Transportation Conformity Guidance

for Quantitative Hot-spot Analyses in

PM2.5 and PM10 Nonattainment and
Maintenance Areas

Appendices

Transportation and Regional Programs Division

Office of Transportation and Air Quality
U.S. Environmental Protection Agency

EPA-420-B-10-040
December 2010
APPENDIX A: CLEARINGHOUSE OF WEBSITES, GUIDANCE, AND OTHER TECHNICAL
RESOURCES FOR PM HOT-SPOT ANALYSES ......................................................... A-1
APPENDIX B: EXAMPLES OF PROJECTS OF LOCAL AIR QUALITY CONCERN .........................B-1
APPENDIX C: HOT-SPOT REQUIREMENTS FOR PM10 AREAS WITH PRE-2006 APPROVED
CONFORMITY SIPS .......................................................................................................C-1
APPENDIX D: CHARACTERIZING INTERSECTION PROJECTS FOR MOVES ............................. D-1
APPENDIX E: EXAMPLE QUANTITATIVE PM HOT-SPOT ANALYSIS OF A HIGHWAY
PROJECT USING MOVES AND CAL3QHCR .............................................................. E-1
APPENDIX F: EXAMPLE QUANTITATIVE PM HOT-SPOT ANALYSIS OF A TRANSIT
PROJECT USING MOVES AND AERMOD .................................................................. F-1
APPENDIX G: EXAMPLE OF USING EMFAC FOR A HIGHWAY PROJECT .................................. G-1
APPENDIX H: EXAMPLE OF USING EMFAC TO DEVELOP EMISSION FACTORS FOR A
TRANSIT PROJECT ....................................................................................................... H-1
APPENDIX I: ESTIMATING LOCOMOTIVE EMISSIONS .................................................................. I-1
APPENDIX J: ADDITIONAL REFERENCE INFORMATION ON AIR QUALITY MODELS AND
DATA INPUTS ..................................................................................................................J-1
APPENDIX K: EXAMPLES OF DESIGN VALUE CALCULATIONS FOR PM HOT-SPOT
ANALYSES ..................................................................................................................... K-1

i
 A ppendix A :
Clearinghouse of Websites, Guidance, and Other Technical
Resources for PM Hot-spot Analyses

A.1 INTRODUCTION

This appendix is a centralized compilation of documents and websites referenced in the

guidance, along with additional technical resources that may be of use when completing
quantitative PM hot-spot analyses. Refer to the appropriate sections of the guidance for
complete discussions on how to use these resources in the context of completing a
quantitative PM hot-spot analysis. The references listed are current as of this writing;
readers are reminded the check for the latest versions when using them for a particular
PM hot-spot analysis.

A.2 TRANSPORTATION CONFORMITY AND CONTROL MEASURE GUIDANCE

The EPA hosts an extensive library of transportation conformity guidance online at:
www.epa.gov/otaq/stateresources/transconf/policy.htm (unless otherwise noted). The
following specific guidance documents, in particular, may be useful references when
implementing PM hot-spot analyses:

• “Policy Guidance on the Use of MOVES2010 for SIP Development and

Transportation Conformity, and Other Purposes,” EPA-420-B-09-046 (December
2009). This document describes how and when to use the MOVES2010
emissions model for SIP development, transportation conformity determinations,
and other purposes.

• “EPA Releases MOVES2010 Mobile Source Emissions Model: Questions and

Answers,” EPA-420-F-09-073 (December 2009).

• “EPA Releases MOVES2010a Mobile Source Emissions Model Update:

Questions and Answers,” EPA-420-F-10-050 (August 2010).

• “Technical Guidance on the Use of MOVES2010 for Emission Inventory Prepara-

tion in State Implementation Plans and Transportation Conformity,” EPA-420-B-
10-023 (December 2009). This document provides guidance on appropriate input
assumptions and sources of data for the use of MOVES2010 in SIP submissions
and regional emissions analyses for transportation conformity purposes.

• EPA and FHWA, “Transportation Conformity Guidance for Qualitative Hot-spot

Analyses in PM2.5 and PM10 Nonattainment and Maintenance Areas,” EPA-420-
B-06-902 (March 2006).

A-1
 • EPA and FHWA, “Guidance for the Use of Latest Planning Assumptions in
Transportation Conformity Determinations,” EPA-420-B-08-901 (December
2008).

• “Guidance for Developing Transportation Conformity State Implementation

Plans,” EPA-420-B-09-001 (January 2009).

• EPA-verified anti-idle technologies (including technologies that pertain to trucks)

can be found at: www.epa.gov/otaq/smartway/transport/what-smartway/verified-
technologies.htm#idle.

• For additional information about quantifying the benefits of retrofitting and

replacing diesel vehicles and engines for conformity determinations, see EPA’s
website for the most recent guidance on this topic:
www.epa.gov/otaq/stateresources/transconf/policy.htm.

• For additional information about quantifying and using long duration truck idling
benefits for conformity determinations, see EPA’s website for the most recent
guidance on this topic: www.epa.gov/otaq/stateresources/transconf/policy.htm.

FHWA’s transportation conformity site has additional conformity information, including

examples of qualitative PM hot-spot analyses. Available at:
www.fhwa.dot.gov/environment/air_quality/conformity/practices/.

A.3 MOVES MODEL TECHNICAL INFORMATION AND USER GUIDES

MOVES, any future versions of the model, the latest user guides, and technical
information can be found at www.epa.gov/otaq/models/moves/index.htm, including the
following:

• “User Guide for MOVES2010a.” This guide provides detailed instructions for
setting up and running MOVES2010a. Available at
www.epa.gov/otaq/models/moves/index.htm.

Policy documents and Federal Register announcements related to the MOVES model can
be found on the EPA’s website at:
www.epa.gov/otaq/stateresources/transconf/policy.htm#models.

Guidance on using the MOVES model at the project level, as well as illustrative
examples of using MOVES for quantitative PM hot-spot analyses, can be found in
Section 4 of the guidance and in Appendices D, E and F.

A-2
A.4 EMFAC2007 MODEL TECHNICAL INFORMATION, USER GUIDES, AND
OTHER GUIDANCE

EMFAC2007, its user guides, and any future versions of the model can be downloaded
from the California Air Resources Board website at:
www.arb.ca.gov/msei/onroad/latest_version.htm.

Policy documents and Federal Register announcements related to the EMFAC model can
be found on the EPA’s website at:
www.epa.gov/otaq/stateresources/transconf/policy.htm#models.

Supporting documentation for EMFAC, including the technical memorandum “Revision

of Heavy Heavy-Duty Diesel Truck Emission Factors and Speed Correction Factors”
cited in Section 5 of this guidance, can be found at
www.arb.ca.gov/msei/supportdocs.htm#onroad.

Instructions on using the EMFAC model at the project level, as well as examples of using
EMFAC for quantitative PM hot-spot analyses, can be found in Section 5 of the guidance
and in Appendices G and H.

A.5 DUST EMISSIONS METHODS AND GUIDANCE

Information on calculating emissions from paved roads, unpaved roads, and construction
activities can be found in AP-42, Chapter 13 (Miscellaneous Sources). AP-42 is EPA’s
compilation of data and methods for estimating average emission rates from a variety of
activities and sources from various sectors. Refer to EPA’s website to access the latest
versions of AP-42 sections and for more information about AP-42 in general:
www.epa.gov/ttn/chief/ap42/index.html.

Guidance on calculating dust emissions for PM hot-spot analyses can be found in Section
6 of the guidance.

A.6 LOCOMOTIVE EMISSIONS GUIDANCE

The following guidance documents, unless otherwise noted, can be found on or through
the EPA’s locomotive emissions website at: www.epa.gov/otaq/locomotives.htm:

• “Procedure for Emission Inventory Preparation - Volume IV: Mobile Sources,”

Chapter 6. Available online at: www.epa.gov/OMS/invntory/r92009.pdf. Note
that the emissions factors listed in Volume IV have been superseded by the April
2009 publication listed below for locomotives certified to meet EPA standards.

A-3
 • “Emission Factors for Locomotives,” EPA-420-F-09-025 (April 2009). Available
online at: www.epa.gov/otaq/regs/nonroad/locomotv/420f08014.htm.

• “Control of Emissions from Idling Locomotives,” EPA-420-F-08-014 (March

2008).

• “Guidance for Quantifying and Using Long Duration Switch Yard Locomotive
Idling Emission Reductions in State Implementation Plans,” EPA-420-B-04-002
(January 2004). Available online at:
www.epa.gov/otaq/smartway/documents/420b04002.pdf.

• EPA-verified anti-idle technologies (including technologies that pertain to

locomotives) can be found at: www.epa.gov/otaq/smartway/transport/what-
smartway/verified-technologies.htm#idle.

Guidance on calculating locomotive emissions for PM hot-spot analyses can be found in

Section 6 of the guidance and in Appendix I.

A.7 AIR QUALITY DISPERSION MODEL TECHNICAL INFORMATION AND

USER GUIDES

The latest version of “Guideline on Air Quality Models” (Appendix W to 40 CFR Part
51) (dated 2005 as of this writing) can be found on EPA’s SCRAM website at:
www.epa.gov/scram001/guidance_permit.htm.

Both AERMOD and CAL3QHCR models and related documentation can be obtained
through EPA’s Support Center for Regulatory Air Models (SCRAM) web site at:
www.epa.gov/scram001. In particular, the following guidance may be useful when
running these models:

• AERMOD Implementation Guide

• AERMOD User Guide (“User’s Guide for the AMS/EPA Regulatory Model –
AERMOD”)

• CAL3QHCR User Guide (“User’s Guide to CAL3QHC Version 2.0: A Modeling

Methodology for Predicting Pollutant Concentrations Near Roadway
Intersections”)

• MPRM User Guide

• AERMET User Guide

A-4
Information on locating and considering air quality monitoring sites can be found in 40
CFR Part 58 (Ambient Air Quality Surveillance), particularly in Appendices D and E to
that part.

Guidance on selecting and using an air quality model for quantitative PM hot-spot
analyses can be found in Sections 7 and 8 of the guidance and in Appendix J. Illustrative
examples of using an air quality model for a PM hot-spot analysis can be found in
Appendices E and F.

A.8 TRANSPORTATION DATA AND MODELING CONSIDERATIONS

The following is a number of technical resources on transportation data and modeling

which may help implementers determine the quality of their inputs and the sensitivity of
various data.

A.8.1 Transportation model improvement

The FHWA Travel Model Improvement Program (TMIP) provides a wide range of
services and tools to help planning agencies improve their travel analysis techniques.
Available online at: http://tmip.fhwa.dot.gov/.

A.8.2 Speed

“Evaluating Speed Differences between Passenger Vehicles and Heavy Trucks for
Transportation-Related Emissions Modeling.” Available online at:
www.ctre.iastate.edu/reports/truck_speed.pdf.

