Emission Factor Documentation for AP-42,
Section 13.2.1
Paved Roads
Measurement Policy Group
Office of Air Quality Planning and Standards
U.S. Environmental Protection Agency
January 2011
-------
Emission Factor Documentation for AP-42,
Section 13.2.1
Paved Roads
Measurement Policy Group
Office of Air Quality Planning and Standards
U.S. Environmental Protection Agency
January 2011
-------
NOTICE
The information in this document has been funded by the Office of Air Quality
Planning and Standards, U.S. Environmental Protection Agency (EPA). This final report
has been subjected to the Agency's review, and it has been approved for publication as an
EPA document. Mention of trade names or commercial products is not intended to constitute
endorsement or recommendation for use.
-------
CONTENTS
List of Figures v
List of Tables vi
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 2-1
2.2.1 September 1985 through January 1995 2-1
2.2.2 January 1995 through October 2002 2-4
2.2.3 October 2002 through December 2003 2-5
2.2.4 December 2003 through November 2006 2-5
2.2.5 November 2006 through May 2010 2-6
2.2.6 May 2010 2-6
3. General Data Review and Analysis 3-1
3.1 Literature search and screening 3-1
3.2 Emission data quality rating system 3-1
3.3 Emission factor quality rating system 3-3
3.4 Methods of emission factor determination 3-3
3.4.1 Mass Emission Measurements 3-4
3.4.2 Emission Factor Derivation 3-5
3.5 Emission factor quality rating scheme used in this study 3-6
4. AP-42 Section Development 4-1
4.1 Revisions to section narrative 4-1
4.2 Pollutant emission factor development 4-1
4.2.1 Review of Specific Data Sets 4-2
4.2.2 Emissions Factor Development 4-36
4.3 Development of other material in AP-42 section 4-69
4.4 References for Section 4 4-70
Appendix A Response to Comments A-l
IV
-------
LIST OF FIGURES
Number Page
4-1 PMio Emissions Factor Data Base by Silt Loading (93 test runs) 4-45
4-2 PMio Emissions Factor Data Base by Average Vehicle Weight (93 test runs). 4-46
4-3 Silt Loading vs. Average Vehicle Weight (93 Test Runs) 4-47
4-4 PMio Emissions Factors by Vehicle Speed 4-48
4-5 Vehicle Speed vs Silt Loading 4-49
4-6 Paved Road Dust Emissions Factors, All Data 4-50
4-7 All Paved Road Data, Silt Loading by Vehicle Weight with EF 4-51
4-8 Paved Road Dust Emissions Factor Data Excluding Z-3 4-52
4-9 Cumulative Distribution of Predicted/Actual Ratios 4-57
4-10 Cumulative Distribution - Predicted/Actual by Silt Loading 4-58
4-11 Cumulative Distribution - Predicted/Actual by Average Vehicle Weight 4-59
4-12 Predicted vs Actual PMIO Emissions Factor by Silt Loading 4-62
4-13 Predicted vs Actual PMIO Emissions Factor by Average Vehicle Weight 4-63
4-14 Predictive Accuracy by Silt Loading (unrestricted range) 4-64
4-15 Predictive Accuracy by Silt Loading (restricted range) 4-65
4-16 Predictive Accuracy by Average Vehicle Weight (unrestricted range) 4-66
4-17 Predictive Accuracy by Average Vehicle Weight (restricted range) 4-67
-------
LIST OF TABLES
Number Page
4-1 Summary Information for Reference 15 4-3
4-2 Summary Information for Reference 17 4-6
4-3 Summary information for Reference 31 4-11
4-4 Detailed Information From Paved Road Tests for Reference 31 4-11
4-5 Summary Information for Reference 8 4-12
4-6 Detailed Information From Paved Road Tests for Reference 8 4-13
4-7 Summary of Paved Road Emission Factors for Reference 7 4-15
4-8 Detailed Information From Paved Road Tests for Reference 7 4-16
4-9 Summary of Paved Road Emission Factors for Reference 6 4-18
4-10 Detailed Information From Paved Road Tests for Reference 6 4-19
4-11 Detailed Information From Paved Road Tests for Reference 30 4-22
4-12 Summary of Emissions Data from MCP's Marshall,
Minnesota Facility (Reference 33) 4-24
4-13 Summary of Emissions Data from MCP's Columbus,
Nebraska Facility (Reference 34) 4-26
4-14 Summary of Emissions Data from Cargill's Blair,
Nebraska Facility (Reference 35) 4-27
4-15 Summary of Emissions Data from ADM's Marshall,
Minnesota Facility (Reference 36) 4-27
4-16 Vehicle Fleet Assumption Used in 2003 MOBILE6.2 Model 4-28
4-17 Final Paved Roads Emissions Factor Data Set 4-41
4-18 Correlation Matrix for log-transformed PMio data 4-53
4-19 Regression Analysis using Silt Loading and Weight 4-55
4-20 Comparison of Previous and New Equations for Estimating Paved Road Dust
Emissions 4-61
VI
-------
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 "particulate matter" (PM), with special
emphasis placed on "PMio" - parti culate matter no greater than lOumA (micrometers in
aerodynamic diameter) and PM2.5. PMio and PM2.5 form the basis for the current National
Ambient Air Quality Standards (NAAQSs) for particulate matter. PMio and PM2.5 thus represent
the two size ranges of particulate 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 Particulate, as measured by the standard high-volume (hi-vol) air
sampler. TSP was the basis for the previous NAAQSs for particulate 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 Particulate, 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 Particulate, 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
1-1
-------
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 "PMi5."
FP Fine Particulate, defined as PM no greater than 2.5umA. FP also may be denoted
as "PM2.5."
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.
1-2
-------
SECTION 2
SOURCE DESCRIPTION
Particulate 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, carry out 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 particulate 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 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. Emission
factors are given in the form of the following equation:
2-1
-------
E = k(sL/0.5)P (2-1)
where: E = particulate emission factor (g/VKT)
s = surface material content silt, defined as particles < 75 um
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
Particle size k (g/VKT) p_
fraction
TSP 5.87 0.9
2.54 0.8
2.28 0.8
PM2.5 1.02 0.6
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 11.2.6, "Industrial Paved Roads," was first published in 1983 and was slightly
modified in Supplement B (1988) to the 4th Edition. Section 11.2.6 contained three distinct
sets of emission factor models as described below.
'4Y s V /
E = 0.02211-I-II — II — | (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 report and was originally included in
Supplement 14 to AP-42 (May 1983). The version used in AP-42 was slightly revised in that
2-2
-------
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 1=1 and "D" otherwise.
For smaller particle size ranges, models somewhat similar to those in Eq. (2-1) were
recommended:
0.3
E = k (sL/12) (2-3)
where: E = emission factor (kg/VKT)
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 k (g/VKT)
fraction
0.28
2.22
PM2.5 0.081
These models represented by Equation 2-3 were first developed in 1984 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 1 1 .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/m ):
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 k (g/VKT)
fraction
0.12
0.093
The single-valued emission factors was quality rated "C."
During the time that AP-42 had four methods for estimating emissions from paved roads
(Sections 11.2.5 and 11.2.6, AP-42 Fourth Edition, 1993), users of AP-42 noted difficulty selecting the
appropriate emission factor model to use in their applications. For example, inventories of industrial
2-3
-------
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 PM10 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
29
characteristics. Confirmatory evidence was obtained in a 1989 field program 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.14 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
update8, 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.
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:
E = k(sL/2f65(W/3)L5 (2-5)
where: E = particulate 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.
2-4
-------
Particle Size Multipliers for Paved Road Equation
Size Range
PM2.5
PMio
PM15
PM30
Multiplier k
g/VKT
2.1
4.6
5.5
24
g/VMT
o o
J.J
7.3
9.0
38
Ib/VMT
0.0073
0.016
0.020
0.082
2.2.3 October 2002 through December 2003
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 jim 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|im 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 particulate 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 particulate emissions factors. The AP-42 equations
prior to December 2003 estimated PM emissions from re-entrained road dust, and vehicle
exhaust, brake wear and tire wear emissions. In the December 2003 revision of the section,
the component of emissions due to exhaust, brake wear and tire wear 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's
Mobile6.2 model, estimates of PM emissions from exhaust, brake wear and tire wear 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 particulate matter emissions from the exhaust of
newer vehicle models and fuel 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 particulate matter emissions. A technical memorandum
provided the basis for estimating PM emissions due to exhaust, break wear and tire wear. The
2-5
-------
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 for PM30, PMi5 and PMi0 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
particulate 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 MRI's 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.s 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 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 PMioConcentration 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.
Unlike the development of earlier paved road equations, the equation development for this
version adjusts the individual test data measured emissions by excluding exhaust emissions,
tire wear emissions and brake wear emissions prior to the equation development. As a result,
different values are subtracted from the results of each test based upon the average vehicle
weights, average vehicle speed, ambient temperature, year of test and estimated mix of light
duty and heavy duty vehicles.
2-6
-------
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 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).
3-1
-------
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
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
3-2
-------
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 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—B el ow 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
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
810
applicable measurement techniques. More detail can be found in earlier AP-42 updates.'
3-3
-------
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 particulate emissions from most open dust sources. These two methods are
discussed separately below.
The basic procedure of the upwind-downwind method involves the measurement of
parti culate 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 back calculate 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 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
3-4
-------
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.
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 particulate 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.
Particulate 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 particulate (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 particulate 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.
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
3-5
-------
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
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
particulate 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
3-6
-------
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
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.
3-7
-------
SECTION 4
AP-42 SECTION DEVELOPMENT
4.1 REVISIONS TO SECTION NARRATIVE
The 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 particulate 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 parti culate 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.
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.
4-1
-------
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 PMio 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 PMio 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.
In addition to the PMio 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 PMio 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 CALINE3 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.
4-2
-------
TABLE 4-1. SUMMARY INFORMATION FOR REFERENCE 15
Operation Location
Vehicle traffic Colfax
Vehicle traffic York St.
Vehicle traffic Belleview
Vehicle traffic 1-225
Vehicle traffic Evans
Vehi cl e traffi c Loui si ana
PMio emission factor (g/VKT)
State Test dates No. of tests Geom. mean Range
Colorado 3-4/89 17 1.33 0.53-9.01
Colorado 4/89 1 1.07 1.07
Colorado 4/89 4 1.62 1.10-4.77
Colorado 4/89 9 0.31 0.17-0.51
Colorado 5-6/89 29 1.06 0.21-7.83
Colorado 6/89 7 0.96 0.42-1.73
-------
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 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 report6noted 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 back calculated 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.)
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
4-4
-------
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
1 o
factors were obtained by back-calculation from the CALINE3 dispersion model 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.
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.
4-5
-------
TABLE 4-2. SUMMARY INFORMATION FOR REFERENCE 17
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
O
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
-------
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.
4.2.1.3 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, ref 06cl3s0201 2011.pdf)
This study evaluated paved road control techniques at two different iron and steel
plants. (See Tables 9 and 10 in Reference 8.) 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 PMio 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 45)
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
No PMio emission factors were reported; results were presented for total particulate
(TP) and suspended particulate (SP, or PM30). Data were quality rated "A" in the 1987 report.
4-7
-------
Because no PMio 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 PMio 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 46)
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.
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.
(Reference 31, ref_31c!3s0201 2011.pdf).
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"
4-8
-------
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.
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 particulate 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 parti culate 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 presents emission factors for total paniculate (TP), total suspended particulate (TSP)
and PMio, for the ten paved road emission tests conducted.
Reference 31 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 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
models in Section 11.2.5 showed a slight tendency to underpredict, the Section 11.2.6
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.
4-9
-------
4.2.1.7 Midwest Research Institute, Paved Road Particulate Emissions - Source
Category Report, for U.S. EPA, July 1984. (Reference 8, ref_08cl3s0201_2011.pdf)
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:
PMis (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 PMi5
concentrations. A high- volume sampler with a SSI and a cascade impactor was also located
downwind to characterize particle size distribution within the PMis 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.
4-10
-------
TABLE 4-3. SUMMARY INFORMATION FOR REFERENCE 31
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 emission
Geom. mean
0.61
0.033
0.078
factor, Ib/VMT
Range
0.39-0.96
0.012-0.12
0.033-0.30
PMio emissi
Geom. mean
0.16
0.0095
0.022
on factor, Ib/VMT
Range
0.14-0.18
0.0009-0.036
0.0071-0.036
1 Ib/VMT =281.9g/VKT.
TABLE 4-4. DETAILED INFORMATION FROM PAVED ROAD TESTS FOR REFERENCE 31
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
PMio 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
speed3
(27)
25
29
(27)
(27)
(27)
15
16
(15)
(15)
Silt
loading,
g/m2
0.42
0.52
0.23
0.23b
0.26b
0.15b
4.0
4.0
2.2
1.3
Silt, %
10
12
9.7
8.6
7.7
9.9
17
17
18
15
a Value in parentheses is the average speed measured for test road during the field exercise.
b Test conducted on a paved road surface vacuum-swept five times per week.
c Mean TSP/TP or PMio/TP ratio applied.
1 Ib/VMT = 281.9 g/VKT.
1 g/m2 =1.434gr/ft2
-------
TABLE 4-5. SUMMARY INFORMATION FOR REFERENCE 8
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
PMio 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.
-------
TABLE 4-6. DETAILED INFORMATION FROM PAVED ROAD TESTS FOR REFERENCE 8
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
PM10
emission
factor,
Ib/VMT
0.0110
0.00340
0.00781
0.00712
0.000400
0.00153
0.00304
0.000680
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
-------
4.2.1.8 Midwest Research Institute, Size Specific Particulate Emission Factors for
Uncontrolled Industrial and Rural Roads, for U. S. EPA, January 1983. (Reference 7,
ref_07c!3s0201 2011.pdf).
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: PMi5, PMi0, and PM2 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
PMis concentrations and the particle size distribution within the PMi5 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.
4-14
-------
TABLE 4-7. SUMMARY OF PAVED ROAD EMISSION FACTORS FOR REFERENCE 7
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
PMis, 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
PMio, 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
PM2.5, 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.
4-15
-------
TABLE 4-8. DETAILED INFORMATION FROM PAVED ROAD TESTS FOR REFERENCE 7
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
PM10
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/m2
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 = 281.9g/VKT.
Ig/m2=1.434gr/ft2
a Not measured
4-16
-------
4.2.1.9. Midwest Research Institute, Iron and Steel Plant Open Source Fugitive Emission
Control Evaluation, for U. S. EPA, August 1983, (Reference 6,
ref_06c!3s0201 2011.pdf).
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 particulate 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 1 5 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 1 1 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 PMio 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
achieved emission reductions ranging from 30 percent to 70 percent. Flushing at 0.48 gal/yd2
combined with broom sweeping resulted in emission reductions ranging from 35 percent to
90 percent.
4-17
-------
TABLE 4-9. SUMMARY OF PAVED ROAD EMISSION FACTORS FOR REFERENCE 6
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
PM15, 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.
-------
TABLE 4-10. DETAILED INFORMATION FROM PAVED ROAD TESTS FOR REFERENCE 6
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
PMio 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.
0 PM10 emission factors were calculated from the PM15
VS = Vacuum sweeping; WF = Water flushing; FBS =
1.434gr/ft2
and PM2 5 data using logarithmic interpolation.
Water flushing and broom sweeping; 1 Ib/VMT = 281.9 g/VKT; 1 g/m2 =
-------
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. (Reference 30, ref_30c!3s0201 2011.pdf).
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 particulate (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 particulate integrated over the time of measurement or, equivalently, the net
particulate 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 PMi0 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
particulate 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 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.s to PMio ratio than were measured by the
cyclone/impactor samplers. The PM2.5/ PMio ratios measured by the dichotomous samplers
are presented to the right of the PMio emissions factors column in Table 4-11. Where two
4-20
-------
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 PMi0 constant.
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 32, ref_32c!3s0201 2011.pdf).
The Corn Refiners Association (CRA) funded four paved road PMio 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 32 compiles test data and information from references 33, 34, 35 & 36. In
addition, reference 32 proposes an expansion of the allowable speed parameters supported in
the paved road equation. Lastly, reference 32 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 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 PMio 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.
4-21
-------
TABLE 4-11. DETAILED INFORMATION FROM PAVED ROAD TESTS FOR REFERENCE 30
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
PMio
emission
factor,
(g/VKT)
1.08
0.102
-
-
-
4.68
0.301
1.94
0.57
0.44
-
PM25/
PM10 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
to
to
Road Speed is the posted speed limit for the road segment.
-------
In addition to PMio 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 32 states that the four test programs conducted by CRA produced 14 and 8
PMio 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 PMio 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.
4.2.1.12 Midwest Research Institute, Emission Tests of Paved Road Traffic at
Minnesota Corn Processors Marshall, Minnesota Facility, McVehil-Monnett Associates,
July 6, 2001. (Reference 33, ref 33cl3s0201 2011.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 PMio 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.
4-23
-------
Table 4-12. Summary of Emissions Data from MCP's Marshall, Minnesota Facility
(Reference 33)
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 PMi0
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.
4.2.1.12. Midwest Research Institute, Emission Tests of Paved Road Traffic at Minnesota
Corn Processors Columbus, Nebraska Facility, McVehil-Monnett Associates, July
13, 2001. (Reference 34, ref_34c!3s0201 2011.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. (Reference 35, ref 35cl3s0201 2011.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
4-24
-------
provided a limited amount of "drone" traffic. The test data are assigned an "A" rating.
Eight PMio emission tests were attempted. The test report describes difficulty
encountered in isolating net PMio 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 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 PMio
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 at ADM's
Marshall, Minnesota Facility, McVehil-Monnett Associates, December 5, 2003.
(Reference 36, ref_36c!3s0201 2011.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 PMio 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.
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 PMio
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.
4-25
-------
Table 4-13. Summary of Emissions Data from MCP's Columbus, Nebraska Facility (Reference 34)
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 PMi0 emission
factor (Ib/VMT)
0.011
0.043
0.036
0.14
0.0024
0.051
0.0068
0.036
to
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.
-------
Table 4-14. Summary of Emissions Data from Cargill's Blair, Nebraska Facility (Reference 35)
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
-
-
-f^
to
a 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.
c "-" 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 (Reference 36)
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
a 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.
-------
4.2.1.15. E.H. Pechan & Associates, Inc., Recommendations for Emission Factor
Equations in AP-42 Paved Roads Section: TECHNICAL MEMORANDUM August 21,
2003. (Reference 28, ref_28c!3s0201 2011.pdf).
This technical memorandum documents the procedure that was used to separate the
various components of paved road particulate matter emissions into two components. One
component includes the emissions from exhaust, brake wear and tire wear. The other
component includes the particulate matter reentrained from the road surface. The combined
paved road particulate 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
VehicleType
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
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 tirewear by 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 for PM2 5. 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 PM2 5 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
4-28
-------
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 37, ref 37cl3s0201 2011.pdf).
This e-mail communication and spreadsheet file concerns estimates
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 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 mph but that they were stopped most of the time.
PM2.5 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 1 1.06035 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. E-mail communication between Ron Myers of EPA/OAQPS/SPPD/MPG,
RTP, NC and Gary Dolce, David Brzezinski and Rudolph Kapichak of
EPA/OTAQ/ASD/HDOC re. vehicle exhaust, tire and brake wear for urban
unrestricted road-types; October 2010 through December 2010. (Reference 39,
ref_39c!3s0201 2011.pdf).
This e-mail communication and spreadsheet files concern improved estimates of
PMio emissions associated engine exhaust, tire wear and brake wear for free flowing traffic.
The purpose of the estimates was to update the emissions estimates produced by E. H.
Pechan using the 2003 version of MOBILE6.2. The emissions model used for this updated
emissions estimates was the 2010 version of the MOVES model. Like the MOBILE6.2
model, the emissions predicted with the MOVES model provide a means for disaggregating
the emissions measured during the paved road field studies that measured emissions due to
road surface dust, vehicle exhaust, break wear and tire wear.
It is explained in the documentation that in order to develop an equation for road dust
alone, estimates of the parti culate emissions from vehicle exhaust, brake wear and tire wear
4-29
-------
were required. The e-mail documentation states that the MOVES model includes significant
new data about PM emissions from both light duty and heavy duty on-road vehicles which
allow MOVES to account for the influences of ambient temperature, vehicle speed, and
vehicle deterioration on emissions. The documentation further states that none of those
factors are accounted for in MOBILE6.2.
Documentation includes information provided to OTAQ on the test date (sometimes
month and year, sometimes just year), vehicle speed, ambient temperature, and average
vehicle weight for each of the paved road field studies. The documentation states that OTAQ
created a MOVES2010a model input file that approximated the information for the paved
road field studies as closely as possible. The documentation also states that since
MOVES2010a provides output for calendar years 1990 and 1999-2050 alternative scenarios
were developed to estimate emissions for years which MOVES2010a is not programmed to
provide.
The documentation states that the speed and ambient temperature measured during
the field study provided additional independent variables used in the MOBILE2010a model
to estimate emissions. The documentation indicates that an emissions estimate was produced
for each of the individual tests by allocating all of the vehicle activity to a single
MOBILE2010a speed bin which included the vehicle speed observed in the test. To reduce
the number of number of total runs needed, temperatures for the individual tests were
rounded to the nearest multiple of 5 degrees. In a small number of cases, vehicle speed or
temperature data were not available for particular tests. In those cases, a vehicle speed of 25
mph or an ambient temperature of 75 degrees was used. All other inputs to MOVES were
national defaults.
All vehicle and fuel type combinations (except for electric vehicles) were included.
Emissions were generated only for the urban unrestricted road-type. Emissions were
generated for all PM10 pollutants (primary exhaust PMi0 total, primary PMi0 brake wear,
and primary PMio tire wear. Only running exhaust and crankcase running exhaust processes
were included in the exhaust emissions calculations as the test sites did not include any
starting or idling activity. Inventory results generated by MOVES source type (vehicle type)
were divided by VMT to get emission factors by source type for each speed and temperature
bin in the original test data.
Emissions estimates for free flowing light duty vehicles and trucks are assigned an B
rating since most of the test data were for model years which an alternative emissions
scenario (year, vehicle mix and assumed degradion level) was used as the independent
variables used in the MOVES model input file. While it is likely that vehicle emissions prior
to 1990 had tailpipe emissions very similar to the 1990 model year, this can not be verified.
Also, while the emissions for each test are comprised of a large number of vehicles and the
emissions factor produced by the MOVES model are based upon a large number of
supporting tests, it is unclear that the MOVES model is an accurate and precise indication of
the vehicle exhaust, tire wear and brake wear emissions during each test series.
4-30
-------
4.2.1.18. 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 43, ref 43cl3s0201 2011.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, PM2.5 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 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 PM2.5 concentration and
the collocated PMio concentration. It was intended that any variation in PM2.5/ PMio 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.s 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
4-31
-------
study included no paved road surface material and was weighted toward higher particulate
matter concentrations. Since the study was a laboratory study, did not include any paved road
surface materials, and was weighted toward higher particulate 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/ PMio ratio to decrease with increasing PMio concentration. At PMio
concentrations above 1.0 mg/m3 the PM2.5/ PMio ratio was between 0.1 and 0.15. The
PM2.5/ PMio ratio increased to about 0.35 as the PMio concentration approached about 0.5
mg/m3.
4.2.1.19. 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.
(Reference 44, 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 PM2.5/ 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
PM2.5/ PMio ratios were made for use in revising existing AP-42 emissions factor parameters
for PM2.5 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.s/PMio ratio of 0.15 was recommended for the
paved roads emissions factor.
4.2.1.20. Technical Memorandum from William B. Kuykendal to File, Subject:
Decisions on Final AP-42 Section 13.2.1 "Paved Roads", October 10, 2002. (Reference
38, ref_38c!3s0201 2011.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 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 and 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 =
4-32
-------
number of hours in period).
4.2.1.21. Clark County (Nevada) Paved Road Dust Emission Studies in Support of Mobile
Monitoring Technologies; R. Langston, R.S. Merle Jr, V. Etyemezian, H. Kuhns, J.
Gillies, D. Zhu, D. Fitz, K. Bumiller, D.E. James and H. Teng; Clark County
Department of Air Quality and Environmental Management, Desert Research Institute,
University of California, Riverside, University of Nevada, Las Vegas; December 22,
2008.
(Reference 42, http://www.epa.gov/ttn/chief/ap42/ch!3/related/Final_Test_Report.pdf).
This report documents the fourth phase of a study by Clark County to investigate
alternative ways of estimating PMio emissions of surface dust entrained from paved roads. A
new vehicle-mounted mobile sampling technology was tested in comparison with the
traditional AP-42 method and its associated road surface sampling. In addition, the plume
flux profiling method, was used to calibrate the mobile monitoring technology.
Two versions of the mobile monitoring technology were tested—TRAKER and
SCAMPER. Both technologies involve on-board sampling of the dust plume generated by a
test vehicle. Both use continuous optical based PMio particle monitors in conjunction with
GPS systems, so that dust plume concentrations can be mapped on to the road system
traveled by the test vehicle. The SCAMPER samples the plume in the wake of the test
vehicle. The TRAKER I and II test vehicles sample the plumes from the front wheel wells of
the respective vehicles. TRAKER II has a dilution system to provide for use on unpaved
roads. All three units have samplers that monitor the PMIO concentration in front of the
vehicle so that "background" PMio can be subtracted.
The referenced study evaluated mobile monitoring technologies in comparison with
the traditional AP-42 methodology, but in a controlled measurement environment that
included restricted vehicle movement, controlled vehicle speeds and controlled road surface
material loadings. This was accomplished by dedicating half of a divided roadway as the test
course for the 5-day field study. The stated specific study objectives were as follows:
• Comparison of SCAMPER and TRAKER system measurements with emission
measurements using a downwind flux tower.
