence the emissions equation, the test data are
assigned an A rating. Table 4-9 presents summary test data and Table 4-10 presents detailed test
information. The PM-10 emission factors presented in Table 4-10 were calculated from the
PMis and PM2 5 data using logarithmic interpolation.
After vacuum sweeping, emissions were reduced slightly more than 50 percent for two
test runs and less than 16 percent for two test runs. Water flushing applied at 0.48 gal/yd2
4-18
-------
TABLE 4-9. SUMMARY OF PAVED ROAD EMISSION FACTORS FROM REFERENCE 4
Control
method
None
Vacuum
Sweeping
Water
Flushing
Flushing &
Broom
Sweep
None
Location
A,D,F,J
A
D,L
K,L,M
L,M
State
OH
OH
TX
TX
TX
Test date
7/80,
10/80, &
11/80
10/80&
11/80
6/81
6/81
6/81
No. of
tests
7
4
4
4
4
TP, Ib/VMT
Geo mean
1.22
0.87
1.43
0.96
3.12
Range
0.29-5.50
0.53-1.46
1.30-1.74
0.54-2.03
0.83-5.46
PMis, Ib/VMT
Geo mean
0.38
0.45
0.47
0.20
0.92
Range
0.13-2.14
0.27-0.87
0.32-0.65
0.10-0.49
0.31-1.83
PM2.5, Ib/VMT
Geo mean
0.10
0.14
0.08
0.07
0.26
Range
0.04-0.52
0.08-0.26
0.08-0.09
0.04-0.13
0.06-0.62
1 Ib/VMT = 281.9 g/VKT.
4-19
-------
TABLE 4-10. DETAILED INFORMATION FOR PAVED ROAD TESTS FROM REFERENCE 4
Site
A
A
A
A
A
A
D
D
D
F
F
J
K
L
L
L
L
L
L
M
M
M
M
Test
Run No.
F-34
F-35
F-36
F-37
F-38
F-39
F-61
F-62
F-74
F-27
F-45
F-32
B-52
B-50
B-51
B-54
B-55
B-56
B-58
B-53
B-57
B-59
B-60
Control
method
None
None
VS
VS
VS
VS
None
None
WF
None
None
none
FBS
FBS
FBS
WF
WF
WF
None
FBS
0.554
0.993
1.18
PM-10 emission
factor, (Ib/VMT)
0.536
0.849
0.147
0.209
0.430
0.686
1.35
0.929
1.32
0.357
0.608
0.144
0.0946
0.230
0.435
0.268
0.575
0.398
1.08
0.161
None
None
None
Duration
(min.)
62
127
335
241
127
215
108
77
205
91
135
259
60
104
93
101
82
61
96
81
101
114
112
Temp.,
(°F)
90
90
50
50
50
50
40
45
50
100
50
90
90
90
90
90
90
90
90
90
90
90
90
Mean wind
speed, (mph)
4.2
7.5
5.9
4.8
4.5
6.4
11.0
12.1
9.0
9.5
4.0
5.8
2.9
5.6
4.2
5.4
8.5
6.3
6.7
5.3
3.6
6.1
5.0
No. of
vehicle passes
79
130
263
199
141
190
93
94
67
158
172
301
119
123
127
118
98
118
67
72
68
67
50
Mean
vehicle weight,
(tons)
28
25
8.3
17
18
18
40
36
29
14
16
14
12
9.4
11
10
11
9.2
18
20
12
11
12
Silt loading,
(g/m2)
2.79
2.03
0.202
0.043
0.217
0.441
17.9
14.4
5.59
17.7
5.11
0.117
7.19
13.6
13.6
3.77
6.29
2.40
10.4
-
2.32
2.06
3.19
Silt, %
16
10.4
18.3
26.4
27.9
19.6
21.0
20.3
9.45a
35.7
28.4
13.4
34.3
28.2b
28.2b
22.6
19.6a
11.2
17.9
9.94
6.45a
14.0a
13.5
aAverage of 2+ values
bSample used for more than 1 run.
°PM-10 emission factors were calculated from the PM-
VS = Vacuum sweeping; WF = Water flushing; FBS =
15 and PM-2.5 data using logarithmic interpolation.
Water flushing and broom sweeping; 1 Ib/VMT = 281.9 g/VKT; 1 g/m2 = 1.434 gr/ft2
4-20
-------
achieved emission reductions ranging from 30 percent to 70 percent. Flushing at 0.48 gal/yd
combined with broom sweeping resulted in emission reductions ranging from 35 percent to
90 percent.
4.2.1.10. Midwest Research Institute, Fugitive Particulate Matter Emissions for U.S.
Environmental Protection Agency, Emission Factor and Inventory Group, April
15,1997. References
This reference documents the performance of six field studies characterizing the vehicle
emissions from three unpaved roads and three paved roads. Testing of unpaved roads was
performed in Kansas City, MO; Raleigh, NC; and Reno, NV. Testing of paved roads was
performed in Denver, CO; Raleigh, NC; and Reno, NV. Midwest Research Institute measured
the emission rates for PMio and PM2 5 at all six locations based upon a plume profiling
methodology. The test data are assigned an A rating.
Plume profiling calculates emission rates using a conservation of mass approach. The
passage of airborne paniculate (i.e., the quantity of emissions per unit of source activity) is
obtained by spatial integration of distributed measurements of exposure (mass/area) over the
effective cross section of the plume. Exposure is the point value of the flux (mass/area time) of
airborne paniculate integrated over the time of measurement or, equivalently, the net paniculate
mass passing through a unit area normal to the mean wind direction during the test. The steps in
the calculation procedure are as follows. The concentration of PMio measured by a sampler is
compared to the wind speed and corrected to standard conditions. The concentration for each
sampler is multiplied by the wind velocity and sampling duration to obtain the exposure for each
sampling height. The exposure is integrated over the plume-effective cross section. The
quantity obtained represents the total passage of airborne paniculate matter (i.e., mass flux) due
to the source. The exposure is set to zero at the maximum effective height of the plume where
the net concentration equals zero). The maximum effective height of the plume is found by
linear extrapolation of the uppermost net concentrations to a value of zero. Although at ground
level the wind velocity is zero, for calculation, the exposure value at ground level is set equal to
4-21
-------
the value at a height of 1 m. The integration is then performed from 1 m to the plume height, H,
using Simpson's approximation.
Testing in Denver CO was conducted to characterize emissions from a high speed (55
mph speed limit) limited access interstate road and a medium speed (40 mph speed limit) one
lane road (two lanes with a wide median). For this part of the study, a profiler with four or five
sampling heads located at heights of 1, 3, 5 and 7 m were deployed. One high-volume cascade
impactor with cyclone preseparators (cutpoint of 10 umA) and two dichotomous samplers were
used to measured the downwind particle size distribution. All of the particle sizing samplers
were located at 2 m above ground level. A single set of the same sampling equipment was
located at 2 m above ground level and upwind for measurement of background concentrations of
TSP, PMio and PM2 5. To the extent possible, each of the emission tests was performed during
periods following snowfall, after the test road surface had dried. In most cases, sand application
was ordered, because the relatively light snow conditions characteristic of the 1996 winter did
not trigger routine sand application.
This test program also assessed the potential bias associated with particle sizing using the
historical impactors that followed the cyclone pre-separator. The use of the dichotomous
samplers consistently yielded a lower ratio of PM2 5 to PMio ratio than were measured by the
cyclone/impactor samplers. The PM2.s/PMio ratios measured by the dichotomous samplers are
presented to the right of the PMio emissions factors column in Table 4-11. Where two values are
presented in the column, these are the ratios measured at two different heights. The ratios range
from 0.26 to 0.37. As a result of this study, the constant in the PM2 5 emissions factor equation
was revised to 25% of the PMio constant.
4-22
-------
TABLE 4-11. DETAILED INFORMATION FOR PAVED ROAD TESTS FROM REFERENCE 5
Site
CO
CO
CO
CO
CO
CO
NC
NC
NC
NC
NV
NV
NV
NV
Test
Run No.
BH-1
BH-2
BH-3
BH-4
BH-5
BH-6
BJ-6
BJ-7
BJ-9
BJ-10
BJ-11
BK-7
BK-8
BK-9
Road
Speed '
55
55
55
55
40
40
45
45
45
45
45
45
45
45
PM10
emission
factor,
(g/VKT)
1.08
0.102
-
-
-
4.68
0.301
1.94
0.57
0.44
-
PM25/PMio
Ratio
0.20
0.34
0.16
0.03
0.27/0.34
0.44/0.44
0.6/0.14
0.44/0.33
0.68/0.47
0.29/0.33
0.26/0.34
0.13/0.38
Duration,
min.
163
360
360
Blank
Blank
240
450
143
178
288
387
420
270
240
Temp.,
°F
18
37
46
-
-
48
71
68
71
68
75
89
87
90
Mean wind
speed, mph
2.7
17.0
17.2
-
-
3.1
8.2
9.4
5.3
3.7
5.1
7.3
6.1
2.6
No. of
vehicle passes
6,561
17,568
14,616
-
-
3,112
14,670
3,748
4,616
10,218
13,216
7,394
5,747
4,622
Mean
vehicle weight,
tons
2.2
2.2
-
-
-
2.2
2.2
2.2
2.2
2.2
2.2
2.2
2.2
-
Silt loading,
/ 2
g/m
0.184
0.0127
0.0127
-
-
1.47
0.060
0.060
0.060
0.060
0.060
0.082
0.082
0.082
Silt, %
9.4
41.0
41.0
-
-
1.2
52
52
52
52
52
3.4
3.4
3.4
Road Speed is the posted speed limit for the road segment.
4-23
-------
4.2.1.11. Paved Road Modifications to AP-42, Background Documentation For Corn
Refiners Association, Inc. Washington, DC 20006 MRI Project No. 310842, May
20,2008. Reference 6
The Corn Refiners Association (CRA) funded four paved road PM-10 test programs
because site conditions did not match source conditions underlying the AP-42 emission factor
equation. The sites enforce speed limits of 5 or 15 mph and employ road sweeping programs to
manage the build up of silt on the roadways. In addition, plants experience traffic queues (i.e.,
stop-and-go traffic) during periods with high corn receipts. The combination of heavy trucks
(delivering corn to the facilities) and fairly low silt loading (sL) values on the plant roads was
not typical of the AP-42 data base. Given these differences, the member companies undertook
testing to develop more representative emission factors. Midwest Research Institute designed
and conducted the test programs at all four facilities.
Reference 6 compiles test data and information from references 7, 8, 9 & 10. In addition,
reference 6 proposes an expansion of the allowable speed parameters supported in the paved
road equation. Lastly, reference 6 proposes a revised equation for paved roads to reflect the
expanded test information. The data upon which the proposed equation was based included
emissions associated with the trucks (engine exhaust, tire wear and brake wear) and with
material deposited on the roadway. Since testing documented in references 7 through 10 were
conducted at facilities with very similar operating conditions using test procedures that were
nearly identical, the following description provides background for all four test programs.