A.8.3 Project level planning

“NCHRP 255: Highway Traffic Data for Urbanized Area Project Planning and Design.”
Available online at:
http://tmip.fhwa.dot.gov/sites/tmip.fhwa.dot.gov/files/NCHRP_255.pdf.

A.8.4 Traffic analysis

Traffic Analysis Toolbox website: http://ops.fhwa.dot.gov/trafficanalysistools/.

“Traffic Analysis Toolbox Volume I: Traffic Analysis Tools Primer.” Federal Highway
Administration, FHWA-HRT-04-038 (June 2004). Available online at:
http://ops.fhwa.dot.gov/trafficanalysistools/tat_vol1/vol1_primer.pdf.

The Highway Capacity Manual Application Guidebook. Transportation Research Board,

Washington, D.C., 2003. Available online at: http://hcmguide.com/.

A-5
The Highway Capacity Manual 2000. Transportation Research Board, Washington,
D.C., 2000. Not available online; purchase information available at:
http://144.171.11.107/Main/Public/Blurbs/Highway_Capacity_Manual_2000_152169.asp
x. As of this writing, the 2000 edition is most current; the most recent version of the
manual, and the associated guidebook, should be consulted when completing PM hot-
spot analyses.

A-6
 A ppendix B :
Examples of Projects of Local Air Quality Concern

B.1 INTRODUCTION

This appendix gives additional guidance on what types of projects may be projects of
local air quality concern requiring a quantitative PM hot-spot analysis under 40 CFR
93.123(b)(1). However, as noted elsewhere in this guidance, PM10 nonattainment and
maintenance areas with approved conformity SIPs that include PM10 hot-spot provisions
from previous rulemakings must continue to follow those approved conformity SIP
provisions until the SIP is revised; see Appendix C for more information.

B.2 EXAMPLES OF PROJECTS THAT REQUIRE PM HOT-SPOT ANALYSES

EPA noted in the March 2006 final rule that the examples below are considered to be the
most likely projects that would be covered by 40 CFR 93.123(b)(1) and require a PM2.5
or PM10 hot-spot analysis (71 FR 12491). 1

Some examples of projects of local air quality concern that would be covered by 40 CFR
93.123(b)(1)(i) and (ii) are:
• A project on a new highway or expressway that serves a significant volume of
diesel truck traffic, such as facilities with greater than 125,000 annual average
daily traffic (AADT) and 8% or more of such AADT is diesel truck traffic;
• New exit ramps and other highway facility improvements to connect a highway or
expressway to a major freight, bus, or intermodal terminal;
• Expansion of an existing highway or other facility that affects a congested
intersection (operated at Level-of-Service D, E, or F) that has a significant
increase in the number of diesel trucks; and,
• Similar highway projects that involve a significant increase in the number of
diesel transit busses and/or diesel trucks.

Some examples of projects of local air quality concern that would be covered by 40 CFR
93.123(b)(1)(iii) and (iv) are:
• A major new bus or intermodal terminal that is considered to be a “regionally
significant project” under 40 CFR 93.101 2; and,
1
EPA also clarified 93.123(b)(1)(i) in the January 24, 2008 final rule (73 FR 4435-4436).
2
40 CFR 93.101 defines a “regionally significant project” as “a transportation project (other than an
exempt project) that is on a facility which serves regional transportation needs (such as access to and from
the area outside of the region, major activity centers in the region, major planned developments such as
new retail malls, sports complexes, etc., or transportation terminals as well as most terminals themselves)
and would normally be included in the modeling of a metropolitan area’s transportation network, including
at a minimum all principal arterial highways and all fixed guideway transit facilities that offer an
alternative to regional highway travel.”

B-1
 • An existing bus or intermodal terminal that has a large vehicle fleet where the
number of diesel buses increases by 50% or more, as measured by bus arrivals.

A project of local air quality concern covered under 40 CFR 93.123(b)(1)(v) could be any
of the above listed project examples.

B.3 EXAMPLES OF PROJECTS THAT DO NOT REQUIRE PM HOT-SPOT

ANALYSES

The March 2006 final rule also provided examples of projects that would not be covered
by 40 CFR 93.123(b)(1) and would not require a PM2.5 or PM10 hot-spot analysis (71 FR
12491).

The following are examples of projects that are not a local air quality concern under 40
CFR 93.123(b)(1)(i) and (ii):
• Any new or expanded highway project that primarily services gasoline vehicle
traffic (i.e., does not involve a significant number or increase in the number of
diesel vehicles), including such projects involving congested intersections
operating at Level-of-Service D, E, or F;
• An intersection channelization project or interchange configuration project that
involves either turn lanes or slots, or lanes or movements that are physically
separated. These kinds of projects improve freeway operations by smoothing
traffic flow and vehicle speeds by improving weave and merge operations, which
would not be expected to create or worsen PM NAAQS violations; and,
• Intersection channelization projects, traffic circles or roundabouts, intersection
signalization projects at individual intersections, and interchange reconfiguration
projects that are designed to improve traffic flow and vehicle speeds, and do not
involve any increases in idling. Thus, they would be expected to have a neutral or
positive influence on PM emissions.

Examples of projects that are not a local air quality concern under 40 CFR
93.123(b)(1)(iii) and (iv) would be:
• A new or expanded bus terminal that is serviced by non-diesel vehicles (e.g.,
compressed natural gas) or hybrid-electric vehicles; and,
• A 50% increase in daily arrivals at a small terminal (e.g., a facility with 10 buses
in the peak hour).

B-2
 A ppendix C :
Hot-Spot Requirements for PM10 Areas with Pre-2006
Approved Conformity SIPs

C.1 INTRODUCTION

This appendix describes what projects require a quantitative PM10 hot-spot analysis in
those limited cases where a state’s approved conformity SIP is based on pre-2006
conformity requirements. 1 The March 10, 2006 final hot-spot rule defined the current
federal conformity requirements for what projects require a PM hot-spot analysis (i.e.,
only certain highway and transit projects that involve significant levels of diesel vehicle
traffic or any other project identified in the PM SIP as a local air quality concern). 2
However, there are some PM10 nonattainment and maintenance areas where PM10 hot-
spot analyses are required for different types of projects, as described further below.

This appendix will be relevant for only a limited number of PM10 nonattainment and
maintenance areas with pre-2006 approved conformity SIPs. This appendix is not
relevant for any PM2.5 nonattainment or maintenance areas, since the current federal
PM2.5 hot-spot requirements apply in all such areas. Project sponsors can use the
interagency consultation process to verify applicable requirements before beginning a
quantitative PM10 hot-spot analysis.

C.2 PM10 AREAS WHERE THE PRE-2006 HOT-SPOT REQUIREMENTS APPLY

Prior to the March 2006 final rule, the federal conformity rule required some type of hot-
spot analysis for all non-exempt federally funded or approved projects in PM10
nonattainment and maintenance areas. These pre-2006 requirements are in effect for
those states with an approved conformity SIP that includes the pre-2006 hot-spot
requirements.

In PM10 areas with approved conformity SIPs that include the pre-2006 hot-spot
requirements, a quantitative PM10 hot-spot analysis is required for the following types of
projects:
• Projects which are located at sites at which PM10 NAAQS violations have
been verified by monitoring;
• Projects which are located at sites which have vehicle and roadway emission
and dispersion characteristics that are essentially identical to those of sites

1
A “conformity SIP” includes a state’s specific criteria and procedures for certain aspects of the
transportation conformity process (40 CFR 51.390).
2
See Section 2.2 and Appendix B of this guidance and the preamble of the March 2006 final rule (71 FR
12491-12493).

C-1
 with verified violations (including sites near one at which a violation has been
monitored); and
• New or expanded bus and rail terminals and transfer points which increase the
number of diesel vehicles congregating at a single location.

This guidance should be used to complete any quantitative PM10 hot-spot analyses.

In addition, a qualitative PM10 hot-spot analysis is required in the pre-2006 hot-spot

requirements for all other non-exempt federally funded or approved projects. For such
analyses, consult the 2006 EPA-FHWA qualitative hot-spot guidance. 3

These pre-2006 hot-spot requirements continue to apply in PM10 areas with approved
conformity SIPs that include them until the state acts to change the conformity SIP. The
conformity rule at 40 CFR 51.390 states that conformity requirements in approved
conformity SIPs “remain enforceable until the state submits a revision to its [conformity
SIP] to specifically remove them and that revision is approved by EPA.”

C.3 REVISING A CONFORMITY SIP

EPA strongly encourages affected states to revise pre-2006 provisions and take advantage
of the streamlining flexibilities provided by the current Clean Air Act. EPA’s January
2008 final conformity rule significantly streamlined the requirements for conformity SIPs
in 40 CFR 51.390. 4 As a result, conformity SIPs are now required to include only three
provisions (consultation procedures and procedures regarding written commitments)
rather than all of the provisions of the federal conformity rule.

EPA recommends that states with pre-2006 PM10 hot-spot requirements in their
conformity SIPs act to revise them to reduce the number of projects where a hot-spot
analysis is required. In affected PM10 areas, the current conformity rule’s PM10 hot-spot
requirements at 40 CFR 93.123(b)(1) and (2) will be effective only when a state either:
• Withdraws the existing provisions from its approved conformity SIP and EPA
approves this SIP revision, or
• Revises its approved conformity SIP consistent with the requirements found at 40
CFR 93.123(b) and EPA approves this SIP revision.

3
“Transportation Conformity Guidance for Qualitative Hot-spot Analyses in PM2.5 and PM10
Nonattainment and Maintenance Areas,” EPA420-B-06-902, found on EPA’s website at:
www.epa.gov/otaq/stateresources/transconf/policy/420b06902.pdf.
4
“Transportation Conformity Rule Amendments to Implement Provisions Contained in the 2005 Safe,
Accountable, Flexible, Efficient Transportation Equity Act: A Legacy for Users (SAFETEA-LU); Final
Rule,” 73 FR 4420.

C-2
Affected states should contact their EPA Regional Office to proceed with one of these
two options. For more information about conformity SIPs, see EPA’s “Guidance for
Developing Transportation Conformity State Implementation Plans (SIPs),” EPA-420-B-
09-001 (January 2009); available online at:
www.epa.gov/otaq/stateresources/transconf/policy/420b09001.pdf.

C-3
 A ppendix D:
Characterizing Intersection Projects for MOVES

D.1 INTRODUCTION

This appendix expands upon the discussion in Section 4.2 on how best to characterize
links when modeling an intersection project using MOVES. The MOVES emissions
model allows users to represent intersection traffic activity with a higher degree of
sophistication compared to previous models. This appendix provides several options to
describe vehicle activity to take advantage of the capabilities MOVES offers to complete
more accurate PM hot-spot analyses of intersection projects. MOVES is the approved
emissions model for PM hot-spot analyses in areas outside of California.