• Determination of the relationship between roadway silt loading and SCAMPER and
TRAKER measurements at several standard vehicle speeds (25, 35 and 45 mph).
• Comparison of SCAMPER and TRAKER measurements to AP-42 emission
estimates.
• Characterization of road surface silt depletion rate as a function of the number of
vehicle passes.
• Characterization of quantified emissions vs. quantified silt loading mass.
• Data assessment and review for recommendations on performance specifications for
vehicle-mounted mobile sampling systems.
Particle concentration measurements formed the basis for the mobile monitoring
technologies as well as the roadside emission flux measurements. A continuously recording
optical light scattering particle monitor (DustTrak Model 8520, TSI Inc., Shoreview MN)
was the basic instrument used for PMio readings. A collocated mass-based reference monitor
4-33
-------
was used to correct the DustTrak readings to equivalent PMio mass-based concentrations,
using a plume profiling tower with various reference, reference equivalent and DustTrak
monitors at different heights. Canister vacuum cleaners with hard-floor inlets were used to
recover applied soil from the roadway sites into pre-tared vacuum bags. Three soil recovery
techniques were used during the study. Road dust emission factors were then calculated for
the silt loadings using the 2006 AP-42 emission factor equation. A weight of 2.88 tons,
based on the arithmetic average of the reported weights of the three mobile source vehicles
was used to calculate the AP-42 emission factors from the silt loadings.
Thirteen different experimental test conditions were performed. Most consisted of
approximately 30 vehicle passes, with each pass identified by the mobile sampling
technology. Each run consisted of three passes by each mobile sampling technology. Cross-
comparisons were performed to determine the ratio between the DustTrak reading and the
PMio mass-based concentration measured by a collocated reference sampler. The correlation
between the DustTrak and TEOM showed that DustTrak values would have to be multiplied
by a factor of 2.8 ± 0.6 to obtain mass-equivalent PMio. A controlled laboratory tests was
also used to obtain a relationship between the DustTrak measurements and mass-based
measurements. These tests generated a DustTrak correction multiplier of 2.4, which was
chosen for use in this program.
Two conclusions were made from the test results obtained in the study, when
comparing mobile monitoring technologies with the AP-42 methodology:
• The calibrated mobile methods measured emission factors that were about 1.5 times
higher than found with the AP-42 methodology when higher silt loadings were
applied to the test road.
• The mobile methods tracked each other quite well under most conditions.
It was concluded that a different silt mobilization process occurred as a result of silt
being distributed on top the embedded road surface aggregates and hence being more easily
entrained by vehicle mechanical and aerodynamic shear. It was also stated that aged silt
found on most roads is more likely to be embedded between the road surface aggregates.
Another conclusion identified in the field study was that implementation of mobile
monitoring technologies provide for much easier representation of spatially distributed
roadway emission characteristics, while eliminating the need to divert traffic.
4.2.1.22. Technical Support Document for Mobile Monitoring Technologies; Prepared
For Clark County Department of Air Quality and Environmental Management;
Chatten Cowherd; Midwest Research Institute; January 9, 2009. (Reference 41,
http://www.epa.gov/ttn/chief/ap42/chl3/related/Mobile Monitoring TSD 010909.pdf).
This report states that it documents a peer review process conducted to determine
whether the mobile monitoring method is a suitable alternative to the traditional AP-42
method for developing road dust emission factors. The report identifies seven individuals
which were requested to review the series of Clark County test reports and to judge the value
of mobile monitoring technologies in relation to the traditional approach for determining
paved road dust emission factors.
The items addressed in this document include:
4-34
-------
• A summary of road dust entrainment dynamics,
• A brief discussion of the basis of the current road dust emissions estimating method.
Also described were the methods used to characterize the road surface silt loadings,
the statistical methods used in developing the AP-42 emission factor equations and
the use of roadside plume exposure profiling to quantify mass emissions rates.
• A brief discussion of the methods used to estimate independent variables required for
the AP-42 emissions factor equations, associated restrictions and the resulting
limitations and a subjective assessment of the uncertainties.
• A more in depth discussion of the two mobile monitoring technologies (the Desert
Research Institute (DRI) and the CE-CERT version) is provided. The report
identifies the presence of high background dust concentration and high wind speeds
as two restrictions for the use of mobile monitoring. The report discusses the
subjectively established calibration requirements for mobile monitoring. Calibration
requirements identified include determining the relationship between concentrations
measured by the instrument used for mobile monitoring and the Federal Register
Measurement Method, the relationship between the concentrations measured at
different vehicle speeds, different road dust characteristics and different vehicle
weight during mobile monitoring and mass emissions measured by plume profiling.
• The report provides a discussion comparing of the implementation of the traditional
application of the emissions factor and the use of mobile monitoring to develop
emissions inventories.
• Lastly, the report provides the charge provided to the reviewers, an overview of
comments by the reviewers and an indication of what changes will be made to
address the reviewers concerns in a Specification for Mobile Monitoring document.
While this document states that the purpose is to demonstrate that mobile monitoring
is equivalent or superior to the traditional AP-42 methodology, it provides only subjective
opinions of the author and the selected reviewers. While there were no quantitative
indicators to compare the precision or accuracy of the mobile monitoring technologies over
the normal range of road conditions (silt loadings, mix of vehicle weights, vehicle speed) and
resultant emissions produced, the author and the majority of the reviewers concluded that the
method was more accurate and precise than the traditions measurement and monitoring
methods. The review does reveal that there is an understanding that there is a lack of
precision and understanding of independent variables other than silt loading, weight and
speed which influence road dust emissions. Several reviewers highlight the potential of
mobile monitoring methods to replace or supplement the resource intensive and dangerous
collection of representative silt loading information. Several reviewers also highlight the
need for further development and standardization of mobile monitoring such that the method
could be used for managing the road dust emissions where required.
4.2.1.23. Mobile Monitoring Method Specifications; Prepared For Clark County
Department of Air Quality and Environmental Management; Chatten Cowherd;
Midwest Research Institute; February 6, 2009. (Reference 40,
http://www.epa.gov/ttn/chief/ap42/chl3/related/MM_Method_Specifications_020609.pdf).
This document provides instructions for performing a standardized methodology for
the construction of a mobile sampling platform, specifications for instrumentation used with
4-35
-------
Federal Register Methods for PM10 or PM2.5, calibrations required to correlate the
combined sampling platform and instrumentation with standardized plume profiling testing
used to quantify mass emissions from roads and procedures for collecting information for use
in road surface characteristics or emissions.
4.2.2 EMISSIONS FACTOR DEVELOPMENT.
A total of 103 individual tests are available. All tests quantified PMio emissions.
Lastly, plume profiling was the test method. Of these, 81 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 particulate 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 particulate 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 particulate 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 average PMio emissions factor of 0.2119
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 using the MOVES model 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 data base,
2. the August 1983 EPA data base,
3. the July 1984 EPA data base,
4. the May 1990 USX data base,
4-36
-------
5. the April 1997 EPA data base, 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 of vehicle class includes gas and diesel heavy duty trucks.
Other assumptions used to estimate vehicle associated emissions include:
• The test fleet includes a mixture of light duty vehicles, heavy duty gas trucks and
heavy duty diesel trucks when the average vehicle weight is less than 23 tons.
The test fleet includes a mixture of light duty vehicles and heavy duty diesel
trucks when the average vehicle weight is between 23 tons and 35 tons.
• The test fleet includes only heavy duty diesel trucks 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
and VMT distribution presented in Table 4-16 are used to calculate the average vehicle
weight. 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 trucks, 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 trucks, 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
trucks, the average gross vehicle weight is 26.09 tons.
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
4-37
-------
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)
Run specific emissions estimates for vehicle exhaust, brake wear and tire wear are
estimated using the EPA Office of Transportation and Air Quality MOVES (MOtor Vehicle
Emission Simulator) 2010 model29. For all tests with vehicle speed greater than 10 mph only
emissions for freely moving traffic is calculated. Emissions for a representative mix of light
duty vehicles and for a representative mix of heavy duty trucks are calculated. For each test
series, information on the date of the test, the location of the test program, ambient
temperature during the test, average vehicle speed, and other general information required to
generate a valid PMio emissions calculation with the MOVES model. While the MOVES
model has the ability to generate start up emissions, all test conditions are assumed to include
only vehicles which have achieved normal operating temperatures. For all test series with
average vehicle speeds greater than 10 mph, the MOVES model calculated only running
exhaust, tire wear and brake wear emissions. For heavy duty vehicles, the running emissions
ranged from 0.645 g/VMT to 4.896 g/VMT. For light duty vehicles, the running emissions
ranged from 0.0196 g/VMT to 0.1324 g/VMT. For test series with average vehicle speeds
below 9.9 mph, in addition to running exhaust, tire wear and brake wear emissions; exhaust
emissions during acceleration and idling are included. A separate MOVES model run
estimated the average emissions for the non steady state emissions at 11.06 g/hour. The
emissions factor for this driving condition was calculated by dividing the hourly emissions
by the average vehicle speed. Summing the product of emissions factors from heavy duty
trucks and light duty vehicles and the ratio of heavy duty vehicles and light duty vehicles
provides an estimate of the total engine exhaust; tire wear and brake wear emissions for the
test run.
The 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 0.01 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. There were 10 test runs of the 103
available data where downwind emissions were not measureable. Six of the data were
associated with low speed traffic at corn refining facilities and four of the data were high or
moderate speed urban traffic. None of these ten data were included in the data analyzed to
estimate the predictive emissions factor equation. There were 3 out of the 103 available data
sets where the estimated emissions from engine exhaust, tire wear and break wear were equal
to or comparable to the measured emissions. Two of the three test runs were on roads where
the average vehicle speed was 55 mph. Emissions of two additional test runs with vehicle
speeds of 55 mph had engine exhaust, tire wear and break wear emissions greater than 160%
of the road emissions. The silt level for one of the 55 mph test runs was greater than all
4-38
-------
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 PMio data base are shown in Figures 4-1 through
4-5. Because of the large range of silt loadings and estimated emissions factors, the data are
plotted on a logarithmic scale for the first three figures. Figure 4-1 presents the data base by
silt loading with five ranges of average vehicle weight depicted with different shape and
color data points. The figure shows that with increasing silt loading there is an increase in
the PMio emissions factor. Figure 4-2 presents the data base by average vehicle weight with
seven ranges of silt loading depicted with different shape and color data points. Although
there is a significant overlap of the different vehicle weight data, there appears to be some
relationship between average vehicle weight and the PMio emissions factor. As with silt
loading, it appears that the PMio emissions factor increases with increasing vehicle weight.
The wider spread of the data around the center line of the data makes the relationship more
difficult to discern. Figure 4-3 presents the relationship between silt loading and average
vehicle weight with eight ranges of emissions factors depicted with different shape and color
data points. Although very poor, there appears to be a weak relationship between silt
loading and vehicle weight. The cause of this relationship is probably due to the selection of
the test location and parameters than any physical force that would cause this relationship.
Figure 4-4 presents the relationship between average vehicle speed and the PMio emissions
factor. It appears that between 10 and 55 mph, the emissions factor decreases with
increasing speed. Below 10 mph there does not appear to be a speed relationship. Figure 4-5
presents the relationship between silt loading and vehicle speed with five ranges of PMio
emissions factors. The silt loading appears to decrease with increasing speed above 10 mph.
In addition, there seems to be a clear increase in PMio emissions factor as silt loading
increases and speed decreases. Figure 4-6 presents a three dimensional view of the silt
loading, vehicle weight and PMio emissions factors. One data point seems to be very
uncharacteristic of the general trend of the data. Figure 4-7 provides a two dimensional view
of the data with the data identifier in the label. For three data points, the PMio emissions
factor is also included in the label. The point which has the uncharacteristic emissions is
point Z-3 with a PMio emissions factor of 1819 g/VMT. While this value is the highest
emissions factor of all of the 92 test data, both the vehicle weight and silt loading for this run
are near other data which are under 100 g/VMT. As a result, this data was flagged as a
potential outlier. This data was reassessed following log transformation and the variation
was determined to be comparable with other data and was included in the final data set used
to estimate the predictive equation. Figure 4-8 presents the three dimensional view of the
test data with silt loading, vehicle weight and PMio emissions factor with test run Z-3
removed. With point Z-3 removed, there appears to be two regimes of the data. Most of the
data had silt loadings below 20 g/m2 with few gaps down to 0.013 g/m2. There are ten data
with silt loadings spread out from 50 g/m2 to almost 400 g/m2 with no data between these
two regimes. There appears to be one incline associated with the lower silt loading data and
a significantly greater incline for the higher silt loading data. This greater incline is the result
of a small number of data collected prior to 1983. These data have higher silt loadings that
the default silt loading for the peak additive contribution value for roads with average daily
4-39
-------
traffic volume counts of less than 500. While there may be a very small number of streets
that reach this silt loading level, these are believed to be unrepresentative of typical well
managed urban or rural roads during any season. As a result, these data are flagged as
extreme values and were not included in the final data set used to estimate the predictive
equation.
4-40
-------
Table 4-17. Final Paved Roads Emissions Factor Data Set
Estimated Estimated
Estimated Engine, brake, PMi0 Road
Downwind Measured PM10 Fraction tire emission Dust Emission
Silt loading Speed Weight Concentration Emission factor Heavy Duty factor factor
Reference Run ID (g/m2) (mph) (tons) mg/m3 (g/VMT) Vehicles (g/VMT) (g/VMT)
USX 5/1990
EPA 7/1984
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
M-18
M-19
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
0.731
0.929
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
30
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
2
2.4
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
0.080
0.050
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
0.37
0.177
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
0.009
0.025
0.3298
0.3537
0.1941
0.3961
0.1653
0.0936
0.9337
0.3709
0.1804
0.1804
0.3610
0.2486
0.2845
0.0927
0.0749
0.1043
0.1146
0.1063
0.2190
0.1798
0.2009
0.1403
0.0988
0.1044
0.0886
0.1581
0.0501
0.0501
0.0791
1.920
15.746
15.106
3.304
0.237
7.786
2.286
10.229
15.920
8.830
4.629
1.301
3.256
0.084
0.617
1.276
4.105
11.094
3.021
0.010
0.121
0.010
0.207
1.266
1.381
0.083
2.590
0.320
0.098
-------
Table 4-17. (Continued)
Reference
EPA 1/1983
EPA 8/1983
Run ID
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
Silt loading
(g/m2)
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
Speed
(mph)
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
Weight
(tons)
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
Downwind
Concentration
mg/m3
0.552
0.057
0.134
0.163
0.301
0.177
0.531
0.138
0.327
0.744
0.294
0.114
Measured PM10
Emission factor
(g/VMT)
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
Estimated
Fraction
Heavy Duty
Vehicles
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
Estimated
Engine, brake,
tire emission
factor
(g/VMT)
0.2274
0.2359
0.2443
0.2359
0.6096
0.5697
0.5697
0.4090
0.4852
0.1466
.8114
.8114
.8114
.4388
.2790
0.5320
.1617
.2339
.2339
0.9292
0.9292
1.8261
1.0896
1.8226
1.5012
Estimated
PM10 Road
Dust Emission
factor
(g/VMT)
116.773
181.764
36.056
199.764
316.390
739.430
1819.430
1749.591
1419.515
612.853
1478.189
340.189
231.189
186.561
296.721
54.168
76.038
165.766
251.766
129.071
52.171
461.174
210.910
315.177
543.499
-f^
to
-------
Table 4-17. (Continued)
Reference
EPA 8/1983
EPA 4/1997
CRA 5/2008
Run ID
B50
B51
B52
B54
B55
B56
B58
B57
B59
B60
BH1
BH2
BH3
BH6
BJ6
BJ7
BJ9
BJ10
BJ11
BK7
BK8
CE-1
CE-2
CE-11
CE-3
CE-1 5
CE-1 6
CE-1 7
CE-1 9
Silt loading
(g/m2)
13.6
13.6
7.19
3.77
6.3
2.4
10.4
2.32
2.06
3.19
0.184
0.0127
0.0127
1.47
0.06
0.06
0.06
0.06
0.06
0.082
0.082
1.16
0.86
1.34
0.86
1.91
1.41
2.93
0.76
Speed
(mph)
NR
NR
NR
NR
NR
NR
NR
NR
NR
NR
55
55
55
40
45
45
45
45
45
45
45
1
1
5
1
5
5
5
5
Weight
(tons)
9.4
11
12
10
11
9.2
18
12
11
12
2.2
2.2
2.2
2.2
2.2
2.2
2.2
2.2
2.2
2.2
2.2
36
36
12
39
40
40
40
38
Downwind
Concentration
mg/m3
0.225
0.410
0.102
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
Measured PM10
Emission factor
(g/VMT)
82.1
140
35.4
93.3
183
126
368
195
348
439
1.08
0.102
0
4.68
0.301
1.94
0
0
0
0.57
0.44
27
64
154
45
63.5
77.1
41.3
18.6
Estimated
Fraction
Heavy Duty
Vehicles
0.313
0.379
0.420
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
0.017
0.017
0.017
0.017
.000
.000
0.420
.000
.000
.000
.000
.000
Estimated
Engine, brake,
tire emission
factor
(g/VMT)
0.5936
0.7108
0.7836
0.6379
0.7108
0.5794
1.2221
0.7836
0.7108
0.7836
0.0306
0.0306
0.0305
0.0343
0.0336
0.0336
0.0305
0.0305
0.0305
0.0336
0.0336
11.06
11.06
2.212
11.06
2.212
2.212
2.212
2.212
Estimated
PM10 Road
Dust Emission
factor
(g/VMT)
81.506
139.289
34.616
92.662
182.289
125.421
366.778
194.216
347.289
438.216
1.049
0.071
4.646
0.267
1.906
0.536
0.406
15.940
52.940
151.788
33.940
61.288
74.888
39.088
16.388
-------
Table 4-17. (Continued)
Reference
CRA 5/2008
Run ID
CE-12
CF-1N
CF- I/South
CF-2N
CF-2/South
CF-3N
CF-3/South
CF-4N
CF-5
CI-1
CI-2
CI-3
CI-4
CI-7
CI-8
CI-11
CI-1 2
CM-1
CM-2
CM-4
Silt loading
(g/m2)
1.34
0.97
0.97
0.81
0.81
0.63
0.63
1.1
1.4
0.06
0.06
0.06
0.06
0.05
0.05
0.025
0.25
0.72
0.72
0.7
Speed
(mph)
5
5
1
5.3
1
5.1
1
4.7
1
15.1
14.85
13.15
14.5
15.3
15.3
13.1
13.1
5
1
5
Weight
(tons)
13
40
40
41
41
41
41
41
41
26
26
27
27
27
27
27
27
39.8
39.6
39.5
Downwind
Concentration
mg/m3
0.085
0.035
0.040
0.044
0.080
0.015
0.025
0.019
0.030
0.030
0.030
0.035
0.050
0.035
Measured PM10
Emission factor
(g/VMT)
23.1
4.99
19.5
16.3
63.5
1.09
23.1
3.08
16.3
0
0
0
0
1.63
2.99
0
0
6.35
63.5
7.26
Estimated
Fraction
Heavy Duty
Vehicles
0.461
.000
.000
.000
.000
.000
.000
.000
.000
0.729
0.729
0.759
0.759
0.759
0.759
0.759
0.759
1.000
1.000
1.000
Estimated
Engine, brake,
tire emission
factor
(g/VMT)
2.212
2.212
11.06
2.0868
11.06
2.1686
11.06
2.3532
11.06
.0008
.0008
.0410
.0410
.0409
.0409
.0410
.0410
2.212
11.06
2.212
Estimated
PM10 Road
Dust Emission
factor
(g/VMT)
20.888
2.778
8.440
14.213
52.440
0.010
12.040
0.727
5.240
0.589
1.949
4.138
52.440
5.048
-------
1 AAAA
10000
1 AAA
1000
£~ N 1 A. A
H 10°
|
^^ i f\
&1
L.
O
** i
7j 1
W
53
u.
fl n i
Q U.I
•PN
&
5C
• P«
M ^ \ f\ i
w °-01
OAA 1
.001
0.
PM10 Emission Factor ( by silt loading)
* D A ^
.• .V.if ^ •"
• A»ngunn*^^ H
• ^a* ^
^ n^ A
B *
i m"-li— i
n^ • Avg Weight
^+ ^H ^ ("tout
W rn v
U g *
^^^ B 4- 4- 2 to 3
^ to 5
• • •
A 5 to 10
• 10 to 20
i — i
u D Over 20
I I / I
)1 0.1 1 10 100 10
Silt Loading (g/m2)
00
Figure 4-1. PMio Emissions Factor Data Base by Silt Loading (93 test runs).
-------
t
ON
PM10 Emission Factor by Weight
i nnnn
1UUUU
i nnn
1UUU
i An -
1UU
X-v
H
^H
^
i"**" i'
- — 1U
b£.
PN
a
% ,
« i ~
fc
=
o
«FH
S5 n i
i
H
On i
.Ul
n no i
U.UUl
]
D D°
D O n ^
0 A& 0 a CS
R A^O0' & 1°
,(J" •
n ' o' A^
A A • o - ^
^i-i t?
. ^ " •
A» .0 4
/3 • •
* *
4." .
>^ 4
•^ A
/A
4
Silt Loading
+ 0.01 to. 1
• 0.1 to 0.5
0.5 to LO
1.0 to 2.0
A 2.0 to 4.0
O 4.0 to 50
D Over 50
L 10 100
Average Vehicle Weight (ton)
Figure 4-2. PMio Emissions Factor Data Base by Average Vehicle Weight (93 test runs).
-------
Silt Loading vs Average Vehicle Weight
1000 "
1 AA
100
/— \
M
E
hr, 1 A -
3 10
DC
d
*v*i
•u
S
3 1 -
H 1
•W
•p*
5/3
01 .
.1
OA 1
.01
.
D n
3 D
n r~i
<*> $
RAO 0 ^ 0°
" °' o" D
• A *. v
° <$ 0° A
J> ^1 -i
»/ ^S
A^ ^- A>
4. A ^4 * A 0
* A f A
| t
1 • •
A Aft*
A
Emission Factor
A 0 to 0,5
• 0,5 to 3.0
* 3.0 to 10
• 10 to 50
A 50 to 100
O 100 to 200
O 200 to 500
D Over 500
i
I 10 100
Average Vehicle Weight (tons)
Figure 4-3. Silt Loading vs. Average Vehicle Weight (93 Test Runs).
-------
10000
1000
100
PM10 Emissions Factor by Vehicle Speed
0
"u
I ]
K
1
J5 0.1
0.01
A
i
• sL 0.1-0.3
A sL 0.3-0.75
OsL0.75- 1.3
• sL 1.3 -3.0
*3L?-20
A sL 20-200
• Over 200
n
O
r-\
0
0
-o-
0.001 -
1
10
Average Vehicle Speed (mph)
Figure 4-4. PMio Emissions Factors by Vehicle Speed.
100
-------
1000
Vehicle Speed vs Silt Loading
100
I 10
60
s^
*
•6
«
6
h-1
£ i
IK
tu
u
£
(/!
•a
I "-1 '
«
0.01
Emissions Factor
• 1.0
* 1.0-4.0
A 4.0-15
• 15-100
• > 100
I-
^
10
Average Vehicle Speed (niph)
Figure 4-5. Vehicle Speed vs Silt Loading.
100
-------
Paved Road Emissions Test Data
All Data - Normal Scale
1800
1705.556
lf.ll III
1516.667
1422222
1327.778
123? 333
II3X XX')
1044444
950
855.5556
7(.| Illl
572,2222
477.777X
383.3333
28KR880
I'M 4444
< 10(1
Figure 4-6. Paved Road Dust Emissions Factors, All Data.
4-50
-------
Paved Road Emissions Test Data
All Data - Normal Scale
Silt Loading (g/m2)
Figure 4-7. All Paved Road Data, Silt Loading by Vehicle Weight with EF.
4-51
-------
Paved Road Emissions Test Data
Less one Data - Normal Scale
40
< 100
Figure 4-8. Paved Road Dust Emissions Factor Data Excluding Z-3.
4-52
-------
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 PMip data.
PMio Emission factor (g/VMT)
Silt loading (g/m2)
Weight (tons)
Speed (mph)
PMio Emission
factor (g/VMT)
1
0.8010
0.3280
-0.4066
Silt loading
(g/m2)
1
-0.1841
-0.2785
Weight
(tons)
1
-0.7784
Speed
(mph)
1
Initially several regression analysis were performed using the Data Analysis tools in
MS Excel to evaluate a range of independent variables. The independent variables included
silt loading, average vehicle weight, the product of silt loading and vehicle weight, the square
of silt loading (after log transformation) and the square of the vehicle weight (after log
transformation). In addition, the influence of including and excluding flagged test runs were
explored. The primary criteria for selecting the most appropriate form and supporting data set
was the predictive performance of the equation using the combination of the correlation
coefficient, the P-value and the relative percent difference from the actual emissions factor for
the test series with silt loadings and vehicle weights in the range of default values used in the
national inventory. The stepwise regression was first performed using the "Regression"
function in the "Analysis Tool" of Excel. It was determined that the use of the speed term
either produced equations with P-values greater than 0.1 or produced equations with
independent parameter relationships that were illogical (i.e. increased emissions with
decreased weight). It was also determined that the inclusion of data with silt loadings greater
2
than 20 g/m produced equations which uniformly overestimated test data with lower silt
loadings without a significant improvement in estimating the high silt loading data. Also, the
exclusion of the ten data with high silt loadings did not significantly change the predictive
accuracy of the equation for the ten high silt loading test runs. The 93 test data with positive
measured emissions were provided to a statistician for subsequent analysis with SAS.