All four testing programs employed the same exposure profiling method used to develop
the test data underlying the emission factor predictive equations for both paved and unpaved
roads. In each program, a test plan was submitted to the state agency for comment and review
prior to the start of testing. The final test reports and supporting information were also submitted
to state agencies. Because low emission levels were expected (due to low sL and slow speeds),
several precautions were taken to assure reliable quantification. First, long sampling durations
were employed. Samplers were operated up to 5 hours to collect adequate sample mass.
Second, to ensure adequate traffic during test periods, the facilities provided "drone" passes by
corn semi-trailers. Drone traffic mimicked the actual traffic except those trucks returned to
4-24
-------
staging areas without emptying corn. In addition, testing applied "lessons learned" throughout
the programs. For example, when it became apparent how difficult it could be to separate net
PM-10 concentrations (i.e., due to traffic on the road) from background (upwind) concentrations,
changes were made in equipment deployment. The use of identical upwind and downwind
vertical sampling arrays permitted better definition of the net contribution of roadway emissions.
In addition to PM-10 concentrations, each sampling program samples included:
• Measurement of average wind speeds at two heights and wind direction at one
height for 5-minute intervals throughout the test period.
• Manual recording of traffic counts by vehicle type. The host facilities provided
information on vehicle weights and corn receipts.
• Collection of road surface material by vacuums with disposable paper bags. The
material collected within the bag was sieved to determine the surface silt loading.
Reference 6 states that the four test programs conducted by CRA produced 14 and 8 PM-
10 emission factor values for slowly moving and stop-and-go traffic, respectively. Other
observations in this report includes: that in all but one of the 22 cases, the AP-42 emission factor
overestimated the measured value; that for some tests, "stop-and-go" emission factors were
substantially greater than the "slowly moving" factor (presumably because of the diesel exhaust
as trucks moved from a dead stop) but that there was no significant difference between "slowly
moving" and "stop-and-go" results on average.
Furthermore, Tables 4-12, 4-13, and 4-15 use bold font to indicate those tests that used
identical upwind and downwind vertical sampling arrays. Those tests provided better definition
of net PM-10 mass thus producing more accurate emission factors. Although these test results
tended to be lower than the other emission factors, the two sets on average did not differ
significantly.
4-25
-------
4.2.1.12 Midwest Research Institute, Emission Tests of Paved Road Traffic at Minnesota
Corn Processors Marshall, Minnesota Facility, McVehil-Monnett Associates, July
6,2001. Reference 7 (ref_29cl3s0201_2010.pdf)
Truck traffic flow at the Minnesota Corn Processor's (MCP's) Marshall, Minnesota facility was
characterized as either slowly moving (5 mph enforced speed limit) or stop-and-go in nature. In
this testing program, data was collected over 5 days during April of 2001. During this period,
three stop-and-go traffic situations and six slowly moving traffic instances were examined.
Truck traffic progressing through the test site was held to two lanes for queued traffic. Silt
content (sL, measured by MCP), truck weight, and number of passes, along with other pertinent
data was recorded for each run. For all runs, a vertical network of samplers was operated
downwind. The last test period used a vertical array of samplers upwind to better characterize
upwind concentrations and to provide a more accurate calculation of the net PM-10 emission
factor.
The results of this testing program are summarized in Table 4-12. The test data are
assigned an A rating. The test report remarked that the emission factors obtained were far below
the value (0.453 Ib/VMT) used in the plant emission inventory. Use of test-specific silt loading
and vehicle weight did not significantly improve the predictive accuracy of the AP-42 factor.
The tests found no discernable relationship between emission levels and either silt loading or
vehicle weight. Finally, it was noted that the shape of the exposure profile was more likely due
to diesel exhaust than re-entrained road dust.
4-26
-------
Table 4-12. Summary of Emissions Data from MCP's Marshall, Minnesota Facility (Test Report 1)
Run
CE-1
CE-2
CE-11
CE-3
CE-1 3
CE-1 5
CE-1 6
CE-1 7
CE-19
Test condition
Stop-and-go
Stop-and-go
Slowly moving
Stop-and-go
Slowly moving
Slowly moving
Slowly moving
Slowly moving
Slowly moving
Traffic rate
(veh/hr)
38
32
35
47
48
30
28
29
61
Traffic speed
(mph)a
NA
NA
5
NA
5
5
5
5
5
Mean vehicle weight,
W (tons)
36
36
12
39
13
40
40
40
38
Surface silt loading,
sL (g/m2)
1.16
0.86
1.34
0.86
1.34
1.91
1.41
2.93
0.76
Measured PM-10 emission
factor (Ib/VMT)
0.059
0.14
0.34
0.10
0.051
0.14
0.17
0.091
0.041
Vehicle speed was maintained at the plant limit of 5 mph. NA = Not applicable.
Bold entries indicate that identical vertical sampling arrays were used to better isolate the source contribution.
4-27
-------
4.2.1.12. Midwest Research Institute, Emission Tests of Paved Road Traffic at Minnesota
Corn Processors Columbus, Nebraska Facility, McVehil-Monnett Associates, July
13,2001. Reference 8. (ref_32cl3s0201_2010.pdf)
Truck traffic flow at MCP's Columbus, Nebraska facility was characterized as either
slowly moving (5 mph enforced speed limit) or stop-and-go in nature. Between June 12 and 15,
2001, four tests each of stop-and-go and slowly moving traffic were performed. Trucks entered
by the north gate and traveled past a vertical sampling array en route to a staggered queue at
which a second vertical sampling array was positioned. In this way, testing evaluated both source
conditions (stop-and-go and slowing moving) at once. Building on experience from testing at
the MCP Marshall facility, the last two runs, CF-4 and CF-5, used identical upwind and
downwind vertical sampling arrays to better characterize background concentrations. In that
case, only one condition could be evaluated during a test. The results of the MCP Columbus test
program are summarized in Table 4-13. The test data are assigned an "A" rating.
4.2.1.13. Midwest Research Institute, Emission Tests of Paved Road Traffic at Cargill
Sweeteners North America Blair, Nebraska Facility, McVehil-Monnett Associates,
November 27,2002. Referenc 9, (ref_30cl3s0201_2010.pdf)
This report describes a testing program conducted at Cargill's Blair, Nebraska facility
during August 2002. The plant used a regular sweeping program to reduce surface loadings on
paved roads. Testing relied on regular corn truck traffic at the site, although the plant provided a
limited amount of "drone" traffic. The test data are assigned an "A" rating.
Eight PM-10 emission tests were attempted. The test report describes difficulty
encountered in isolating net PM-10 mass due to traffic on the test road. During test plan review,
the Nebraska Department of Environmental Quality requested a change in test site to allow two
trucks to pass by at the same time. The original site would have permitted upwind monitoring in
the immediate vicinity of the tests road, but this was not possible at the second location.
Furthermore, steeply sloping ground on the upwind side of the test road prevented use of a
4-28
-------
Table 4-13. Summary of Emissions Data from MCP's Columbus, Nebraska Facility (Test Report 2)
Runa
CF-l/N
CF-l/S
CF-2/N
CF-2/S
CF-3/N
CF-3/S
CF-4/N
CF-5/N
Test condition
Low Speed
Stop-and-go
Slowly moving
Stop-and-go
Slowly moving
Stop-and-go
Slowly moving
Stop-and-go
Traffic rate
(veh/hr)
47
47
66
66
54
54
86
52
Traffic speed
(mph)b
5.0
NA
5.3
NA
5.1
NA
4.7
NA
Mean vehicle weight,
W (tons)
40
40
41
41
41
41
41
41
Surface silt loading,
sL (g/m2)
0.97
0.97
0.81
0.81
0.63
0.63
1.1
1.4
Measured PM-10 emission
factor (Ib/VMT)
0.011
0.043
0.036
0.14
0.0024
0.051
0.0068
0.036
Suffix indicates whether tests was conducted on the North or South portion of the corn haul road. Trucks were held in a queue
toward the south; trucks entering the north gate traveled passed the north sampling array to reach the queue.
Speed of moving trucks determined by accumulating time required to travel a measured distance. NA = not applicable.
Bold entries indicate that identical vertical sampling arrays were used to better isolate the source contribution.
4-29
-------
vertical background sampling array (as used at the two MCP plants) to better isolate the source
contribution.
The results are summarized in Table 4-14. Only two tests (CI-7 and CI-8) had net mass
attributed to the source. In the remaining instances, the measured downwind PM-10
concentrations were lower than upwind values. It was stated that this was believed to be an
undesired result from moving the test source. Runs CI-7 and CI-8 showed the measured emission
factor to be much lower than that predicted by the AP-42 equation. Comments in the report
indicated that exposure profiles showed a maximum more likely due to diesel exhaust than from
re-entrained surface road dust.
4.2.1.14. Midwest Research Institute, Emission Tests of Paved Road Traffic atADM's
Marshall, Minnesota Facility, McVehil-Monnett Associates, December 5,2003.
Reference 10. (ref_31cl3s0201_2010.pdf)
The test program at ADM's Marshall MN facility represented the last test by the Corn
Refiners Association. By September 2003, the Marshall facility had implemented a road
sweeping program. Three tests of PM-10 emissions were conducted, one from stop-and-go
traffic and two from slowly moving traffic. Because of experience gained from the earlier tests,
identical vertical networks of samplers were operated downwind and upwind during each test.
The results of this testing program are summarized in Table 4-15. The test data are
assigned an A rating. Measured emission factors were all significantly lower than that predicted
by the AP-42 equation. The test report also remarked that the measured emission rates were
independent of traffic rate, while the AP-42 factor implies a linear dependency between the
emission and traffic rates.
4-30
-------
Table 1-14. Summary of Emissions Data from Cargill's Blair, Nebraska Facility (Test Report 3)
Run
CI-1
CI-2
CI-3
CI-4
CI-7
CI-8
CI-11
CI-1 2
Test condition
Low Speed
Low Speed
Slowly
moving
Low Speed
Slowly
moving
Low Speed
Low Speed
Low Speed
Traffic
rate
(veh/hr)
45
45
60d
60d
47
47
56
56
Traffic speed
(mph)a
13.4/16.8
12.8/16.9
13.6/12.7
13.5/15.5
15.2/16.2
13.6/16.1
13.5/12.7
Mean vehicle weight,
W (tons)
26
26
27
27
27
27
27
27
Surface silt loading,
sL (g/m2)b
0.06
0.06
0.06
0.06
0.05
0.05
0.025
0.25
Measured PMio emission
factor (lb/VMT)c
-
-
-
-
0.0036
0.0066
-
-
3 Vehicle speed for inbound (loaded) /outbound (empty) trucks determined by accumulating time required to travel a
measured distance.
b Surface silt loading sample information provided by Cargill.