Exhibit D-1 is an example of a simple signalized intersection showing the links

developed by a project sponsor to represent the two general categories of vehicle activity
expected to take place at this intersection (approaching the intersection and departing the
intersection).

Exhibit D-1. Example of Approach and Departure Links for a Simple Intersection

D-1
When modeling an intersection, each approach link or departure link can be modeled as
one or more links in MOVES depending on the option chosen to enter traffic activity.
This guidance suggests three possible options for characterizing activity on each
approach and departure link (such as those shown in Exhibit D-1):
• Option 1: Using average speeds
• Option 2: Using link drive schedules
• Option 3: Using Op-Mode distributions

While Option 1 may need to be relied upon more during the initial transition to using
MOVES, as more detailed data are available to describe vehicle activity, users are
encouraged to consider using the Options 2 and 3 to take full advantage of the
capabilities of MOVES.

Once a decision has been made on how to characterize links, users should continue to
develop the remaining MOVES inputs as discussed in Section 4 of the guidance.

D.2 OPTION 1: USING AVERAGE SPEEDS

The first option is for the user to estimate the average speeds for each link in the
intersection based on travel time and distance. Travel time should account for the total
delay attributable to traffic signal operation, including the portion of travel when the light
is green and the portion of travel when the light is red. The effect of a traffic signal cycle
on travel time includes deceleration delay, move-up time in a queue, stopped delay, and
acceleration delay. Using the intersection example given in Exhibit D-1, each approach
link would be modeled as one link to reflect the higher emissions associated with vehicle
idling through lower speeds affected by stopped delay; each departure link would be
modeled as one link to reflect the higher emissions associated with vehicle acceleration
through lower speeds affected by acceleration delay.

Project sponsors can determine congested speeds by using appropriate methods based on
best practices for highway analyses. Some resources are available through FHWA’s
Travel Model Improvement Program (TMIP). 1 Methodologies for computing
intersection control delay are provided in the Highway Capacity Manual. 2 All
assumptions, methods, and data underlying the estimation of average speeds and delay
should be documented as part of the PM hot-spot analysis.

1
See FHWA’s TMIP website: http://tmip.fhwa.dot.gov/.
2
Users should consult the most recent version of the Highway Capacity Manual. As of the release of this
guidance, the latest version is the Highway Capacity Manual 2000, which can be obtained from the
Transportation Research Board (see http://144.171.11.107/Main/Public/Blurbs/152169.aspx for details).

D-2
D.3 OPTION 2: USING LINK DRIVE SCHEDULES

A more refined approach is to enter vehicle activity into MOVES as a series of link drive
schedules to represent individual segments of cruise, deceleration, idle, and acceleration
of a congested intersection. A link drive schedule defines a speed trajectory to represent
the entire vehicle fleet via second-by-second changes in speed and highway grade.
Unique link drive schedules can be defined to describe types of vehicle activity that have
distinct emission rates, including cruise, deceleration, idle, and acceleration.

Exhibit D-2 illustrates why using this more refined approach can result in a more detailed
emissions analysis. This exhibit shows the simple trajectory of a single vehicle
approaching an intersection during the red signal phase of a traffic light cycle. This
trajectory is characterized by several distinct phases (a steady cruise speed, decelerating
to a stop for the red light, idling during the red signal phase, and accelerating when the
light turns green). In contrast, the trajectory of a single vehicle approaching an
intersection during the green signal phase of a traffic light cycle is characterized by a
more or less steady cruise speed through the intersection.

Exhibit D-2. Example Single Vehicle Speed Trajectory Through a Signalized

Intersection
50

45

40
Green Light
35

30
Speed (mph)

25 Red Light

20
Cruise Decelerate Idle Accelerate Cruise
15

10

0
-100 -80 -60 -40 -20 0 20 40 60 80 100
Distance (m)

For the example intersection in Exhibit D-1, link drive schedules representing the
different operating modes of vehicle activity on the approach and departure links can be
determined. For approach links, the length of a vehicle queue is dependent on the
number of vehicles subject to stopping at a red signal. Vehicles approaching a red traffic

D-3
signal decelerate over a distance extending from the intersection stop line back to the
stopping distance required for the last vehicle in the queue. The average stopping
distance can be calculated from the average deceleration rate and the average cruise
speed. Similarly, for the departure links, vehicles departing a queue when the light turns
green accelerate over a distance extending from the end of the vehicle queue to the
distance required for the first vehicle to reach the cruise speed, given the rate of
acceleration and cruise speed. Exhibit D-3 provides an illustration of how the different
vehicle operating modes may be apportioned spatially near this signalized intersection.

Exhibit D-3. Example Segments of Vehicle Activity Near a Signalized Intersection

There are other considerations with numerous vehicles stopping and starting at an
intersection over many signal cycles during an hour. For instance, heavy trucks
decelerate and accelerate at slower rates than passenger cars. Drivers tend not to
decelerate at a constant rate, but through a combination of coasting and light and heavy
braking. Acceleration rates are initially higher when starting from a complete stop at an
intersection, becoming progressively lower to make a smooth transition to cruise speed.

In the case of an uncongested intersection, the rates of vehicles approaching and

departing the intersection are in equilibrium. Some vehicles may slow, and then speed up
to join the dissipating queue without having to come to a full stop. Once the queue
clears, approaching vehicles during the remainder of the green phase of the cycle will
cruise through the intersection virtually unimpeded.

In the case of a congested intersection, the rate of vehicles approaching the intersection is
greater than the rate of departure, with the result that no vehicle can travel through
without stopping; vehicles approaching the traffic signal, whether it is red or green, will
have to come to a full stop and idle for one or more cycles before departing the
intersection. The latest Highway Capacity Manual is a good source of information for

D-4
vehicle operation through signalized intersections. All assumptions, methods, and data
underlying the development of link drive schedules should be documented as part of the
PM hot-spot analysis.

The MOVES emission factors for each segment of vehicle activity obtained via
individual link drive schedules are readily transferable to either AERMOD or
CAL3QHCR, as discussed further in Section 7 of the guidance. There will most likely be
a need to divide the cruise and the acceleration segments to account for differences in
approach and departure traffic volumes.

Note: For both free-flow highway and intersection links, users may directly enter output
from traffic simulation models in the form of second-by-second individual vehicle
trajectories. These vehicle trajectories for each road segment can be input into MOVES
using the Link Drive Schedule Importer and defined as unique LinkIDs. There are no
limits in MOVES as to how many links can be defined; however, model run times
increase as the user defines more links. A representative sampling of vehicles can be
used to model higher volume segments by adjusting the resulting sum of emissions to
account for the higher traffic volume. For example, if a sampling of 5,000 vehicles
(5,000 links) was used to represent the driving patterns of 150,000 vehicles, then the sum
of emissions would be adjusted by a factor of 30 to account for the higher traffic volume
(i.e., 150,000 vehicles/5,000 vehicles). Since the vehicle trajectories include idling,
acceleration, deceleration, and cruise, separate roadway links do not have to be
explicitly defined to show changes in driving patterns. The sum of emissions from each
vehicle trajectory (LinkID) represents the total emission contribution of a given road
segment.

D.4 OPTION 3: USING OP-MODE DISTRIBUTIONS

A third option is for a user to generate representative Op-Mode distributions for approach
and departure links by calculating the fraction of fleet travel times spent in each mode of
operation. For any given signalized intersection, vehicles are cruising, decelerating,
idling, and accelerating. Op-Mode distributions can be calculated from the ratios of
individual mode travel times to total travel times on approach links and departure links.
This type of information could be obtained from Op-Mode distribution data from (1)
existing intersections with similar geometric and operational (traffic) characteristics, or
(2) output from traffic simulation models for the proposed project or similar projects.
Acceleration and deceleration assumptions, methods, and data underlying the activity-to-
Op-Mode calculations should be documented as part of the PM hot-spot analysis.

The following methodology describes a series of equations to assist in calculating vehicle

travel times on approach and departure links. Note that a single approach and single
departure link should be defined to characterize vehicles approaching, idling at, and
departing an intersection (e.g., there is no need for an “idling link,” as vehicle idling is
captured as part of the approach link).

D-5
D.4.1 Approach links

When modeling each approach link, the fraction of fleet travel times in seconds (s) in
each mode of operation should be determined based on the fraction of time spent
cruising, decelerating, accelerating, and idling:

Total Fleet Travel Time (s) = Cruise Time + Decel Time + Accel Time +
Idle Time

The cruise travel time can be represented by the number of vehicles cruising multiplied
by the length of approach divided by the average cruise speed:

Cruise Time (s) = Number of Cruising Vehicles * (Length of Approach (mi) ÷

Average Cruise Speed (mi/hr)) * 3600 s/hr

The deceleration travel time can be represented by the number of vehicles decelerating
multiplied by the average cruise speed divided by the average deceleration rate:

Decel Time (s) = Number of Decelerating Vehicles * (Average Cruise Speed

(mi/hr) ÷ Average Decel Rate (mi/hr/s))

The acceleration travel time occurring on an approach link can be similarly represented.
However, to avoid double-counting acceleration activity that occurs on the departure link,
users should multiply the acceleration time by the proportion of acceleration that occurs
on the approach link (Accel Length Fraction on Approach):

Accel Time (s) = Number of Accelerating Vehicles * (Average Cruise Speed

(mi/hr) ÷ Average Accel Rate (mi/hr/s)) * Accel Length Fraction on
Approach

The idle travel time can be represented by the number of vehicles idling multiplied by the
average stopped delay (average time spent stopped at an intersection):

Idle Time (s) = Number of Idling Vehicles * Average Stopped Delay (s)

Control delay (total delay caused by an intersection) may be used in lieu of average
stopped delay, but control delay includes decelerating and accelerating travel times,
which should be subtracted out (leaving only idle time).

After calculating the fraction of time spent in each mode of approach activity, users
should select the appropriate MOVES Op-Mode corresponding to each particular type of
activity (see Section 4.5.7 for more information). The operating modes in MOVES
typifying approach links include:
• Cruise/acceleration (OpModeID 11-16, 22-25, 27-30, 33, 35, 37-40);
• Low and moderate speed coasting (OpModeID 11, 21);
• Braking (OpModeID 0, 501);

D-6
 • Idling (OpModeID 1); and
• Tire wear (OpModeID 400-416).

The relative fleet travel time fractions can be allocated to the appropriate Op-Modes in
MOVES. The resulting single Op-Mode distribution accounts for relative times spent in
the different driving modes (cruise, deceleration, acceleration, and idle) for the approach
link. A simple example of deriving Op-Mode distributions for a link using this
methodology is demonstrated in Step 3 of Appendix F for a bus terminal facility.