4-53
-------
Several additional assessments were performed to determine an equation that provided a high
correlation coefficient, a low average percent error for test series with targeted independent
variables and which provided a reasonable level of predictive accuracy for test series where
the independent variables were outside the targeted range. The equation which produced the
highest correlation coefficient was one which forced the intercept to zero. This equation
performed well and was consistent with engineering assessments of the physical influences
on emissions. This equation used only silt loading and average vehicle weight as the
independent variables. It was decided that the traditional scaling factors of 2 for silt loading
and 3 for average vehicle weight were no longer required and resulted in simpler calculation
of paved roads emissions factors. The resulting equation for PMio is:
T-"T-" i r\ t T V).912 A-17V-021
EF = 1.0(sL) (W)
Table 4-19 shows the statistical output. The predicted exponents for silt and weight are
0.912 and 1.021 respectively and have a coefficient of determination (R2) of 0.72. The standard
error associated with the silt and weight terms are 0.12 and 0.08 respectively. As a result, it is
expected that 95% of future data would fall within equations with exponents of 0.677 and 1.14 for
the silt term and 0.852 and 1.19 for the weight term.
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)
4-54
-------
Table 4-19. Regression Analysis using Silt Loading and Weight.
SUMMARY OUTPUT
All positive test data, sL < 20 force 0, sL W
Regression Statistics
Multiple R 0.848347765
R Square 0.71969393
Adjusted R Square 0.703887682
Standard Error 1.921751464
Observations 83
ANOVA
Regression
Residual
Total
Intercept
Weight (tons)
Silt loading (g/m2)
df
2
81
83
Coefficients
0
1.0212836
0.911843675
SS
768.0593789
299.1434238
1067.202803
Standard Error
#N/A
0.084774552
0.117787966
MS
384.0296894
3.693128689
tStat
#N/A
12.04705393
7.741399277
F
103.9849195
P-value
#N/A
9.58964E-20
2.42283E-11
Significance
F
5.61978E-23
Lower 95%
#N/A
0.852608836
0.677482574
Upper 95%
#N/A
1.189958364
1.146204776
-------
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 particulate matter.
Another complication is that PM emissions of the vehicle exhaust were not measured during
the tests and a modeled average emission factor or rate was subtracted to arrive at the road
dust emissions.
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
two parameter equation, the average predicted to actual ratio is 49. This is significantly
lower than the average predicted to actual ratio of 315 for the previous equation when
applied to the existing data. When limited to silt loading levels of 20 g/m2, the new
equation produces average predicted vs actual ration of 38 compared to the previous
equations ration of 221. It should be noted that the previous equation subtracted 0.2119
g/VMT (the estimated national average engine exhaust, brake wear and tire wear emissions
factor) from the previous equation which was based upon measured emissions. The new
equation subtracts the estimated engine; brake wear and tire wear emissions estimated for
each test run. These emissions average 1.565 g/VMT and range from 0.031 to 11.06 g/VMT
depending on meteorological conditions, vehicle speed and vehicle weight determined during
the test. Figure 4-9 depicts the cumulative distribution of the predicted to actual ratios for
both the previous equation and the new equation. Figure 4-10 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 4-11 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 range of the predicted vs actual ratio of the older
equation is evident. The new equation appears to demonstrate an improved performance
compared to the previous equation.
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. Two 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 order of magnitude improvement in estimating the
actual measured emissions over the previous equation. Associated with the reduction in the
percent difference from actual emissions is a 47 percent reduction in the emissions factor.
When the performance of the equation is evaluated within classes of the independent
variables of silt loading, average vehicle weight and speed; the new equation shows
comparable or improved performance in all groups of the variables except two.
4-56
-------
100%
001
Cumulative Distribution - Predicted/Actual Ratios
-• t-e—e-
• New Equation
O 2006 Equation
1
1000
10 100
Predicted/Actual Ratio
Figure 4-9. Cumulative Distribution of Predicted/Actual Ratios.
10000
100000
-------
Cumulative Distribution - Predicted/Actual Ratio by Silt Loading
J^
00
1UUTO '
90%
onai
OU'O
7nai
iT^^ A ^ °°7
• J?^ A n , o"
VT A >.
.JL-^ ffl^ *
ng (gin2)
3,07 •
0.5 A
-10 ^
-5.0 •
iO A
0.01 0.1 1 10 100 1000 10000 1001
Predicted/ActiMl Ratio
Figure 4-10. Cumulative Distribution - Predicted/Actual by Silt Loading.
-------
11 OS
t
a
5
i
0.01
Cumulative Distribution
Predicted/Actual Ratio by Average Vehicle Weight
0.1
10 100
Predicted/Actual Ratio
1000
10000
100000
Figure 4-11. Cumulative Distribution - Predicted/Actual by Average Vehicle Weight.
-------
Figure 4-12 through Figure 4-9 provide graphical indications of the performance of
the updated equation to estimate the actual emissions. The first figure shows the relationship
of emissions to the road surface silt loading. Included in this figure is information on the
average vehicle weight through the use of a different shape and color for different ranges of
vehicle weight. While not shown, the previous equation had a greater spread than the new
equations estimates. Figure 4-13 shows the influence of vehicle weight on the emissions
factors. For all weight ranges, the spread of the data is much greater than is demonstrated in
the figures with silt as the ordinate. Included in this figure is information on the silt loading
associated with the test. One can see a general increase in emissions with silt loading. This
is probably due to the greater correlation between silt loading and PMio emissions factors
than between average vehicle weight and PMio emissions factors. 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.
4-60
-------
Table 4-20. 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 Difference
Old Equation vs
Actual
31,378
By Classes of Silt Loading (g/m )
<0.2
0.2-0.75
0.75-1.5
1.5 - 50
>50
33,601
102,647
3,236
221
248
New Equation
vs Actual
3,142
3,858
17,049
669
47
253
By Classes of Average Weight (ton)
2-3
3-5
5-10
10-40
>40
467
718
-2
38,248
128,217
333
289
-41
4,906
21,549
Old Equation vs
New Equation
-47
-71
-62
-61
-45
73
40
350
53
74,840
68,550
By Classes of Average Speed (mph)
< 10
10-25
25-45
45
55
90,216
54,063
293
1,041
1,404
15,112
7,034
170
662
467
-79
-6
-41
-34
-114
Relative Standard Deviation
Old Equation
vs Actual
5.77
2.12
3.71
2.48
3.57
1.81
2.09
1.71
-394.3
3.06
4.27
New Equation
vs Actual
5.84
1.38
11.54
0.41
0.11
0.27
0.43
0.72
0.37
24.73
112.84
Old Equation vs
New Equation
-1.2
-0.62
-0.35
-0.46
-0.46
1.20
-1.62
9.91
-0.74
-0.18
-0.22
4.30
2.17
2.41
1.28
1.57
20.67
4.94
0.139
0.198
0.114
-0.07
-15.11
-0.45
-0.05
-0.60
-------
innnn -,
1UUUU
i nnn
i on
P
id
i
"iv i n
M 1«
Spi?
o
•M
O
fi
,_ 1
e "
—
_O
"8
Js
M
^ 01
U.I
0 01
0,1
EF - Predicted vs Actual - Silt loading
0
0°
V
0 /•* *
** .A
0 * ft ^
,_,
* A^A^Pa** 0
°Un * D> l^
AU^
R • D
,S»5' A D
Jl 1 — 1
D A u
•
|—| A 1 1
1 1 1
)1 0.1 1 10
Silt Loading (g/m2)
Oo
: A
^ A
A
A
Ave Velucle Weight
Actual Predicted
D 2 *
J -^
D 5-10 •
10-20
D Over 20 •
100 1
0
00
Figure 4-12. Predicted vs Actual PMio Emissions Factor by Silt Loading.
-------
OJ
i niuvi -,
i oon
/~v
Ml A A
giUH
:_
O
ission Fat
s
w
A 1 .
n ni
1
Silt Loading
Actual Predicted
n ooi-o.o^ m
(} D" n j.
O 0.4- O.S •
A O.S-T.O A
D r.0-/0 •
A //") //")/) A
A jy it/it
O Cher 100 %
n
^
•
LI
EF- Predicted vs Actual (Weight)
A
2 o - *-\
® • D D An
I-i— i ^ •' ''• 1
A 3p| H;) B A
j < • o i"1'i n:^
F o« ^ ^
*
n u *
' A
1
• .-fl
u
i i i i
i 1 i i i 1 1 M \J
10 H
Average Vehicle Weight (ton)
0
Figure 4-13. Predicted vs Actual PMio Emissions Factor by Average Vehicle Weight.
-------
•CAAAO, _
-KflllO
40TO7B'
fl
4
JUUU o^
o
£
ft i ^flAO .• -
O
B
s
r, -jrifino
IB
• •M
Qi \niio
1?UU o
-M
flj
7\ 1^1 AAO -
U 1UUU o
u
PH
cn/w)
Tvv o
00.^ .
• 0
-500%
0.
01
Weigh!
+ <2J
• 2.5 - 5
A 5-15
• 15-40
• 1 A
-40
• •
m
Predictive Accuracy by Silt Loading
•
•
•
* A
A
> t
*«
I " * •
t* " m*j* bJm^< +f »A* •• A A
0 1 1 10 100 10
Silt Loading (g/m)
00
Figure 4-14. Predictive Accuracy by Silt Loading (unrestricted range).
-------
Pflfl°'n -i
ifinno „
93
•4-1
u anno ,• .
E
p
«*• ^00° ji -
V
u
c
V
u
t£ 400% -
4-i
• P4
(^
M
+J
« ~>00°'o -
y -UU 0
P4
V
PH
no/.
^OfF'fl -
0.
Predictive Accuracy by Silt Loading
t
>
•
• •
• •
•
•
• A A
: "M • i*'1:^ -
' iXiTT^TS^ %
31 0.1 1 10 100
Silt Loading (g/m )
4 Weight < 2. 5
• Weight 2.5-?
A Weight 5 -15
• Weight b-40
• Weight > 40
•
A A
1C
00
Figure 4-15. Predictive Accuracy by Silt Loading (restricted range).
-------
Predictive Accuracy by Vehicle Weight
jUUU/b
IxAAO -
43UU 0
/innno .
4UUU o
.5 3 sfino .
*J JjUUTO
2 JM M H ) ij
9
-ISAAO .
it _X'Uuo
o
qj
^ U U U '0
1
a
^ lUyUyb
$
w
C AAO .
M)U o
0% -
•\ ("1 M rt
A
A
t
• •
•
Silt Loading
+ <0.25
• 0.25-0,75
A 0.75-2.5
• 2.5-20
• - 20
A
i^U ^ ^lAl A
• W 'r •• •• •$ $•
OUU o ii I I
0 5 10 15 20 25 30 35
Average Vehicle Weight (ton)
• A
A Ali
1
40
45
Figure 4-16. Predictive Accuracy by Average Vehicle Weight (unrestricted range).
-------
1200%
1000%
E
o
s.
u
s
•V
o
0%
Predictive Accuracy by Vehicle Weight
rv.^:
10
15 20 25 30
Average Vehicle Weight (ton)
35
• sL 0.25-0.75
AsL 0.75-2.5
• sL>20
40
45
Figure 4-17. Predictive Accuracy by Average Vehicle Weight (restricted range).
-------
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,
• EPA 4/97 - 3 tests (speed 55, 45), 3 locations, and
• 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
further data set groupings. 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 preliminary regressions produced equations with varying constants
and exponents with correlation coefficient below 60%. By excluding the high silt loading data
and forcing a zero intercept, the correlation coefficient (R2) for the final equation is 72%.
This indicates that approximately 72% of the variations in the emissions factors are due to the
silt level and average vehicle weight. 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.s.
While several of the reports include measurements of PM2.5, 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/ PMio ratio to decrease with increasing
4-68
-------
concentration. At PMio concentrations above 1.0 mg/m3 the PM2.5/ PMio ratio was
between 0.1 and 0.15. The PM2.5/ 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 particulate 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 PM2.5/ PMio ratios during field tests. The range of PM2.5/ PMio ratios was 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.
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 PMso 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 38, 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 PMio 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 PMio 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
4-69
-------
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 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.
4.4 References for Section 4
1. D. R. Dunbar, Resuspension Of Particulate Matter, EPA-450/2-76-031, U. S.
Environmental Protection Agency, Research Triangle Park, NC, March 1976.
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 Particulate Emission Factors For Uncontrolled Industrial And Rural
Roads, EPA Contract No. 68-02-3158, Midwest Research Institute, Kansas City, MO,
September 1983.
4-70
-------
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 Particulate 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 Particulate 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 Particulate Emission Factors For
Industrial And Rural Roads, EPA-600/7-85-051, 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 Particulate Matter Emission Inventory — Sampling And Analysis,
Contract No. 68-02-4395, Midwest Research Institute, Kansas City, MO, May 1988.
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.
4-71
-------
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. C. Cowherd, Background Document for Revisions to Fine Fraction Ratios Used for
AP-42 Fugitive Dust Emission Factors. Prepared by Midwest Research Institute for
Western Governors Association, Western Regional Air Partnership, Denver, CO,
February 1, 2006.
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.
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.
29. EPA, 2009, MOVES2010 User Guide, United States Environmental Protection
Agency, Office of Transportation and Air Quality. EPA420B-09-041, December 2009.
30. Fugitive Particulate Matter Emissions, U.S. Environmental Protection Agency,
Research Triangle Park, NC, Midwest Research Institute Project No. 4604-06, April
15, 1997.
31. Midwest Research Institute, 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.
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
4-72
-------
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-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.
38. Technical Memorandum from William B. Kuykendal to File, Subject: Decisions on
Final AP-42 Section 13.2.1 "Paved Roads", October 10, 2002.
39. E-mail communication between Ron Myers of EPA/OAQPS/SPPD/MPG, RTF, NC
and Gary Dolce and Rudolph Kapichak of EPA/OTAQ/ASD/HDOC re. Paved Road
Test Data; October 12, 2010 through December 16, 2010.
40. C. Cowherd, Mobile Monitoring Method Specifications, Prepared by Midwest
Research Institute for Clark County Department of Air Quality and Environmental
Management, Las Vegas, NV, February 6, 2009.
41. C. Cowherd, Technical Support Document for Mobile Monitoring Technologies,
Prepared by Midwest Research Institute for Clark County Department of Air Quality
and Environmental Management, Las Vegas, NV, January 9, 2009.
42. R. Langston, R. S. Merle Jr., et al., Clark County (Nevada) Paved Road Dust Emission
Studies in Support of Mobile Monitoring Technologies, Clark County Department of
Air Quality and Environmental Management, Las Vegas, NV, December 22, 2008.
43. 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.
44. 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.
45. 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
4-73
-------
46. 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.
4-74
-------
Appendix A
Response to Comments
A-l
-------
Comments and responses on 2010 proposed revision of Section 13.2.1 Paved
Roads.
Commenter Page
Chatten Cowherd of Midwest Research Institute on behalf of the Center for the Study of Open Source
Emissions (CSOSE) 1
Rebecca Kies and Courtney Bokenkroger Senior Statistician of Midwest Research Institute, Kansas City,
MO 5
Greg Muleski of Midwest Research Institute 7
Camille Sears forthe Sierra Club 9
David E. James, PhD PE; Associate Vice Provost for Academic Programs; UNLV, Las Vegas, NV 14
Steve Zemba of Cambridge Environmental Inc forthe National Asphalt Pavement Association 16
Catharine Fitzsimmons, Chief, Air Quality Bureau and Lori Hanson Iowa Department of Natural
Resources 16
Pat Davis of MARAMA forthe States of New Jersey, Delaware, Maryland and Massachusetts 17
Julie McDill (MARAMA), David Fees (Delaware), Julie Rand (New Jersey) 18
Gary Garman of McVehil-Monnett 19
-------
Comments and Responses on June 2010 Draft Revision
of AP-42 Section 13.2.1 for Paved Roads.
Chatten Cowherd of Midwest Research Institute on behalf of the Center for the Study of
Open Source Emissions (CSOSE)
Comment: The general consensus among the Center for the Study of Open Source
Emissions (CSOSE) participants who have worked in this field is that the proposed
equation does not offer improved predictive capability but introduces additional data
requirements to the paved road emission inventory process.
Response: We disagree that the proposed equation does not offer improved predictive
capability. The predictive equation published in November 2006 produced negative
PMio emissions at very low silt loadings and negative PM2.5 emissions estimates
whenever a silt loading of less than 0.06 and average vehicle weight of 3.75 tons (or silt
loading of 0.1 and vehicle weight of 3 tons). As presented in Table 4-23 of the draft
background report, the 2006 equation had an average relative percent error of over
27,000 compared to the proposed equation with a relative percent error of 1,200. Part of
the error imbedded in the 2006 equation is due to the use of the estimated 1980's fleet
average vehicle emissions (average vehicle weight of 3.75 tons) for adjustment of the
equation presented in the 2003 revision of the AP-42 section. This average
underestimated the vehicle emissions of the fleets measured in almost 2/3 of the paved
road emissions test (58 of the 93 tests had average vehicle weights over 5 tons). Since
the proposed revision provided a correction to each test series based upon the average
vehicle weight presented in the test report and the correction used in the final revision
includes variations in speed, ambient temperature, year of vehicle fleet; this error has
been reduced. Combining the reduction in error of the test data with the use of a more
traditional revised stepwise regression of the paved road emissions data, we believe the
revised equation will provide a superior basis than the 2006 equation.
Comment: There is also the broader issue of adopting mobile monitoring as the basis for
more realistic emission inventorying of paved roads.
Response: EPA agrees that the adoption of mobile monitoring to estimate either the silt
loading of the road system or the emissions factor provides a significant advance in
characterizing the system wide emissions and the variation that exists with different
roads. The use of mobile monitoring offers the ability to characterize road classes which
have been problematic in the past due to resource constraints and safety issues. The
ability of mobile monitoring to provide a temporally and spatially resolved emissions
estimates and to characterize significantly more miles of roadways than were possible by
the traditional vacuuming, screening and weighing techniques is a distinct advantage. In
addition, the mobile monitoring method provides an excellent means for tracking system
wide management controls instituted to provide emissions reductions from roadway
emissions.
January 2011 Page 1
-------
Comments and Responses on June 2010 Draft Revision
of AP-42 Section 13.2.1 for Paved Roads.
In the final version of the AP-42 section we describe the mobile monitoring technique
along with a brief assessment that mobile monitoring provides significant improvements
in the estimation of road dust emissions caused by vehicle traffic.
Comment: The proposed equation has a significant new data input requirement (vehicle
speed) that increases the difficulty of generating paved road emission inventories.
Response: We disagree; access to the average vehicle speed of road segments is an
existing requirement for the accurate estimation of vehicle exhaust emissions in the
MOVES model. While the incorporation of the vehicle speed for every road segment
may increase the complexity of emissions inventory development, for most road systems
emissions estimates can be assembled by grouping of road segments into a limited
number of groups.
The assessment of the influence of the speed term on the predictive accuracy of the
resulting equation is a better criterion to determine whether this term should be used in
the equation. Limited improvement (or degradation) in the predictive accuracy of the
equation provides a more compelling rationale to exclude the speed term in the final
equation than the alleged difficulty of generating the emissions inventory. The
reassessment of the form of the emissions factor equation included the assessment of the
influence of speed on the predictive accuracy of the equation, the improvement of the
equation to address the variance which may be due to the independent parameters, and
the statistical significance of each variable in predicting the dependent variable.
Comment: Based on our discussions of the proposed equation and the technical analyses
presented by EPA, we find the scientific foundation for the revision unconvincing.
Response: The foundation upon which EPA proposed a revision of the paved road
equation was a proposal by the Corn Refiners Association (CRA) to perform emissions
tests to support the extension of the applicable source conditions. The Corn Refiners
retained the services of Midwest Research Institute (MRI) in Kansas City to perform the
emissions testing at lower average vehicle speeds to support the extension of the
applicable source conditions. Twenty two usable profiling tests were performed. In
addition to designing and conducting the emissions tests, MRI provided EPA with three
options for incorporating the new data into the paved roads section. The Agency decided
that returning to multiple estimation methods would recreate the problems that existed
prior to 1995 when there was two AP-42 sections for paved roads and multiple methods
within these two sections.
When MRI drafted the AP-42 section that included the CRA data, it was highlighted by
the Office of Transportation and Air Quality (OTAQ) that the proposal and adoption of a
revised equation had conformity implications that needed to be addressed. Several issues
associated with conformity were raised. These included the situation that areas
January 2011 Page 2
-------
Comments and Responses on June 2010 Draft Revision
of AP-42 Section 13.2.1 for Paved Roads.
containing low volume rural roads were predicted to have greater emissions of PMi0 than
the previous equation predicted. Another situation was that the revised equation may
result in greater predicted emissions of PM2.5 under some conditions. The greater
predicted emissions were the result of the existing equation generating negative
emissions for high volume roads. In an assessment to understand the extent and
significance of these issues, it was revealed that the vehicle exhaust, tire wear and break
wear emissions components were not addressed properly. The estimates of vehicle
exhaust, tire wear and break wear used in the 2003 revision did not account for the
significant differences in these emissions during the available tests and in addition
significantly mis characterized for the additional data provided by the corn refiners. For
the historical data, the proposed revision incorporated test specific emissions estimates as
calculated by MOBIL 6.2 and based upon the average vehicle weight reported for each
test. For the CRA data, the proposed revision incorporated test specific emissions
estimates as calculated by the MOVES model and based upon the average vehicle weight,
vehicle speed and estimated acceleration rates. For consistency and for improved
accuracy in predicting vehicle exhaust emissions, MOVES model estimates were
calculated for the historical data. While the incorporation of the data provided by the
Corn Refiners Association extended the capabilities of the equation to 1 mph, the Corn
Refiners Association data highlighted the variable significance of exhaust emissions and
the need to address these emission on a test by test basis. An additional advantage of
determining road emissions prior to developing the road emissions equation is that the
equation never predicted negative PMio or PM2.5 emissions.
Comment: Besides the problems stated above, we find difficulty in understanding the
scientific basis for replacing the existing PM2.s/PMio ratio published in 2006 with the
ratio that was previously used by EPA. The ratio in the existing equation was accepted by
EPA as an outcome of an experimental program supported by the Western Regional Air
Partnership (WRAP).
Response: In evaluating the data underlying the equation proposed in this revision, all of
the data were assessed to understand the basis and representativeness of the data. The
WRAP laboratory study was evaluated and was found to focus primarily on categories of
emissions that would generate very large concentrations of dust emissions and focused
primarily on western sources of these emissions. These types of emissions sources have
a high probability of overloading air sampling devices that depend on impaction to
collect particles of differing sizes. These sources are also predominately dominated by
sources where the emissions may have large variations over time depending on the
repetition rate of the activity which generates the emissions. Paved roads, especially
those with high traffic volumes and those that have neared their normal aged equilibrium
state generate dust emissions of greater consistency in concentration and particle size
characteristics. Not only are these emissions more consistent, the emissions
January 2011 Page 3
-------
Comments and Responses on June 2010 Draft Revision
of AP-42 Section 13.2.1 for Paved Roads.
concentrations are much lower except when the roads silt loading is very high. These
high silt loadings are not typical of public roads except for periods where sand is applied
as anti skid material, natural forces exacerbate the normal soil loading on the road or in
areas where there is a large track out of dirt from an adjacent unpaved area.
The WRAP study included the collection of seven soil samples. The samples included
sediment from Alaska, Alluvial Channel from Phoenix AZ, Agricultural Soil from
Phoenix AZ, Road Dust from the Las Cruces Landfill in New Mexico, Grazing Soil from
Radium Springs in New Mexico, Shoreline Soils from the Salton Sea in California and a
Barrow Pit from Thunder Basin Mine in Wyoming. In addition, three additional samples
which were used in the first Phase of the study were also used in the second Phase of the
study. These three samples included a Standard fine Arizona Test dust, a Standard coarse
Arizona Test dust and Lakebed Soil from Owens Dry Lake in California. For each of
these samples, the WRAP study states that two five gallon containers of soil were
collected. To collect this volume of sample from paved roads which are in equilibrium
would require sweeping or vacuuming of multiple miles of roadway. Additionally, none
of these samples are representative of aged material deposited on paved roads except for
paved roads which have had anti-skid abrasives (such as sand) applied during winter or
where significant windblown dust or track out dirt is deposited on paved roads.
Most of the laboratory tests performed to assess the revised PMi0/PM2.5 ratio to assign to
historical data was conducted at PMio concentrations above 2.5 mg/m3. The greatest
downwind concentration measured in tests used to support the paved road equation
development was 0.74 mg/m3 in run ID F45. Of the tests conducted in the wind tunnel
laboratory, only 15 percent of the samples were performed at concentrations below 0.74
mg/m3. The lowest PMio concentration measured during the laboratory study was 0.381
mg/m3. Of the 80 profiling tests used to support the paved road emissions factor equation
and where the downwind concentrations were available, only five had concentrations
greater than 0.358 mg/m3. In addition, over 80% of the profiling tests had downwind
concentrations less than 0.2 mg/m3 and 60% had downwind concentrations less than 0.1
mg/m3. In the wind tunnel laboratory studies, the only particulate used to challenge the
sampling devices was the material collected for the studies. The emissions measured
during the paved road profiling tests was a combination of emissions from the road
surface, engine exhaust, break wear and tire wear emissions. As presented in Table 4-17
of the draft background document, vehicle emissions can be a significant component of
the emissions measured by the profiling samplers. In three cases, the estimated exhaust,
break wear and tire wear emissions exceed the measured emissions and were assigned an
emissions factor of 0.01 g/VMT (see test runs M10, M12 and CF-3N). In an additional
10% of the profiling tests, about half of the measured emissions were estimated to be
exhaust break wear and tire wear emissions. And in approximately 35% of the profiling
January 2011 Page 4
-------
Comments and Responses on June 2010 Draft Revision
of AP-42 Section 13.2.1 for Paved Roads.
tests, the measured emissions were more than 10% exhaust; break wear and tire wear
emissions.