0 "-" indicates that no net mass was attributed to the test road traffic.
d Twenty of 238 total passes were by "drone" trucks
Table 4-15. Summary of Emissions Data from ADM's Marshall, Minnesota Facility (Test Report 4)
Run
CM-1
CM-2
CM-4
Test Condition
Slowly moving
Stop-and-go
Slowly moving
Traffic rate
(veh/hr)
154
42
156
Traffic speed
(mph)a
NA
NA
5
Mean vehicle
weight, W (tons)
40
40
40
Surface silt loading,
sL (g/m2)
0.72
0.72
0.70
Measured PMio emission
factor (Ib/VMT)
0.014
0.14
0.016
1 Vehicles speeds maintained at plant limit of 5 mph. NA = not applicable.
Bold entries indicate that identical vertical sampling arrays were used to better isolate the source contribution.
4-31
-------
4.2.1.15. E.H. Pechan & Associates, Inc., Recommendations for Emission Factor Equations
in AP-42 Paved Roads Section: TECHNICAL MEMORANDUM August 21, 2003;
Reference 12,
(http://www.epa.gov/ttn/chief/ap42/chl3/related/pavedroadstechmemo 082103.pdf)
This technical memorandum documents the procedure that was used to separate the
various components of paved road paniculate matter emissions into two components. One
component includes the emissions from exhaust, brake wear and tire wear. The other component
includes the paniculate matter reentrained from the road surface. The combined paved road
paniculate matter emissions were estimated with the empirical equation published in the October
2002 AP-42 Section for Paved Roads. The vehicle exhaust, brakewear and tirewear emission
factors were obtained from the MOBILE6.2 model. A typical vehicle fleet and fuel source from
1980 was utilized for the model runs. The assumption included a vehicle fleet for July 1980, a
gasoline sulfur content of 300 ppm, a diesel sulfur content of 500 ppm and no use of
reformulated gas. The vehicle fleet assumptions used in the analysis are presented in Table 4-16.
The model was run to estimate PMio and PM2 5 emission factors in g/VMT for each vehicle class
at speeds of 25, 30, 35, 40, 45, 50, 55, and 60 mph. Within vehicle classes, the greatest standard
deviation was lower than 0.04% of the emissions factor. Based on the low relative standard
deviation, it was assumed that the vehicle speed was not a factor in exhaust, brakewear and
tirewear PM emissions. Table 4-16 presents the vehicle fleet characteristics used in the model
and the calculated average PMio and PM2 5 emission factors for exhaust, brakewear and tirewear
for each class of vehicle.
Table 4-16: Vehicle Fleet Assumptions Used in 2003 MOBILE6.2 Model
Vehicle Type
GVWR
VMT Distribution
PMio Emissions Factor
PM2.5 Emissions
Factor
LDGV LDGT12 LDGT34 LDGT HDGV LDDV LDDT HDDV
3,075
0.6748
0.1053
0.0686
4,105
0.1477
0.1061
0.0690
7,000
0.0758
0.2746
0.1851
0.1632
0.1084
35,000
0.0365
0.3825
0.2576
3,705
0.0088
0.7206
0.6519
6,000
0.0118
0.7206
0.6521
70,000
0.0352
2.1227
1.9272
MC
550
0.0094
0.0922
0.0590
4-32
-------
The contractor developed "AP-42 Composite" PMio and PM2 5 emission factors using the
October 2002 AP-42 paved roads emission factor equation with the mean vehicle weight set at
3.74 tons (a value they indicated was typical of the 1980 paved road vehicle fleet. The
contractor used silt loadings ranging from 0.02 to 400 g/m2 for calculating the emissions factors.
The contractor also calculated the fleet average PMio and PM2 5 emission factors for exhaust,
brakewear and tirewearby summing the products of the VMT Distribution ratio and the PMio
and PM2 5 emission factors for each vehicle class. The calculated fleet average values were
0.2119 for PMio and 0.1617 forPM25. The contractor then subtracted the fleet average
emissions factors for exhaust, brakewear and tirewear from the "AP-42 Composite" emissions
factors to produce an emission factor for only the re-entrained road dust component. The
contractor noted that the while the stated applicable silt loadings for the October 2002 AP-42
paved road equation ranged from 0.02 to 400 g/m2, the PM25 emissions factor became negative
at silt loadings less than 0.029 g/m2. They stated that since negative emissions were not
physically possible, the equation they recommended was only valid for silt loading ranging from
0.03 to 400 g/m2. While no test data are associated with this report, the report does provide
estimates of engine exhaust, tire wear and brake wear derived from an EPA emissions model
which is based upon emissions testing by a validated test method on multiple vehicles for each
type of vehicle. As a result, emissions estimates by vehicle class are assigned an A rating.
Because the use of a national average vehicle fleet emissions estimate does not provide
emissions that are representative of the mix of vehicle classes measured during the above test
reports, the composite emissions estimates are assigned a C rating.
4.2.1.16. E-mail communication between Ron Myers of EPA/OAQPS/SPPD/MPG, RTF,
NC and Prashanth Gururaja and Ed Glover of EPA/OTAQ/ASD/HDOC re.
Diesel exhaust, tire and brake wear for low speed stop and go traffic; January
2009 through May 2009. Reference 13, (OTAQ_Diesel_PMresult.pdf).
This e-mail communication and spreadsheet file concerns estimates of PMio emissions
associated with slow moving and stop and go diesel engine semi-trailer trucks. The purpose of
the request was to provide a means to disaggregate the consolidated PM emissions measured of
trucks during delivery of product at corn storage and transfer facilities. The request stated that
4-33
-------
the trucks were 18 wheel semitrailers of about ten years of age, were queued for the delivery of
their load to a transfer or processing facility and that the estimated vehicle speed averaged about
1 mphbut that they were stopped most of the time. PM25 emissions were estimated using the
MOVES mobile source emissions model. The trucks modeled were approximately ten years old,
traveling at an average of 1.5 mph on level pavement. Emissions were estimated at 11.06035
g/hour or 8.789778 g/VMT. PMio emissions were estimated to be approximately 3% greater
than PM2 5 emissions. While no test data are associated with this report, the report does provide
estimates of engine exhaust, tire wear and brake wear derived from an EPA emissions model
which is based upon emissions testing by a validated test method on multiple vehicles for the
specific type of vehicle measured during the Corn Refiners Association Studies. As a result,
emissions estimates for slow moving trucks are assigned an A rating.
4.2.1.17. Midwest Research Institute; Analysis of the Fine Fraction of Particulate Matter in
Fugitive Dust; Western Governors' Association - Western Regional Air
Partnership (WRAP); October 12,2005. Reference 14,
http://www.epa.gov/ttn/chief/ap42/chl3/related/mri final fine fraction dust report.pdf)
This project was conducted by Midwest Research Institute for the Western Regional Air
Partnership to provide more accurate PM2 5 and PMio fugitive dust emissions inventories for
regional haze regulatory purposes to address the significant contribution of fugitive dust to
visibility impairment. The results of this project were expected to affect the quantity of dust
apportioned to the fine versus coarse size modes. It was stated that the results would be helpful
in developing accurate emission inventories for PM nonattainment, maintenance, and action plan
areas in the WRAP region. Finally, it was stated that the results may be used to seek
modifications to the EPA's AP-42 emission factors to ensure widespread availability of the
information developed in the study.
During the first testing phase of the project, PM25 measurements using the high-volume
cascade impactors were compared to simultaneous measurements obtained using EPA reference-
method samplers for PM2 5. The tests were conducted in a flow-through wind tunnel and
exposure chamber, where concentration level and uniformity were controlled. With the same
4-34
-------
test setup, a second phase of testing was performed with reference method samplers, for the
purpose of measuring PM2 5 to PMio ratios for fugitive dust from different geologic sources in
the West. The testing provided information on the magnitude and variability of PM2 5 to PMio
ratios for source materials that were recognized as problematic with regard to application of
mitigative dust control measures.
Three dust source materials were tested under the first Phase of the study. The three dust
source materials included an Owens Dry Lake surface soil, and two Arizona road dust reference
standards (one coarse and one fine fraction material). Fixed PMio concentration levels in the
range of 1, 2.5, and 5 milligrams per cubic meter (each with its naturally occurring PM2 5 level)
were tested. It was stated that those PMio concentration levels were selected as representative of
dust plume concentrations under which major particle mass contributions to plume samples
occur in emission factor development. The ratios of PM2 5 to PMio for fugitive dust from
different geologic soil types were measured. A total of seven source materials were tested. The
materials included Alaska river bed sediment, Arizona alluvial channel, Arizona agricultural soil,
New Mexico unpaved landfill road dust, New Mexico grazing soil, California Salton Sea
shoreline soil, and Wyoming unpaved road surface material. Test results included the
calculation of the average PM25 concentration and the collocated PMio concentration. It was
intended that any variation in PM2 5/PMi0 ratio be evaluated as a function of the test soil
properties (for example, position in soil texture triangle).
A total of 100 individual tests were performed, including 17 blank runs (for quality
assurance purposes). The results of the testing are well documented and the documentation is
sufficient to assess that the study was well designed and implemented. This was a laboratory
study designed to assess those emissions sources that were considered to have the greatest
influence in PMio and PM2 5 non attainment areas. As a result, the study is assigned a quality
rating of B when applied within the bounds of the type of surface material that was available and
for dust generation characteristics comparable to those used in the study. The study included no
paved road surface material and was weighted toward higher paniculate matter concentrations.
Since the study was a laboratory study, did not include any paved road surface materials, and
4-35
-------
was weighted toward higher paniculate concentrations, it is assigned a quality rating of "D"
when used for paved roads.
The results of the Phase I testing indicated that the PM2 5 concentrations measured by the
cyclone/impactor system were consistently biased by a factor of about 2 relative the PM2 5
concentrations measured by the Partisol samplers. While there was some data separation of
different test materials, the second phase testing showed a tendency of the measured PM2.5/PMi0
ratio to decrease with increasing PMio concentration. At PMio concentrations above 1.0 mg/m3
the PM2.s/PMio ratio was between 0.1 and 0.15. The PM2.s/PMio ratio increased to about 0.35 as
the PMio concentration approached about 0.5 mg/m3.
4.1.2.18. Midwest Research Institute; Background Document for Revisions to Fine Fraction
Ratios Used for AP-42 Fugitive Dust Emission Factors; Western Governors'
Association - Western Regional Air Partnership (WRAP); November 1,2006.
Referenc 15, (http://www.epa.gov/ttn/chief/ap42/chl3/bgdocs/bl3s02.pdf)
This report summarizes the results of the October 2005 WRAP study which evaluated the
PM2 s/PMio ratio measured by the cyclone/impactor system and measured by the Partisol
samplers. While no additional analyses of the laboratory study were performed, suggested
PM2 s/PMio ratios were made for use in revising existing AP-42 emissions factor parameters for
PM25 dust emissions factor equations in Sections 13.2.1 (paved roads), 13.2.2 (unpaved roads),
13.2.3 (material transfer and storage piles), 13.2.4 (windblown dust) and 13.2.5 (industrial wind
erosion). A revised PM2 5/PMi0 ratio of 0.15 was recommended for the paved roads emissions
factor.