D.4.2 Departure links

When modeling each departure link, the fraction of fleet travel times spent in each mode
of operation should be determined based on the fraction of time spent cruising and
accelerating:

Total Fleet Travel Time (s) = Cruise Time + Accel Time

The cruise travel time can be represented by the number of vehicles cruising multiplied
by the travel distance divided by the average cruise speed:

Cruise Time (s) = Number of Cruising Vehicles * (Length of Departure (mi) ÷

Average Cruise Speed (mi/hr)) * 3600 s/hr

The acceleration travel time occurring during the departure link can be represented by the
number of vehicles accelerating multiplied by the average cruise speed divided by the
average acceleration rate. However, to avoid double-counting acceleration activity that
occurs on the approach link, users should multiply the resulting acceleration time by the
proportion of acceleration that occurs on the departure link (Accel Length Fraction on
Departure):

Accel Time (s) = Number of Accelerating Vehicles * (Average Cruise Speed

(mi/hr) ÷ Average Accel Rate (mi/hr/s)) * Accel Length Fraction on
Departure

After calculating fraction of time spent in each mode of departure activity, users should
select the appropriate MOVES Op-Mode corresponding to each particular type of activity
(see Section 4.5.7 for more information). The operating modes typifying departure links
include:
• Cruise/acceleration (OpModeID 11-16, 22-25, 27-30, 33, 35, 37-40); and
• Tire wear (OpModeID 401-416).

The relative fleet travel time fractions can be allocated to the appropriate Op-Modes. The
resulting single Op-Mode distribution accounts for relative times spent in the different
driving modes (cruise and acceleration) for the departure link.

D-7
 A ppendix E :
Example Quantitative PM Hot-spot Analysis of a Highway
Project using MOVES and CAL3QHCR

E.1 INTRODUCTION

The purpose of this appendix is to demonstrate the procedures for completing a hot-spot
analysis using MOVES and CAL3QHCR following the basic steps described in Section
3. Readers should reference the appropriate sections in the guidance as needed for more
detail on how to complete each step of the analysis. This example is limited to showing
the build scenario; in practice, project sponsors may also have to analyze the no-build
scenario. While this example calculates emission rates using MOVES, EMFAC users
may find the air quality modeling described in this appendix helpful.

Note: The following example of a quantitative PM hot-spot analysis is highly simplified

and intended only to demonstrate the basic procedures described in the guidance. This
example uses default data in places where the use of project-specific data in a real-world
situation would be expected. In addition, actual PM hot-spot analyses could be
significantly more complex and are likely to require more documentation of data and
decisions. In this example, the interagency consultation process is used as needed for
evaluating and choosing models, methods, and assumptions, according to the
requirements of 40 CFR 93.105(c)(1)(i).

E .2 P ROJECT DESCRIPTION AND CONT E XT

The proposed project is the construction of a highway interchange connecting a four-lane

principle arterial with a six-lane freeway through on-and-off ramps (see Exhibit E-1).
The project is being built to allow truck access to local businesses. The project is located
in an area that was designated nonattainment for the 2006 24-hour PM2.5 NAAQS and
1997 annual PM2.5 NAAQS.

The following is some additional pertinent data about the project:

• The project is located in a medium-sized city (within one county) in a state other
than California.
• The project is expected to take less than a year to complete and has an estimated
completion date of 2013. The year of peak emissions is expected to be 2015,
when considering the project’s emissions and background concentrations.
• In 2015, the average annual daily traffic (AADT) at this location is expected to
exceed 125,000 vehicles and greater than eight percent of the traffic will be
heavy-duty diesel trucks.
• The area surrounding the proposed project is primarily residential, with no nearby
sources that need to be included in air quality modeling.

E-1
 • The state does not have an adequate or approved SIP budget for either PM2.5
NAAQS, and neither the EPA nor the state air agency has made a finding that
road dust is a significant contributor to the PM2.5 nonattainment problem.

Exhibit E-1. Simple Diagram of the Proposed Highway Project

E.3 DETERMINE NEED FOR PM HOT-SPOT ANALYSIS (STEP 1)

The proposed project is determined to be of local air quality concern under the
conformity rule because it is a new freeway project with a significant number of diesel
vehicles (see 40 CFR 93.123(b)(1)(i) and Sections 2.2 and Appendix B of the guidance).
Therefore, a quantitative PM hot-spot analysis is required.

E-2
E.4 DETERMINE APPROACH, MODELS, AND DATA (STEP 2)

E.4.1 Determining geographic area and emission sources to be covered by the analysis

First, the interagency consultation process is used to ensure that the project area is
defined so that the analysis includes the entire project, as required by 40 CFR
93.123(c)(2). As previously noted, it is also determined that, in this case, there are no
nearby emission sources to be included in air quality modeling (see Section 3.3.2).

E.4.2 Deciding the general analysis approach and analysis year(s)

Second, the project sponsor determines that the preferred approach in this case is to
model the build scenario first, completing a no-build scenario only if necessary.

In addition, it is determined that the year of peak emissions (within the timeframe of the
current transportation plan) is mostly likely to be 2015. Therefore, 2015 is selected as the
year of the analysis, and the analysis considers traffic data from 2015 (see Section 3.3.3).

E.4.3 Determining the PM NAAQS to be evaluated

Because the area has been designated nonattainment for both the 2006 24-hour PM2.5
NAAQS and 1997 annual PM2.5 NAAQS, the results of the analysis will have to be
compared to both NAAQS (see Section 3.3.4). All four quarters are included in the
analysis in order to estimate a year’s worth of emissions for both NAAQS.

E.4.4 Deciding on the type of PM emissions to be modeled

Next, the following directly-emitted PM2.5 emissions are determined to be relevant for
estimating the emissions in the analysis (see Section 2.5):
• Vehicle exhaust 1
• Brake wear
• Tire wear

E.4.5 Determining the models and methods to be used

Since this project is located outside of California, MOVES is used for emissions
modeling. In addition, it is determined that, since this is a highway project with no
nearby sources that need to be included in the air quality modeling, either AERMOD or
CAL3QHCR could be used for air quality modeling (see Section 3.3.6). In this case,
CAL3QHCR is selected. Making the decision on what air quality model to use at this
stage is important so that the appropriate data are collected, among other reasons (see
next step).

1
Represented in MOVES as PMtotal running and PMtotal crankcase running.

E-3
E.4.6 Obtaining project-specific data

Finally, the project sponsor compiles the data required to use MOVES, including project
traffic data, vehicle types and age, and temperature and humidity data for the months and
hours to be modeled (specifics on the data collected are described in the following steps).
In addition, information necessary to use CAL3QHCR to model air quality is gathered,
including meteorological data and information on representative air quality monitors.
The sponsor also ensures the latest planning assumptions are used and that data used for
the analysis are consistent with that used in the latest regional emissions analysis, as
required by the conformity rule (see Section 3.3.7).

E.5 ESTIMATE ON-ROAD MOTOR VEHICLE EMISSIONS (STEP 3)

Having completed the analysis preparations described above, the project sponsor then
follows the instructions provided in Section 4 of the guidance to use MOVES to estimate
the project’s on-road emissions:

E.5.1 Characterizing the project in terms of links

As described in Section 4.2 of the guidance, links are defined based on the expected
emission rate variability across the project. Generally, a highway project like the one
proposed in this example can be broken into four unique activity modes:
• Freeway driving at 55 mph;
• Arterial cruise at 45 mph;
• Acceleration away from intersections to a cruising speed of 45/55 mph; and
• Cruise, deceleration, and idle/cruise (depending on light timing) at intersections.

Following the guidance given in Section 4.2, 20 links are defined for MOVES and
CAL3QHCR modeling, each representing unique geographic and activity parameters (see
Exhibits E-2 and E-3). Each LinkID is defined with the necessary information for air
quality modeling: link length, link width, link volume, as well as link start and end points
(x1, y1, x2, y2 coordinates).

Decisions on how to best define links are based on an analysis of vehicle activity and
patterns within the project area. AADT is calculated from a travel demand model for
passenger cars, passenger trucks, intercity buses, short haul trucks, and long haul trucks.
From these values, both an average-hour and peak-hour volume is calculated. The
average and peak-hour vehicle counts for each part of the project are shown in Exhibit E-
3.

E-4
Exhibit E-2. Diagram of Proposed Highway Project Showing Links

Based on the conditions in the project area, for this analysis peak traffic is assumed to be
representative of morning rush hour (AM: 6 a.m. to 9 a.m.) and evening rush hour (PM: 4
p.m. to 7 p.m.), while average hour traffic represents all other hours: midday (MD: 9 a.m.
to 4 p.m.), and overnight (ON: 7 p.m. to 6 a.m.) Identical traffic volume and speed
profiles are assumed for all quarters of the year. Quarters are defined as described in
Section 3.3.4 of the guidance: Q1 (January-March), Q2 (April-June), Q3 (July-
September), and Q4 (October-December).

E-5
Exhibit E-3. Peak-Hour and Average-Hour Traffic Counts for Each Project Link

Freeway Peak Hour Count Average Hour Count Fraction of Total

Passenger Cars 2260 452 0.45
Passenger Trucks 1760 352 0.35
Intercity Buses 36 7 0.01
Short Haul Trucks (gas) 60 12 0.01
Long Haul Trucks (diesel) 944 189 0.19
Total 5060 1012 1.00

Exit Ramps Peak Hour Count Average Hour Count Fraction of Total
Passenger Cars 124 25 0.22
Passenger Trucks 124 25 0.22
Intercity Buses 8 2 0.01
Short Haul Trucks (gas) 12 2 0.02
Long Haul Trucks (diesel) 300 60 0.53
Total 568 114 1.00

Entrance Ramps Peak Hour Count Average Hour Count Fraction of Total
Passenger Cars 176 35 0.29
Passenger Trucks 148 30 0.24
Intercity Buses 0 0 0.00
Short Haul Trucks (gas) 16 3 0.03
Long Haul Trucks (diesel) 276 55 0.45
Total 616 123 1.00

Arterial Road Peak Hour Count Average Hour Count Fraction of Total
Passenger Cars 124 25 0.22
Passenger Trucks 116 23 0.20
Intercity Buses 12 2 0.02
Short Haul Trucks (gas) 0 0 0.00
Long Haul Trucks (diesel) 316 63 0.56
Total 568 114 1.00

A significant amount of traffic using the project is expected to be diesel trucks. While
the freeway contains approximately 19% diesel truck traffic, traffic modeling for the on-
and off-ramps connecting the freeway to the arterial road suggests approximately half of
vehicles are long-haul diesel trucks.

The average speeds on the freeway, arterial, and on/off-ramps are anticipated to be
identical in the analysis year for both peak and average hours and assumed to
approximately reflect the speed limit (55 mph, 45 mph, and 45 mph, respectively).
Traffic flow through the two intersections north and south of the freeway is controlled by
a signalized light with a 60% idle time for vehicles exiting the freeway and 40% idle time

E-6
for traffic entering the freeway from the arterial road or traveling north and south on the
arterial road passing over the freeway. The total project emissions, therefore, are
determined to be a function of:
• Vehicles traveling east and west on the freeway at a relatively constant 55 mph;
• Exiting vehicles decelerating to a stop at either the north or south signalized
intersection (or continuing through if the light is green);
• Vehicles accelerating away from the signalized intersections north and south, as
well as accelerating to a 55 mph cruise speed on the on-ramps;
• Idling activity at both intersections during the red phase of the traffic light; and
• Vehicles traveling between the north and south intersections at a constant 45 mph.