Based upon a more careful and thorough examination of the experimental design of the
WRAP study and the profile measurements conducted to characterize paved road
emissions it is concluded that EPA mistakenly accepted the conclusion that the PM2.5 to
PMio ratio for paved roads should be estimated at 15%. While the WRAP study provides
a reasonable indicator that past measurements of the particle size distributions below 10
jam are unreliable due to particle bounce and re-entrainment associated with impactors, it
does not discredit PM2 5 to PMio ratios established by field studies which used FRJVI or
equivalent monitors for measuring PM2 5 to PMio concentrations. While there were only
twelve test runs conducted during the profiling tests documented in the April 15, 1997
report by MRI for EPA, the PM2.5 to PMio ratios determined at these three locations
provide a superior estimate of a national ratio for estimating PM2.5 emissions than an
extrapolation from the WRAP laboratory study.
Rebecca Kies and Courtney Bokenkroger Senior Statistician of Midwest Research
Institute, Kansas City, MO.
Comment: The approach used by EPA to calculate the proposed paved road equation
differs from standard least-squares regression procedures. MRI recommends that
ordinary least squares regression procedures be used.
Response: EPA used the non standard approach in an attempt to provide an improved
predictor of emissions than the exponential form traditionally used for this section. In the
traditional form of regressing the equation, the log transformed data would be regressed
and include an intercept. Then when returned to normal space, the inverse log of the
intercept constant would be the multiplier for the silt and weight terms. The regression
terms for silt and weight would then be the exponents for those terms in the final
equation. More sophisticated statistical software and individuals with more thorough
knowledge in the application of stepwise non linear regression were not available at the
time but were used in the equation development for the published final section. EPA
used SAS which is more robust statistical software than Excel for developing the
equation used in the final AP-42 section. With guidance from the statistician, EPA used
Excel to explore limited alternative forms of the equation that could potentially provide
an equation with better predictive accuracy. EPA assessed the influence of test data that
potentially would adversely influence the resulting equation and assessed the use of
composite factors in an attempt to alleviate the additional problems identified by MRI's
statisticians. These assessments led EPA to exclude ten test data where the silt loadings
were greater than 20 g/m2 and to exclude test data where field measurements could not
quantify emissions due to traffic on the road. Additional regression methods available in
January 2011 Page 5
-------
Comments and Responses on June 2010 Draft Revision
of AP-42 Section 13.2.1 for Paved Roads.
SAS were evaluated following the exploratory assessment within Excel. The equation
which had the best predictive accuracy was based upon the traditional least squares
regression of the log transformed data with the intercept forced to zero.
Comment: Additional concerns about gaps in the range of data surfaced during our
statistical analysis. Notice the major holes highlighted by the circles in the speed-silt
loading and speed-weight boxes. The dataset is missing low silt loading, low speed; low
silt loading, high speed data; and low weight, low speed data. Ideally, the boxes relating
silt loading, weight, and speed should be completely filled with data points in order to
cover all ranges of possible occurrences and consider them to be independent factors in
the model.
Response: It is recognized that there are gaps in the data. In most cases, the contractor
performing the study (MRI in all cases) and the studies sponsor (EPA, industry) was
interested only in un-managed road systems at the test location. In some of these
instances, the condition highlighted would not be expected due to the physical forces
influencing the independent variables. For example, low silt loading would not be a
normal condition when the average vehicle speeds are low since the aerodynamic energy
imparted on the road surface would not be great enough to move the silt to the road
shoulder. This situation of low silt loading and low average speed may be a possibility
should there be active management of the silt loading on the road. Either the active
management of the road silt loading lacks the frequency to achieve lower silt loadings or
there was not a need to achieve these lower silt loadings. In other cases, the data may be
missing due to safety concerns associated with the collection of one or more pieces of
information. For example, the collection of data at roads with high speed and low silt
loading requires extensive time to collect sufficient material to quantify the low silt
loading. Should resources become available in the future improving the emissions factor
for paved roads, the collection of test data to fill in these data gaps will be suggested. In
addition, mobile monitoring methods may be a viable alternative to the vacuuming of
roads to estimate the silt loading of roads where there are safety concerns.
Comment: It is recommended that different modeling options be explored to find the
best fit and set of predictors for the data provided. Two such options are:
• Look at low speed and high speed models separately, potentially excluding
vehicle speeds under 5 mph from equation development.
• Use a composite factor of weight and speed together with either weight or speed
as independent variables in the regression. This helps alleviate the problems due
to multicollinearity between weight and speed seen in these data.
January 2011 Page 6
-------
Comments and Responses on June 2010 Draft Revision
of AP-42 Section 13.2.1 for Paved Roads.
Response: EPA assessed different modeling options to find a best fit. A return to
multiple sets of equations or values as predictors which introduce multiple results for
similar independent variables has been shown to create confusion, "results in shopping
for a fortuitous estimate" and adversarial debates. Any set of predictors should have
nearly identical results for comparable independent parameters where there multiple
predictors could be used.
EPA evaluated the exclusion of atypical independent parameter conditions such as the
very low speed conditions. Other conditions that were evaluated were very high silt
loading conditions. It was decided to exclude emissions tests with silt loading levels over
20 g/m2 due to the potential complexity of an equation needed to incorporate the different
characteristics that these few data present. While these high silt loadings may have been
representative of conditions which would be tolerated by the sources (or regulatory
authorities) in the mid to early 1980's, they are unusual conditions and may not be
reasonable to use in developing or assessing the best predictor for the more representative
and dominant situations. It is believed that management practices would be implemented
by sources and regulatory authorities to address extended durations of high silt loading
conditions. Additionally, an assessment of the final equations ability to estimate the
emissions of the ten tests with high silt loading. While there were changes in the percent
difference from actual emissions for individual test runs, the average percent difference
from actual emissions was almost the same as the 2006 equation.
Greg Muleski of Midwest Research Institute
Comment: The measured emission factor for CM-2 should be "63.5" rather than "6.35"
so the independent variable in Table 4-17 should have been about 52 g/vmt (rather than
the default value of 0.02 g/vmt).
Response: The measured PMio emissions factor in Table 4-15 was checked against the
value reported in the test report. The value of 0.14 Ib/VMT in the table was consistent
with the value in the submitted test report. As indicated in the comment, there was an
error in transcribing or units conversion to transfer the value from Table 4-15 to Table 4-
17. The emissions factor for the Corn Refiners Association test numbered CM-2 was
revised from 6.35 grams/VMT to 63.5 grams/VMT in Table 4-17. As a result, the
subtraction of the estimated vehicle exhaust, tire wear and break wear resulted in Road
Dust Emissions of 52.44 grams/VMT rather than 0.02 grams/VMT.
Comment: The two-step regression process described in Section 4.2.2.2 differs from
standard stepwise multiple regression used in the past AP-42 updates. It is not clear how
R-squared values at each step can be combined to obtain a meaningful value.
January 2011 Page 7
-------
Comments and Responses on June 2010 Draft Revision
of AP-42 Section 13.2.1 for Paved Roads.
Response: As indicated by several comments from individuals at MRI, the multi-step
regression used by EPA does not conform to traditional stepwise multiple regression
techniques. More traditional techniques were used in the development of the equation
used in the final section and SAS (which is more robust software for statistical analysis)
was employed to assess the predictive accuracy of the final equation.
Comment: The high degree of correlation between speed and weight precludes both
being included as independent terms in the emission factor equation.
Response: It was believed that the large number of tests where the road surface silt
loading was artificially changed through either the addition of sand or through removal
with mechanical means altered the normal correlation between the vehicle speed and the
road silt loading. With the use of more robust statistical software, the presence of inter
correlation between speed and silt loading was re assessed. In addition, the more robust
software allows a better determination of the potential improvement of an equation which
includes speed to predict road dust emissions. This assessment revealed that the use of
the speed term was contraindicated and the final equation contains only silt loading and
average vehicle weight as independent variables.
Comment: The goal should be to develop a predictive tool for situations without
measured emissions rather than trying to get the best fit for the set of measured
emissions.
Response: The use of Excel to generate the predictive equation made an evaluation of
the capability for the equation to predict data that was not part of the existing data set
difficult and labor intensive. The use of SAS allows for a more reliable assessment of the
equations predictive capabilities.
Comment: The geometric mean is the better choice than the arithmetic average when
working with the predicted/observed ratios.
Response: It is assumed that the use of the geometric mean is a metric to evaluate the
predictive accuracy of the equation through the use of the average predicted to observed
values. With the use of SAS, several indicators of the predictive capabilities of the
resulting equation were evaluated.
Comment: The document would have benefitted from a thorough review/edit prior to
being posted on the CHIEF web site.
Response: Prior to posting the final background report, the AP-42 Section and
background report was reviewed and edited more thoroughly and the Table of contents
was updated to provide an accurate indication of the contents of the chapters.
January 2011 Page 8
-------
Comments and Responses on June 2010 Draft Revision
of AP-42 Section 13.2.1 for Paved Roads.
Camille Sears for the Sierra Club
Comment: I have a few concerns regarding USEPA's proposed revision to AP-42
Section 13.2.1:
• USEPA's multiple regression analysis incorporating vehicle speed excludes a
valuable data set for assessing paved road PM emissions from industrial facilities.
• USEPA's proposed revision to AP-42 Section 13.2.1 results in a very significant
reduction in PMio and PM2.5 emission factors from paved roads in industrial
settings.
• It is unclear whether USEPA's proposed revision to AP-42 Section 13.2.1
improves upon predictive performance of the existing 2006 emission factor.
Response: The performance of the multiple stepwise regression of the data recognized
that incorporation of the speed term involved the exclusion of 22 test runs. EPA
recognized that the exclusion of these data could affect the resulting equation and decided
to include the speed term since the correlation coefficient showed a modest improvement.
Another commenter indicated that there are better software and process available than
were used by EPA to develop the equation. EPA employed software more suited for
stepwise multiple nonlinear regression than Microsoft Excel in the final equation
development. EPA used this improved software for a more rigorous assessment of the
influence of incorporating the speed term in the equation in this reassessment. (In EPA's
reassessment, it was revealed that the speed term provided no improvement in the
predictive accuracy of the resulting equation. As a result, the equation published in the
final AP-42 section includes only silt loading and average vehicle weight).
While EPA is cognizant of potential impact of any changes that may result in revising the
emissions factors in AP-42, the primary goal of emissions factors development is to
provide factors that provide as accurate of a prediction of the target population as
possible. The underlying data has considerable variation even when several of the
independent parameters are nearly identical. With the increased number of independent
parameters, it is possible that some situations where emissions will be greater than the
previous equation and some where emissions will be less.
While there may be some situations where the predictive performance of the proposed
equation performed poorer at predicting the underlying data, there were others where the
predictive performance was improved. Several measures were used to assess the
predictive performance of the revised equation and the final equation performs better than
the previous equation.
January 2011 Page 9
-------
Comments and Responses on June 2010 Draft Revision
of AP-42 Section 13.2.1 for Paved Roads.
Comment: USEPA excluded 22 tests performed at two integrated iron and steel plants
due to lack of vehicular speed data. These iron and steel plant source tests are crucial for
calculating fugitive dust emissions from industrial facilities, and excluding these data has
a very significant impact on predicted paved road emission rates. As discussed in the
following section, USEPA's proposed revision to the paved road emission factor will
reduce particulate emission calculations at typical industrial sites by roughly an order of
magnitude. This large, and perhaps unrealistic, reduction in calculated industrial paved
road emissions is an artifact of trying to develop an emission factor based on tests that
must include vehicle speed data.
Response: The exclusion of the 22 tests performed at iron and steel facilities did not
significantly bias the equation. An evaluation of the predictive precision of the equation
in the November 2006 version of the AP-42 Section for Paved Roads reveals that on
average the equation over predicted the 92 individual data by over 11,000%. While
approximately 50% of the predicted estimates underestimated the measured emissions
and 50% overestimated emissions, overestimates were significantly greater than the
underestimates. The 25 percentile value underestimated actual emissions by 54% while
the 75 percentile value overestimated actual emissions by 713%. The equation using
only silt loading and average vehicle weight which was rejected for the equation that
included speed overestimates actual emissions by 1,429%. The equation that was
proposed and includes the speed term overestimates actual emissions by only 890%. For
both the previously published equation and the proposed equation, the majority of the
overestimation appears to be associated with the lowest speeds, silt loading in the middle
third of the data and in the highest average vehicle weights. In these categories, it
appears that the previously published equation overestimates emissions more than the
proposed equation. With respect to roads with greater average vehicle weights such as
may be present at industrial facilities, the equation in the November 2006 AP-42 section
tended to overestimate emissions more than the proposed equation. Table 1 below
presents the independent parameter variables, estimated measured emissions, predicted
emissions by the 2006 AP-42 equation, the equation considered in the proposal that
includes only silt loading and average vehicle weight and the equation proposed that
includes silt loading, average vehicle weight and speed (with an average speed of 35 mph
assigned for unrecorded speeds). For those test conditions where average vehicle weight
was greater than 8 tons, the 2006 AP-42 equation tended to overestimate actual emissions
factors by about 350%. The equation that considered only silt loading and average
vehicle weight tended to overestimate actual emissions factors by about 3%. The
equation that considered silt loading, average vehicle weight and speed tended to
underestimate actual emissions factors by about 12%. A comparison between the
equation proposed for use and the equation that was considered but did not include the
speed term shows that the exclusion of the 22 test data that were missing the average
January 2011 Page 10
-------
Comments and Responses on June 2010 Draft Revision
of AP-42 Section 13.2.1 for Paved Roads.
speed did not adversely affect the average predictive capabilities of the equation. As
stated elsewhere, a more rigorous and capable statistical software package was used to
develop the final equation used in the AP-42 section.
For the equation published in the 2011 final AP-42 section, the predictive accuracy is
slightly improved over the equation proposed in the draft AP-42 section. As presented in
Table 2, the equation published in the final section provides a moderately better or worse
predictor of actual emissions for a few tests, but does not provide a significantly different
accuracy that the equation in the draft AP-42 section. While the equation presented in the
AP-42 section published in 2006 overestimates actual emissions factors by 350%, the
equation presented in the final 2011 section overestimates actual emissions by an average
of 77%.
January 2011 Page 11
-------
Comments and Responses on June 2010 Draft Revision
of AP-42 Section 13.2.1 for Paved Roads.
Table 1. Performance of 2006 AP-42 equation and equations considered for 2010 draft section revision.
ID#
ADI
F61
ADS
AD2
F62
F74
F34
CI-7
CI-8
F35
F38
F39
B58
F37
F45
F32
F27
B57
B60
B52
AUE1
B59
B55
B51
B54
B50
B56
F36
Silt
Loading
94.8
17.9
52.9
63.6
14.4
5.59
2.78
0.05
0.05
2.03
0.218
0.441
10.4
0.417
5.11
0.117
14.8
2.32
3.19
7.19
4
2.06
6.3
13.6
3.77
13.6
2.4
0.201
Average
Speed
23
NR
23
23
NR
NR
NR
15.3
15.3
NR
NR
NR
NR
NR
NR
NR
NR
NR
NR
NR
15
NR
NR
NR
NR
NR
NR
NR
Average
Weight
42
40
40
39
36
29
28
27
27
25
18
18
18
17
16
14
14
12
12
12
12
11
11
11
10
9.4
9.2
8.3
Measured EF
(g/VMT)
1478.189
461.174
231.189
340.807
315.177
543.498
186.561
0.589
1.949
296.721
165.766
251.766
366.778
76.038
210.911
52.170
129.070
194.216
438.216
34.616
2.286
347.289
182.289
139.289
92.662
81.506
125.421
54.168
Predicted Emissions (g/VMT)
Old AP-42
(sL,W)
4696.25
1476.91
2987.29
3241.81
1094.65
427.73
257.62
17.71
17.71
177.11
25.19
39.95
313.08
35.33
165.22
11.42
270.08
64.10
78.89
133.95
91.43
52.04
107.84
177.97
66.87
140.54
43.92
7.33
Rejected Proposal
(sL,W)
1575.71
325.20
881.27
1020.25
241.87
83.19
42.37
1.02
1.02
28.62
2.73
5.21
95.42
4.70
44.58
1.22
105.02
16.59
22.24
46.98
27.39
13.74
38.43
78.02
21.97
67.62
13.44
1.25
Proposed
(sL,W,s)
1156.43
235.28
636.21
751.93
179.78
63.42
31.39
0.53
0.53
21.73
2.05
4.09
90.51
3.76
42.38
0.98
111.94
16.78
22.93
50.85
24.99
14.26
42.66
90.68
24.52
83.43
15.07
1.26
Percent
Old AP-42
(sL,W)
218%
220%
1192%
853%
247%
-21%
38%
2907%
809%
-40%
-85%
-84%
-15%
-54%
-22%
-78%
109%
-67%
-82%
287%
3900%
-85%
-41%
28%
-28%
72%
-65%
-86%
Average 358%
difference from Measured
Rejected Proposal
(sL,W)
7%
-29%
281%
200%
-23%
-85%
-77%
73%
-48%
-90%
-98%
-98%
-74%
-94%
-79%
-98%
-19%
-91%
-95%
36%
1098%
-96%
-79%
-44%
-76%
-17%
-89%
-98%
3%
Proposed
(sL,W,s)
-22%
-49%
175%
121%
-43%
-88%
-83%
-11%
-73%
-93%
-99%
-98%
-75%
-95%
-80%
-98%
-13%
-91%
-95%
47%
993%
-96%
-77%
-35%
-74%
2%
-88%
-98%
-12%
January 2011
Page 12
-------
Comments and Responses on June 2010 Draft Revision
of AP-42 Section 13.2.1 for Paved Roads.
Table 2. Performance of 2006 AP-42 equation, equation proposed in 2010 draft section and Final 2010 section.
ID#
ADI
F61
ADS
AD2
F62
F74
F34
CI-7
CI-8
F35
F38
F39
B58
F37
F45
F32
F27
B57
B60
B52
AUE1
B59
B55
B51
B54
B50
B56
F36
Silt
Loading
94.8
17.9
52.9
63.6
14.4
5.59
2.78
0.05
0.05
2.03
0.218
0.441
10.4
0.417
5.11
0.117
14.8
2.32
3.19
7.19
4
2.06
6.3
13.6
3.77
13.6
2.4
0.201
Average
Speed
23
NR
23
23
NR
NR
NR
15.3
15.3
NR
NR
NR
NR
NR
NR
NR
NR
NR
NR
NR
15
NR
NR
NR
NR
NR
NR
NR
Average
Weight
42
40
40
39
36
29
28
27
27
25
18
18
18
17
16
14
14
12
12
12
12
11
11
11
10
9.4
9.2
8.3
Measured EF
(g/VMT)
1478.189
461.174
231.189
340.807
315.177
543.498
186.561
0.589
1.949
296.721
165.766
251.766
366.778
76.038
210.911
52.170
129.070
194.216
438.216
34.616
2.286
347.289
182.289
139.289
92.662
81.506
125.421
54.168
Predicted Emissions (j
Old AP-42
(sL, W)
4696.25
1476.91
2987.29
3241.81
1094.65
427.73
257.62
17.71
17.71
177.11
25.19
39.95
313.08
35.33
165.22
11.42
270.08
64.10
78.89
133.95
91.43
52.04
107.84
177.97
66.87
140.54
43.92
7.33
Proposed
(sL, W, s)
1156.43
235.28
636.21
751.93
179.78
63.42
31.39
0.53
0.53
21.73
2.05
4.09
90.51
3.76
42.38
0.98
111.94
16.78
22.93
50.85
24.99
14.26
42.66
90.68
24.52
83.43
15.07
1.26
g/VMT)
Final
(sL, W)
2886.277
600.570
1613.169
1859.513
442.254
149.639
76.359
1.886
1.886
51.060
4.773
9.073
161.994
8.133
75.113
2.093
172.829
27.253
36.436
76.446
44.785
22.375
62.006
125.074
35.222
106.525
21.429
2.010
Percent difference from
Old AP-42
(sL, W)
218%
220%
1192%
853%
247%
-21%
38%
2907%
809%
-40%
-85%
-84%
-15%
-54%
-22%
-78%
109%
-67%
-82%
287%
3900%
-85%
-41%
28%
-28%
72%
-65%
-86%
Average 358%
Proposed
(sL, W. s)
-22%
-49%
175%
121%
-43%
-88%
-83%
-11%
-73%
-93%
-99%
-98%
-75%
-95%
-80%
-98%
-13%
-91%
-95%
47%
993%
-96%
-77%
-35%
-74%
2%
-88%
-98%
-12%
Measured
Final
(sL, W)
95%
30%
598%
447%
40%
-72%
-59%
220%
-3%
-83%
-97%
-96%
-56%
-89%
-64%
-96%
34%
-86%
-92%
121%
1859%
-94%
-66%
-10%
-62%
31%
-83%
-96%
77%
January 2011
Page 13
-------
Comments and Responses on June 2010 Draft Revision
of AP-42 Section 13.2.1 for Paved Roads.
Comment: USEPA prepared a consequence analysis of the National Emission Inventory
("NEI") resulting from their proposed revision.8 USEPA found that their revised paved
road emission factor will significantly reduce PMio emissions in the NEI (up to 200%
reduction), while PM2.5 emissions are only slightly affected (some NEI calculations
increase, some decrease). USEPA, however, did not examine the affect of their draft
revised paved road equation on fugitive dust emissions from industrial sources.
Response: The estimated impact on State Emissions Inventories and the NEI was
performed as a tool for decisions which may need to be made to address conformity
requirements. The Agency may provide States with extensions of times for adopting
revised emissions estimates in their SIP and Transportation plans. These estimates were
also produced to assist State and local agencies understand the potential impact that the
revised emissions factors may have on their PMio and PM2 5 inventories which are being
prepared to address non attainment conditions and required SIP plan development. The
emissions inventory impact estimates were not produced as a decision criteria for revision
of the emissions factor equation. The only criteria used in assessing the proper equation
to publish are the representativeness of the underlying test data and the comparison of the
equation to the actual measured emissions. Although not presented in the background
report, the performance of the equation was made by ordering the available test data by
silt loading, average vehicle weight and by speed to evaluate whether there was any
systematic bias which was driven by one or more outlying data. Table 4-23 of the
background report for the proposed revision did include the average percent error for the
2006 equation and the proposed equation. When arranged by weight, the 2006 equation
produced errors of about 70,000 percent for vehicle weights of over 10 tons while the
proposed equation produced errors of about 2,500 percent. The equation published in the
final section produces errors of 5,000 percent for vehicle weights between 10 and 40 tons
and errors of about 20,00 percent for vehicle weights over 40 tons. Although when
limited to these high weight classes the performance appears to be worse, for lower
weight classes the new equation demonstrates superior performance to both the previous
published equation and the proposed equation.
David E. James, PhD PE; Associate Vice Provost for Academic Programs; UNLV, Las
Vegas, NV.
Comment: In many parts of the country where there is significant rain or a rainy season,
rain days may considerably effect estimated PMio emissions in the inventory. However,
for Las Vegas and other places like it in arid places, I tend to use a 'pessimistic' approach
that doesn't include the rain days, since rain occurs sporadically, and what rain does fall is
often very light. For the desert southwest, I think that it is best to look at the data without
rain adjustments.
January 2011 Page 14
-------
Comments and Responses on June 2010 Draft Revision
of AP-42 Section 13.2.1 for Paved Roads.
Response: It is recognized that the mitigation adjustment for rain events in AP-42 is
imperfect. It is recognized that with very light rain events, the silt loading on paved roads
may increase due to the removal of soil on the under carriage of vehicles. For most areas
of the US, these very light rain events are offset with heavier rain events. Over a month
to a year, these enrichment and mitigation events balance out. It should also be noted that
the mitigation level is not based upon any measured data and is an "engineering or expert
elicitation" estimate.
The emissions factors and the adjustment factors in AP-42 are educated estimates of the
national average value and do not include variations that may occur due to local and
regional influences. While some variation in the emissions factors for paved road has
been reduced through the incorporation of the independent variables silt loading, vehicle
weight and number of rain events, the remaining variation is still substantial. EPA does
not prohibit the use of alternative emissions factors or adjustments when accompanied by
a scientifically credible rationale and supporting data.
Comment: With locally derived data, we obtain results that are different from those that
might be predicted using default silt loading data. The actual impact on total estimated
PMio emissions in an inventory or SIP would depend on how much VMT was assigned to
each roadway category.
Response: It is recognized that the default silt loading information presented in AP-42
does not provide the precision and accuracy that may be needed to properly represent the
influence of emissions from paved or unpaved roads. It is also recognized that the
resources required collecting representative silt loadings for large numbers of roads is
substantial. However, where roads are believed to be significant contributors to the
levels of ambient air particulate matter, obtaining this information is valuable to
accurately estimate emissions. To address the needs to obtain this information in a cost
effective manner, we have included a discussion of the potential advantages of mobile
monitoring to develop temporally and spatially resolved silt loading (or emissions)
information.
Comment: I also ran a hypothetical sensitivity analysis comparing arbitrary
combinations of vehicle weight and silt loading, to see what the impacts of the new PMio
equation might be.
Response: It is recognized that different road classes may have different silt loadings
and the vehicles using these roads may have different average vehicle weights. These
variables will have differing influences on the predicted emissions from these roads. As
a result, the use of locally derived silt loading information is strongly encouraged.
January 2011 Page 15
-------
Comments and Responses on June 2010 Draft Revision
of AP-42 Section 13.2.1 for Paved Roads.
Steve Zemba of Cambridge Environmental Inc for the National Asphalt Pavement
Association.