4.1.2.19. Technical Memorandum from William B. Kuykendal to File, Subject: Decisions on
Final AP-42 Section 13.2.1 "Paved Roads", October 10, 2002. Reference 32,
(http://www.epa.gov/ttn/chief/ap42/chl3/related/rel c 13s0201 .pdf)
This technical memorandum to the files summarizes and responds to comments on an
October 2001, EPA proposed revision of Section 13.2.1 "Paved Roads" for AP-42 and request
4-36
-------
for comments. The memorandum also presents EPA's decisions and rational supporting these
decisions for the final changes leading to the final section. The proposed revisions to the section
included an adjustment for rain events (comparable to the adjustment in the unpaved road
section) which in essence "zeroed" the emissions on days that more than 0.01 inch of rain was
recorded. In addition, the proposed revisions included the separation of vehicle engine exhaust,
breakwear an tirewear as recommended in the E. H. Pechan Technical Memorandum of August
21, 2003. The memorandum includes attachments with the detailed comments that lead to the
final revision of the emissions factor equation. The final changes to the emissions factor
equation included:
• the subtraction of 0.2119 g/VMT for engine exhaust, brakewear and tirewear,
• an adjustment of (1- (P/4N)) for rain events (P = number of rain days and N = number of
days in period), and
• an adjustment of (1- (1.2P/N)) for rain events (P = number of rain hours and N = number
of hours in period).
422 EMISSIONS FACTOR DEVELOPMENT.
A total of 93 individual tests are available. All tests quantified PMio emissions. Lastly,
plume profiling was the test method. Of these, 71 emissions tests included mean vehicle weight,
road silt loading, and vehicle speed. The remaining tests included all of these parameters except
vehicle speed. These emissions tests measured PMio emissions associated with engine exhaust,
tire wear, brake wear and material deposited on the road surface. Policy decisions within EPA
make it necessary to separate paniculate matter emissions associated with the operation of the
vehicles (engine exhaust, tire wear and brake wear) and those associated with the road surface
characteristics. These policy decisions are based in part on the recent and future efforts to
control engine exhaust emissions. Many of the emissions tests performed to quantify paniculate
matter emissions from paved roads were conducted in the mid 1980's to middle 1990's. Several
of the emissions studies have experienced comparable upwind and downwind concentrations
with downwind paniculate that appears to consist of a large percentage of organic or
carbonaceous material. The first separation of vehicle associated emissions and pavement
associated emissions was in the 2003 update. This update used the national VMT weighted fleet
4-37
-------
average PMio emissions factor of 0.1317 g/VMT to subtract from the existing emissions factor
equation as a means of separating the emissions from engine exhaust, tire wear and brake wear
from the composite paved road emissions factor. A fleet average vehicle weight of 3.75 tons is
associated with this emissions factor. Since the average vehicle weight used in the development
of the paved road emissions factor equation was about 10 tons, the PMIO emissions factor for
engine exhaust, tire wear and brake wear probably underestimated these emissions. In addition,
because of the range and variation in mean vehicle weight, the use of an average for adjustment
value introduces excessive error in the estimated road dust emissions estimates. Improved test
specific adjustments for vehicle exhaust, tire wear and brake wear can be made since (1) average
vehicle weights are available for each test series, (2) PMio emissions factors estimates for each
vehicle class are available from the 2003 Technical Report and (3) PMio emissions estimates for
slowly moving and stop and go truck traffic are available. By subtracting the estimated test
specific vehicle emissions from the measured emissions prior to performing the stepwise
multiple regression, emissions associated with the road surface material will be isolated.
4.2.2.1. Compilation and Adjustment of Final Data Base.
In keeping with the results from the data set review, a final data base was compiled by
combining the following sets:
1. The January 1983 EPA database,
2. the August 1983 EPA database,
3. the July 1984 EPA database,
4. the May 1990 USX data base,
5. the April 1997 EPA database, and
6. the May 2008 CRA data base.
While several of the test reports include detailed information on the number of light duty
vehicles, moderate weight trucks and heavy weight trucks, none provide detailed information on
vehicle class as used to estimate emissions of vehicle exhaust, tire wear and break wear. For this
assessment the vehicle classes will be separated into two vehicle classes. One group of vehicle
class will include the six classes of light duty vehicles/trucks and motorcycles. The other group
4-38
-------
of vehicle class includes gas and diesel heavy duty tracks. Other assumptions used to estimate
vehicle associated emissions include:
• The test fleet includes a mixture of light duty vehicles, heavy duty gas tracks and heavy
duty diesel tracks when the average vehicle weight is less than 23 tons.
• The test fleet includes a mixture of light duty vehicles and heavy duty diesel tracks when
the average vehicle weight is between 23 tons and 35 tons.
• The test fleet includes only heavy duty diesel tracks when the average vehicle weight is
more than 35 tons.
First, the average vehicle weight and emissions are determined for the two classes of
vehicles used to estimate the adjustment for the measured emissions. The vehicle weights, VMT
distribution and emissions presented in Table 4-16 are used to calculate the average vehicle
weight and emissions. The VMT adjusted gross vehicle weight is calculated for each class of
vehicle by multiplying the VMT distribution by the average gross vehicle weight for the class.
The individual vehicle class VMT adjusted gross vehicle weights are summed to arrive at the
two VMT adjusted gross vehicle weights used in this assessment. For light duty vehicles, the
VMT adjusted gross vehicle weight is 3320 pounds. For heavy duty tracks, the VMT adjusted
gross vehicle weight is 3742 pounds. The sums of the VMT distributions for these two classes
of vehicles are obtained by summing the individual VMT distributions for the two classes of
vehicles used in this assessment. For light duty vehicles, the VMT distribution is 0.928. For
heavy duty tracks, the VMT distribution is 0.0717. Dividing the VMT adjusted gross vehicle
weights by the VMT distributions and converting to tons yields the average vehicle weights for
the two classes of vehicles. For light duty vehicles, the average gross vehicle weight is 1.79
tons. For the combination of heavy duty gas and diesel tracks, the average gross vehicle weight
is 26.09 tons. The average emissions for the two classes of vehicle are calculated in a similar
manner. The emissions factors for the two classes are calculated by summing the individual
classes. For light duty vehicles, the VMT adjusted emissions factor is 0.1232 g/VMT. For
heavy duty tracks, the VMT adjusted emissions factor is 0.0887 g/VMT. Dividing the VMT
adjusted emissions factor by the VMT distributions yields the average emissions factor for the
4-39
-------
two classes of vehicles. For light duty vehicles, the average emissions factor is 0.1328 g/VMT.
For a mixture of heavy duty gas and diesel trucks, the average emissions factor is 1.2368
g/VMT. For only heavy duty diesel trucks, the emissions factor is 2.1227 g/VMT (from Table 4-
16).
Next, an algorithm is developed to provide test run specific ratios of light duty vehicles
and heavy duty trucks. The algorithm is developed by solving the following two equations.
Wt = (RLD x WLD) + (RHD x RHD)
I.OO = RLD+ RHD
where: Wt = Test report average vehicle weight
WLD = Average Light Duty Vehicle Weight (1.78848 tons)
RHD = Average Heavy Duty Truck Weight (26.09135 tons)
RLD = Light duty vehicle ratio
RHD = Heavy duty truck ratio
For test runs where the average vehicle weight is less than 23 tons, the resulting
algorithm to estimate the ratio of heavy duty gas/diesel trucks in each test series is:
RHD = (Wt - 1.78848) / (26.09135 - 1.78848)
For tests where the average vehicle weight is more than 23 tons, the resulting algorithm
to estimate the ratio of heavy duty diesel trucks in each test series is:
RHD = (Wt - 1.78848) / (35 - 1.78848)
Summing the emissions factors from heavy duty trucks and light duty vehicles provides
an estimate of the total engine exhaust; tire wear and brake wear emissions for the test run.
Multiplying the ratio of heavy duty trucks in a test series by 1.2368 g/VMT when average
vehicle weight is less than 23 tons or 2.1227 g/VMT when average vehicle weight is over 23
tons yields the portion of emissions due to heavy duty trucks. Multiplying the ratio of light duty
vehicles in a test series by 0.1328 g/VMT yields the portion of emissions due to light duty
vehicles.
4-40
-------
During very low vehicle speeds or when vehicles are moving intermittently, emissions
are estimated from the hourly emissions estimates for heavy duty diesel trucks. By dividing the
hourly emissions estimate by the average vehicle speed, an estimated emissions factor for the
test run is produced. This test run specific emissions factor estimate for engine exhaust, tire
wear and brake wear is subtracted from the test run measured emissions factor to produce the test
run specific emissions factor due to road surface material. To allow log transformation of the
data, values of zero or less were set to 0.02 g/VMT. Table 4-17 presents the final dependent and
independent variables for all of the useable test series that were assembled for developing the
paved road emissions factor equation. It should be noted that there were 10 out of the 93
available data sets where the estimated emissions from engine exhaust, tire wear and break wear
were equal to or comparable to the measured emissions. Tests of high speed traffic dominated
the lowest site specific emissions factor tests. Four of the six test runs where the average vehicle
speed was 55 mph had estimated engine exhaust, tire wear and break wear emissions greater than
the measured emissions. The emissions of another test run with vehicle speeds of 55 mph were
over 80% engine exhaust, tire wear and break wear emissions. The silt level for the last run was
greater than all other 55 mph data sets and was performed to characterize emissions from a road
that had been sanded for traction control. For slightly slower moving traffic (40 - 45 mph), three
of the five test runs had significant percentage of engine exhaust, tire wear and brake wear
emissions. One of the remaining two runs had silt levels greater than 60% of the entire data set
and the test was performed to characterize emissions from a road that had been sanded for
traction control.
Graphical presentations of the final PMi0 data base are shown in Figures 4-1 and 4-2.
Because of the large range of silt loadings and estimated emissions factors, the data are plotted
on a logarithmic scale. Figure 4-1 presents the data base by silt loading with nine ranges of
average vehicle weight depicted with different shape and color data points. Figure 4-2 presents
the data base by average vehicle weight with eight ranges of silt loading depicted with different
shape and color data points. From the graphical presentation, it is evident that PM10 emissions
factors are influenced by both road silt loading and vehicle weight. However, it appears that
emissions factors are influenced more by the road surface silt loading than by vehicle weight.