As there is no new parking associated with the project (e.g., parking lots), there are no
start emissions to be considered. Additionally, there are no trucks parked or “hoteling” in
extended idle mode anywhere in the project area, so extended idle emissions do not need
to be calculated.

E.5.2 Deciding how to handle link activity

As discussed in Section 4.2 of the guidance, MOVES offers several options for users to
apply activity information to each LinkID. For illustrative purposes, based on the
available information for the project (in this case, average speed, link average and peak
volume, and red-light idle time) several methods of deriving Op-Mode distributions are
employed in this example, as described below.

The links parameter table in Exhibit E-4 shows the various methods that activity is
entered into MOVES for each link. The column “MOVES activity input” describes how
the Op-Mode distribution is calculated for each particular link:
• Freeway links (links 1 and 4) are defined through a 55 mph average speed input,
from which MOVES calculated an Op-Mode distribution (as described in
Appendix D.2).
• Arterial cruise links (links 12 and 18) and links approaching an intersection queue
(links 2, 5, 9 and 15) are defined through a link-drive schedule with a constant
speed of 45 mph; indicating vehicles are cruising at 45 mph, with no acceleration
or deceleration (as described in Appendix D.3).
• Links representing vehicles accelerating away from intersections (links 7, 8, 11,
14, 17, 20) are given “adjusted average speeds” calculated from guidance in the
Highway Capacity Manual 2000, based on the link cruise speed (45 mph or 55
mph), red-light timing, and expected volume to capacity ratios. The adjusted
average speeds (16.6 mph or 30.3 mph) are entered into MOVES, which
calculates an Op-Mode distribution to reflect the lower average speed and
subsequent higher emissions (as described in Appendix D.2).
• Queue links (links 3, 6, 10, 13, 16, and 19) are given an Op-Mode distribution that
represents vehicles decelerating and idling (red light) as well as cruising through
(green light) (as described in Appendix D.4).
Step 1. First, an Op-Mode distribution is calculated for the link average speed
(45 mph).

E-7
 Step 2. Because this does not adequately account for idling at the intersection,
the Op-Mode fractions are re-allocated to add in idling. For instance, after
consulting the Highway Capacity Manual 2000, for this project scenario the
red light timing corresponds to approximately 40% idle time. A fraction of
0.4 for Op-Mode “1” is therefore added to the Op-Mode distribution
calculated from the 45 mph average speed in Step 1. The resulting Op-Mode
distribution represents all activity on a queuing intersection link.

The length of the queue links are estimated as a function of the length of three trucks, one
car, and one passenger truck with two meters in between each car and five meters in
between each truck.

Departure links on the arterial road are assumed to have a link length of 125 meters
(estimated to be the approximate distance that vehicles accelerate to a 45 mph cruising
speed). The departure links from the intersection to the on-ramp are assumed to have a
link length of 200 meters (estimated to be the approximate distance that vehicles
accelerate to a 55 mph cruising speed).

Exhibit E-4. Link Parameters (Peak Traffic)

E-8
E.5.3 Determining the number of MOVES runs

Following the guidance given in Section 4.3, it is determined that 16 MOVES runs
should be completed to produce emission factors that show variation across four hourly
periods (12 a.m., 6 a.m., 12 p.m., and 6 p.m., corresponding to overnight, morning,
midday, and evening traffic scenarios, respectively) and four quarterly periods
(represented by the months of January, April, July, and October; see Section 3.3).
MOVES will calculate values for all project links for the time period specified in each
run. The 16 emission factors produced for each link are calculated as grams/vehicle-
mile, which will then be paired with corresponding traffic volumes (peak or average
hour, depending on the hour) and used in CAL3QHCR.

E.5.4 Developing basic run specification inputs

When configuring MOVES for the analysis, the project sponsor follows Section 4.4 of
the guidance, including, but not limited to, the following:
• From the Scale panel, selecting the “Project” domain; in addition, choosing output
in “Emission Rates,” so that emission factors will be in grams/vehicle-mile as
needed for CAL3QHCR (see Section 4.4.2).
• From the Time Spans panel, the appropriate year, month, day, and hour for each
run is selected (see Section 4.4.3).
• From the Geographic Bounds panel, the custom domain is selected (see Section
4.4.4).
• From the Vehicles/Equipment panel, appropriate Source Types are selected (see
Section 4.4.5).
• From the Road Types panel, Urban Restricted and Urban Unrestricted road types
are selected (see Section 4.4.6).
• From the Pollutants and Processes panel, the appropriate pollutant/processes are
selected according to Section 4.4.7 of the guidance for “highway links.”
• In the Output panel, an output database is specified with grams and miles selected
as units (see Section 4.4.10).

E.5.5 Entering project details using the Project Data Manager

Meteorology

As described previously, it is determined that MOVES should be run 16 times to reflect

the following scenarios: 12 a.m., 6 a.m., 12 p.m., and 6 p.m. (corresponding to overnight,
morning, midday, and evening traffic scenarios, respectfully) for the months of January,
April, July, and October. Temperature and humidity data from a representative
meteorological monitoring station are obtained and confirmed to be consistent with data
used in the regional emissions analysis from the currently conforming transportation plan
and TIP (see Section 4.5.1). Average values for each hour and month combination are
used for each of the 16 MOVES runs. As an example, temperature and humidity values
for 12 a.m. January are shown in Exhibit E-5.

E-9
Exhibit E-5. Temperature and Humidity Input (January 12 a.m.)

Age Distribution

Section 4.5.2 of the guidance specifies that default data should be used only if an
alternative local dataset cannot be obtained and the regional conformity analysis relies on
national defaults. However, for the sake of simplicity only, in this example the national
default age distribution for 2015 is used for all vehicles and all runs (see Exhibit E-6).

Exhibit E-6. Age Distribution Table (Partial)

E-10
Fuel Supply and Fuel Formulation

In this example, it is determined appropriate to use the default fuel supply and
formulation (see Exhibits E-7 and E-8). The default fuel supply and formulation are
imported for each respective quarter (January, April, July, and October) and used for the
corresponding MOVES runs.

Exhibit E-7. Fuel Supply Table

Exhibit E-8. Fuel Formulation Table

E-11
Inspection and Maintenance (I/M)

As there is no PM emissions benefit in MOVES for I/M programs, this menu item is
skipped (see Section 4.5.4).

Link Source Type

The distribution of vehicle types on each link is defined in the Link Source Type table
(Exhibit E-9) following the guidance in Section 4.5.5. The fractions are derived from the
vehicle count estimates in Exhibit E-3.

Exhibit E-9. Link Source Type Table

Links

The Links input table shown in Exhibit E-10 is used to define each individual project link
in MOVES. Road Types 4 and 5 indicate Urban Restricted (freeway) and Urban
Unrestricted (arterial) road types, respectively; these correspond to the two road types
represented in this example. The average speed is entered for all links, but only used to
calculate Op-Mode distributions for links 1, 4, 7, 8, 11, 14, 17, and 20 (other links are
explicitly defined with a link-drive schedule or Op-Mode distribution). Link length and
link volume is entered for each link; however, since the “Emission Rates” option is
selected in the Scale panel, MOVES will produce grams/vehicle-mile. The volume and
link length will become relevant when running the air quality model later in this analysis.

E-12
Exhibit E-10. Links Input (AM Period)

The remaining links are defined with an Op-Mode distribution (Exhibit E-11) calculated
separately, as discussed earlier. Operating modes used in this analysis vary by both link
and source type, but not by hour or day.

Exhibit E-11. Operating Mode Distribution Table (Partial)

E-13
Off-Network

As it was determined that there are no off-network links (such as parking lots or truck
stops) that would have to be considered using the Off-Network Importer, there is no need
to use this option in this example.

E.5.6 Generating emission factors for use in air quality modeling

After generating the run specification and entering the required information into the
Project Data Manager as described above, MOVES is run 16 times, once for each unique
hour/month combination. Upon completion of each run, the MOVES output is located in
the MySQL output database table “rateperdistance” and sorted by Month, Hour, LinkID,
ProcessID, and PollutantID. An aggregate PM2.5 emission factor is then calculated by the
project sponsor for each Month, Hour, and LinkID combination using the following
equation and the guidance given in Section 4.4.7 of the guidance: 2

PMaggregate total = (PMtotal running) + (PMtotal crankcase running) + (brake wear) + (tire wear)

The 16 resulting grams/vehicle-mile emission factors (Exhibit E-12) for each link are
then ready to be used as input into the CAL3QHCR dispersion model to predict future
PM2.5 concentrations.

2
EPA is considering creating one or more MOVES scripts that would automate the summing of aggregate
emissions when completing project-level analyses. These scripts would be made available for download on
the MOVES website (www.epa.gov/otaq/models/moves/tools.htm), when available.

E-14
Exhibit E-12. Grams/Vehicle-Mile Emission Factors Calculated from MOVES
Output by Link, Quarter, and Hour

0.098 0.103 0.090

0.371 0.372 0.371
0.246 0.249 0.242
4 0.098 0.103 0.090
5 0.371 0.372 0.371
6 0.246 0.249 0.242
7 0.498 0.507 0.484
8 0.498 0.507 0.484
9 0.396 0.397 0.396
10 0.316 0.320 0.309
11 0.346 0.350 0.340
12 0.396 0.397 0.396
13 0.316 0.320 0.309
141 0.350t 0.340!
15 0 34: f
0.396 0.397 0.396
16 0.316 0.320 0.309
17 0.346 0.350 0.340
18 0.396 0.397 0.396
19 0.316 0.320 0.309
20 0.346 0.350

0.096 0.099 0.088

0.370 0.371 0.370
0.244 0.247 0.240
4 0.096 0.099 0.008
5 0.370 0.371 0.370
6 0.244 0.247 0.240
7 0.489 0.495 0.475
8 0.489 0.495 0.475
9 0.395 0.396 0.394
10 0.313 0.316 0.300
11 0.340 0.343 0.334
12 0.395 0.396 0.394
13 0.313 0.316 0.306
0.340 0.343 0.334
0.395 0.396 0.394
0.313 0.316 0.306
0.340 0.343 0.334
0.395 0.396 0.394
0.31 3 0.316 0.306

E-15
E.6 ESTIMATE EMISSIONS FROM ROAD DUST, CONSTRUCTION, AND
ADDITIONAL SOURCES (STEP 4)

E.6.1 Estimating re-entrained road dust

In this case, this area does not have any adequate or approved SIP budgets for either
PM2.5 NAAQS, and neither the EPA nor the state air agency have made a finding that
road dust emissions are a significant contributor to the air quality problem for either
PM2.5 NAAQS. Therefore, PM2.5 emissions from road dust do not need to be considered
in this analysis (see Sections 2.5.3 and 6.2).