Comment: The recommended default values for silt loading in draft Table 13.2.1-3, and
particularly that for asphalt batching, may be too high for typical current applications. The
recommended value is 120 g/m2, but, as you know, in EPA's 2000 Emission Assessment Report
for Hot Mix Asphalt Plants, a silt-loading value 3 g/m2 is suggested for paved roads at typical
hot-mix asphalt production facilities. Also, site-specific measurements at a hot mix asphalt
facility in Alexandria, Virginia in 2005 (using the sampling and analytical methods described in
AP42 Appendix C) found a silt loading level of 0.5 g/m2. This facility, which we analyzed in
detail for the City of Alexandria, employs aggressive dust suppression techniques.
Response: Values presented in Table 13.2.1-3 are based upon road dust samples collected in the
mid to late 1970's through the mid to late 1980's. It is unclear whether any management
practices were used at these facilities to control the silt loading of the roads where these samples
were collected. It is possible that current normal maintenance practices would achieve lower silt
loadings than are presented in the table. Statements in the documentation included in the reports
by the Corn Refiners Association and several other test programs used in the equation
development indicate that there was active management of the road surface dust levels. As a
result, the silt loading data collected during those test programs are lower than they would be
otherwise. While there is no requirement to use the silt loading values provided in the tables of
AP-42 updated silt loading data can be collected by any individual as long as they follow the
procedures presented in the AP-42 appendices. It is recommended that in addition to
documenting the sampling and analyses, the documentation include normal housekeeping
practices and special monitoring and maintenance practices at the collection sites. While we
cannot guarantee rapid incorporation of new silt loading data into the table, any reports
submitted will be posted for use by subsequent users.
Catharine Fitzsimmons, Chief, Air Quality Bureau and Lori Hanson Iowa Department of
Natural Resources.
Comment: The DNR supports the revision of this section to incorporate new data from
corn wet mills and to account for mean vehicle speeds below 10 miles per hour.
Response: Thanks for your support.
Comment: The proposed form of the equation requires that a mobile source emissions
model be run in order to determine a paved road emission factor. Obtaining the
emissions factor for vehicle emissions in this manner will be problematic as the DNR
does not have the resources to generate specific emissions factors for vehicle emissions
January 2011 Page 16
-------
Comments and Responses on June 2010 Draft Revision
of AP-42 Section 13.2.1 for Paved Roads.
by running MOVES20IO for every construction permitting project that includes a paved
haul road. The DNR suggests that either the empirical equation be developed to include
vehicle emissions from engine exhaust, tire and brake wear, or that a table of default
values be included in the section to account for vehicle emissions as an alternative to
running a mobile source emission model.
Response: While vehicle exhaust emissions may have been relatively stable for the last
twenty or thirty years, several regulatory programs which cover mobile source emissions
are expected to produce decreasing exhaust emissions over the next five to ten years. In
addition, engine exhaust like road dust emissions is highly dependent on the road
characteristics, meteorological conditions, vehicle speed, vehicle class and other
environmental conditions. As a result, a default engine exhaust equation will result in
unknown errors and may lead to incorrect decisions on different programs. While
decisions for many programs may not require the accuracy that would occur with
individual selection of the requisite parameters needed for the most accurate emissions
estimates, this would be a decision that should be made for each application. While State
agencies (Department of Transportation or Air Quality) may not have the resources or
time to generate a project specific emissions estimate for every project, individual States
are in a better position to develop default parameters (engine exhaust, silt and average
vehicle weight) which is appropriate for use for projects with different sensitivities.
Pat Davis of MARAMA for the States of New Jersey, Delaware, Maryland and
Massachusetts.
Comment: We have been examining the ERTAC/PECHAN emission factors for Road Dust
and Maryland noticed that the PM2.5 emission factors were zeroed out for the following road
types:
• Urban Collector
• Urban Minor Arterial
• Urban Other Principal Arterial
• Urban Other Freeways and Expressways
• Urban Interstate
Emission factors for PMio were found and there was no mention in the documentation of
why the PM2.5 emission factors were zeroed out, so we are bit confused.
Response: As a result of a revision of the ratio of the PM2.5 to PMio recommended by the
Western States Air Resources Council (WESTAR) from 25% to 15%, the multiplier k in the
predictive equation for PM2.5 was revised from 1.8 (for grams/VMT) to 1.1 (for grams/VMT)
in the 2006 revision of the paved roads AP-42 Section. With a constant emissions factor of
January 2011 Page 17
-------
Comments and Responses on June 2010 Draft Revision
of AP-42 Section 13.2.1 for Paved Roads.
0.1617 subtracted for the vehicle exhaust break wear and tire wear emissions, these
emissions result in a negative calculated road dust emission when one enters an average
vehicle weight of 4 tons or less and a silt loading of 0.2 grams/square meter or less. While
the k value used in the previous version of the equation resulted in negative emissions
whenever the silt loading was less than 0.03 grams/square meter, this affected only Freeways,
Expressways and Interstates and was believed to be rational since roadways with average
speeds of 55 mph (and the normal level of silt for that speed) had a high number of tests with
low measured emissions and were considered to be composed primarily of exhaust
emissions.
In the equation presented in the final version of this update, the estimated exhaust component
was subtracted from each source test prior to the stepwise regressions of the test data to
develop the predictive equation. As a result of the absence of vehicle exhaust, tire wear and
break wear in the predictive equation, there are no conditions that will result in negative
emissions for the road dust emissions.
Julie McDill (MARAMA), David Fees (Delaware), Julie Rand (New Jersey).
Comment: Here is Delaware's paved road dust spreadsheet for 2007, using the new
equation. We got very detailed with this category; estimating emissions by month.
Regarding the new equation, PMio was reduced by 58% from the emissions submitted to
MACTEC; while PM2.5 increased by 48%. I believe the PM2.5 increase is caused by two
factors-first, the PM2.5/ PMio ratio was increased to 25% (previously 15%). The second
reason is that under the old equation, one had to apply a correction factor, C, to remove the
exhaust, brake, and tire wear from the front part of the equation. By subtracting C at the end
of the equation, the resulting PM2.5 value went negative for several roadway types. Of course
we zeroed these out, but with the new method there is never a situation where the emission
factor value can go negative. Having negative emission factors result from the use of the old
equation was obviously a flaw in the method, so I expect the new equation is more accurate.
I look forward to NT's results when they apply the new equation, to see if they get changes
similar to mine.
New Jersey has similar results, but even more drastic for PM2.5. An increase in PM2.5 of
350% and a decrease in PMio of 46% I think one big cause is the difference in k factor,
among other changes. The k factor for PM2.5 went down from the 2003 AP-42 to the 2006
AP-42, and back up again in this new draft. We guessed at the new vehicle speed
requirement, but a slight variation in speeds will not make that much of a difference.
Response: It is correct that the k value and the C value both influence the predictive value
for the emissions factor. In addition, the exponents associated with the silt loading and the
average vehicle weight also influence the emissions estimates. It is also correct that the
January 2011 Page 18
-------
Comments and Responses on June 2010 Draft Revision
of AP-42 Section 13.2.1 for Paved Roads.
updated equation will not generate a negative emissions factor since the vehicle emissions,
tire wear and break wear will not be included in the equation development. Based upon an
assessment of the predicted to actual emissions factor for each of the available emissions
tests, the updated equation provides an improved estimate of the emissions compared to the
previous equation. It is also believed that the return to the PM2.5 to PMio ratio of 25% is a
better indicator of the PM2.5 than the 15% ratio that was based upon laboratory assessment
conducted for WESTAR.
Gary Garman of McVehil-Monnett.
Comment: It's good to see the paved road section is being revised. Thanks. It has been a
challenge in the past explaining to industrial clients that paving a road would actually result
in higher predicted emissions than if the road is left unpaved. I think we'll see more paving
and actual emission reductions as a result of the new equation. A few editorial comments on
the draft paved road section:
Page 13.2.1-1, third paragraph, first sentence..change to "The particulate emission factors
presented in a previous version.."
Page 13.2.1-5, third paragraph, last sentence..change "Table 13.2.1-3" to "Table 13.2.1-2"
Page 13.2.1-8, fifth paragraph, first sentence..change "Table 13.2.1-3" to "Table 13.2.1-
2"
Page 13.2.1-9, second paragraph, second sentence..remove hyphen between "not" and
"suggest"
Table 13.2.1-3...the page number this table is on should be changed to 13.2.1.10. Also,
total loading range for iron and steel should be 0.006-4.77, not 43.0-64.0.
Page 13.2.1-11, first paragraph, fourth sentence..remove hyphen between "any" and "of
Thanks again. I look forward to this draft being finalized.
Response: An assessment of the paved verses unpaved road equation performance will be
conducted. A statement will be added to the paved road section explaining that under some
high silt loading conditions the equation may predict higher emissions than for an unpaved
road and that for these conditions the unpaved road equation should be used. The
typographical errors will be corrected in the final version.
January 2011 Page 19
-------
Comments on Proposed Paved Road Equation
Cowherd, Chatten Ron Myers 08/31/201003:OOPM
"Kies, Rebecca", "Muleski, Greg"
This message has been forwarded.
Hello Ron,
Thank you for the opportunity to comment on the proposed revision to the paved road dust equation in
AP-42 section 13.2.1. The attached letter presents comments developed on behalf of the Center for the
Study of Open Source Emissions (CSOSE).
As you know, the revised equation (proposed by EPA as a replacement for the existing equation) and its
technical foundation were topics of discussion during the August 18 teleconference hosted by the
CSOSE. During this teleconference and in related information exchanges, the general consensus among
CSOSE participants who have worked in this field is that the proposed equation does not offer improved
predictive capability but introduces additional data requirements to the paved road emission inventory
process.
There is also the broader issue of adopting mobile monitoring as the basis for more realistic emission
inventorying of paved roads. In previous conversations, I believe that you have acknowledged the clear
advantages of mobile monitoring over the traditional AP-42 method for determining paved road dust
emissions with its reliance on limited and difficult measurements of silt loading.
We believe that the CSOSE constitutes a substantial resource in resolving these issues and in assisting
EPA with the goal of developing improved emission factors such as those applicable to paved road dust
emissions.
Please contact me with any questions or comments.
Sincerely,
Chat Cowherd
Chatten Cowherd, Jr., Ph.D.
Midwest Research Institute
425 Volker Boulevard
Kansas City, MO 64110
(816) 753-7600 ext. 1586
(816) 360-5346 direct dial
-------
Center for Study of
Open Source Emissions
Chatten Cowherd, Jr., Ph.D.
Director
ccowherd@mriresearch.org
(816)360-5346
August 31,2010
Mr. Ron Myers
U.S. Environmental Protection Agency
Research Triangle Park NC 27711
RE: Proposed Revision to AP-42 Emission Factor Equation for Paved Road Dust
Dear Mr. Myers:
The Center for Study of Open Source Emissions (CSOSE) is pleased with the opportunity
to submit comments in response to EPA's proposed revision of the emission factor equation in
AP-42 Section 13.2.1. It should be noted that these comments were prepared by the undersigned
as Director of CSOSE, taking into account verbal and written communications from interested
members of the Center, including those provided during a presentation and discussion of this
topic in the August 18 teleconference hosted by the Center. However, this letter was not
circulated to CSOSE participants for review prior to submission.
One of the goals of CSOSE is to promote transparency and collaboration in the
documentation of test results and in the use of those results to derive effective tools for
compliance with air quality standards. We believe that this goal is consistent with EPA's stated
goal to develop a self-sustaining emissions factors program that produces high quality, timely
emissions factors, better indicates the precision and accuracy of emissions factors, encourages
the appropriate use of emissions factors, and ultimately improves emissions quantification (see
EPA's Advance Notice of Proposed Rulemaking on "Emission Factors Program Improvements,"
Oct. 14, 2009).
We acknowledge the concerns of various parties related to the scientific foundation for the
proposed equation as well as the increased effort required in developing vehicle speed data to
include in paved road emission inventories. CSOSE participants have presented analyses
demonstrating that the proposed equation does not provide an improved predictive capability
above that provided by the current equation. In addition the proposed equation has a significant
new data input requirement (vehicle speed) that increases the difficulty of generating paved road
emission inventories and that has possible implications on projected effectiveness of current SIP-
mandated control strategies.
Based on our discussions of the proposed equation and the technical analyses presented by
EPA, we find the scientific foundation for the revision unconvincing. This leads us to question
the process used in advancing this proposed equation. Our understanding of the rationale for
revision of the existing equation might be clarified if there were evidence of an internal review
process within EPA that raised issues and resolved them appropriately.
-------
Besides the problems stated above, we find difficulty in understanding the scientific basis
for replacing the existing PM-2.5/PM-10 ratio published in 2006 with the ratio that was
previously used by EPA. The ratio in the existing equation was accepted by EPA as an outcome
of an experimental program supported by the Western Regional Air Partnership (WRAP). That
experimental program included regular progress updates in WRAP teleconferences with
participation from EPA representatives. To our knowledge, WRAP was never directly informed
in advance that the stated conclusions of their study are now being discounted.
We have encouraged others to present comments on the proposed equation that are
supportive of the goal of providing improved emission factors. At the time of this writing, we
are aware that separate comments are being submitted by Midwest Research Institute (Ms.
Courtney Bokenkroger and Dr. Greg Muleski), by the Clark County Department of Air Quality
and Environmental Management (Mr. Rodney Langston) and by the University of Nevada at Las
Vegas (Dr. David James).
We trust that EPA will publish all comments as well as the responses to each comment.
This will be of great assistance to all in moving toward the best possible use of the test data in
supporting a meaningful and appropriate emission factor equation for entrained dust from paved
roads.
In summary, we conclude that there is no compelling scientific justification for adopting the
proposed emission factor equation as a replacement for the existing equation. This problem is
compounded by the requirement for additional input data and the potential impact on current and
future emission inventories as tools for compliance determination. We conclude that an internal
EPA review may not have been performed prior to posting the proposed equation for public
comment. Finally we emphasize the importance of publishing all comments submitted to EPA
along with EPA's responses to each comment.
If you have questions about these comments submitted on behalf of CSOSE, please contact
the undersigned by email (ccowherd(a),mriresearch.org) or by telephone (816) 360-5346. We
look forward to your responses to these comments. We believe that CSOSE constitutes a
substantial resource in resolving the above issues and in assisting EPA with the goal of
developing improved emission factors for open sources. Thank you again for the opportunity to
submit comments on the proposed revision to the current AP-42 equation for paved road dust
emissions.
Sincerely yours,
CENTER FOR STUDY OF OPEN SOURCE EMISSIONS
Chatten Cowherd, Jr., Ph.D.
Director
L*\ ,
-------
From: "Kies
To: Ron Myers/RTP/USEPA/US@EPA
Date: Tuesday, August 31, 2010 11:17AM
Subject: statistical Comments on Draft AP-42 Section 13.2.1
History: - This message has been forwarded.
Ron,
Thank you for the opportunity to comment on the proposed AP-42 paved roads section 13.2.1.
Attached to this email are MRI's comments resulting from statistical analysis of the proposed
changes to the paved road equation by MRI Senior Statistician, Courtney Bokenkroger. These
comments have been reviewed by myself, Chat Cowherd, and Greg Muleski.
Please feel free to respond with any questions or comments.
Sincerely,
Becky Kies
Rebecca Kies
Assistant Scientist
Midwest Research Institute
425 Volker Blvd. KCMO 64110
(816) 360-3825 (direct)
(816) 753-7600x1818
rkies@mriresearch.org
This message is intended exclusively for the individual or entity to which it is addressed.
This communication may contain information that is confidential, proprietary, privileged or otherwise legally exempt from disclosure.
If you have received this message in error, please notify the sender immediately by facsimile, e-mail or phone and delete all copies of the message.
Attachments:
Comments in Response to EPA Proposed Section 13.2.1 Paved Road Equation.pdf
-------
MIDWEST RESEARCH INSTITUTE
Courtney Bokenkroger
Senior Statistician
816-360-5303
August 31,2010
Mr. Ron Myers
U.S. Environmental Protection Agency
Research Triangle Park NC 27711
RE: Draft AP-42 Section 13.2.1 Paved Roads
Dear Mr. Myers:
Midwest Research Institute (MRI) is pleased with the opportunity to submit comments in response to
EPA's proposed draft revisions to AP-42 Section 13.2.1 Paved Roads and corresponding background
documents. We applaud EPA's effort to improve the quality of the emission factor model for paved roads
and appreciate your consideration of external comments.
MRI has a productive history of work in air pollutant source testing, process characterization, and
development of emission factors for EPA's Emission Factor Handbook (AP-42). Besides serving for
more than 25 years as an EPA contractor in the testing of ducted sources and in associated methods
development, we have made unique contributions to the development and application of test methods for
open (non-ducted) sources. The open sources that we have tested over the past 35 years include
agricultural operations, paved and unpaved roads, construction activities, surface mining activities,
military training operations, and open area wind erosion. Because of the large natural variability of these
sources, MRI pioneered the concept of predictive emission factor equations rather than relying on simple
averaging of test results for fugitive dust sources. This approach reduced the uncertainty of emission
factor estimates for unpaved roads—as the largest contributor to the national PM-10 emission total—by up
to two orders of magnitude.
Our comments on the draft AP-42 Section 13.2.1 Paved Roads focus on a statistical analysis of the
data set and procedure used to calculate the proposed new paved road emission factor equation and can be
summarized as follows:
• The approach used by EPA to calculate the proposed paved road equation differs from
standard least-squares regression procedures. MRI recommends that ordinary least-
squares regression procedures be used.
• In using ordinary least squares regression to compare models for only the field
measurements that included vehicle speed, we find that inclusion of speed in the model
takes away from the explanation of variance of the model (R2) and that vehicle speed
does not have a statistically significant relation to emission factor.
• It is recommended that different modeling options be explored to find the best predictive
equation from the data provided. Two such options are:
o Look at low speed and high speed models separately, potentially excluding
vehicle speeds under 5 mph from equation development.
-------
Use a composite factor of weight and speed together with either weight or speed
as independent variables in the regression. This helps alleviate the problem of the
multicollinearity of weight and speed seen in these data.
Model Comparison
The data set used by EPA to develop the proposed paved road equation included emission factor, silt
loading, weight, and speed. Out of 93 total observations, 71 included speed data. The 71 observations that
included speed data were the ones used by MRI for model comparison.
It is not reasonable to compare the proposed model with other possible models for the data using the
approach taken by EPA to calculate the proposed model. The double-regression approach used renders
two different R-square values (one for each regression), neither of which accurately represent the
proportion of variability explained by the final resulting model.
The resulting equations obtained from running least-squares regression on the log transformed,
normalized values with and without inclusion of speed on the set of 71 data points appear below.
Regression without speed:
Regression including speed:
EF = 6.51 *(siltloading/2)097* (weight/3)036
EF = 6.41 * (silt loading/2)097* (weight/3)027 * (speed/30)'012
Regression
without speed
Regression
including speed
Variance Explained
by Model
R2= 0.6335
R2= 0.6288
Variable
Silt loading
Weight
Silt loading
Weight
Speed
p-value
< 0.0001
0.0739
< 0.0001
0.3892
0.7140
"Proportion of
Variance Explained"
0.62673
0.04621
0.62673
0.04621
0.00202
The R-square value from a standard least-squares regression represents the proportion of variability
explained by the model. When speed is included in the regression, the R-square is slightly lower than
when speed is not included. This means that the model explains less of the variance seen in emission
factor when speed is included than when it is not.
The column labeled p-value represents the statistical significance of the factor in the prediction of the
dependent variable (the lower the p-value, the greater the significance). In order to be considered
statistically significant for inclusion in the model, generally p-values are less than or equal to 0.15. Note
that the p-values for the equation that includes speed indicate that speed and weight are both statistically
insignificant (this is because there is likely a relationship between weight and speed). When speed is not
included, weight is statistically significant.
The column labeled "proportion of variance explained" is the proportion of R-square that is explained
by each individual variable. Speed contributes almost no additional "explanation of variance" to the
model (i.e. speed doesn't add much to the predictive power of the model).
-------
Gaps in the Data
Additional concerns about gaps in the range of data surfaced during our statistical analysis. Notice
the major holes highlighted by the circles in the speed-silt loading and speed-weight boxes. The dataset is
missing low silt loading, low speed; low silt loading, high speed data; and low weight, low speed data.
Ideally, the boxes relating silt loading, weight, and speed should be completely filled with data points in
order to cover all ranges of possible occurrences and consider them to be independent factors in the
model.
Conclusions and Recommendations
The proposed approach used by EPA to calculate the proposed paved road equation differs from
standard regression procedures. The two-regression approach used results in two different R-square
values, neither of which accurately represent the proportion of variability explained by the final resulting
model. Additionally, different data sets were used to develop the two models.
In using ordinary least-squares regression to compare data models for the same data, inclusion of
speed in the model does not significantly add to the explanation of variance in emission factor. Also,
speed does not have a statistically significant relationship with emission factor.
The low-speed data (< 5 mph) have an un-proportionally large effect on the fit of the model. This is
of concern because there are not enough low speed data to represent all ranges of weight and silt loading.
-------
Because the correlation between the log-transformed, normalized weight and speed in the model is
approximately -0.78, inclusion of both factors introduces issues related to multicollinearity. The problem
with having highly correlated variables in a model is that the coefficients are easily influenced by the
dataset used in estimation and may not be meaningfully interpreted because they are not independent.
It is recommended that different modeling options be explored to find the best fit and set of predictors
for the data provided. Two such options are:
• Look at low speed and high speed models separately, potentially excluding vehicle
speeds under 5 mph from equation development.
• Use a composite factor of weight and speed together with either weight or speed as
independent variables in the regression. This helps alleviate the problems due to
multicollinearity between weight and speed seen in these data.
If you have questions or comments about this evaluation of the proposed paved road equation in EPA
AP-42 Section 13.2.1, please contact the undersigned by email (cbokenkroger@mriresearch.org) or
telephone (816- 360-5303). We look forward to your response on this matter.
Sincerely yours,
MIDWEST RESEARCH INSTITUTE
Courtney Bokenkroger
Senior Statistician
-------
Comments on Section 13.2.1 draft
Muleski, Greg to: Ron Myers 08/30/2010 02:58 PM
This message has been forwarded.
Ron
Thank you for the opportunity to comment on the draft paved road emission
factor. Based on my analysis for the Corn Refiner Association member
companies, I know that the revision moves the power on the "mean vehicle
weight term" in the right direction.
My specific comments are as follows:
1. The measured emission factor for CM-2 should be "63.5" rather than "6.35"
so the independent variable in Table 4-17 should have been about 52 g/vmt
(rather than the default value of 0.02 g/vmt).
2. The two-step regression process described in Section 4.2.2.2 differs from
standard stepwise multiple regression used in the past AP-42 updates. It is
not clear how R-squared values at each step can be combined to obtain a
meaningful value.
3. The high degree of correlation between speed and weight precludes both
being included as independent terms in the emission factor equation.
Furthermore, it is not clear what inclusion of speed does for the model. The
goal should be to develop a predictive tool for situations without measured
emissions rather than trying to get the best fit for the set of measured
emissions.
4. The geometric mean is the better choice than the arithmetic average when
working with the predicted/observed ratios.
5. The draft background document is in rough shape. It would have been better
to have posted only Section 4 to avoid confusion arising from the table of
contents, references, etc. The document would have benefitted from a thorough
review/edit prior to being posted on the CHIEF web site.
Please feel free to contact me with any questions or comments.
Greg Muleski
This message is intended exclusively for the individual or entity to which it
is addressed.
This communication may contain information that is confidential, proprietary,
privileged or otherwise legally exempt from disclosure.
If you have received this message in error, please notify the sender
immediately by facsimile, e-mail or phone and delete all copies of the
message.
-------
FW: Message from KMBT_421
Muleski, Greg to: Ron Myers 08/26/201009:52 AM
This message has been replied to and forwarded.
Ron
Sorry I missed your phone call. I've attached 2 annotated pages from your
draft background document that show the problem.
Original Message
From: copier211h@mriresearch.org [mailto:copier211h@mriresearch.org]
Sent: Thursday, August 26, 2010 8:50 AM
To: Muleski, Greg
Subject: Message from KMBT_421
This message is intended exclusively for the individual or entity to which it
is addressed.
This communication may contain information that is confidential, proprietary,
privileged or otherwise legally exempt from disclosure.
If you have received this message in error, please notify the sender
immediately by facsimile, e-mail or phone and delete all copies of the
message.
-------
Ta
ble 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-ii
CI-12
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)1
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
Measured PM10 emission
factor (lb/VMTf
_
0.06 j
0.06
0.06
0.05
0.05
0.025
0,25
-
_
0.0036
0.0066
_
.
a Vehicle speed for inbound (loaded) /outbound (empty) trucks determined by accumulating time required to travel a
measured distance.
Q
Surface silt loading sample information provided by Cargill.
c "-" 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
GM-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 emjifiion
factor (Ib/VMT)
.JLQUJL
C 0.14 ^
"Trwr"-"
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
-------
Table 4-4. Measurement-Based PM-1Q Emission Rates/Factors
Run
CM-1 (low speed)
CM-2 (slop/go)
CM-4 (tow speed)
Traffic
rate*
184
41.9
156
Mean
vehicle®
ws[ght
39.8
39.8
39.5
Silt loading
0.72
0.72
0,70
Measured
line source
emrssfon rate
(a/mlle-s)
0,27°
0.76 °
0.31"
Measured
par vehicle
emission factor
(Ib/vrnt)
dSSb>
0.016
AP-42
predicted
emission8
factor (Ib/vmt)
0,40
0,40
0.39
Vehicle walghts based on the following values (Ibs) In MCP's Title V application:
Empty Loaded
Straight Truck 10,000 28,000
Tandem 16,000 45,000
Semi 2T,000 80,000
Afl trucks ware Inbound and full (including "drone" passes), AP-42 factor based on value of 40 tons.
Based on 2 lines of queued traffic.