4-41
-------
Table 4-17. Final Paved Roads Emissions Factor Data Set
Downwind Estimated Estimated Estimated PM- 10
Silt Concentration Measured PM- 10 Fraction Engine, brake, Road Dust
loading Speed Weight mg/m Emission factor Heavy Duty tire emission Emission factor
Reference Run ID (g/m2) (mph) (tons) (g/VMT) Vehicles factor (g/VMT) (g/VMT)
USX 5/1 990
EPA 7/1 984
AUC3
AUC4
AUC5
AUC6
AUC7
AUC8
AUE1
AUE2
AUE3
AUE4
M-l
M-2
M-3
M-4
M-5
M-6
M-7
M-8
M-9
M-10
M-ll
M-12
M-13
M-14
M-15
M-16
M-17
0.42
0.52
0.23
0.23
0.26
0.15
4
4
2.2
1.3
0.46
0.26
0.147
0.432
1.01
0.716
0.59
2.48
0.293
0.022
0.022
0.022
0.11
0.079
0.049
0.022
0.809
27
25
29
27
27
27
15
16
15
15
30
30
30
35
35
30
35
20
30
55
55
55
35
35
35
55
30
5.5
6
3.9
6.2
3
2
12
5.1
2.6
2.6
5.6
3.8
4.5
2.1
2.2
2.1
2.3
2.2
4.1
4.5
4.8
3.8
2.7
2.7
2.7
4.3
2
0.011
0.04
0.07
0.03
0.01
0.03
0.01
0.6
0.08
0.06
0.124
0.033
0.070
0.030
0.090
0.063
0.130
0.120
0.130
0.104
0.080
0.080
0.065
0.030
0.090
0.060
0.056
2.25
16.1
15.3
3.7
0.402
7.88
3.22
10.6
16.1
9.01
4.99
1.55
3.54
0.177
0.692
1.38
4.22
11.2
3.24
0.177
0.322
0.084
0.306
1.37
1.47
0.241
2.64
0.153
0.173
0.087
0.182
0.050
0.009
0.420
0.136
0.033
0.033
0.157
0.083
0.112
0.013
0.017
0.013
0.021
0.017
0.095
0.112
0.124
0.083
0.038
0.038
0.038
0.103
0.009
0.3014
0.3241
0.2287
0.3332
0.1878
0.1424
0.5967
0.2832
0.1696
0.1696
0.3059
0.2242
0.2560
0.1469
0.1515
0.1469
0.1560
0.1515
0.2378
0.2560
0.2696
0.2242
0.1742
0.1742
0.1742
0.2469
0.1424
1.949
15.776
15.071
3.367
0.214
7.738
2.623
10.317
15.930
8.840
4.684
1.326
3.284
0.030
0.541
1.233
4.064
11.049
3.002
0.020
0.052
0.020
0.132
1.196
1.296
0.020
2.498
4-42
-------
Downwind Estimated Estimated Estimated PM- 10
Silt Concentration Measured PM- 10 Fraction Engine, brake, Road Dust
loading Speed Weight mg/m3 Emission factor Heavy Duty tire emission Emission factor
Reference Run ID (g/m2) (mph) (tons) (g/VMT) Vehicles factor (g/VMT) (g/VMT)
EPA 1/1 983
EPA 8/1 983
M-18
M-19
Yl
Y2
Y3
Y4
Zl
Z2
Z3
AC4
ACS
AC6
ADI
AD2
AD3
F34
F35
F36
F37
F38
F39
F27
F32
F61
F45
F62
F74
B52
0.731
0.929
90.7
76.1
193
193
11.3
12.4
12.4
287
188
399
94.8
63.6
52.9
2.78
2.03
0.201
0.417
0.218
0.441
14.8
0.117
17.9
5.11
14.4
5.59
7.19
30
30
10
10
10
10
10
15
15
10
15
20
23
23
23
NR
NR
NR
NR
NR
NR
NR
NR
NR
NR
NR
NR
NR
2
2.4
3.6
3.7
3.8
3.7
8
8
8
5.7
7
3.1
42
39
40
28
25
8.3
17
18
18
14
14
40
16
36
29
12
0.080
0.050
0.552
0.057
0.134
0.163
0.301
0.177
0.531
0.138
0.327
0.744
0.294
0.114
0.102
0.37
0.177
117
182
36.3
200
317
740
1820
1750
1420
613
1480
342
233
188
298
54.7
77.2
167
253
130
53.1
463
212
317
545
35.4
0.009
0.025
0.075
0.079
0.083
0.079
0.256
0.256
0.256
0.161
0.214
0.054
1.000
1.000
1.000
0.789
0.699
0.268
0.626
0.667
0.667
0.502
0.502
1.000
0.585
1.000
0.819
0.420
0.1424
0.1606
0.2151
0.2196
0.2242
0.2196
0.4150
0.4150
0.4150
0.3105
0.3695
0.1924
2.1227
2.1227
2.1227
1.7033
1.5235
0.4286
0.8238
0.8692
0.8692
0.6875
0.6875
2.1227
0.7784
2.1227
1.7632
0.5967
0.228
0.016
116.785
181.780
36.076
199.780
316.585
739.585
1819.585
1749.690
1419.630
612.808
1477.877
339.877
230.877
186.297
296.476
54.271
76.376
166.131
252.131
129.312
52.412
460.877
21 1 222
314.877
543.237
34.803
4-43
-------
Downwind Estimated Estimated Estimated PM- 10
Silt Concentration Measured PM- 10 Fraction Engine, brake, Road Dust
loading Speed Weight mg/m3 Emission factor Heavy Duty tire emission Emission factor
Reference Run ID (g/m2) (mph) (tons) (g/VMT) Vehicles factor (g/VMT) (g/VMT)
EPA 4/1 997
CRA 5/2008
B50
B51
B54
B55
B56
B58
B57
B59
B60
BH1
BH2
BH6
BJ6
BJ7
BK7
BK8
CE-1
CE-2
CE-11
CE-3
CE-1 5
CE-1 6
CE-1 7
CE-1 9
CE-1 2
CF-1N
CF-1 /South
CF-2N
13.6
13.6
3.77
6.3
2.4
10.4
2.32
2.06
3.19
0.184
0.0127
1.47
0.06
0.06
0.082
0.082
1.16
0.86
1.34
0.86
1.91
1.41
2.93
0.76
1.34
0.97
0.97
0.81
NR
NR
NR
NR
NR
NR
NR
NR
NR
55
55
40
45
45
45
45
1
1
5
1
5
5
5
5
5
5
1
5.3
9.4
11
10
11
9.2
18
12
11
12
2.2
2.2
2.2
2.2
2.2
2.2
2.2
36
36
12
39
40
40
40
38
13
40
40
41
0.225
0.410
0.187
0.295
0.229
0.190
0.358
0.149
0.339
0.233
0.030
0.300
0.045
0.130
0.033
0.033
0.050
0.075
0.200
0.070
0.065
0.050
0.040
0.040
0.085
0.035
0.040
0.044
82.1
140
93.3
183
126
368
195
348
439
1.08
0.102
4.68
0.301
1.94
0.57
0.44
27
64
154
45
63.5
77.1
41.3
18.6
23.1
4.99
19.5
16.3
0.313
0.379
0.338
0.379
0.305
0.667
0.420
0.379
0.420
0.017
0.017
0.017
0.017
0.017
0.017
0.017
1.000
1.000
0.420
1.000
1.000
1.000
1.000
1.000
0.461
1.000
1.000
1.000
0.4786
0.5512
0.5058
0.5512
0.4695
0.8692
0.5967
0.5512
0.5967
0.1515
0.1515
0.1515
0.1515
0.1515
0.1515
0.1515
11.06
11.06
2.212
11.06
2.212
2.212
2.212
2.212
2.212
2.212
11.06
2.0868
81.621
139.449
92.794
182.449
125.531
367.131
194.403
347.449
438.403
0.929
0.020
4.529
0.150
1.789
0.419
0.289
15.940
52.940
151.788
33.940
61.288
74.888
39.088
16.388
20.888
2.778
8.440
14.213
4-44
-------
Downwind Estimated Estimated Estimated PM- 10
Silt Concentration Measured PM- 10 Fraction Engine, brake, Road Dust
loading Speed Weight mg/m3 Emission factor Heavy Duty tire emission Emission factor
Reference Run ID (g/m2) (mph) (tons) (g/VMT) Vehicles factor (g/VMT) (g/VMT)
CF-2/South
CF-3N
CF-3/South
CF-4N
CF-5
CI-7
CI-8
CM-1
CM-2
CM-4
0.81
0.63
0.63
1.1
1.4
0.05
0.05
0.72
0.72
0.7
1
5.1
1
4.7
1
15.3
15.3
5
1
5
41
41
41
41
41
27
27
39.8
39.6
39.5
0.080
0.015
0.025
0.019
0.030
0.030
0.030
0.035
0.050
0.035
63.5
1.09
23.1
3.08
16.3
1.63
2.99
6.35
6.35
7.26
1.000
1.000
1.000
1.000
1.000
0.759
0.759
1.000
1.000
1.000
11.06
2.1686
11.06
2.3532
11.06
1.6434
1.6434
2.212
11.06
2.212
52.440
0.020
12.040
0.727
5.240
0.020
1.347
4.138
0.020
5.048
4-45
-------
Emissions Factor vs. Sift Level
10000
1000
100
"
0.1
0.01
0.01
0.1 1 10
A -
^~
-
• *
Q
• * 0
A 4nfej
I
^TS
A ' -
L A
Q\-
^ ^
Vtc
Di. i.
n
£ -
1 —
1 •
D
4
•
ii f|
^V A
^
B
A
A
A
Weight
(tons)
- 2to3
A 3to4
• 4to6
* 6toLO
• 10 to 15
• 15 to 10
4 20 to 30
A 30 to 40
PI over 40
100
1000
Figure 4-1 PMio Emissions Factor Data Base (93 test runs).
4-46
-------
PM10 Emissions Factor vs. Vehicle Weight
lUUUU.U
^000.0
$ 100.0
1
810.0
W. 1.0
0.1
nm
nno
n o
.^
n
£ . /' *'
*!.-
A
D
_~ A /K
g* • 25
10
Vehfcte Weight (torn)
Figure 4-2 PMio Emissions Factor Data Base (93 test runs).
100
4-47
-------
4.2.2.2. Emission Factor Development.
Stepwise multiple linear regression was used to develop a predictive model with the final
data set. The potential correction factors included:
- silt loading, sL
- mean vehicle weight, W
- mean vehicle speed, S
All variables were log-transformed in order to obtain a multiplicative model as in the
past. Table 4-18 presents the correlation matrix of the log-transformed independent and
dependent variables. The most notable feature of the correlation matrix is the high degree of
correlation between silt loading and emissions factors. The correlation between emissions
factor, weight and speed is much lower than with silt loading. The high correlation between
weight and speed is believed to be the result of the large data collected by the corn refiners
association to characterize emissions at terminals. This suggests that obtaining accurate silt
loading information is the most important independent variable to obtain for accurately
estimating emissions factors.
Table 4-18 Correlation Matrix for log-transformed PM10 data.