E.6.2 Estimating transportation-related construction dust

The construction of this project will not occur during the analysis year. Therefore,
emissions from construction dust are not included in this analysis (see Sections 2.5.5 and
6.4).

E.6.3 Estimating additional sources of emissions in the project area

It is determined that the project area in the analysis year does not include locomotives or
other nearby emission sources that have to be considered in the air quality modeling (see
Section 6.6).

E.7 SELECT AN AIR QUALITY MODEL, DATA INPUTS, AND RECEPTORS

(STEP 5)

E.7.1 Characterizing emission sources

As discussed previously, the CAL3QHCR model is selected to estimate PM2.5

concentrations for this analysis (see Section 7.3). Each link is defined in CAL3QHCR
with coordinates and dimensions matching the project parameters (shown in Exhibit E-4).
The necessary inputs for link length, traffic volume, and corresponding link emission
factor are also added using the CAL3QHCR Tier II approach. Each MOVES emission
factor (12 a.m., 6 a.m., 12 p.m., and 6 p.m.) and traffic volume (average or peak) for each
link is applied to multiple hours of the day, as follows:
• Morning peak (AM) emissions based on traffic data and meteorology occurring
between 6 a.m. and 9 a.m.;
• Midday (MD) emissions based on data from 9 a.m. to 4 p.m.;
• Evening peak (PM) emissions based on data from 4 p.m. to 7 p.m.;
• Overnight (ON) emissions based on data from 7 p.m. to 6 a.m.

In addition, these factors are applied to each of the four quarters being modeled.

E-16
CAL3QHCR scenarios are built to model traffic conditions for all 24 hours of a weekday
in each quarter (partial elements of the CAL3QHCR input file can be found in Exhibits
E-13a and 13b, as file requires one to scroll down the screen): in all, four separate
scenarios.

Exhibit E-13a. CAL3QHCR Quarter 1, 6 a.m. Input File (Partial)

~ highway_jan.INP- Notepad GJ(QJ(g)

File Edit Help
Format View
'H01:-Sp01: Highway Example' 60. 175. 0. 0. 41 1 0 ~
1 1 98 12 31 98
94823 98 94823 98
1 1 , u'
'1'
'2'
-42.9
-28.8
-20
-3.1
1.8
1.8
d
'3' -16.5 29 1.8
'4' -16.5 48.8 1.8
'5' 29.6 31.8 1.8
'6' 12.7 -101 1.8
'7' -14.6 -100.1 1.8
'8' 15. 5 -152.9 1.8
'9' -14.6 -154.7 1.8
'10' -12.7 -220.7 1.8
'11' 17.4 -205.6 1.8
'12' -11.8 -265.9 1.8
'13' 21.2 -257.4 1.8
'14' 19.3 -334. 7 1.8
'15' 35.3 -333.7 1.8
'16' 34.4 -317.7 1.8
'17' -21.2 -395.9 1.8
'18' -23.1 -349.7 1.8
'19' 24 -18.1 1.8
'20' -31.6 15.8 1.8
'21' -3.3 -43 5. 5 1.8
'22' 12.7 -7.8 1.8
'23' 24 -379 1.8
'24' 28.7 -411 1.8
'2 5' 4 5. 7 -3 53. 5 1.8
'26' -9.9 -360.1 1.8
'27' -12.7 -320.5 1.8
'28' -42.9 -365.8 1.8
'29' -19.3 -21 1.8
'30' -22.2 -57.7 1.8
'31' 10.8 19.6 1.8
'32' 46.6 20. 5 1.8
'33' 10.8 52. 5 1.8
'34' 13.6 -32.3 1.8
'3 5' 48. 5 -6.8 1.8
'36' -7.1 -386.5 1.8
'37' -43.8 -394 1.8
'38' -6.2 -410 1.8
'39' 43.8 -387.4 1.8
'40' -33. 5 -38.9 1.8
'41' 33.4 -432.7 1. 81
,3
~ .::

E-17
Exhibit E-13b. CAL3QHCR Quarter 1, 6 a.m. Input File (Partial)

Section 7.5 of the guidance recommends that users run the air quality model for five
years of meteorological data when site-specific meteorology data is not available. Since
CAL3QHCR can only process one year of meteorological data for each run, each
quarterly scenario is run for five years of meteorological data for a total of 20 runs. 3

3
As explained in Section 7, AERMOD allows five years of meteorological data to be modeled in a single
run (see Section 7.5.3)

E-18
E.7.2 Incorporating meteorological data

A representative set of meteorology data, as well as an appropriate surface roughness, are

selected (see Section 7.5). The recommended five years of meteorological data are
obtained from a local airport for calendar years 1998-2002. A surface roughness of 175
cm is selected for the site, consistent with the recommendations made in the Section 7.

E.7.3 Placing receptors

Using the guidance given in Section 7.6, receptors are placed at appropriate locations
within the area substantially affected by the project (Exhibit E-14). 4 Note that this grid is
shown for illustrative purposes only; placement, location, and spacing of actual receptors
should follow the guidance in Section 7.6. Receptor heights are set at 1.8 meters.
Additionally, a background concentration of “0” is input into the model. Representative
background concentrations are added later (see Step 7).

CAL3QHCR is then run with five years of meteorological data (1998 through 2002) and
output is produced for all receptors for each of the five years of meteorological data.

Exhibit E-14. Receptor Locations for Air Quality Modeling

4
The number and arrangement of receptors used in this example are simplified for ease of explanation.

E-19
E.8 DETERMINE BACKGROUND CONCENTRATIONS FROM NEARBY AND
OTHER EMISSION SOURCES (STEP 6)

Through the interagency consultation process, a nearby upwind PM2.5 monitor that has
been collecting ambient data for both the annual and 24-hour PM2.5 NAAQS is
determined to be representative of the background air quality at the project location. The
most recent data set is used (in this case, calendar year 2008 through 2010) and average
24-hour PM2.5 values are taken in a four-day/three-day measurement interval. As
previously noted, no nearby sources needing to be included in the air quality model are
identified.

Note: This is a highly simplified situation for illustrative purposes; refer to Section 8 of
the guidance for additional considerations for how to most accurately reflect background
concentrations in a real-world scenario.

E-20
E.9 CALCULATE DESIGN VALUES AND DETERMINE CONFORMITY (STEP 7)

With both CAL3QHCR outputs and background concentrations now available, the
project sponsor can calculate the design values. For illustrative purposes, calculations for
a single receptor with the highest modeled concentrations for the build scenario are
shown in this example.5 In this step, the guidance from Sections 9.3.2 and 9.3.3 are used
to calculate design values from the modeled results and the background concentrations
for comparison with the annual and 24-hour PM2.5 NAAQS.

E.9.1 Determining conformity to the annual PM2.5 NAAQS

First, average background concentrations are determined for each year of monitored data
(shown in Exhibit E-15).

Exhibit E-15. Annual Average Background Concentration for Each Year

Annual Average
Monitoring
Background
Year
Concentration
2008 13.348
2009 12.785
2010 13.927

The three-year average background concentration is then calculated (see Exhibit E-16).

Exhibit E-16. Calculation of Annual Design Value (At Highest Receptor)

Annual Average Annual Average

Background Modeled Sum of
Concentration Concentration Background + Annual Design Value
(Three-year (Five-year Project
Average) Average)
13.353 1.580 14.933 14.9

5
In an actual PM hot-spot analysis, design values would be calculated at additional receptors as described
in Section 9.3.

E-21
To determine the annual PM2.5 design value, the annual average background
concentration is added to the five-year annual average modeled concentration (at the
receptor with the highest annual average concentration from the CAL3QHCR output).
This calculation is shown in Exhibit E-16. The sum (project + background) results in a
design value of 14.9 µg/m3. This value at the highest receptor is less than the 1997
annual PM2.5 NAAQS of 15.0 µg/m3. It can be assumed that all other receptors with
lower modeled concentrations will also have design values less than this NAAQS. In this
example it is unnecessary to determine appropriate receptors in the build scenario (per
Section 9.4 of the guidance) or develop a no-build scenario for the annual PM2.5 NAAQS,
since the build scenario demonstrates that the hot-spot analysis requirements in the
transportation conformity rule are met at all receptors.

E.9.2 Determining conformity to the 24-Hour PM2.5 NAAQS

The next step is to calculate a design value to compare with the 2006 24-hour PM2.5
NAAQS through a “Second Tier” analysis as described in Section 9.3.3. For ease of
explanation, this process has been divided into individual steps, consistent with the
guidance.

Step 7.1
The number of background measurements is counted for each year of monitored data
(2008 to 2010). Based on a 4-day/3-day measurement interval, the dataset has 104 values
per year.

Step 7.2
For each year of monitored concentrations, the eight highest daily background
concentrations for each quarter are determined, resulting in 32 values (4 quarters; 8
concentrations/quarter) for each year of data (shown in Exhibit E-17).

Step 7.3
Identify the highest-predicted modeled concentration resulting from the project in each
quarter, averaged across each year of meteorological data used for air quality modeling.
For illustrative purposes, the highest average concentration across five years of
meteorological data for a single receptor in each quarter is shown in Exhibit E-18. Note
that, in a real-world situation, this process would be repeated for all receptors in the build
scenario.

E-22
Exhibit E-17. Highest Daily Background Concentrations for Each Quarter and
Each Year

2008
Rank Q1 Q2 Q3 Q4
1 20.574 21.262 22.354 20.434
2 20.152 20.823 22.042 20.016
3 19.743 20.398 21.735 19.611
4 19.346 19.985 21.434 19.218
5 18.961 19.584 21.140 18.837
6 18.588 19.196 20.851 18.467
7 18.226 18.819 20.568 18.109
8 17.874 18.454 20.291 17.761
2009
Rank Q1 Q2 Q3 Q4
1 20.195 20.867 21.932 20.058
2 19.784 20.440 21.628 19.651
3 19.386 20.026 21.329 19.257
4 19.000 19.624 21.037 18.875
5 18.625 19.235 20.750 18.504
6 18.262 18.857 20.469 18.145
7 17.910 18.490 20.194 17.796
8 17.568 18.135 19.924 17.457
2010
Rank Q1 Q2 Q3 Q4
1 21.137 21.847 22.980 20.990
2 20.698 21.390 22.655 20.556
3 20.272 20.948 22.336 20.135
4 19.860 20.519 22.023 19.726
5 19.459 20.102 21.717 19.330
6 19.071 19.698 21.417 18.945
7 18.694 19.307 21.123 18.572
8 18.329 18.927 20.834 18.211

Exhibit E-18. Five-year Average 24-hour Modeled Concentrations for Each Quarter
(At Example Receptor)

Q1 Q2 Q3 Q4
Five Year Average
Maximum
10.42 10.62 10.74 10.61
Concentration (At
Example Receptor)

E-23
Step 7.4
The highest modeled concentration in each quarter (from Step 7.3) is added to each of the
eight highest monitored concentrations for the same quarter for each year of monitoring
data (from Step 7.2). As shown in Exhibit E-19, this step results in eight concentrations
in each of four quarters for a total of 32 values for each year of monitoring data. As
mentioned, this example analysis shows only a single receptor’s values, but project
sponsors should calculate design values at all receptors in the build scenario (see
Section 9.3 of the guidance).