4.2 Discussion and Recommendation
The PM-1Q emission factors developed in the 2003 testing program provide further
evidence that Equation 1-1 produces highly overestimated predictions for PM-10
emissions from paved road traffic at the Marshall facility. At least two features of the
AP-42 modeling approach fail to describe the emissions observed at Marshall,
First, re-entrained surface road dust is not nearly as dominant in the emissions
measured at Marshall as compared to the AP-42 emission factor database. This was first
noted in the 2001 test report [3], The 2003 program provides no evidence that measured
emissions exhibit a dependency on silt loading even remotely similar to that found hi
Equation 1-1,
Just as importantly, the 2003 test results point up a second shortcoming of AP-42 in
modeling emissions at the Marshall plant. The predicted emission rate using AP-42 is
found by multiplying Equation 1 -1 by the traffic volume. In other words, the emission
rate varies linearly with traffic rate. For example, if twice as many vehicles pass during
one hour compared to the next, then the first how's emission rate should be twice that of
the second.
However, measured emission rates are remarkably constant over flie range of traffic
rates considered during the two test programs. Figure 4-1 presents the line source
emission rate measured in both 2001 and 2003 for fie inbound corn haul route. The
emission rate is plotted against the traffic rate. Also included for comparison are the
predicted values using AP-42. Measured emission rates show ao significant relationship
with traffic rate,
4-4
Emission Tests of Paved Road Traffic
at ADM's Marshall, Minnesota Facility
Test Report
For
McVehil-Monnett Associates
MR! Project No, 310479.1.001
Decembers, 2003
-------
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) («/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
6.35t
-3-»d 6.35"
^* 7.26
1.000
1.000
1.000
1.000
1.000
0.759
0.759
1.000
J 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
-«=4J_1S,
Cc=o(ffl(»
<£$48
/" /
4-45
-------
Page 1 of 1
From: "Camille Sears"
To: Ron Myers/RTP/USEPA/US@EPA
Date: Monday, August 30, 2010 11:53PM
Subject: Re: AP-42 13.2.1
History: - This message has been forwarded.
Hi Ron,
Attached are our comments. Please let me know if you have questions.
Your help is greatly appreciated!
Best wishes,
Camille
Attachments:
SC-13,2.1. comments.
https://rtairmaill.rtp.epa.gov/mail/rmyers.nsf/9ff539ale24f5aaf852577890046a8f6/56708... 10/22/2010
-------
Camille Sears 502 W. Lomita Ave., Ojai, CA 93023
Tel: (805) 646-2588 e-mail: camille.marie@sbcglobal.net
August 3 0,2010
Mr. Ron Myers
U.S. Environmental Protection Agency
Office of Air Quality Planning and Standards
Research Triangle Park, NC 27711
Subject: Proposed Revision to AP-42 Section 13.2.1 - Paved Roads
Dear Mr. Myers,
Thank you for the opportunity to provide comments on the proposed revision to AP-42 Section
13.2.1. I have reviewed the proposed AP-42 revisions and associated reference documents and on
behalf of Sierra Club offer the following brief comments.
1. Introduction
The existing USEPA air pollution emission factor for fugitive dust from vehicle traffic on paved
roads is as follows:1
E=k(Siy2)°'65*(W/3)L5-C
Where: E = annual or other long-term average emission factor in the same units as k
k = particle size multiplier (from Table 13.2.1-1, k = 0.0024 Ib/vehicle mile traveled (VMT) for
PM2.5 and 0.016 Ib/VMT for PMio)
SL = road surface silt loading (g/m2)
W = average weight of vehicles (tons)
C = emission factor for 1980s vehicle fleet exhaust, brake wear, and tire wear (from Table 13.2.1-2,
C = 0.00036 Ib/VMT for PM2.5 and 0.00047 Ib/VMT for PMio)
The existing version of AP-42 Section 13.2.1 appears to be based on 64 source tests performed prior
to 1995, the date when the paved road emission factor first takes its current form.
In July 2008, the Corn Refiners Association ("CRA") proposed a revision to AP-42 Section 13.2.1.
CRA's proposed revision is based on 22 additional source tests performed at ethanol plants in 2001
1 USEPA, Office of Air Quality Planning and Standards, AP-42. Section 13.2.1. Paved Roads. November 2006, p.
13.2.1-4.
-------
Revisions to AP-42, 13.2.1
August 30, 2010
Page - 2
r\
through 2003. CRA recalculated a paved road emission factor including the 64 source tests used by
USEPA as the bases for the existing emission factor, plus the 22 additional CRA source tests (for a
total of 86 tests). Based on their regression analyses, CRA proposed a revised paved road emission
factor with the following form:3
E=k(Si/2)0'8*(W/3)°-8-C
CRA also proposed a revised particle size multiplier (k), where k = 0.0034 Ib/VMT for PM2.5 and
0.023 Ib/VMT for PMio)
The additional 22 tests performed by the CRA include:
• Nine tests performed on roads at the Minnesota Corn Processors facility, Marshall,
Minnesota, during April 2001;
• Eight tests performed on roads at the Minnesota Corn Processors facility, Columbus,
Nebraska, during June 2001;
• Two tests performed on roads at the Cargill Sweeteners North America facility, Blair,
Nebraska, during August 2002;
• Three tests performed on roads at ADM's facility, Marshall, Minnesota, during September
2003 (this is the same facility as the April 2001 tests).
In May 2010, USEPA developed and proposed a revision to AP-42 Section 13.2.1, "Paved Roads."
From USEPA:
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.4
USEPA's update to AP-42 Section 13.2.1, however, incorporates other data than that collected by
the Corn Refiners Association, and, more importantly, USEPA's update excludes important data that
have been used in developing the existing paved road emission factor. In summary, USEPA's 2010
update to AP-42 Section 13.2.1 incorporates the following data base changes:
• Including the 22 CRA tests performed in 2001 through 2003;
• Including three tests performed on public roads in Denver, Colorado, during March 1996;
2 Corn Refiners Association, Paved Road Modifications at AP-42, Background Documentation, MRI Project No.
310842, July 18, 2008, p. 4.
3 Id., p. 20.
4 USEPA, Emission factor Documentation for AP-42, Section 13.2.1, Paved Roads, Draft, June 2010, p. 2-9.
-------
Revisions to AP-42, 13.2.1
August 30, 2010
Page - 3
• Including two tests performed on public roads in Raleigh, North Carolina, during April 1996;
• Including two tests performed on public roads in Reno, Nevada, during June 1996;
• Excluding 22 tests performed at two integrated iron and steel plants - one located in Houston,
Texas, and the other in Middletown, Ohio (during 1980 and 1981).5
USEPA developed a proposed multiple regression equation based on paved road silt loading, mean
vehicle weight, and vehicle speed. The existing version of AP-42 Section 13.2.1 is based on
regression analyses of silt loading and mean vehicle weight. Since vehicle speed was not measured
at the 22 tests from the two integrated iron and steel plants (Houston, Texas and Middletown, Ohio
during 1980 and 1981), these tests were excluded from the data set.
USEPA's proposed revision to AP-42 Section 13.2.1, which is based on 71 individual source tests,
takes the form:6
E= k(SL/2)°'98 * (W/3)0'53 * (S/30)0'16
Where: E = annual or other long-term average emission factor in the same units as k
k = particle size multiplier; k = 0.0037 Ib/VMT for PM2.5 and 0.015 Ib/VMT for PMio
SL = road surface silt loading (g/m2)
W = average weight of vehicles (tons)
S = average vehicle speed (miles per hour)
This equation does not incorporate emissions from engine exhaust and brake and tire wear, which
will need to be estimated and added using USEPA's MOBILE6.2 or MOVES2010 models.
I have a few concerns regarding USEPA's proposed revision to AP-42 Section 13.2.1:
• USEPA's multiple regression analysis incorporating vehicle speed excludes a valuable data
set for assessing paved road PM emissions from industrial facilities.
• USEPA's proposed revision to AP-42 Section 13.2.1 results in a very significant reduction in
PMio and PM2.5 emission factors from paved roads in industrial settings.
• It is unclear whether USEPA's proposed revision to AP-42 Section 13.2.1 improves upon
predictive performance of the existing 2006 emission factor.
5 Id., p. 4-18.
6 USEPA, Office of Air Quality Planning and Standards, AP-42, Draft Section 13.2.1, Paved Roads, p. 13.2.1-4.
-------
Revisions to AP-42, 13.2.1
August 30, 2010
Page - 4
2. Key Industrial Source Tests are Excluded from USEPA's Revised Factor
USEPA's proposed revision to the paved road emission factor includes a third variable, mean vehicle
speed. Vehicle speed, however, does not appear to be an important predictive aid to the overall
emission factor equation. This is evidenced by vehicle speed having a small (0.16) exponential term
in USEPA's proposed paved road emission factor. Furthermore, the CRA, in their analyses of the
source test data, state:
Taken together, these observations indicate that (a) silt loading and vehicle weight
may be used as independent variables and that (b) inclusion of speed would add very
little to the predictive capability of the model.7
I understand that USEPA has been asked to include vehicle speed in the revised paved road emission
factor. Doing so, however, excludes valuable source tests that were performed without measuring
vehicle speed. In particular, USEPA is excluding 22 tests performed at two integrated iron and steel
plants due to lack of vehicular speed data. These iron and steel plant source tests are crucial for
calculating fugitive dust emissions from industrial facilities, and excluding these data has a very
significant impact on predicted paved road emission rates. As discussed in the following section,
USEPA's proposed revision to the paved road emission factor will reduce particulate emission
calculations at typical industrial sites by roughly an order of magnitude. This large, and perhaps
unrealistic, reduction in calculated industrial paved road emissions is an artifact of trying to develop
an emission factor based on tests that must include vehicle speed data.
3. USEPA's Proposed Update will Result in a Roughly Order of Magnitude Emission
Reduction at Industrial Sites
In addition to developing an updated paved road emission factor, USEPA prepared a consequence
analysis of the National Emission Inventory ("NEI") resulting from their proposed revision.8
USEPA found that their revised paved road emission factor will significantly reduce PMio emissions
in the NEI (up to 200% reduction), while PM2.5 emissions are only slightly affected (some NEI
calculations increase, some decrease). USEPA, however, did not examine the affect of their draft
revised paved road equation on fugitive dust emissions from industrial sources.
I prepared two tables that compare the existing paved road emission factor with USEPA's proposed
revision - one for PMio (Table 1 A), and one for PM2.5 (Table IB). These tables include a range of
silt loading, vehicle weight, and vehicle speed conditions. For each set of silt loading, weight, and
speed, I calculated the emission factor using both the existing and proposed paved road emission
factor. As can be seen, the reduction in calculated emissions for industrial sites using the revised
factor is very large - about an order of magnitude lower for PMio and somewhat less for PM2.5.
7 Corn Refiners Association, Paved Road Modifications at AP-42, Background Documentation, MRI Project No.
310842, July 18, 2008, p. 15.
8 See Excel spreadsheet: Impact_of_revised_paved_roads_pm_emission_factors_on_NEI.xls.
-------
Revisions to AP-42, 13.2.1
August 30, 2010
Page - 5
USEPA's choice to go ahead with their proposed paved road emission factor would have serious
ramifications for NAAQS and PSD increment compliance. This is particularly true for proposed
major sources of PMio and PM2.5 which have paved haul road emission sources. Using USEPA's
proposed revision, sources that are currently being scrutinized for PMio PSD increment and PM2.5
NAAQS compliance would most likely be well below any regulatory design concentrations, even
with significantly relaxed control measures. Again, USEPA's proposed revision is largely due to
excluding a significant portion of the existing industrial source test data base, and is not due to any
tests that contradict the excluded data. In effect, USEPA's revision would be "sweeping under the
rug" what is perhaps the greatest impact caused by many industrial sources.
In terms of the modeling analyses for NAAQS and PSD increment compliance, the 24-hour
PSD increment, which is 30 micrograms per cubic meter "|ig/m3," is almost always the most
problematic regulatory design concentration. Proposed industrial sources, such as coal-fired power
plants, pig iron facilities, coal-to-liquid operations, coal-to-synthetic gas plants, and lime production
facilities, often cause air impacts that are quite close to exceeding the 24-hour PMio PSD increment.
It is common to see proposed PSD permit application modeled impacts consuming some 80 to 99%
of the allowable 24-hour PMio PSD increment. The majority of this modeled impact is caused by
low-level open source fugitive emissions, including paved haul roads.
There is no basis to assume that the existing paved road emission factor overpredicts fugitive dust
emissions from these major sources. And as we discussed earlier, it is common for major sources of
emissions to be permitted without any PSD pre-construction or post-construction ambient air
monitoring, even when such requirements are triggered by PMio significant monitoring
concentrations identified in 40 CFR 52.21(i). Thus, there is no current way to verify whether source
PMio impacts at the fenceline are realistically handled by the applied fugitive dust emission factor
and subsequent air dispersion modeling.
I have also prepared two tables that compare the existing paved road emission factor with the CRA's
proposed revision - one for PMio (Table 2A), and one for PM2.5 (Table 2B). While CRA's proposed
revision results in lower industrial source PMio and PM2.5 emission factors, they are not nearly as
severe as the changes proposed by USEPA.
The CRA source tests, however, include an apparent contradiction. CRA's source tests were
designed for low vehicular speeds and stop-and-go conditions.9 But CRA also acknowledges that
"inclusion of speed would add very little to the predictive capability of the model."10 So, the basis
for including CRA's source tests in AP-42 Section 13.2.1 seems unnecessary.
Revising AP-42 Section 13.2.1, using either CRA's or USEPA's proposed revisions, will greatly
reduce calculated fugitive dust emissions at most industrial facilities. This would make it easier for
applicants to meet regulatory design concentration compliance, and to do so with fewer emission
controls. These revisions, however, are based on data that are not representative of the majority of
9 Corn Refiners Association, Paved Road Modifications at AP-42, Background Documentation, MRI Project No.
310842, July 18, 2008, p. 4.
10 Id., p. 15.
-------
Revisions to AP-42, 13.2.1
August 30, 2010
Page - 6
major emission sources. For example, the CRA source tests are for ethanol plants with low to very
low silt loading levels. These conditions are not representative of the scores of proposed coal-fired
power projects that have recently submitted permit applications to State agencies. And USEPA's
modification of the source test data base, to add public road source tests and to eliminate the
integrated steel plant tests, probably makes things even worse. The silt loading levels (and
associated emission factors) measured at the integrated steel plant sites are representative of many
industrial facilities.11 Excluding these data will weigh the equation in a manner that reduces
predictive performance for most industrial plants.
4. USEPA's Proposed Update may not Improve Predictive Performance
As part of the proposed revision to AP-42 Section 13.2.1, it would be helpful if USEPA presented
performance analyses of both the existing and proposed paved road emission factors. Furthermore, it
would be helpful if USEPA presented performance criteria for sub-categories of emission sources,
such as public roads, industrial roads with low silt loading levels, and perhaps industrial roads with
higher silt loading levels. From this analysis, USEPA and the reviewing public could get a better
idea of whether the proposed changes will provide better predictive capability than does the existing
method. And just as important, would be information on predictive performance for each sub-
category of emission sources. In other words, we could tell whether improving performance for one
source category, ethanol plants for example, would have a detrimental effect on emission prediction
for other industrial sources with higher silt loadings.
Likewise, focusing on performance of public roads, with vehicle speed included, greatly affects
industrial source emission rates. But what effect does it have on the predictive performance of
industrial sources? As we discussed earlier, the coefficient of determination (r2) is not particularly
great for the proposed revision (all data sets included). It would be useful to examine the predictive
performance on various subsets of the existing data base, with both the existing and proposed
emission factors.
5. Other Factors Affecting USEPA's Paved Road Emission Factor
Following are a few observations that will affect the predictive emission factor equation when used
on industrial sources. I believe USEPA should address these concerns prior to revising their existing
paved road emission factor.
• The paved road emission factor should consider whether the road shoulder is paved and
whether there is a source of dust fallout present. For example, facilities with dust-generating
storage piles, and truck traffic moving between these piles, are likely to have high particulate
emission rates. This is particularly true for facilities with unpaved road shoulders.
USEPA, Emission factor Documentation for AP-42, Section 13.2.1, Paved Roads, Draft, June 2010, pp. 4-42 to 4-45.
-------
Revisions to AP-42, 13.2.1
August 30, 2010
Page - 7
• Some vehicles have exhaust pipes pointing skyward, others are parallel to the ground, and
still others pointing down to the ground. Downward-pointing exhaust can exacerbate
resuspension of dust, as I have often observed with forklifts and delivery vehicles with such
an exhaust configuration. It is unclear whether industrial vehicles with downward-pointing
exhaust are accounted for in the paved road emission factor.
• In developing the revised emission factor, USEPA subtracted a "C" term from the CRA
source tests. This results in very small or even negative emission rates for certain tests.12
Given the plume rise of exhaust from the slow-moving CRA test vehicles, it is possible that
most, if not all, of the exhaust plume passed above the downwind air samplers. In other
words, the "C" term used by USEPA may be too large for the CRA (and other) source tests.
USEPA should reevaluate to what extent, if any, exhaust, and brake and tire wear impact the
downwind profile measurements.
6. Concluding Remarks
USEPA's proposed revision to AP-42 Section 13.2.1 excludes a valuable industrial source paved
road data base simply because vehicle speed was not included in the study. USEPA's revised
emission factor will result in a roughly order of magnitude emission reduction in industrial source
paved road emissions. This very significant change may not be appropriate given that a key data set
was excluded from the regression analyses.
USEPA may be trying to fit too many source categories into a one-size-fits-all emission factor.
Under the umbrella of "paved roads" fits urban freeways, local street traffic, industrial sites with a
wide-range of truck sizes and weights, parking lots, and all shapes and sizes of vehicles using these
paved surfaces. I understand that USEPA has a very difficult task in developing a paved road
emission factor that meets the needs of all affected sources. It is likely that "clean" roads are
downward-biasing the emission factor for high-emitting facilities. And the opposite is also true -
industrial roads with high silt loading are likely upward-biasing the emission factor for cleaner roads
with lighter vehicles.
I offer the suggestion that USEPA should consider developing separate paved road emission factors
for public and industrial roads. It may not be too far-fetched to examine separate emission factors
for sub-categories of industrial source paved road emissions as well. Also, USEPA may want to
focus on silt loading and vehicle weight, as variability in vehicle speed seems to have a less
significant impact on predicted emission performance.
Until USEPA has addressed whether the severe reduction in industrial source paved road emission
calculations is warranted, I believe that the existing AP-42 paved road emission factor should
continue to be used.
-------
Revisions to AP-42, 13.2.1
August 30, 2010
Page - 8
I greatly appreciate your help in reviewing and commenting on the proposed revisions to AP-42
Section 13.2.1. Please contact me if you have any questions or require additional information.
Sincerely,
Camille Sears
-------
Table 1A
AP-42 Section 13.2.1: Paved Roads
Comparison of Existing and Draft Paved Road PM10 Emission Factors
(Silt loading resuspension only)
Setting
Public
Public
Public
Public
Public
Public
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
sL (g/m2)
0.2
0.2
0.2
0.6
0.6
0.6
0.6
0.6
0.6
0.6
0.6
0.6
0.6
0.6
0.6
2.0
2.0
2.0
2.0
2.0
2.0
2.0
2.0
2.0
5.0
5.0
5.0
5.0
5.0
5.0
5.0
5.0
5.0
10.0
10.0
10.0
10.0
10.0
10.0
10.0
10.0
10.0
W (tons)
3.75
3.75
3.75
3.75
3.75
3.75
10
10
10
20
20
20
30
30
30
10
10
10
20
20
20
30
30
30
10
10
10
20
20
20
30
30
30
10
10
10
20
20
20
30
30
30
S (mph)
25
35
45
25
35
45
5
15
25
5
15
25
5
15
25
5
15
25
5
15
25
5
15
25
5
15
25
5
15
25
5
15
25
5
15
25
5
15
25
5
15
25
Draft E
(Ib/VMT)
0.0017
0.0018
0.0019
0.0050
0.0053
0.0055
0.0066
0.0078
0.0085
0.0095
0.0113
0.0122
0.0117
0.0140
0.0152
0.0213
0.0254
0.0276
0.0308
0.0367
0.0398
0.0382
0.0455
0.0494
0.0523
0.0624
0.0677
0.0756
0.0901
0.0977
0.0937
0.1117
0.1212
0.1032
0.1230
0.1335
0.1490
0.1777
0.1928
0.1847
0.2203
0.2390
Existing E
(Ib/VMT)
0.0045
0.0045
0.0045
0.0098
0.0098
0.0098
0.0441
0.0441
0.0441
0.1255
0.1255
0.1255
0.2309
0.2309
0.2309
0.0969
0.0969
0.0969
0.2749
0.2749
0.2749
0.5055
0.5055
0.5055
0.1762
0.1762
0.1762
0.4992
0.4992
0.4992
0.9174
0.9174
0.9174
0.2767
0.2767
0.2767
0.7835
0.7835
0.7835
1.4398
1.4398
1.4398
Draft E /
Existing E
0.38
0.40
0.42
0.52
0.55
0.57
0.15
0.18
0.19
0.08
0.09
0.10
0.05
0.06
0.07
0.22
0.26
0.28
0.11
0.13
0.14
0.08
0.09
0.10
0.30
0.35
0.38
0.15
0.18
0.20
0.10
0.12
0.13
0.37
0.44
0.48
0.19
0.23
0.25
0.13
0.15
0.17
Notes:
E = resuspension emission factor; calculations exclude vehicle exhaust, brake wear, and tire wear emissions
sL = silt loading; W = mean vehicle weight; S = mean vehicle speed
For comparison purposes, no rain adjustments or control efficiencies applied
-------
Table 1B
AP-42 Section 13.2.1: Paved Roads
Comparison of Existing and Draft Paved Road PM2.5 Emission Factors
(Silt loading resuspension only)
Setting
Public
Public
Public
Public
Public
Public
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
sL (g/m2)
0.2
0.2
0.2
0.6
0.6
0.6
0.6
0.6
0.6
0.6
0.6
0.6
0.6
0.6
0.6
2.0
2.0
2.0
2.0
2.0
2.0
2.0
2.0
2.0
5.0
5.0
5.0
5.0
5.0
5.0
5.0
5.0
5.0
10.0
10.0
10.0
10.0
10.0
10.0
10.0
10.0
10.0
W (tons)
3.75
3.75
3.75
3.75
3.75
3.75
10
10
10
20
20
20
30
30
30
10
10
10
20
20
20
30
30
30
10
10
10
20
20
20
30
30
30
10
10
10
20
20
20
30
30
30
S (mph)
25
35
45
25
35
45
5
15
25
5
15
25
5
15
25
5
15
25
5
15
25
5
15
25
5
15
25
5
15
25
5
15
25
5
15
25
5
15
25
5
15
25
Draft E
(Ib/VMT)
0.0004
0.0004
0.0005
0.0012
0.0013
0.0014
0.0016
0.0019
0.0021
0.0023
0.0028
0.0030
0.0029
0.0034
0.0037
0.0053
0.0063
0.0068
0.0076
0.0091
0.0098
0.0094
0.0112
0.0122
0.0129
0.0154
0.0167
0.0186
0.0222
0.0241
0.0231
0.0275
0.0299
0.0255
0.0303
0.0329
0.0368
0.0438
0.0476
0.0456
0.0543
0.0590
Existing E
(Ib/VMT)
0.0004
0.0004
0.0004
0.0012
0.0012
0.0012
0.0063
0.0063
0.0063
0.0185
0.0185
0.0185
0.0343
0.0343
0.0343
0.0142
0.0142
0.0142
0.0410
0.0410
0.0410
0.0755
0.0755
0.0755
0.0261
0.0261
0.0261
0.0746
0.0746
0.0746
0.1373
0.1373
0.1373
0.0412
0.0412
0.0412
0.1172
0.1172
0.1172
0.2157
0.2157
0.2157
Draft E /
Existing E
1.08
1.14
1.19
1.06
1.12
1.16
0.26
0.30
0.33
0.13
0.15
0.16
0.08
0.10
0.11
0.37
0.44
0.48
0.19
0.22
0.24
0.12
0.15
0.16
0.49
0.59
0.64
0.25
0.30
0.32
0.17
0.20
0.22
0.62
0.74
0.80
0.31
0.37
0.41
0.21
0.25
0.27
Notes:
E = resuspension emission factor; calculations exclude vehicle exhaust, brake wear, and tire wear emissions
sL = silt loading; W = mean vehicle weight; S = mean vehicle speed
For comparison purposes, no rain adjustments or control efficiencies applied
-------
Table 2A
AP-42 Section 13.2.1: Paved Roads
Comparison of Existing and CRA-Proposed Paved Road PM10 Emission Factors
(Silt loading resuspension only)
Setting
Public
Public
Public
Public
Public
Public
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
sL (g/m2)
0.2
0.2
0.2
0.6
0.6
0.6
0.6
0.6
0.6
0.6
0.6
0.6
0.6
0.6
0.6
2.0
2.0
2.0
2.0
2.0
2.0
2.0
2.0
2.0
5.0
5.0
5.0
5.0
5.0
5.0
5.0
5.0
5.0
10.0
10.0
10.0
10.0
10.0
10.0
10.0
10.0
10.0
W (tons)
3.75
3.75
3.75
3.75
3.75
3.75
10
10
10
20
20
20
30
30
30
10
10
10
20
20
20
30
30
30
10
10
10
20
20
20
30
30
30
10
10
10
20
20
20
30
30
30
S (mph)
25
35
45
25
35
45
5
15
25
5
15
25
5
15
25
5
15
25
5
15
25
5
15
25
5
15
25
5
15
25
5
15
25
5
15
25
5
15
25
5
15
25
CRAE
(Ib/VMT)
0.0039
0.0039
0.0039
0.0100
0.0100
0.0100
0.0225
0.0225
0.0225
0.0396
0.0396
0.0396
0.0549
0.0549
0.0549
0.0598
0.0598
0.0598
0.1044
0.1044
0.1044
0.1447
0.1447
0.1447
0.1250
0.1250
0.1250
0.2179
0.2179
0.2179
0.3016
0.3016
0.3016
0.2179
0.2179
0.2179
0.3797
0.3797
0.3797
0.5254
0.5254
0.5254
Existing E
(Ib/VMT)
0.0045
0.0045
0.0045
0.0098
0.0098
0.0098
0.0441
0.0441
0.0441
0.1255
0.1255
0.1255
0.2309
0.2309
0.2309
0.0969
0.0969
0.0969
0.2749
0.2749
0.2749
0.5055
0.5055
0.5055
0.1762
0.1762
0.1762
0.4992
0.4992
0.4992
0.9174
0.9174
0.9174
0.2767
0.2767
0.2767
0.7835
0.7835
0.7835
1.4398
1.4398
1.4398
CRAE/
Existing E
0.86
0.86
0.86
1.03
1.03
1.03
0.51
0.51
0.51
0.32
0.32
0.32
0.24
0.24
0.24
0.62
0.62
0.62
0.38
0.38
0.38
0.29
0.29
0.29
0.71
0.71
0.71
0.44
0.44
0.44
0.33
0.33
0.33
0.79
0.79
0.79
0.48
0.48
0.48
0.36
0.36
0.36
Notes:
E = resuspension emission factor; calculations exclude vehicle exhaust, brake wear, and tire wear emissions
sL = silt loading; W = mean vehicle weight; S = mean vehicle speed
CRA = Corn Refiners Association
For comparison purposes, no rain adjustments or control efficiencies applied
-------
Table 2B
AP-42 Section 13.2.1: Paved Roads
Comparison of Existing and CRA-Proposed Paved Road PM2.5 Emission Factors
(Silt loading resuspension only)
Setting
Public
Public
Public
Public
Public
Public
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
Industrial
sL (g/m2)
0.2
0.2
0.2
0.6
0.6
0.6
0.6
0.6
0.6
0.6
0.6
0.6
0.6
0.6
0.6
2.0
2.0
2.0
2.0
2.0
2.0
2.0
2.0
2.0
5.0
5.0
5.0
5.0
5.0
5.0
5.0
5.0
5.0
10.0
10.0
10.0
10.0
10.0
10.0
10.0
10.0
10.0
W (tons)
3.75
3.75
3.75
3.75
3.75
3.75
10
10
10
20
20
20
30
30
30
10
10
10
20
20
20
30
30
30
10
10
10
20
20
20
30
30
30
10
10
10
20
20
20
30
30
30
S (mph)
25
35
45
25
35
45
5
15
25
5
15
25
5
15
25
5
15
25
5
15
25
5
15
25
5
15
25
5
15
25
5
15
25
5
15
25
5
15
25
5
15
25
CRAE
(Ib/VMT)
0.0003
0.0003
0.0003
0.0012
0.0012
0.0012
0.0030
0.0030
0.0030
0.0056
0.0056
0.0056
0.0078
0.0078
0.0078
0.0085
0.0085
0.0085
0.0151
0.0151
0.0151
0.0211
0.0211
0.0211
0.0182
0.0182
0.0182
0.0319
0.0319
0.0319
0.0443
0.0443
0.0443
0.0319
0.0319
0.0319
0.0558
0.0558
0.0558
0.0774
0.0774
0.0774
Existing E
(Ib/VMT)
0.0004
0.0004
0.0004
0.0012
0.0012
0.0012
0.0063
0.0063
0.0063
0.0185
0.0185
0.0185
0.0343
0.0343
0.0343
0.0142
0.0142
0.0142
0.0410
0.0410
0.0410
0.0755
0.0755
0.0755
0.0261
0.0261
0.0261
0.0746
0.0746
0.0746
0.1373
0.1373
0.1373
0.0412
0.0412
0.0412
0.1172
0.1172
0.1172
0.2157
0.2157
0.2157
CRAE/
Existing E
0.73
0.73
0.73
1.02
1.02
1.02
0.48
0.48
0.48
0.30
0.30
0.30
0.23
0.23
0.23
0.60
0.60
0.60
0.37
0.37
0.37
0.28
0.28
0.28
0.70
0.70
0.70
0.43
0.43
0.43
0.32
0.32
0.32
0.77
0.77
0.77
0.48
0.48
0.48
0.36
0.36
0.36
Notes:
E = resuspension emission factor; calculations exclude vehicle exhaust, brake wear, and tire wear emissions
sL = silt loading; W = mean vehicle weight; S = mean vehicle speed
CRA = Corn Refiners Association
For comparison purposes, no rain adjustments or control efficiencies applied
-------
Fw: Dave's comments on the Excel workbook - do not need to be mentioned
^^w on the call
dave.james to: Ron Myers 08/18/201003:07 PM
Dear Ron,
Please find attached some comments on the proposed new AP42 paved road equation
A) I think that, on tab PM10 Paved Roads EF's, column Z, the
column labeled "Percent Total Emissions Factor Increase" uses the formula (column x - column s) /
column x
to calculate percent changes. I think this should be, instead (column x - column s) / column s, so that
the percent change is calculated relative to the 2006 emissions factor equation instead
of the proposed new 2010 emissions factor equation
Column AD is the recalculated percent reduction for the rain corrected EF's based on this suggested
equation revision
B) For the desert southwest, I think that it is best to look at the data without rain adjustments
C) In my edited tab "PM10 Paved Roads EF's" I have added several columns, AB, AC, and AD
(1) Column AB is the calculated raw reduction of 2010 dry EF's compared to 2006 dry EF's. (column u -
column o)
(2) Column AC is the calculated percent reduction of 2010 dry EF's compared to 2006 dry EF's using
the equation (column u - column o)/column o
D) based on the 7,632 row data set in the tab PM10 Paved Roads EF's
(1) The new 2010 dry EF's are much lower overak than the 2006 dry EF's. seethe chart in the new tab
labeled "compareNewOLDPM 10EF's"
(2) The reductions of dry 2010 EF's compared to dry 2006 EF's linearly increase in magnitude with the
magnitude of the original 2006 emissions factor (see the chart in the new tab labeled "reductions" -
calculated in column AB)
(3) When I plot the percentage changes of the dry 2010 PM 10 EF's calculated) above against 2006
emissions factors, they are all around 70-80% (see the chart new tab labeled "percent reductions")
E) Athough national data might show reductions, since the new equation
1) raises the influence of silt loading (new exponent 0.98, old exponent 0.65)
2) lowers the influence of vehicle weight (new exponent 0.53, old exponent 1.5)
3) adds in an influence of vehicle speed,
4) eliminates the influence of the correction factor for exhaust brake and tire wear,
I would recommend that any assessment of the impact of the proposed new equation be
based on locally sampled data and not use the national data.