PM-10 Emission factor (g/VMT)
Silt loading (g/m2)
Weight (tons)
Speed (mph)
PM-10 Emission
factor (g/vmt)
1
0.7918
0.2745
-0.3339
Silt loading
(g/m2)
1
0.1838
-0.2783
Weight
(tons)
1
-0.7784
Speed
(mph)
1
The initial regression analysis evaluated the independent variables of silt level and
average vehicle weight. The stepwise regression was first performed using the "Regression"
function in the "Analysis Tool" of Excel. Since the speed term was not included, the entire data
set of 93 test runs were included in the analysis. Table 4-19 shows the output from Excel. The
predicted exponents for silt and weight are 0.92 and 0.91 respectively and have an adjusted
coefficient of determination (R2) of 0.57. The second regression step, used the adjusted
emissions factors as the dependent variables and a single independent variable that was the silt
loading (divided by 2) and weight (divided by 3) raised to their respective exponents. Table 4-
4-48
-------
20 shows the output from Excel. The predicted multiplier for the predictive equation is 4.10 and
the associated adjusted R2 is 0.32. As a result, if only silt loading and average vehicle weight
were used, the predictive equation would be:
/ T \0.92 / \0.91
•*-" °(f) (f)
While it is expected that speed and silt loading would be highly inter-correlated under
normal conditions. Many of the paved roads in the data base had external influences affecting
the silt loading. In some cases, sand was used to improve traction during cold weather. In other
cases, there was active removal of deposited silt from the road. It is believed that these external
influences affecting the silt loading were the principal reasons that the correlation between these
two independent variables was only -0.28. To evaluate the viability of including speed as an
additional term in the predictive equation, the 1983 EPA data set was excluded from the analysis
and the remaining 71 test data runs were used in the stepwise regression analysis. Table 4-21
shows the output from Excel. The predicted exponents for silt, weight and speed are 0.98, 0.53
and 0.16 respectively and have an adjusted R2 of 0.61. The second regression step, used the
adjusted emissions factors as the dependent variables and a single independent variable that was
the silt loading (divided by 2), weight (divided by 3) and speed (divided by 30) raised to their
respective exponents. Table 4-22 shows the output from Excel. The predicted multiplier for the
predictive equation is 6.79 and the associated adjusted R2 is 0.44. While the addition of the
speed term produces little difference in the adjusted R2 for the slope (exponent portion) of the
predictive equation, the speed term does provide a modest improvement of the adjusted R2 for
the final equation. As a result, the three parameter equation is selected for the new predictive
equation. When silt loading, average vehicle weight and speed in the development process, the
predictive equation would be:
EF = 6.79
2 3 30
Where: EF is in g/VMT,
sL is in g/m2,
W is in tons and
S is in mph.
4-49
-------
Table 4-19. Initial Regression Analysis using Silt Loading and Weight.
SUMMARY OUTPUT
Regression Statistics
Multiple R
R Square
Adjusted R Square
Standard Error
Observations
ANOVA
Regression
Residual
Total
Intercept
Silt loading (g/m2)
Weight (tons)
0.7632
0.5824
0.5668
2.0682
93
df
2
91
93
Coefficients
0
0.9240
0.9051
ss
542.8875
389.2504
932.1379
Standard Error
#N/A
0.0934
0.0883
MS
271.4438
4.2775
tStat
#N/A
9.8923
10.2489
F
63.4588
P-value
#N/A
0.0000
0.0000
Significance F
6.42153E-18
Lower 95%
#N/A
0.7385
0.7297
Upper 95%
#N/A
1.1096
1.0806
4-50
-------
Table 4-20. Second Regression Analysis using Silt Loading and Weight.
SUMMARY OUTPUT
Regression Statistics
Multiple R
R Square
Adjusted R Square
Standard Error
Observations
0.5782
0.3343
0.3234
282.63
93
ANOVA
Regression
Residual
Total
Intercept
Silt/Weight
df
1
92
93
Coefficients
0
4.098226
SS
3690191.28
7348732.32
11038923.60
Standard Error
#N/A
0.4725
MS
3690191.28
79877.53
tStat
#N/A
8.6738
F
46.20
P-value
#N/A
1.41E-13
Significance F
1.086E-09
Lower 95%
#N/A
3.1598
Upper 95%
#N/A
5.0366
4-51
-------
Table 4-21. Initial Regression Analysis using Silt Loading, Weight and Speed.
SUMMARY OUTPUT
Regression Statistics
Multiple R
R Square
Adjusted R Square
Standard Error
Observations
ANOVA
Regression
Residual
Total
Intercept
Silt loading (g/m2)
Weight (tons)
Speed (mph)
0.8002
0.6404
0.6151
1.9308
71
df
3
68
71
Coefficients
0
0.9804
0.5298
0.1594
ss
451.4629
253.5105
704.9734
Standard Error
#N/A
0.0951
0.1269
0.1057
MS
150.4876
3.7281
tStat
#N/A
10.3065
4.1770
1.5074
F
40.3658
P-value
#N/A
1.57E-15
8.61E-05
0.1363
Significance F
5.12E-15
Lower 95%
#N/A
0.7906
0.2767
-0.0516
Upper 95%
#N/A
1.1702
0.7830
0.3704
4-52
-------
Table 4-22. Second Regression Analysis using Silt Loading, Weight and Speed.
SUMMARY OUTPUT
Regression Statistics
Multiple R 0.6764
R Square 0.4575
Adjusted R Square 0.4432
Standard Error 285.3151
Observations 71
ANOVA
df SS MS F Significance F
Regression 1 4804841.24 4804841.24 59.0241 7.60E-11
Residual 70 5698331.03 81404.72
Total 71 10503172.27
Coefficients Standard Error t Stat P -value Lower 95%
Intercept 0 #N/A #N/A #N/A #N/A
Silt/weight/speed 6.7871 0.7779 8.7245 8.47E-13 5.2355
Upper 95%
#N/A
8.3386
4-53
-------
The range of conditions which existed at the test sites used in developing the equation
was as follows:
Silt loading: 0.03 - 400 g/m2
0.01 - 570 grains/square foot (ft2)
Mean vehicle weight: 1.8 - 38 megagrams (Mg)
2.0 - 42 tons
Mean vehicle speed: 1-88 kilometers per hour (kph)
1-55 miles per hour (mph)
An assessment of the performance of the predictive equation is difficult since the range
of silt loadings and the associated emissions factors spans five orders of magnitude. This is
further complicated by the focus of many of the field tests. Approximately half of the field test
locations were selected either due to concerns that these sources were major contributors to air
quality impacts, or were selected because of elevated road silt levels to allow the measurement of
a difference from background concentrations of paniculate matter. Another complication is that
PM emissions of the vehicle exhaust were not measured during the tests and a modeled national
average emission factor or rate was subtracted to arrive at the road dust emissions. Lastly, the
use of three independent variables makes it difficult to show the performance characteristics
graphically.
One can assess the performance of the predictive equation by calculating the average
predicted to actual ratio and producing the cumulative distribution of these ratios. For the three
parameter equation, the average predicted to actual ratio is 13.8. This is significantly lower than
the average predicted to actual ratio of 277 for the previous two parameter equation when
applied to the existing data. It should be noted that the previous equation subtracted the national
average engine exhaust from the resulting equation. Figure 4-3 depicts the cumulative
distribution of the predicted to actual ratios for both the previous equation and the new equation.
Figure 4-4 presents this same information but with ranges of silt loading depicted through the
use of different shapes and colors for the markers of the data. Figure 4-5 is this same
information but with ranges of vehicle weights depicted with different markers. It is difficult to
discern any differences below the ratio of 1.0. Above the ratio of 1.0 the increased variation of
the older
4-54
-------
Predicted vs Actual Cumulative Distribution
1
n o
u.y
0.8
0.7
n a
tl.O
O.S
0.4
0.3
0.2
0.1
n
•m*
c
Cn
O
q
o
*s
• •*
u-
***
•n^ tt
• o»*
>°
•
1
2> 1
o 0 0 •
"/"«
^{ a °o
0* o o°
1
•0* »°
of' So 0
0
(^p
*>
ONew
Equation
o Previofos Equation
* *_j
* .11
4
€
-a
5
^
0.01
0.1
1
10
100
1000
10000
Predictive vs. Actual EF
Figure 4-3 Cumulative Distribution of Predicted vs. Actual PMio EFs.
4-55
-------
Predicted TO Actual Cumulative Distribution
tf *
. ^ .....
D
D
~i
%**
^B
A>'
• n
<**°
•
.
° *~ "
A
V. o °o
» ^
1
Jl 2
SL* D-X> °
o D
Q
^6>
* o
—•9
Silt Loading
NewBq-
DOldEq-
NewEq-0.1
sL80
sL>80
0.9
0.8
0.7
0,6
!O.S
3.4
03
0.1
0.001
0.01
0.1
10
100
1000
10000
Figure 4-4 Cumulative Distribution of Predicted vs. Actual PMio EFs with Silt Loadings.
4-56
-------
Predicted vs Actual Cumulative Distribution
.,#!
0
o
o
oE
•
£
/:
• o •
v°
&
*
«
0^ ^«
o
0 H^-^l
m •
. • D
* * D °
3 *ocP
•
• rj
, '• ^ * D
J1 8? "
D D
D
535]
Vehicl
«NewEq
OOldEq-
• NewEq
noldEq-
NewEq
OldEq-
• NewEq
OOldEq-
• NewEq
D OldEq
t Weight
W<3
W<3
- 3
-------
equation is evident. Middle and lower silt loadings appear to have a greater variation in the
predicted to actual ratios when the older equation is used. Vehicle weights appear to have a
measurable influence on spread of the cumulative distribution when the older equation is used.
Another means of assessing the performance of the regression equations is to compare
the calculated results of the equations to the actual value measured. With a large range of
measured emissions factors, comparing the relative percent difference between the results of the
equation and the measured value places the differences in the smallest measured value and the
largest measured value on comparable terms. Three comparisons were made to assess the
relative predictive performance of the existing equation to the previous equation. As shown with
the average percent error for the entire population in Table 4-23, the new equation provides an
improved estimate of the measured emissions over the previous equation. When the
performance of the equation is evaluated within classes of the independent variables, the new
equation shows comparable or improved performance in all groups of the variables except two.
Figure 4-6 through Figure 4-9 provide graphical indications of the performance of the
two equations to estimate the actual emissions. The first two figures show the relationship of
emissions to the road surface silt loading. Figure 4-6 provides this information with the addition
of average vehicle weight and Figure 4-7 provides this information with the addition of average
vehicle speed. In both figures the estimated emissions factors predicted by the previous
equations show a greater spread than the new equations estimates. Figure 4-8 shows the
influence of vehicle weight on the emissions factors. The spread of the data is much greater than
is demonstrated in the figures with silt as the ordinate. One can see a general increase in
emissions with silt loading. This is probably due to the greater influence that silt loading has on
the emissions. Figure 4-9 shows the influence of speed on the emissions factors. As with
vehicle weight, there is a greater spread of the emissions factor than when silt is the primary
dependent variable graphed. One can also see a weak relationship between silt loading and
average vehicle speed.
4-58
-------
Table 4-23. Comparison of Previous and New Equations for Estimating Paved Road Dust Emissions.