Exhibit E-19. Sum of Background and Modeled Concentrations at Example

Receptor for Each Quarter

2008
Rank Q1 Q2 Q3 Q4
1 31.084 31.902 32.994 31.074
2 30.662 31.463 32.682 30.656
3 30.253 31.038 32.375 30.251
4 29.856 30.625 32.074 29.858
5 29.471 30.224 31.780 29.477
6 29.098 29.836 31.491 29.107
7 28.736 29.459 31.208 28.749
8 28.384 29.094 30.931 28.401
2009
Rank Q1 Q2 Q3 Q4
1 30.705 31.507 32.572 30.698
2 30.294 31.080 32.268 30.291
3 29.896 30.666 31.969 29.897
4 29.510 30.264 31.677 29.515
5 29.135 29.875 31.390 29.144
6 28.772 29.497 31.109 28.785
7 28.420 29.130 30.834 28.436
8 28.078 28.775 30.564 28.097
2010
Rank Q1 Q2 Q3 Q4
1 31.647 32.487 33.620 31.630
2 31.208 32.030 33.295 31.196
3 30.782 31.588 32.976 30.775
4 30.370 31.159 32.663 30.366
5 29.969 30.742 32.357 29.970
6 29.581 30.338 32.057 29.585
7 29.204 29.947 31.763 29.212
8 28.839 29.567 31.474 28.851

E-24
Step 7.5
As shown in Exhibit E-20, for each year of monitoring data, the 32 values from Step 7.4
are ordered together in a column and assigned a yearly rank for each value, from 1
(highest concentration) to 32 (lowest concentration).

Exhibit E-20. Ranking Sum of Background and Modeled Concentrations at

Example Receptor for Each Year of Background Data

Rank 2008 2009 2010

1 32.994 32.572 33.620
2 32.682 32.268 33.295
3 32.375 31.969 32.976
4 32.074 31.677 32.663
5 31.902 31.507 32.487
6 31.780 31.390 32.357
7 31.491 31.109 32.057
8 31.463 31.080 32.030
9 31.208 30.834 31.763
10 31.084 30.705 31.647
11 31.074 30.698 31.630
12 31.038 30.666 31.588
13 30.931 30.564 31.474
14 30.662 30.294 31.208
15 30.656 30.291 31.196
16 30.625 30.264 31.159
17 30.253 29.897 30.782
18 30.251 29.896 30.775
19 30.224 29.875 30.742
20 29.858 29.515 30.370
21 29.856 29.510 30.366
22 29.836 29.497 30.338
23 29.477 29.144 29.970
24 29.471 29.135 29.969
25 29.459 29.130 29.947
26 29.107 28.785 29.585
27 29.098 28.775 29.581
28 29.094 28.772 29.567
29 28.749 28.436 29.212
30 28.736 28.420 29.204
31 28.401 28.097 28.851
32 28.384 28.078 28.839

E-25
Step 7.6
For each year of monitoring data, the value with a rank that corresponds to the projected
98th percentile concentration is determined. As discussed in Section 9, an analysis
employing 101-150 background values for each year (as noted in Step 7.1, this analysis
uses 104 values per year) uses the 3rd highest rank to represent a 98th percentile. The 3rd
highest concentration (highlighted in Exhibit E-20) is referred to as the “projected 98th
percentile concentration.”

Step 7.7
Steps 7.1 through 7.6 are repeated to calculate a projected 98th percentile concentration at
each receptor based on each year of monitoring data and modeled concentrations.

Step 7.8
For the example receptor, the average of the three projected 98th percentile concentrations
(see Step 7.6) is calculated.

Step 7.9
The resulting value of 32.440 µg/m3 is then rounded to the nearest whole µg/m3, resulting
in a design value at the example receptor of 32 µg/m3. At each receptor this process
should be repeated. In the case of this analysis, the example receptor is the receptor with
the highest design value in the build scenario.

Step 7.10
The design values calculated at each receptor are compared to the NAAQS. In the case
of this example, the highest 24-hour design value (32 µg/m3) is less than the 2006 24-
hour PM2.5 NAAQS of 35 µg/m3. Since this is the design value at the highest receptor, it
can be assumed that the conformity requirements are met at all receptors in the build
scenario. Therefore, it is unnecessary for the project sponsor to calculate design values
for the no-build scenario for the 24-hour NAAQS.

E.10 CONSIDER MITIGATION AND CONTROL MEASURES (STEP 8)

In this case, the project is determined to conform. In situations when this is not the case,
it may be necessary to consider additional mitigation or control measures. If measures
are considered, additional air quality modeling would need to be completed and new
design values calculated to ensure that conformity requirements are met. See Section 10
for more information, including some specific measures that might be considered.

E.11 DOCUMENT THE PM HOT-SPOT ANALYSIS (STEP 9)

The final step is to properly document the PM hot-spot analysis in the conformity
determination (see Section 3.10).

E-26
 A ppendix F :
Example Quantitative PM Hot-spot Analysis of a Transit
Project using MOVES and AERMOD

F.1 INTRODUCTION

The purpose of this appendix is to demonstrate the procedures for completing a hot-spot
analysis using MOVES and AERMOD following the basic steps described in Section 3.
Readers should reference the appropriate sections in the guidance as needed for more
detail on how to complete each step of the analysis. This example is limited to showing
the build scenario; in practice, project sponsors may have to also analyze the no-build
scenario. While this example calculates emission rates using MOVES, EMFAC users
may find the air quality modeling described in this appendix helpful.

Note: The following example of a quantitative PM hot-spot analysis is highly simplified

and intended only to demonstrate the basic procedures described in the guidance. This
example uses default data in places where the use of project-specific data in a real-world
situation would be expected. In addition, actual PM hot-spot analyses could be
significantly more complex and are likely to require more documentation of data and
decisions. In this example, the interagency consultation process is used as needed for
evaluating and choosing models, methods, and assumptions, according to the
requirements of 40 CFR 93.105(c)(1)(i).

F.2 PROJECT DESCRIPTION AND CONTEXT

The proposed project is a new regionally significant bus terminal that would be created
by taking a downtown street segment one block in length and reserving it for bus use
only. It would be an open-air facility containing six “sawtooth” lanes where buses enter
to load and unload passengers. The terminal is designed to handle about 575 diesel buses
per day with up to 48 buses in the peak hour. The project is located in an area designated
nonattainment for the 2006 24-hour PM2.5 NAAQS and 1997 annual PM2.5 NAAQS.

The following is some additional pertinent data about the project:

• The proposed project is located in a medium-size city (within one county) in a
state other than California.
• The project is expected to take less than a year to complete and has an estimated
completion date of 2013. The year of peak emissions is expected to be 2015,
when considering the project’s emissions and background concentrations.
• The area surrounding the proposed project is primarily commercial, with no
nearby sources of PM2.5 that need to be modeled. This assumption is made to
simplify the example. In most cases, transit projects include parking lots with
emissions that would be considered in a PM hot-spot analysis.

F-1
 • The state does not have an adequate or approved SIP budget for either PM2.5
NAAQS, and neither the EPA nor the state air quality agency has made a finding
that road dust is a significant contributor to the PM2.5 nonattainment problem.

F.3 DETERMINE NEED FOR PM HOT-SPOT ANALYSIS (STEP 1)

The proposed project is determined to be of local air quality concern under the
conformity rule because it is a new bus terminal that has a significant number of diesel
vehicles congregating at a single location (see 40 CFR 93.123(b)(1)(iii) and Section 2.2
and Appendix B of the guidance). Therefore, a quantitative PM hot-spot analysis is
required.

F.4 DETERMINE APPROACH, MODELS, AND DATA (STEP 2)

F.4.1 Determining geographic area and emission sources to be covered by the analysis

First, the interagency consultation process is used to ensure that the project area is
defined so that the analysis includes the entire project, as required by 40 CFR
93.123(c)(2). As previously noted, it is also determined that, in this case, there are no
nearby emission sources to be modeled (see Section 3.3.2).

F.4.2 Deciding the general analysis approach and analysis year(s)

The project sponsor then determines that the preferred approach in this case is to model
the build scenario first, completing a no-build scenario only if necessary.

The year of peak emissions (within the timeframe of the current transportation plan) is
determined to be 2015. Therefore, 2015 is selected as the year of the analysis, and the
analysis will consider traffic data from 2015 (see Section 3.3.3).

F.4.3 Determining the PM NAAQS to be evaluated

Because the area has been designated nonattainment for both the 2006 24-hour PM2.5
NAAQS and 1997 annual PM2.5 NAAQS, the results of the analysis will be compared to
both NAAQS (see Section 3.3.4). All four quarters are included in the analysis in order
to estimate a year’s worth of emissions for both NAAQS.

F-2
F.4.4 Deciding on the type of PM emissions to be modeled

Next, the following directly-emitted PM emissions are determined to be relevant for

estimating the emissions in the analysis (see Section 2.5):
• Vehicle exhaust 1
• Brake wear
• Tire wear

F.4.5 Determining the models and methods to be used

Since this project will be located outside of California, MOVES is used for emissions
modeling. In addition, it is determined that, since this is a terminal project, the
appropriate air quality model to use would be AERMOD (see Section 3.3.6). Making the
decision on what air quality model to use at this stage is important so that the appropriate
data are collected, among other reasons (see next step).

F.4.6 Obtaining project-specific data

Finally, having selected a model and a general modeling approach, the project sponsor
compiles the data required to use MOVES, including project traffic data, vehicle types
and age, and temperature and humidity data for the months and hours to be modeled
(specifics on the data collected are described in the following steps). In addition,
information required to use AERMOD to model air quality is gathered, including
meteorological data and information on representative air quality monitors. The sponsor
ensures the latest planning assumptions are used and that data used for the analysis are
consistent with that used in the latest regional emissions analysis, as required by the
conformity rule (see Section 3.3.7).