Thank you for the opportunity to comment.
-------
Sincerely,
Dave
David E.James, PhD PE
Associate Vice Provost for Academic Programs
Office of the Vice Provost for Academic Affairs
Box 451099
4505 South Maryland Parkway
Las Vegas, NV 89154-1099
Direct Line (702) 895-5804 Main Office (702) 895-1267
FAX (702) 895-3670 FDH 704 Mail Stop 1099
email: dave.james@unlv.edu
http://provost.unlv.edu/acadaffairs.html
-------
*Li
Re: Fw: Dave's comments on the EPA Excel workbook - some additional follow
up thoughts
dave.james to: Ron Myers 09/15/2010 02:44 PM
Cc: Rodney Langston, Russell Merle
Hi Ron,
Thank you for your good email below and for the additional information.
I apologize for taking so long to get back to you with my thoughts and responses. Here they are:
A) Understood about the zeros being problematic
B) In many parts of the country where there is significant rain or a rainy season, rain days may
considerably effect estimated PM10 emissions in the inventory. However, for Las Vegas and other places
like it in arid places,
I tend to use a 'pessimistic' approach that doesn't include the rain days, since rain occurs sporadically,
and what rain does fall is often very light.
C) I'm glad that my extra columns in your Excel workbook are helpful
D) Yes, from the default data it looks like many of the estimated EF's would go down with the new
proposed equation
E) Since we last corresponded,
1) I ran some calculations for Clark County's AP42 measured 2003-2006 silt loading data set using
their locally derived fleet weights. Please see the attached file "comparison20062010_AP42 road dust
EFs2003_2006.pdf"
If you examine the bottom-most table on the page, where percentage EF changes are computed, that the
net impact of the new proposed equation on Clark County's estimated paved road dust PM-10 emissions
would be to
a) increase estimated PM10 emissions as grams/VMT23% on local roads,
b) decrease them by 3% on collector roads (probably not significant), and
c) increase them by 1% on minor arterials (also probably not significant).
With locally derived data, we obtain results that are different from those that might be predicted using
default silt loading data. The actual impact on total estimated PM10 emissions in an inventory or SIP
would depend on how much VMT was assigned to each roadway category.
2) I also ran a hypothetical sensitivity analysis comparing arbitrary combinations of vehicle weight and silt
loading, to see what the impacts of the new PM10 equation might be. Please seethe attached file
"new2010EFsensitivityanalysis.pdf"
Table 1 shows the 2006 equation predicted PM10 emissions
Table 2 shows the proposed 2010 equation predicted PM10 emissions
Table 3 shows the changes in predicted emissions (2010 EF - 2006 EF)
and
Table 4 shows the Percentage changes, (2010 EF - 2006 EF)/2006 EF
Table 4 shows that the net effect of using the new proposed 2010 equation is that predicted
-------
PM10 emissions
a) increase for lower silt loadings at all fleet average vehicle weights, and
b) decrease for higher silt loadings, espeically at lower fleet average vehicle weights
I hope that these preliminary calculations are helpful. I have also sent them as PDF and as the original
Excel files to
my research sponsors, Clark County Dept of Air Quality and Environmental Management.
Sincerely,
Dave
David E.James, PhD PE
Associate Vice Provost for Academic Programs
Office of the Vice Provost for Academic Affairs
Box451099
4505 South Maryland Parkway
Las Vegas, NV 89154-1099
Direct Line (702) 895-5804 Main Office (702) 895-1267
FAX (702) 895-3670 FDH 704 Mail Stop 1099
email: dave.james@unlv.edu
http://provost.unlv.edu/acadaffairs.html
From: Myers.Ron@epamail.epa.gov
To: dave.james@unlv.edu
Date: 08/18/2010 06:34 PM
Subject: Re: Fw: Dave's comments on the Excel workbook - do not need to be mentioned on the call
Dave:
Thanks for looking at the proposed revisions of the paved road equation.
First, I was trying to replicate the emissions estimates that are being made
for the 2008 NEI, any rain adjustments or other mitigation that I included in
the spreadsheet are the same as I estimated were used in the NEI emissions
estimates. As with you I would not have included as much mitigation for rain
and "Street Sweeping" and other silt management as there is is used in the NEI
estimates.
A. You are correct. I should have divided by the estimated 2008 emissions as
calculated with the existing AP-42. I think this was a hold over from when I
was just looking at the road dust emissions estimates. When looking only at
road dust emissions, all the zero emissions estimates is problematic since
dividing by zero only generates errors in Excel. I added in the vehicle
emissions when I saw how many 2008 NEI estimates were zero.
-------
B. I would tend to agree with you as there are not many rain days. As I
stated above, I don't know what mitigation is included in the "adjusted"
emissions data in the NEI. Frankly to documentation of the NEI emissions
estimates doesn't help me much to recreate their emissions estimates (see
paved_roads_2294000000_documentation.doc which is attached).
C. Thanks for the calculations. I did these calculation only because a few
internal EPA people suggested that I provide State/local agencies with some
information to provide an indication of how this change might affect their
inventories.
D. My original assessment also showed that the revised equation generates much
lower PM10 estimates than the previous equation. From a combined emisions
inventory perspective and use in the modeling for SIP development this should
get support from inventory developers, modelers and Air Quality Assessors as
it has always been difficult to explain how fugitive dust emissions are the
majority of the emissions in the inventory but comprise less than 10% of the
emissions on PM monitors. This will not get the inventory there but it goes
in the right direction. I agree that for best emissions estimates, locally
derived silt loadings are needed. However, no one wants to develop these and
would rather complain that EPA's default values aren't good enough and they
want better defaults. There is so much variation in silt levels on roads no
single number is good enough for every road.
Ron Myers
U.S. Environmental Protection Agency
Office of Air Quality Planning and Standards
Sector Policy and Programs Division
Monitoring Policy Group, D243-05
RTP NC 27711
Tel. 919.541.5407
Fax 919.541.1039
E-mail myers.ron@epa.gov
dave.James 08/18/2010 03:10:06 PM Dear Ron, This is a resend, using a
compressed version of the Excel file to reduce
From:
dave.j ames@unlv.edu
To:
Ron Myers/RTP/USEPA/US@EPA
Date:
08/18/2010 03:10 PM
Subj ect:
Fw: Dave's comments on the Excel workbook - do not need to be mentioned on the
call
-------
Dear Ron,
This is a resend, using a compressed version of the Excel file to reduce the
file size,
in case my earlier send was rejected.
Please find attached some comments on the proposed new AP42 paved road
equation
A) I think that, on tab PM10 Paved Roads EF's, column Z, the
column labeled "Percent Total Emissions Factor Increase" uses the formula
(column x - column s) / column x
to calculate percent changes. I think this should be, instead (column x -
column s) / column s, so that
the percent change is calculated relative to the 2006 emissions factor
equation instead
of the proposed new 2010 emissions factor equation
Column AD is the recalculated percent reduction for the rain corrected EF's
based on this suggested equation revision
B) For the desert southwest, I think that it is best to look at the data
without rain adjustments
C) In my edited tab "PM10 Paved Roads EF's" I have added several columns, AB,
AC, and AD
(1) Column AB is the calculated raw reduction of 2010 dry EF's compared to
2006 dry EF's. (column u - column o)
(2) Column AC is the calculated percent reduction of 2010 dry EF's compared to
2006 dry EF's using
the equation (column u - column o)/column o
D) based on the 7,632 row data set in the tab PM10 Paved Roads EF's
(1) The new 2010 dry EF's are much lower overal 1 than the 2006 dry EF's. see
the chart in the new tab labeled "compareNewOLDPMlOEF's"
(2) The reductions of dry 2010 EF's compared to dry 2006 EF's linearly
increase in magnitude with the magnitude of the original 2006 emissions factor
(see the chart in the new tab labeled "reductions" - calculated in column AB)
(3) When I plot the percentage changes of the dry 2010 PM 10 EF's calculated
) above against 2006 emissions factors, they are all around 70-80% (see the
chart new tab labeled "percent reductions")
E) Athough national data might show reductions, since the new equation
1) raises the influence of silt loading (new exponent 0.98, old exponent 0.65)
2) lowers the influence of vehicle weight (new exponent 0.53, old exponent
1.5)
3) adds in an influence of vehicle speed,
4) eliminates the influence of the correction factor for exhaust brake and
tire wear,
I would recommend that any assessment of the impact of the proposed new
equation be
based on locally sampled data and not use the national data.
-------
Thank you for the opportunity to comment.
Sincerely,
Dave
David E. James, PhD PE
Associate Vice Provost for Academic Programs
Office of the Vice Provost for Academic Affairs
Box 451099
4505 South Maryland Parkway
Las Vegas, NV 89154-1099
Direct Line (702) 895-5804 Main Office (702) 895-1267
FAX (702) 895-3670 FDH 704 Mail Stop 1099
email: dave.james@unlv.edu
http : //provost . unlv. edu/acadaf fairs . html
ATTACHMENT NOT DELIVERED
*******************
This Email message contained an attachment named
djedit Impact of revised paved roads pm emission factors on NEI.xls.zip
which may be a computer program. This attached computer program could
contain a computer virus which could cause harm to EPA' s computers,
network, and data. The attachment has been deleted.
This was done to limit the distribution of computer viruses introduced
into the EPA network. EPA is deleting all computer program attachments
sent from the Internet into the agency via Email.
If the message sender is known and the attachment was legitimate, you
should contact the sender and request that they rename the file name
extension and resend the Email with the renamed attachment. After
receiving the revised Email, containing the renamed attachment, you can
rename the file extension to its correct name.
For further information, please contact the EPA Call Center at
(866) 411-4EPA (4372). The TDD number is (866) 489-4900.
*********************** ATTACHMENT NOT DELIVERED
*********************** [attachment "paved_roads_2294000000_documentation.doc"
deleted by Dave James/UNLV]
-------
Page 1 of 1
From: Steve Zemba
To: Ron Myers/RTP/USEPA/US@EPA
cc: gfore@hotmix.org, Mike , Laura Green
, HMarks@hotmix.org
Date: Tuesday, August 31, 2010 02:32PM
Subject: Comment on AP42 Paved Roads Draft Section 13.2.1
History: ^ This message has been replied to and forwarded.
Dear Ron,
I write to provide the attached comment on the draft AP42 section on
Paved Road dust emissions. As described in the comment, NAPA (who
sponsored the review) is potentially interested in collecting data to
provide more representative parameters for applications to the asphalt
pavement industry. We would appreciate your advice on how best to
gather these data so that they could be submitted for consideration in
the AP42 section.
Thanks for your help and consideration,
Steve
Stephen G. Zemba, Ph.D., P.E.
Senior Engineer
^Cambridge Environmental Inc*
58 Charles Street
Cambridge, MA 02141
Office: 617-225-0810 x34 M-W 518-306-4603 Th-F
Cell: 339-223-9328
Fax: 617-225-0813
http://www.CambridgeEnvironmental.com
AP42PavedRoadsSectionComment083110.pdf
Type: application/pdf
Name:
AP42PavedRoadsSectionComment(
https://rtairmaill.rtp.epa.gov/mail/rmyers.nsf/9ff539ale24f5aaf852577890046a8f6/8F3D... 10/22/2010
-------
Cambridge Environmental Inc
58 Charles Street Cambridge, Massachusetts 02141
617-225-0810 www.CambridseEnvironmental.com
August 31,2010
Ronald Myers
U.S. Environmental Protection Agency
109 T.W. Alexander Drive
Mail Code: D243-05
Research Triangle Park, NC 27709
Dear Ron,
It was a pleasure speaking with you again recently - thank you for the background information on the
draft update to the AP42 section on Paved Road emissions (Section 13.2. 1).
I have reviewed the draft update on behalf of the National Asphalt Pavement Association (NAP A), and
write to comment on a specific aspect of interest. I believe that the recommended default values for silt-
loading in draft Table 13.2.1-3, and particularly that for asphalt batching, may be too high for typical
current applications. The recommended value is 120 g/m2, but, as you know, in EPA's 2000 Emission
Assessment Report for Hot Mix Asphalt Plants, a silt-loading value 3 g/m2 is suggested for paved roads at
typical hot-mix asphalt production facilities. Also, site-specific measurements at a hot mix asphalt
facility in Alexandria, Virginia in 2005 (using the sampling and analytical methods described in AP42
Appendix C) found a silt loading level of 0.5 g/m2. This facility, which we analyzed in detail for the City
of Alexandria, employs aggressive dust suppression techniques.
More generally, as you know, best management practices (BMPs) such as water spraying and road
sweeping can effectively control dust emissions; by the same token, the absence of these practices can
indeed result in dusty roads. Perhaps the value of 120 g/m2, which appears to be based on older data,
derives from testing at one or more facilities that failed to employ BMPs. If so, then perhaps 120 g/m2
could be considered to be a default value in the absence of BMPs, whereas the value of 3 g/m2, as used in
EPA's Emission Assessment Report, could be a default value in the presence of typical BMPs.
Of course, more data are always better. In that regard, we have spoken with representatives from NAPA ,
and they have expressed potential willingness to coordinate a study to provide updated values for silt
loading and other emission factor parameters that reflect current practices in the hot-mix asphalt industry.
At your convenience, might we schedule a call to discuss whether this would be of interest to you and
your colleagues at the Agency?
Thank you for your consideration, and best regards.
Sincerely,
Stephen G. Zemba, Ph.D., P.E.
Senior Engineer
-------
IDNR comment on proposed AP42 Section 13.2.1 Paved Roads
Hanson, Lori [DNR] to: Ron Myers 08/20/2010 10:45 AM
"McGraw, Jim [DNR]"
History: This message has been forwarded.
Mr. Myers,
I have attached the Iowa Department of Natural Resources comments on the proposed revision toAP42
section 13.2.1 on paved roads. Thank you for the opportunity to provide comments, Lori Hanson
-------
Fields of Opportunities
STATE OF IOWA
CHESTER J. CULVER, GOVERNOR DEPARTMENT OF NATURAL RESOURCES
PATTY JUDGE, LT. GOVERNOR . RICHARD A. LEOPOLD, DIRECTOR
August 20, 2010
U.S. Environmental Protection Agency (EPA)
Measurement Policy Group
Attn: Proposed Revisions to AP42 sectionl3.2,1 Paved Roads
The Iowa Department of Natural Resources (IDNR) is providing comment on the proposed
revision of AP-42 Section 13.2.1 for paved roads. The DNR supports the revision of this section
to incorporate new data from corn wet mills and to account for mean vehicle speeds below 10
miles per hour.
The current AP-42 emission factor (November 2006) includes vehicle emissions (engine exhaust,
tire wear and brake wear) in the empirical equation. Additionally there is a vehicle emission
constant "C" that is subtracted from the equation. This "C" constant accounts for 1980's vehicle
fleet exhaust, brake wear and tire wear and is subtracted from the equation to eliminate the
possibility of double counting emissions and to account for the decrease in paniculate emissions
from improvements related to newer model trucks and cleaner fuels since the empirical equation
was derived. A table of default values for "C" that varied with particle aerodynamic size range is
included in the section.
The proposed empirical equation was developed by linear regression analysis after subtracting
the engine, tire and brake wear estimated using EPA's MOBILE6.2 and MOVES2010 models
from the measured impacts to estimate emissions solely from vehicle travel on the paved roads.
To determine the total paved road emission factor, the emission factor from vehicle emissions
generated by running either EPA's MOBILE6.2 or MOVES2010 models must be added to the
emission factor from the empirical equation. This methodology requires that a mobile source
emissions model be run in order to determine a paved road emission factor.
Obtaining the emissions factor for vehicle emissions in this manner will be problematic as the
DNR does not have the resources to generate specific emissions factors for vehicle emissions by
running MOVES2010 for every construction permitting project that includes a paved haul road.
The DNR suggests that either the empirical equation be developed to include vehicle emissions
from engine exhaust, tire and brake wear, or that a table of default values be included in the
section to account for vehicle emissions as an alternative to running a mobile source emission
model.
7900 Hickman Road, Suite 1 /Windsor Heights, Iowa 50324
515-242-5100 FAX 515-242-5094 htip://www.iowaclBanalr.com/
-------
Thank you for the opportunity to comment on the proposed revision of AP-42 Section 13.2,1 for
paved roads.
Sincerely,
Catharine Fitzsimmons, Chief
Air Quality Bureau
-------
*Li
RE: PECHAN/ERTAC Road Dust Emissions
Pat Davis o Roy Huntley, Ron Myers 07/26/2010 01:22 PM
History: This message has been replied to.
Hi Ron,
Have you had a chance to look into this issue?
To refresh your memory we noticed that a number of the PM2.5 emission factors
were zeroed out for a number of road types. Can you please tell us why the
road types listed below were zeroed out?
Urban Collector
Urban Minor Arterial
Urban Other Principal Arterial
Urban Other Freeways and Expressways
Urban Interstate
Thanks,
Pat Davis
Original Message
From: Huntley.Roy@epamail.epa.gov [mailto:Huntley.Roy@epamail.epa.gov]
Sent: Tuesday, July 13, 2010 1:38 PM
To: Myers.Ron@epamail.epa.gov
Cc: Pat Davis
Subject: Fw: PECHAN/ERTAC Road Dust Emissions
Ron, could you answer Pat question?
Roy Huntley
Environmental Engineer
Emission Inventory and Analysis Group
Mail Drop (C339-02)
Environmental Protection Agency
RTP, NC 27711
Voice - 919 541-1060
Fax - 919 541-0684
Office C341H
Forwarded by Roy Huntley/RTP/USEPA/US on 07/13/2010 01:24 PM
l>
| From: Pat Davis
I
l>
| To: Roy Huntley/RTP/USEPA/US@EPA
I
>
l>
| Cc:
Judy Rand , Julie McDill ,
Pat Davis , "Fees David F. (DNREC)",|
"WRBARNARD@mactec.com" , Walter Simms ,
"kenneth.santlal@state.ma.us"
-------
I >
I Date: 07/13/2010 12:02 PM
I
| Subject: PECHAN/ERTAC Road Dust Emissions
I
>Hi Roy,
We have been examining the ERTAC/PECHAN emission factors for Road Dust
and Maryland noticed that the PM2.5 emission factors were zeroed out for
the following road types:
Urban Collector
Urban Minor Arterial
Urban Other Principal Arterial
Urban Other Freeways and Expressways
Urban Interstate
Emission factors for PM10 were found and there was no mention in the
documentation of why the PM2.5 emission factors were zeroed out, so we
are bit confused.
We were hoping that you might have answer for us, or be able to point us
in the direction of someone who might know why the PM2.5 emission
factors are zeroed out.
Thanks, and I hope you are well!
Pat Davis
-------
FW: [chief] Proposed revisions to AP42 section 13.2.1 Paved Roads
Julie McDill to: Ron Myers 06/23/201003:29 PM
This message has been replied to.
Hello Ron,
I called and left a message about possibly getting on a call with the MARAMA
states in the next couple of weeks to discuss proposed changes to the Paved
Road PM emissions estimation method.