Population Average
Predictive Performance of Paved Road Dust Emissions Equations
Average Relative Percent Error :
Old Equation vs
Actual
27,630
By Classes of Silt Loading (g/m2)
<0.2
0.2-0.75
0.75-1.5
1.5 - 50
>50
1,220
530
104,720
6,160
188,760
New Equation
vs Actual
1,280
1,150
67
3,550
380
7,590
By Classes of Average Weight (ton)
2-3
3-5
5-10
10-40
>40
1,100
590
3.7
77,270
71,200
1,070
240
-50
2,420
2,650
Old Equation vs
New Equation
26,340
70
470
101,170
5,790
181,220
40
350
53
74,840
68,550
By Classes of Average Speed (mph)
<10
10-25
25-45
45
55
87,360
4,530
1,230
-3.8
1,150
2,970
240
1,120
-27
250
84,400
4,300
110
23
890
Relative Standard Deviation
Old Equation
vs Actual
5.3
3.5
2.0
2.7
2.0
0.89
3.6
1.5
29.9
3.2
3.4
New Equation
vs Actual
4.3
3.6
4.0
2.8
1.6
0.77
3.6
1.7
-1.0
3.2
3.4
Old Equation vs
New Equation
5.4
6.0
1.7
2.7
2.1
0.89
5.2
2.4
1.5
3.2
3.4
3.0
4.2
0.047
-15.3
1.0
3.1
2.7
0.039
-1.7
0.8
3.0
4.4
0.13
0.6
1.1
1 Relative Percent Error = (((Result 1 EF)
2 Relative Standard Deviation = Standard
- (Result 2 EF)}/Actual EF)xlOO.
Deviation / Average
4-59
-------
10000
1000
Actual, Old Predicted & New Predicted EF by Silt loading
D
>**
Q
D D Jj
. •*
: • *
• 4t
1 1 1 1 L J I I
nnd
•
* %*l
n
T
1 * A
. 1 *** t
n ^| 1 |4k *fi
? **irfr
jf • ••>
•
•
A
1 1 1 [ 1 1 I 1
?
a
• «
QH
^P
^
* ° *
*
Category Weight
* Actual W < 4
d Old Pred W < 4
• New Pred W<4
A rtnal A
-------
Aetna], Old Predicted & New Predicted by Sitt Loadug
10000
1000
100
10
D
-1°
! v
n n •
i i i i i 1 1
IHM
j
Pl°
• D
1 9
* .
• _
1
.
1 t L 1 1 1 1 L
1 1 I 1 1 1 L L
m
r-m n
m
B n
Speed
DPrevEq- S<
• Actual - S<
• NewEq- S
• Actual - 10
-------
Actual, Old Predicted & New Predicted by Vehicle Weight
10000
1000
100
10
Silt Loading
0.1
0.01
0.001
• Actual - sL<0.15
*NewEq- sL<0.15
D PrevEq- sL<0.15
• Actual - 0.15
-------
Actual, Old Predicted & New Predicted by Vehicle Speed
10000
l
i
E
e
R
0.001
-
.
: • - s
Sm _ •
-— • — B
A • •
* • $ ^
A a "* y tt
1 i
*
•* * * *
;
CiT* T x>vf^1
» Actual - &L<0.2
QOldEq - sL<0.2
»Actual -0.2
-------
4.2.2.3 Emissions Factor Quality Rating Assessment.
All of the source test data used to develop the emissions factor equation were rated A
since the test procedures used were profiling tests and were all well documented. While only six
reports are available that provide documentation of emissions factors for paved roads, these test
reports contain the results of 17 different road conditions. The reports and the number of test
conditions documented in the report are:
• USX 5/1990 - 2 tests (sL ~. 3 & sL > 2),
• EPA 7/1984 - 2 tests (30 mph & 55 mph)
• EPA 1/1983 - 4 tests (<15 mph, >20 mph, W < 3 tons, W 5-8 tons, W > 30 tons)
• EPA 8/1983 - 2 tests for two parameter equation, 0 for three (no speed info)
• EPA 4/97 - 3 tests (speed 55, 45), 3 locations,
• CRA 5/2008 - 4 tests (4 locations, 2 speeds,)
However, since the EPA 8/1983 report does not contain information on the average speed
of the vehicles in the study, none of the tests documented in that report is usable for the three
factor equation. The remaining five reports contain the results of 15 different road conditions.
While all of the tests were performed on paved roads, the ranges of conditions (silt loading,
vehicle speed and vehicle weight) were diverse. An assessment of the variation associated with
the data and the impact of that variation on a single value emissions factor. The average of all
the adjusted emissions factors is 140 g/VMT and the standard deviation is 387. A relative
standard deviation of 3 is greater than many other factors. As a result, the number of tests
needed to achieve the predictive accuracy of the mean is greater. The availability of 15 A or B
rated test reports would normally justify an initial assignment of a factor rating of B. However,
the greater variability of the underlying data justifies a single value factor rating of C.
The stepwise regression of the available data indicated that a large portion of the
variation of the emissions factor was due to the large range of the road silt loading that existed at
the test locations. The first regression step estimates the exponents associated with the three
parameter equation. The adjusted correlation coefficient (R2) for the exponents is 0.62. This
indicates that approximately 62% of the variations in the emissions factors are due to the silt
level, vehicle weight and speed. The second step in the regression identifies the constant
associated with the equation. The adjusted correlation coefficient (R2) for the final equation is
4-64
-------
0.44. While this is not as high as is typically used to justify the use of the equation results, it
follows a regression with a greater correlation coefficient. As a result of the improved ability of
the equation to estimate the measured values over the single value emissions factor, a quality
rating of B is assigned to the equation.
4.2.2.4 Assignment of equation parameters for PM30 and PM2.5.
While several of the reports include measurements of PM25, the WRAP studies suggest
that many of these measurements are in error due to particle bounce issues with the impactor
stages. The results of the WRAP study indicated that the PM2 5 concentrations measured by the
cyclone/impactor system were consistently biased by a factor of about 2 relative the PM2 5
concentrations measured by the Partisol samplers. The second phase of the WRAP showed a
tendency of the measured PM2.5/PMi0 ratio to decrease with increasing PMi0 concentration. At
PMio concentrations above 1.0 mg/m3 the PM25/PMi0 ratio was between 0.1 and 0.15. The
PM2 s/PMio ratio increased to about 0.35 as the PMio concentration approached about 0.5 mg/m3.
While some of the paved road test data encountered concentrations above 1.0 mg/m3 much of
the test data consisted of measured concentrations below 0.5 mg/m3. The paved road emissions
factor for PM2 5 was revised to 15% of the calculated PMio emissions factor in 2008. It is not
clear whether the WRAP study assessed the PMio concentrations measured during the paved
roads testing prior to their recommendations for revising the PM2 5 emissions factors. As shown
in Table 4-17 the PMio concentrations associated with 58 of the 71 test runs used to develop the
three parameter emissions factor equation. Many of these test runs involve traffic volumes that
would produce fairly constant paniculate concentrations. Also, of these 58 test runs, only three
runs were the highest PMio concentrations greater than 0.5 mg/m3. An earlier report (Reference
5) measured PM25/PMi0 ratios during field tests. The range of PM25/PMi0 ratios were from 0.25
to 0.37. Since essentially all of the measured PMio concentrations used for the stepwise
regression were below 0.5 mg/m3 and the ratios measured during field sampling of paved road
emissions were between 0.25 and 0.37, the recommended PM2 5 emissions factor is 25% of the
PMio emissions factor. Since there is little measured PM2 5 data, an emissions factor quality
rating of "D" is assigned.
4-65
-------
While a stepwise regression could be performed to estimate the PM30 emissions factor
equation, it is believed that the number of available data would be significantly less and a
comparable confidence in the resulting equation could not be achieved. The ratio of PM30 to
PMio presented in the present AP-42 section is 5.2 and is proposed for the revised equation.
4.2.2.5. Assignment of a precipitation correction factor.
As is presented in Reference 32, a correction parameter for precipitation events was included in
the revision of the AP-42 section in October 2002. As recommended in the Technical
Memorandum to the files, the correction parameters are retained in this version of the AP-42
section.
4.3 DEVELOPMENT OF OTHER MATERIAL IN AP-42 SECTION
Concurrent with the development of the revised AP-42 section for paved roads, a
separate effort was conducted to assemble a silt loading data base for nonindustrial roads. Over
the past 10 years, numerous organizations have collected silt loading samples from public paved
roads. Unfortunately, uniformity—in sampling and analysis methodology as well as roadway
classification schemes—has been sorely lacking in these studies.
Silt loading data were compiled in the following manner. Persons knowledgeable about
PM-10 at each EPA regional office were asked to identify sL data for public roads. In many
instances, the EPA representatives identified state/local air regulatory personnel who were then
asked to supply the data. Given that the relative importance of PM-10 emissions from public
sources is greater in the western United States, it is not surprising that most of the data are from
that area of the country. What is surprising, perhaps, is that Montana has collected roughly two-
thirds of all data. Furthermore, only Montana had data collected from the same road over
extended periods of time, thus permitting examination of temporal variation.
The assembled data set did not yield any readily identifiable, coherent relationship between silt loading
and road class, average daily traffic (ADT), etc. Much of the difficulty is probably due to the fact that not all
4-66
-------
variables were reported by each organization. Further complicating the analysis is the fact that, in many parts
of the country, paved road silt loading varies greatly over the course of the year. Recall that repeated sampling
at Montana municipalities indicated a very noticeable annual cycle. Nevertheless, it is questionable whether the
seasonal variation noted in the Montana data base could successfully predict variations for many other sites.
While one could possibly expect similar variations for, say, Idaho or Wyoming roads, there is far less reason to
suspect a similar cycle in, say, Maine or Michigan, in the absence of additional information.
Because no meaningful relationship could be established between sL and an independent variable, the
decision was made to directly employ the nonindustrial data base in the AP-42 section. The draft AP-42 section
presents the cumulative frequency distribution for the sL data base, with subdivisions into (a) low-ADT (< 5000
vehicles/day) and high-ADT roads and (b) first and second halves of the year. Suggested default values are
based on the 50th and 90th percentile values.
The second use of the assembled data set recognizes that the end users of AP-42 are the
most capable in identifying which roads in the data base are similar to roads of interest to them.
The draft AP-42 section presents the paved road surface loading values together with the city,
state, road name, collection date (samples collected from the same road during the same month
are averaged), road ADT if reported, classification of the roadway, etc. Readers of AP-42 are
invited to review the data base and to select values that they deem appropriate for the roads and
seasons of interest.
4.3 References for Section 4
1. Roadway Emissions Field Tests at U.S. Steel's Fairless Works, U.S. Steel Corporation,
Fairless Hills, PA, USX Purchase Order No. 146-0001191-0068, May 1990.
2. Paved Road Particulate Emissions—Source Category Report, U. S. Environmental
Protection Agency, Research Triangle Park, NC, EPA Contract No. 68-02-3158,
Assignment 19, July 1984.
3. Iron and Steel Plant Open Source Fugitive Emission Control Evaluation, U.S.
Environmental Protection Agency, Research Triangle Park, NC, EPA Contract No. 68-02-
3177, Assignment 4, August 1983.
4-67
-------
4. Size Specific P articulate Emission Factors for Uncontrolled Industrial and Rural Roads,
U.S. Environmental Protection Agency, Research Triangle Park, NC, EPA Contract
No. 68-02-3158, Assignment 12, January 1983.
5. Fugitive Particulate Matter Emissions, U.S. Environmental Protection Agency, Research
Triangle Park, NC, Midwest Research Institute Project No. 4604-06, April 15, 1997.