Please respond to let me know if and when that might be possible. I can set
up a conference call and distribute a slide set. It would be best sometime
between July 7 and 16th. What follows (and the attachment) are some emails
that give you a flavor of the changes that states are finding as a result of
the new calculations. As you probably know, the PM emission from paved roads
has always posed problems in modeling. In general, modelers take our
inventories and reduced the paved road emissions by about 90% before running
the model.
Thanks for your help.
Julie McDill
MARAMA
From: Judy Rand [Judy.Rand@dep.state.nj.us]
Sent: Wednesday, June 16, 2010 4:43 PM
To: Julie McDill; Pat Davis; rthunell@mde.state.md.us; David.Fees@state.de.us
Cc: Nicholle Worland; WRBARNARD@mactec.com; kenneth.santlal@state.ma.us
Subject: RE: [chief] Proposed revisions to AP42 section 13.2.1 Paved Roads
Thanks Dave. We have come up with similar results, but even more drastic for
PM2.5. An increase in PM2.5 of 350% and a decrease in PM10 of 46% I think
one big cause is the difference in k factor, among other changes. The k
factor for PM2.5 went down from the 2003 AP-42 to the 2006 AP-42, and back up
again in this new draft. We guessed at the new vehicle speed requirement, but
a slight variation in speeds will not make that much of a difference.
See NJ's attached calcs and compare spreadsheet. I won't be in til Monday.
If you want to have a call either Nicholle can cover it tomorrow, or we are in
on Monday.
Judy
>» "Fees David F. (DNREC)" 6/16/2010 2:02 PM >»
Roger,
Here is Delaware's paved road dust spreadsheet for 2007, using the new
equation. We got very detailed with this category; estimating emissions by
month.
Regarding the new equation, PM10 was reduced by 58% from the emissions
submitted to MACTEC; while PM2.5 increased by 48%. I believe the PM2.5
increase is caused by two factors-first, the PM2.5/PM10 ratio was increased to
25% (previously 15%). The second reason is that under the old equation, one
had to apply a correction factor, C, to remove the exhaust, brake, and tire
wear from the front part of the equation. By subtracting C at the end of the
equation, the resulting PM2.5 value went negative for several roadway types.
Of course we zeroed these out, but with the new method there is never a
-------
situation where the emission factor value can go negative. Having negative
emission factors result from the use of the old equation was obviously a flaw
in the method, so I expect the new equation is more accurate.
I look forward to NJ's results when they apply the new equation, to see if
they get changes similar to mine.
If you have any questions about the calculations within the spreadsheet, just
give a call.
Regards,
Dave
David F. Fees, P.E.
Managing Engineer
Emission Inventory Development Program
Air Quality Management Section, DNREC
tel. (302) 739-9402, fax (302) 739-3106
e-mail: david.fees@state.de.us
Blue Skies Delaware; Clean Air for Life
From: Roger Thunell [mailto:rthunell@mde.state.md.us]
Sent: Monday, June 14, 2010 3:00 PM
To: Judy Rand; Julie McDill; Pat Davis
Cc: WRBARNARD@mactec.com; Fees David F. (DNREC); kenneth.santlal@state.ma.us
Subject: RE: [chief] Proposed revisions to AP42 section 13.2.1 Paved Roads
Judy/Dave/Kenneth:
Could any of you send me a spreadsheet calculating emissions in this manner?
I am not sure if we are using the latest methods or not.
Thanks
Roger
>» Pat Davis 6/14/2010 12:54 PM >»
Thanks a lot for sending this along, Judy. Please let us know what you find
when you look at the changes in emissions.
Pat
Original Message
From: Judy Rand [mailto:Judy.Rand@dep.state.nj.us]
Sent: Monday, June 14, 2010 9:16 AM
To: Julie McDill; Pat Davis
Cc: WRBARNARD@mactec.com; rthunell@mde.state.md.us; David.Fees@state.de.us;
kenneth.santlal@state.ma.us
Subject: Fwd: [chief] Proposed revisions to AP42 section 13.2.1 Paved Roads
Pat and Julie,
We are going to look at this to see how it affects emissions. In the past,
each change to this category has changed emission calculations.
Thanks,
Judy
Judy Rand, PE
Environmental Engineer
NJDEP Air Quality Planning
(609) 984-1950
j rand@dep.state.nj.us
-------
FW: Proposed revisions to AP42 section 13.2.1 Paved Roads
Julie McDill to: Ron Myers 06/30/2010 04:25 PM
Hi Ron,
Here is the announcement for our call next week. Can you send me a slide set
by noon next Tuesday and I will distribute it to the group and post it on our
ftp.
Thanks,
Julie
From: Julie McDill
Sent: Tuesday, June 29, 2010 3:21 PM
To: Paul.Bodner@ct.gov; mark.prettyman@state.de.us; David.Fees@state.de.us;
j essica.daniels@dc.gov; melanie.loyzim@maine.gov; rthunell@mde.state.md.us;
kenneth.santlal@state.ma.us; david.healy@des.nh.gov;
judy.rand@dep.state.nj.us; Nicholle.Worland@dep.state.nj.us;
jdbarnes@gw.dec.state.ny.us; rwstanna@gw.dec.state.ny.us; sbogart@state.pa.us;
karen.slattery@dem.ri.gov; jeff.merrell@state.vt.us;
Thomas.Foster@deq.Virginia.gov; laura.boothe@ncdenr.gov;
Robert.J.Betterton@wv.gov; mcconnell.robert@epamail.epa.gov;
Forde.Raymond@epamail.epa.gov; kremer.j anet@epamail.epa.gov;
huntley.roy@epa.gov; Susan Wierman
Cc: cooke.donald@epamail.epa.gov; burkhart.richard@epamail.epa.gov;
Garcia.Ariel@epamail.epa.gov; Kelly.Bob@epamail.epa.gov;
Salomone.Jenna@epamail.epa.gov; Wieber.Kirk@epamail.epa.gov;
Moltzen.Michael@epamail.epa.gov; Laurita.Matthew@epamail.epa.gov;
Feingersh.Henry@epamail.epa.gov; Kremer.Janet@epamail.epa.gov;
Ellsworth.Todd@epamail.epa.gov; Leon-Guerrero.Tim@epamail.epa.gov;
Cripps.Christopher@epamail.epa.gov; Rehn.Brian@epamail.epa.gov;
Kotsch.Martin@epamail.epa.gov; Dolce.Gary@epamail.epa.gov;
Kapichak.Rudolph@epamail.epa.gov; Houyoux.Marc@epamail.epa.gov;
Timin.Brian@epamail.epa.gov; Stackhouse.Butch@epamail.epa.gov;
Broadwell.Valerie@epamail.epa.gov; Ling.Michael@epamail.epa.gov;
Fox.Tyler@epamail.epa.gov; Cook.Leila@epamail.epa.gov;
Spink.Marcia@epamail.epa.gov; Wayland.Richard@epamail.epa.gov;
Hemby.James@epamail.epa.gov; Wilkie.Walter@epamail.epa.gov;
Fernandez.Cristina@epamail.epa.gov; Ruvo.Richard@epamail.epa.gov;
Werner.Raymond@epamail.epa.gov; amold.anne@epamail.epa.gov;
Baker.William@epamail.epa.gov; Arnold.David@epamail.epa.gov;
Conroy.Dave@epamail.epa.gov
Subject: FW: Proposed revisions to AP42 section 13.2.1 Paved Roads
Hello all,
This email is to announce a teleconference on July 7 at 2:30 PM Eastern
concerning the proposed change to the equation used to estimate PM 10 and 2.5
emissions from paved roads. Ron Myers of OAQPS will provide a presentation on
the development of the new equation and will answer your questions. Modellers
and planners from MANE-VU state agencies along with some USEPA regional staff
have been invited. Call in information is as follows:
Number: 866-202-1783
Code: *5743656* - Make sure you press * before and after the number.
Date: July 7
Time: 2:30 - 4:00 P.M. Eastern
BACKGROUND FOR THE CALL
This equation is used to calculate emissions for the area source modeling
inventory. Delaware and New Jersey have already done some preliminary
calculations and find the new equation results in very different values than
the old equation. I attach their spreadsheets for your review. Toward the
bottom of this email are texts of emails discussing the differences. In
addition is the text distributed by NACAA which provides links to materials
-------
for your formal comment to USEPA.
As you are no doubt aware, modellers have applied a transport fraction
reduction to fugitive road dust emissions in the past to bring the calculated
impact on ambient PM in line with measured concentrations. The new equation
may require a revision to the transport fraction calculation. I have invited
our NY modellers to join the call to hear the discussion so that they can
consider any impact on the transport fraction calculation.
The new equation is proposed, so we can decide to use the old calculation
method for our modeling inventory. That is what is in our current draft area
source inventory files. However, States will then face a disconnect with the
model for future emission calculations. At any rate, it seems to me that all
states should use the same methodology so that the inventory is consistant
accross our region.
Julie McDill
MARAMA
Relevant Email texts
From: Judy Rand [Judy.Rand@dep.state.nj.us]
Sent: Wednesday, June 16, 2010 4:43 PM
Thanks Dave. We have come up with similar results, but even more
drastic for PM2.5. An increase in PM2.5 of 350% and a decrease in PM10
of 46% I think one big cause is the difference in k factor, among other
changes. The k factor for PM2.5 went down from the 2003 AP-42 to the
2006 AP-42, and back up again in this new draft. We guessed at the new
vehicle speed requirement, but a slight variation in speeds will not
make that much of a difference.
See NJ's attached calcs and compare spreadsheet. I won't be in til
Monday. If you want to have a call either Nicholle can cover it
tomorrow, or we are in on Monday.
Judy
From: "Fees David F. (DNREC)" 6/16/2010 2:02 PM
Roger,
Regarding the new equation, PM10 was reduced by 58% from the emissions
submitted to MACTEC; while PM2.5 increased by 48%. I believe the PM2.5
increase is caused by two factors-first, the PM2.5/PM10 ratio was
increased to 25% (previously 15%). The second reason is that under the
old equation, one had to apply a correction factor, C, to remove the
exhaust, brake, and tire wear from the front part of the equation. By
subtracting C at the end of the equation, the resulting PM2.5 value went
negative for several roadway types. Of course we zeroed these out, but
with the new method there is never a situation where the emission factor
value can go negative. Having negative emission factors result from the
use of the old equation was obviously a flaw in the method, so I expect
the new equation is more accurate.
I look forward to NJ's results when they apply the new equation, to see
if they get changes similar to mine.
If you have any questions about the calculations within the spreadsheet,
just give a call.
Regards,
Dave
TO: NACAA EMISSIONS & MODELING COMMITTEE
Please information below regarding a proposed revision of the AP-42 paved
roads section. The proposed draft can be found here -
http://www.epa.gov/ttn/chief/ap42/chl3/index.html; scroll down to section
13.2.1, paved roads. EPA will take comments on the draft until July 30, 2010.
For more information, please contact Ron Myers at myers.ron@epa.gov.
-------
Emissions Comparison
AP-42 k factors (g/mile)
2002 2007 (Existing)
Annual
Summer
Winter
Spring
Fall
pm-10 tpy
pm-2.5 tpy
pm-10 tpd
pm-2.5 tpd
pm-10 tpd
pm-2.5 tpd
pm-10 tpd
pm-2.5 tpd
pm-10 tpd
pm-2.5 tpd
37,606.28
3,788.42
115.11
11.56
95.87
9.69
99.94
10.07
101.03
10.18
38,210.45
1,142.03
105.70
3.13
101.69
3.13
105.08
3.11
106.23
3.15
2007(new) %
20,532.18
5,110.37
56.75
14.12
54.74
13.63
56.41
14.04
57.08
14.21
Change
-46%
347%
-46%
351%
-46%
336%
-46%
351%
-46%
352%
PM-10
PM-2.5
2003
7.3000
1.8000
2006
7.3000
1.1000
2010
6.79
1.69
-------
2007 CAP Emissions Calculations
Rural Oth. Princ. Art.
January
February
March
April
May
June
July
August
September
October
November
December
Rural Minor Arterial
January
February
March
April
May
June
July
August
September
October
November
December
Rural Major Collector
January
February
March
April
May
June
July
August
September
October
November
December
PMIO-FIL(TPY)
Kent New Castle Sussex
3.0368
2.8324
3.3463
3.4705
4.3379
4.4049
5.1486
4.8552
4.2558
3.6182
3.2676
2.9585
45.5327
19.7917
19.4746
7.3003
6.7427
7.5361
6.8577
7.5020
7.2545
7.2050
7.0923
6.6536
6.7191
110.1298
17.4130
15.2798
6.6067
6.3955
8.4396
8.9101
8.7305
7.9809
9.0665
8.0543
6.8855
5.8669
109.6292
3.8504
3.6256
4.3680
4.4744
5.4584
5.1675
5.3205
5.4994
5.1069
4.6311
4.3388
4.0132
55.8543
0.9224
0.9004
1 .0547
1.0176
1.1867
1 .0829
1.1643
1.1817
1.1851
1.1383
1.0271
0.9796
12.8408
10.8453
9.2386
4.1449
4.2942
5.6208
5.3072
6.1122
5.1746
5.4730
4.6414
3.8170
3.4370
68.1062
5.6417
5.2929
6.3490
6.6084
8.1201
8.5339
10.4483
10.3512
7.7932
6.9531
6.2977
5.7959
88.1854
2.9397
2.7698
1 .7424
1 .8092
2.1881
2.5017
1 .5249
2.8365
2.0514
1 .7639
1 .5527
1 .4285
25.1086
81.4407
74.5901
31.0497
30.7597
36.4193
11.1730
13.2889
12.3685
33.1489
30.6447
28.0237
27.2418
410.1489
PM2.5-FIL (TPY)
Kent New Castle Sussex
0.7558
0.7050
0.8329
0.8638
1.0797
1 .0964
1.2815
1 .2084
1.0592
0.9006
0.8133
0.7364
11.3329
4.9261
4.8472
1.8170
1 .6782
1.8757
1.7069
1 .8672
1.8056
1.7933
1 .7652
1.6561
1 .6723
27.4108
4.3340
3.8031
1 .6444
1.5918
2.1006
2.2177
2.1730
1 .9864
2.2566
2.0047
1.7138
1 .4602
27.2862
0.9584
0.9024
1.0872
1.1137
1.3586
1 .2862
1 .3243
1.3688
1.2711
1.1527
1.0799
0.9989
13.9019
0.2296
0.2241
0.2625
0.2533
0.2954
0.2695
0.2898
0.2941
0.2950
0.2833
0.2556
0.2438
3.1960
2.6993
2.2994
1.0317
1.0688
1.3990
1.3209
1.5213
1.2879
1 .3622
1.1552
0.9500
0.8555
16.9513
1 .4042
1.3174
1.5802
1 .6448
2.0210
2.1240
2.6005
2.5764
1.9397
1.7306
1.5675
1 .4426
21.9489
0.7317
0.6894
0.4337
0.4503
0.5446
0.6226
0.3795
0.7060
0.5106
0.4390
0.3865
0.3556
6.2494
20.2702
18.5651
7.7281
7.6560
9.0646
2.7809
3.3075
3.0785
8.2506
7.6273
6.9750
6.7804
102.0842
Rural Minor Collector
-------
2007 CAP Emissions Calculations
January
February
March
April
May
June
July
August
September
October
November
December
Rural Local
January
February
March
April
May
June
July
August
September
October
November
December
Urban Interstate
January
February
March
April
May
June
July
August
September
October
November
December
PMIO-FIL(TPY)
Kent New Castle Sussex
7.5825
6.6536
2.8769
2.7849
3.6750
3.8799
3.8017
3.4753
3.9480
3.5072
2.9983
2.5547
47.7380
72.2816
63.4268
20.6871
20.0256
26.4262
9.5065
27.3372
24.9900
9.6733
25.2198
21.5599
18.3704
339.5044
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
3.1843
2.7126
1.2170
1 .2609
1 .6504
1 .5583
1 .7946
1.5194
1 .6069
1 .3628
1.1207
1 .0092
19.9970
14.2450
13.0827
5.3708
4.9021
6.0689
4.9294
5.0483
5.2372
6.0687
5.8458
5.2538
5.0929
81.1457
11.7187
12.5883
13.3265
14.0185
15.5068
14.5005
15.7325
16.9323
14.9430
13.8368
13.6508
12.6716
169.4261
9.1770
8.4051
3.4988
3.4661
4.1039
4.0969
4.8727
4.5352
3.7353
3.4532
3.1578
3.0697
55.5718
182.5225
167.1691
17.8860
17.7190
20.9791
20.9434
24.9096
23.1843
19.0953
17.6527
47.3761
46.0544
605.4913
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
PM2.5-FIL (TPY)
Kent New Castle Sussex
1.8872
1.6561
0.7160
0.6931
0.9147
0.9657
0.9462
0.8650
0.9826
0.8729
0.7463
0.6359
11.8818
17.9906
15.7866
5.1489
4.9843
6.5774
2.3661
6.8041
6.2199
2.4076
6.2771
5.3662
4.5723
84.5011
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.7926
0.6751
0.3029
0.3138
0.4108
0.3878
0.4467
0.3782
0.4000
0.3392
0.2789
0.2512
4.9772
3.5455
3.2562
1.3368
1.2201
1.5105
1 .2269
1 .2565
1.3035
1.5105
1.4550
1.3077
1 .2676
20.1968
2.9167
3.1332
3.3169
3.4891
3.8596
3.6091
3.9158
4.2144
3.7192
3.4439
3.3976
3.1539
42.1694
2.2841
2.0920
0.8708
0.8627
1.0214
1.0197
1.2128
1.1288
0.9297
0.8595
0.7860
0.7640
13.8316
45.4290
41.6076
4.4517
4.4102
5.2216
5.2127
6.1999
5.7705
4.7527
4.3937
11.7917
1 1 .4627
150.7040
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
Urban Oth. Freeway
January
2.2724
2.2051
0.0000
0.5656
0.5488
0.0000
-------
2007 CAP Emissions Calculations
February
March
April
May
June
July
August
September
October
November
December
Urban Oth. Princ. Art.
January
February
March
April
May
June
July
August
September
October
November
December
Urban Minor Arterial
January
February
March
April
May
June
July
August
September
October
November
December
Urban Collector
January
February
PMIO-FIL(TPY)
Kent New Castle Sussex
2.1240
2.6382
2.6584
3.3363
3.5141
4.2085
4.1031
3.2980
2.6345
2.3741
2.1895
35.3512
1.3373
1 .2472
1 .4735
1 .5282
1.9102
1.9397
2.2672
2.1380
1 .8740
1.5933
1 .4389
1 .3028
20.0502
4.3310
4.2617
4.6884
4.3304
4.8399
4.4042
4.8180
4.6591
4.6273
4.5549
4.2731
4.3152
54.1031
31.7118
30.9520
2.3687
2.5076
2.6378
2.9179
2.7285
2.9604
3.1861
2.8118
2.6036
2.5686
2.3844
31.8807
13.5266
13.0598
14.9285
14.5413
16.7660
15.0529
15.1950
15.2299
15.1373
14.6302
13.7450
13.7174
175.5298
4.8764
4.8104
5.7734
5.8589
7.1419
6.2220
6.2667
6.1525
6.2988
5.6747
4.7114
4.3381
68.1252
16.1893
0.0378
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
3.0648
2.8754
3.4491
3.5900
4.4112
4.6360
5.6760
5.6233
4.2336
3.7772
3.4212
3.1486
47.9065
2.4339
2.2835
1 .3887
0.7328
0.9004
0.9463
1.1586
1.1478
0.8642
0.7710
0.6983
1 .2677
14.5932
32.0745
29.3765
PM2.5-FIL (TPY)
Kent New Castle Sussex
0.5287
0.6566
0.6617
0.8304
0.8747
1.0475
1.0213
0.8209
0.6557
0.5909
0.5450
8.7988
0.3328
0.3104
0.3668
0.3804
0.4754
0.4828
0.5643
0.5321
0.4664
0.3966
0.3581
0.3243
4.9904
1.0780
1.0607
1.1669
1.0778
1 .2046
1 .0962
1.1992
1.1596
1.1517
1.1337
1 .0636
1.0740
13.4660
7.8929
7.7038
0.5896
0.6241
0.6565
0.7262
0.6791
0.7368
0.7930
0.6998
0.6480
0.6393
0.5935
7.9349
3.3667
3.2505
3.7156
3.6193
4.1730
3.7466
3.7820
3.7906
3.7676
3.6414
3.4211
3.4142
43.6886
1.2137
1.1973
1.4370
1 .4582
1.7776
1 .5486
1.5597
1.5313
1.5677
1.4124
1.1727
1.0797
16.9561
4.0294
3.4325
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.7628
0.7157
0.8585
0.8935
1.0979
1.1539
1.4127
1.3996
1.0537
0.9401
0.8515
0.7837
11.9237
0.6058
0.5683
0.3456
0.1824
0.2241
0.2355
0.2884
0.2857
0.2151
0.1919
0.1738
0.3155
3.6322
7.9832
7.3117
-------
2007 CAP Emissions Calculations
March
April
May
June
July
August
September
October
November
December
Urban Local
January
February
March
April
May
June
July
August
September
October
November
December
All Roadway Types
January
February
March
April
May
June
July
August
September
October
November
December
PMIO-FIL(TPY)
Kent New Castle Sussex
3.7702
3.4498
3.8650
3.5323
3.6517
1 1 .2659
3.5772
3.5078
10.2321
10.0602
119.5761
43.1321
42.0987
16.6866
15.2685
17.1063
15.6337
16.1621
15.3230
15.8323
15.5252
13.9169
13.6832
240.3684
202.8901
188.3509
70.0744
66.6544
81 .4724
62.5833
83.6273
86.0449
63.3575
75.3075
73.6000
68.0204
1121.9831
9.2060
9.5377
12.4840
1 1 .7876
13.5754
11.4931
12.1557
10.3087
8.4777
7.6337
122.8866
149.3299
137.1463
56.3019
51.3884
63.6201
51.6748
52.9218
54.9018
63.6183
61.2816
55.0759
53.3890
850.6498
230.8934
199.5712
118.1994
113.9317
138.4216
120.0116
126.0918
126.5080
134.4055
125.9550
113.7869
108.6660
1656.4422
3.7580
3.7229
4.4079
4.4004
5.2337
4.8712
4.0121
3.7090
1 1 .0368
10.7289
117.3317
23.6401
21.6515
9.0129
8.9287
10.5715
10.5535
12.5521
1 1 .6827
9.6223
8.8953
8.1345
7.9076
143.1529
342.9350
314.4139
78.1346
77.3368
92.1014
67.7849
79.6648
76.6007
84.5562
77.6201
109.6988
106.6430
1507.4902
PM2.5-FIL (TPY)
Kent New Castle Sussex
0.9384
0.8586
0.9620
0.8792
0.9089
2.8040
0.8903
0.8731
2.5467
2.5039
29.7620
10.7354
10.4782
4.1532
3.8003
4.2577
3.8912
4.0227
3.8138
3.9406
3.8641
3.4639
3.4057
59.8266
50.4984
46.8797
17.4412
16.5900
20.2781
15.5767
20.8145
21.4162
15.7694
18.7437
18.3187
16.9300
279.2565
2.2913
2.3739
3.1072
2.9339
3.3789
2.8606
3.0255
2.5658
2.1101
1.9000
34.0090
37.1675
34.1351
14.0133
12.7903
15.8347
12.8616
13.1720
13.6648
15.8343
15.2527
13.7081
13.2883
211.7228
57.4683
53.0954
29.4193
28.3571
34.4525
29.8703
31.3837
31.4873
33.4529
31 .3496
28.3210
27.0465
415.7040
0.9353
0.9266
1.0971
1.0952
1 .3026
1.2124
0.9986
0.9231
2.7470
2.6704
29.2033
5.8839
5.3890
2.2433
2.2223
2.6312
2.6267
3.1242
2.9078
2.3949
2.2140
2.0246
1 .9682
35.6301
85.3550
78.2562
19.4473
19.2488
22.9236
16.8714
19.8282
19.0656
21.0457
19.3193
27.3035
26.5430
375.2074
STATEWIDE
PM10-PRI PM2.5-PRI
776.7185
702.3359
266.4084
257.9230
311.9955
250.3798
289.3839
289.1537
282.3192
278.8826
297.0857
283.3294
4285.9155
193.3217
178.2313
66.3078
64.1958
77.6543
62.3184
72.0263
71.9690
70.2680
69.4126
73.9433
70.5194
1070.1679
-------
comments to draft AP-42 paved road section
t
GaryGarman o Ron Myers 06/24/201012:58 PM
Please respond to ggarman
This message has been replied to and forwarded.
Ron,
It's good to see the paved road section is being revised. Thanks. It has been a challenge in the
past explaining to industrial clients that paving a road would actually result in higher predicted
emissions than if the road is left unpaved. I think we'll see more paving and actual emission
reductions as a result of the new equation. A few editorial comments on the draft paved road
section:
Page 13.2.1-1, third paragraph, first sentence..change to "The paniculate emission factors
presented in a previous version.."
Page 13.2.1-5, third paragraph, last sentence..change "Table 13.2.1-3" to "Table 13.2.1-2"
Page 13.2.1-8, fifth paragraph, first sentence..change "Table 13.2.1-3" to "Table 13.2.1-2"
Page 13.2.1-9, second paragraph, second sentence..remove hyphen between "not" and "suggest"
Table 13.2.1-3...the page number this table is on should be changed to 13.2.1.10. Also, total
loading range for iron and steel should be 0.006-4.77, not 43.0-64.0.
Page 13.2.1-11, first paragraph, fourth sentence..remove hyphen between "any" and "of
Thanks again. I look forward to this draft being finalized.
Gary
Gary Garman
Environmental Scientist
McVehil-Monnett Associates, Inc.
44 Inverness Drive East, Bldg C
Englewood, CO 80112
303-790-1332
------- |