6. T. Cuscino, Jr., et al., Iron And Steel Plant Open Source Fugitive Emission Control
Evaluation, EPA 600/2 83 110, U. S. Environmental Protection Agency, Cincinnati, OH,
October 1983.
7. Paved Road Modifications to AP-42, Background Documentation For Corn Refiners
Association, Inc. Washington, DC 20006, Midwest Research Institute Project No. 310842,
May 20, 2008.
8. Emission Tests of Paved Road Traffic at Minnesota Corn Processors Marshall, Minnesota
Facility, McVehil-Monnett Associates, Midwest Research Institute Project No.
310212.1.001, July 6, 2001.
9. Emission Tests of Paved Road Traffic at Minnesota Corn Processors Columbus, Nebraska
Facility, McVehil-Monnett Associates, Midwest Research Institute Project No.
310212.1.002. July 13,2001.
10. Emission Tests ofPavedRoad Traffic at Cargill Sweeteners North America Blair, Nebraska
Facility, McVehil-Monnett Associates, Midwest Research Institute Project No.
310395.1.001. November 27, 2002.
11. Emission Tests of Paved Road Traffic at ADM's Marshall, Minnesota Facility, McVehil-
Monnett Associates, Midwest Research Institute Project No. 310479.1.001. December 5,
2003.
12. E.H. Pechan & Associates, Inc., Recommendations for Emission Factor Equations in AP-42
Paved Roads Section: TECHNICAL MEMORANDUM August 21, 2003.
13. E-mail communication between Ron Myers of EPA/OAQPS/SPPD/MPG, RTF, NC and
Prashanth Gururaja and Ed Glover of EPA/OTAQ/ASD/HDOC re. Diesel exhaust, tire and
brake wear for low speed stop and go traffic, January 2009 through May 2009.
14. Midwest Research Institute; Analysis of the Fine Fraction of Particulate Matter in Fugitive
Dust; Western Governors' Association - Western Regional Air Partnership (WRAP);
October 12, 2005.
15. Midwest Research Institute; Background Document for Revisions to Fine Fraction Ratios
Used for AP-42 Fugitive Dust Emission Factors; Western Governors' Association -
Western Regional Air Partnership (WRAP); November 1, 2006.
4-68
-------
16. PEI Associates 1989. Street Sanding Emissions and Control Study, EPA Contract No. 68-
02-4394, Work Assignment No. 27, prepared for U.S. Environmental Protection Agency,
Regions. October 1989.
17.
18. RTF Environmental Associates 1990. Street Sanding Emissions and Control Study,
prepared for the Colorado Department of Health. July 1990.
30. G. E. Muleski, Measurement of Fugitive Dust Emissions from Prilled Sulfur Handling,
Final Report, MRI Project No. 7995-L, Prepared for Gardinier, Inc., June 1984
31. T. F. Eckle and D. L. Trozzo, "Verification of the Efficiency of a Road-Dust Emission-
Reduction Program by Exposure Profile Measurement," Presented at EPA/AISI Symposium
on Iron and Steel Pollution Abatement, Cleveland, Ohio, October 1984.
32. Technical Memorandum from William B. Kuykendal to File, Subject: Decisions on Final
AP-42 Section 13.2.1 "PavedRoads", October 10, 2002.
4-69
-------
References For Section 13.2.1
1. D. R. Dunbar, Resuspension Of Paniculate Matter, EPA 450/2 76 031, U. S. Environmental Protection
Agency, Research Triangle Park, NC, March 1976. (ref01_cl3s0201_2003.pdf)
2. R. Bohn, et al., Fugitive Emissions From Integrated Iron And Steel Plants, EPA 600/2 78 050, U. S.
Environmental Protection Agency, Cincinnati, OH, March 1978.
3. 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, Cincinnati, OH, May 1979.
4. C. Cowherd, Jr., et al., Quantification Of Dust Entrainment From Paved Roadways, EPA 450/3 77 027, U.
S. Environmental Protection Agency, Research Triangle Park, NC, July 1977.
5. Size Specific Paniculate Emission Factors For Uncontrolled Industrial And Rural Roads, EPA Contract No.
68 02 3158, Midwest Research Institute, Kansas City, MO, September 1983.
6. T. Cuscino, Jr., et al., Iron And Steel Plant Open Source Fugitive Emission Control Evaluation, EPA 600/2
83 110, U. S. Environmental Protection Agency, Cincinnati, OH, October 1983.
7. J. P. Reider, Size-specific Paniculate Emission Factors For Uncontrolled Industrial And Rural Roads, EPA
Contract 68 02 3158, Midwest Research Institute, Kansas City, MO, September 1983.
8. C. Cowherd, Jr., and P. J. Englehart, Paved Road Paniculate Emissions, EPA 600/7 84 077, U. S.
Environmental Protection Agency, Cincinnati, OH, July 1984.
9. C. Cowherd, Jr., and P. J. Englehart, Size Specific Paniculate Emission Factors For Industrial And Rural
Roads, EPA 600/7 85 038, U. S. Environmental Protection Agency, Cincinnati, OH, September 1985.
10. Emission Factor Documentation For AP 42, Sections 11.2.5 and 11.2.6 — Paved Roads, EPA Contract No.
68-DO-0123, Midwest Research Institute, Kansas City, MO, March 1993.
11. Evaluation Of Open Dust Sources In The Vicinity Of Buffalo, New York, EPA Contract No. 68 02 2545,
Midwest Research Institute, Kansas City, MO, March 1979.
12. PM 10 Emission Inventory Of Landfills In The Lake Calumet Area, EPA Contract No. 68 02 3891,
Midwest Research Institute, Kansas City, MO, September 1987.
13. Chicago Area Paniculate Matter Emission Inventory — Sampling And Analysis, Contract No. 68 02 4395,
Midwest Research Institute, Kansas City, MO, May 1988.
4-70
-------
14. Montana Street Sampling Data, Montana Department Of Health And Environmental Sciences, Helena, MT,
July 1992.
15. Street Sanding Emissions And Control Study, PEI Associates, Inc., Cincinnati, OH, October 1989.
16. Evaluation Of PM 10 Emission Factors For Paved Streets, Harding Lawson Associates, Denver, CO,
October 1991.
17. Street Sanding Emissions And Control Study, RTF Environmental Associates, Inc., Denver, CO, July 1990.
18. Post-storm Measurement Results — Salt Lake County Road Dust Silt Loading Winter 1991/92
Measurement Program, Aerovironment, Inc., Monrovia, CA, June 1992.
19. Written communication from Harold Glasser, Department of Health, Clark County (NV).
20. PM 10 Emissions Inventory Data For The Maricopa And Pima Planning Areas, EPA Contract No. 68-02-
3888, Engineering-Science, Pasadena, CA, January 1987.
21. Characterization Of PM 10 Emissions From Antiskid Materials Applied To Ice- And Snow-Covered
Roadways, EPA Contract No. 68-DO-0137, Midwest Research Institute, Kansas City, MO, October 1992.
22. Fugitive Particulate Matter Emissions, EPA Contract No. 68-D2-0159, Work Assignment No. 4-06,
Midwest Research Institute, Kansas City, MO, April 1997.
23. Climatic Atlas Of The United States, U.S. Department of Commerce, Washington, D.C., June 1968.
24. C. Cowherd, Jr., et al., Improved Activity Levels for National Emission Inventories of Fugitive Dust from
Paved and Unpaved Roads, Presented at the 11th International Emission Inventory Conference, Atlanta,
Georgia, April 2002.
25. C. Cowherd, Jr., et al., Control Of Open Fugitive Dust Sources, EPA-450/3-88-008, U. S. Environmental
Protection Agency, Research Triangle Park, NC, September 1988.
26. Written communication (Technical Memorandum) from G. Muleski, Midwest Research Institute, Kansas
City, MO, to B. Kuykendal, U. S. Environmental Protection Agency, Research Triangle Park, NC,
September 27, 2001.
27. EPA, 2002b. MOBILE6 User Guide, United States Environmental Protection Agency, Office of
Transportation and Air Quality. EPA420 R 02 028, October 2002.
4-71
-------
28. Written communication (Technical Memorandum) from P. Hemmer, E.H. Pechan & Associates, Inc.,
Durham, NC to B. Kuykendal, U. S. Environmental Protection Agency, Research Triangle Park, NC,
August, 21, 2003.
The following should be added to the section.
29. Midwest Research Institute 1990. "Roadway Emission Field Tests at U.S. Steel's Fairless Works." USX
Purchase Order No. 146-0001191-0068, prepared for United States Steel Corporation. May 1990.
30. G. E. Muleski, Measurement of Fugitive Dust Emissions from Prilled Sulfur Handling, Final Report, MRI
Project No. 7995-L, Prepared for Gardinier, Inc., June 1984.
31. T. F. Eckle and D. L. Trozzo, "Verification of the Efficiency of a Road-Dust Emission-Reduction Program
by Exposure Profile Measurement," Presented at EPA/AISI Symposium on Iron and Steel Pollution
Abatement, Cleveland, Ohio, October 1984.
32. Paved Road Modifications to AP-42, Background Documentation For Corn Refiners Association, Inc.
Washington, DC 20006, Midwest Research Institute Project No. 310842, May 20, 2008.
33. Emission Tests of Paved Road Traffic at Minnesota Corn Processors Marshall, Minnesota Facility,
McVehil-Monnett Associates, Midwest Research Institute Project No. 310212.1.001, July 6, 2001.
34. Emission Tests of Paved Road Traffic at Minnesota Corn Processors Columbus, Nebraska Facility,
McVehil-Monnett Associates, Midwest Research Institute Project No. 310212.1.002. July 13,2001.
35. Emission Tests of Paved Road Traffic at Cargill Sweeteners North America Blair, Nebraska Facility,
McVehil-Monnett Associates, Midwest Research Institute Project No. 310395.1.001. November 27, 2002.
36. Emission Tests of Paved Road Traffic at ADM's Marshall, Minnesota Facility, McVehil-Monnett
Associates, Midwest Research Institute Project No. 310479.1.001. Decembers, 2003.
37. E.H. Pechan & Associates, Inc., Recommendations for Emission Factor Equations in AP-42 Paved Roads
Section: TECHNICAL MEMORANDUM August 21, 2003.
38. E-mail communication between Ron Myers of EPA/OAQPS/SPPD/MPG, RTF, NC and Prashanth Gururaja
and Ed Glover of EPA/OTAQ/ASD/HDOC re. Diesel exhaust, tire and brake wear for low speed stop and
go traffic; January 2009 through May 2009.
39. Midwest Research Institute; Analysis of the Fine Fraction ofParticulate Matter in Fugitive Dust; Western
Governors' Association- Western Regional Air Partnership (WRAP); October 12, 2005.
4-72
-------
40. Midwest Research Institute; Background Document for Revisions to Fine Fraction Ratios Used for AP-42
Fugitive Dust Emission Factors', Western Governors' Association - Western Regional Air Partnership
(WRAP); November 1, 2006.
4-73
------- |