F.5 ESTIMATE ON-ROAD MOTOR VEHICLE EMISSIONS (STEP 3)

Having completed the analysis preparations described above, the project sponsor then
follows the instructions provided in Section 4 of the guidance to use MOVES to estimate
the on-road emissions from this terminal project:

F.5.1 Characterizing the project in terms of links

Using the guidance described in Section 4.2, a series of links are defined in order to
accurately capture the activity at the proposed terminal. As shown in Exhibit F-1, two
one-way running links north and south of the facility (“Link 1” and “Link 2”) are defined
to describe buses entering and exiting the terminal. A third running/idle link (shown as
“Link 3” to the north of the facility), is used to describe vehicles idling at the signalized

1
Represented in MOVES as PMtotal running, PMtotal crankcase running, PMtotal ext. idle, and PMtotal crankcase ext. idle.

F-3
light before exiting the facility. Links 4 through 9 represented bus bays where buses
drop-off and pick-up passengers; these are referred to as the terminal links. 2

Exhibit F-1. Diagram of Proposed Bus Terminal Showing Links

The running links have the following dimensions:

Link 1: 200 feet long by 24 feet wide
Link 2: 160 feet long by 24 feet wide
Link 3: 40 feet long by 24 feet wide

Additionally, the dimensions of the six terminal links (Links 4 through 9) are 60 feet long
by 12 feet wide. These links are oriented diagonally from southwest to northeast. The
queue link (Link 3) is defined with a length of 40 feet, based on the average length of a
transit bus.

After identifying and defining the links, traffic conditions are estimated for the project in
the analysis year of 2015. The terminal was presumed to be in operation all hours of the
year. Based on expected terminal operations, the anticipated future traffic volumes are
available for each hour of an average weekday (see Exhibit F-2). To simplify the
analysis, the sponsor conservatively assumes weekday traffic for all days of the year,
even though the operating plan calls for reduced service on weekends. 3 Identical traffic
volume and activity profiles are assumed for all quarters of the year. Quarters are defined
for this analysis as described in Section 3.3.4 of the guidance: Q1 (January-March), Q2
(April-June), Q3 (July-September), and Q4 (October-December).

2
The project area in this example is not realistic and has been simplified for demonstration purposes.
Analyses of transit facilities will likely include inbound and outbound links beyond what is described in
this simplified example, as well as the surrounding area.
3
This decision is made to save time and effort, as it would result in the need for fewer modeling runs.
More accurate results would be obtained by treating weekends differently and modeling them using the
actual estimated Saturday and Sunday traffic.

F-4
Exhibit F-2. Average Weekday Bus Trips through Transit Terminal for Each Hour

Hour Bus Trips

12am - 1am 7
1am - 2am 6
2am - 3am 6
3am - 4am 6
4am - 5am 7
5am - 6am 9
6am - 7am 27
7am - 8am 48
8am - 9am 39
9am - 10am 29
10am - 11am 26
11am - 12pm 28
12pm - 1pm 30
1pm - 2pm 31
2pm - 3pm 31
3pm - 4pm 39
4pm - 5pm 44
5pm - 6pm 42
6pm - 7pm 26
7pm - 8pm 21
8pm - 9pm 22
9pm - 10pm 17
10pm - 11pm 13
11pm - 12am 10

F.5.2 Deciding on how to handle link activity

As discussed in Section 4.2 of the guidance, MOVES offers several options for users to
apply activity information to each LinkID. For illustrative purposes, based on the
available information for the project (average speed, hourly bus volume, idle time, and
fraction of vehicles encountering a red light) several methods of deriving Op-Mode
distributions are employed in this example, as described below.
• Links 1 and 2 represent buses driving at an average of 5 mph through the
terminal, entering and exiting the bus bays. An average speed of 5 mph is entered
into the MOVES “links” input, which calculates an Op-Mode distribution to
reflect the MOVES default 5 mph driving pattern.
• The queue link (Link 3) is given an Op-Mode distribution that represents buses
decelerating, idling, and accelerating (red light) as well as cruising through (green
light). First, an Op-Mode distribution is calculated for the link average speed (5
mph). Because this does not adequately account for idling at the intersection, the
Op-Mode fractions are re-allocated to add in 50% idling (determined after
consulting the Highway Capacity Manual 2000 to approximate idle time in an
under-capacity scenario) reflecting 50% of buses encountering a red light. A

F-5
 fraction of 0.5 for Op-Mode “1” is added to the re-allocated 5 mph average speed
Op-Mode distribution. The resulting Op-Mode distribution represents all activity
on a queuing intersection link.
• The bus bays (Links 4 through 9) are represented by a single link (modeled in
MOVES as “LinkID 4”) and activity is defined in the Links table by an average
speed of “0”, representing exclusively idle activity.

F.5.3 Determining the number of MOVES runs

Following the guidance given in Section 4.3, it is determined that 16 MOVES runs
should be completed to produce emission factors that show variation across four hourly
periods (12 a.m., 6 a.m., 12 p.m., and 6 p.m., corresponding to overnight, morning,
midday, and evening traffic scenarios, respectively) and four quarterly periods
(represented by the months of January, April, July, and October; see Section 3.3).
MOVES would calculate values for all project links for the time period specified in each
run. Although traffic data is available for 24 hours, the emission factors produced from
the 16 scenarios would be post-processed into grams/vehicle-hour and further converted
to grams/hour emission factors that vary based on the hour-specific vehicle count. This
methodology avoids running 24 hourly scenarios for four quarters (96 runs). Grams/hour
emissions rates are required to use AERMOD.

F.5.4 Developing basic run specification inputs

When configuring MOVES for the analysis, the project sponsor follows Section 4.4 of
the guidance, including, but not limited to, the following:
• From the Scale menu, selecting the “Project” domain; in addition, choosing
output in “Inventory” so that total emission results are produced for each link,
which is equivalent to a grams/hour/link emission factor needed by AERMOD
(see Section 4.4.2).
• From the Time Spans panel, the appropriate year, month, day, and hour for each
run is selected (see Section 4.4.3).
• From the Geographic Bounds panel, the custom domain is selected (see Section
4.4.4).
• From the Vehicles/Equipment panel, Diesel Transit Buses are selected (see
Section 4.4.5).
• From the Road Types panel, the Urban Unrestricted road type is selected (see
Section 4.4.6).
• From the Pollutants and Processes panel, appropriate pollutant/processes are
selected according to Section 4.4.7 of the guidance for “highway links.”
• In the Output panel, an output database is specified with grams and miles selected
as units (see Section 4.4.10).

F-6
F.5.5 Entering project details using the Project Data Manager

Meteorology

As described previously, it is determined that MOVES should be run 16 times to reflect

the following scenarios: 12 a.m., 6 a.m., 12 p.m., and 6 p.m. (corresponding to overnight,
morning, midday, and evening traffic scenarios, respectively) for the months of January,
April, July, and October. Temperature and humidity data from a representative
meteorological monitoring station are obtained and confirmed to be consistent with data
used in the regional emissions analysis from the currently conforming transportation plan
and TIP (see Section 4.5.1). Average values for each hour and month combination are
used for each of the 16 MOVES runs. As an example, temperature and humidity values
for 12 a.m. January are shown in Exhibit F-3.

Exhibit F-3. Temperature and Humidity Input (January 6 a.m.)

F-7
Age Distribution

Section 4.5.2 of the guidance specifies that default data should be used only if an
alternative local dataset cannot be obtained and the regional conformity analysis relies on
national defaults. However, for the sake of simplicity only, in this example the national
default age distribution for 2015 is used for all vehicles and all runs (see Exhibit F-4). As
discussed in the guidance, transit agencies should be able to provide a fleet-specific age
distribution, and the use of fleet-specific data is always recommended (and would be
expected in a real-world scenario) because emission factors vary significantly depending
on the age of the fleet.

Exhibit F-4. Age Distribution Table (Partial)

F-8
Fuel Supply and Fuel Formulation

An appropriate fuel supply and formulation is selected to match the project area’s diesel
use. In MOVES, diesel fuel formulation is constant across all quarters, so one fuel
supply/fuel formulation combination is used for all MOVES runs. Also, it is known that
all transit buses would use the same diesel fuel, so a fraction of 1 is entered for fuel
20011 (ultra-low-sulfur diesel fuel) in the Fuel Supply Table. In the case of this example,
the default fuel supply/formulation matches the actual fuel supply/formulation, so it is
therefore appropriate to use the default in the analysis (see Exhibits F-5 and F-6).

Exhibit F-5. Fuel Supply Table

Exhibit F-6. Fuel Formulation Table

F-9
Inspection and Maintenance (I/M)

As there is no PM emissions benefit in MOVES for I/M programs, this menu item is
skipped (see Section 4.5.4).

Link Source Type

The distribution of vehicle types on each link is defined in the Link Source Type table
following the guidance in Section 4.5.5. Given that the project will be a dedicated transit
bus terminal this analysis assumes only transit buses are operating on all links.
Therefore, a fraction of 1 is entered for Source Type 42 (Transit Buses) for each LinkID
indicating 100% of vehicles using the project are transit buses (see Exhibit F-7).

Exhibit F-7. Link Source Type Table

F-10
Links

The Links table (see Exhibit F-8) is populated with parameters for the four defined links
of the bus terminal: three running links (Links 1-3) and one idle link (representing the
terminal links). The link length is entered in terms of miles for each link. The road type
for the four links is classified as “5” (Urban Unrestricted). The entrance and exit links
(Links 1 and 2) are given an average speed of 5 mph. The queue link (Link 3) is given an
average speed of 2.5 mph, representing 50% of the vehicle operating hours in idling
mode and 50% operating hours traveling at 5 mph. Although MOVES is capable of
calculating emissions from an average speed (as is done for Links 1 and 2), the specific
activity on Link 3 is directly entered with an Op-Mode distribution. LinkID 4 is given a
link average speed of “0” mph, which indicates entirely idle operation. Link volume
(which represents the number of buses per hour) is entered for each link; however, since
the goal of the analysis is to produce an estimate in grams/vehicle-hour, the volume (i.e.,
the number of vehicles) will be divided out during post-processing. Also, because link
volume is arbitrary, the Links table shown in Exhibit F-8 can be used for all 16 MOVES
runs.

Exhibit F-8. Links Table

Describing Vehicle Activity

MOVES can capture details about vehicle activity in a number of ways. In this case, it is
decided to use average speeds for Links 1, 2, and 4 and a detailed Op-Mode distribution
for Link 3 (see Section 4.5.7).

Op-Mode distributions for Links 1 and 2 are calculated within MOVES based on a 5 mph
average speed. The MOVES model calculates a default Op-Mode distribution based on
average speed and road type (for these links, 5 mph on road type 5). Link 3 is given a
unique Op-Mode distribution to better simulate the queuing and idling that occurs prior to
buses exiting the facility at a traffic signal. The sponsor estimates that 50% of buses

F-11
would idle at a red light before exiting the facility, so the idling operation (OpMode ID 1)
is assumed to be 0.5 for Link 3. The remaining 50% is re-allocated based on the default 5
mph Op-Mode distribution calculated for Links 1 and 2 (which includes acceleration,
deceleration, and cruise operating modes). This process requires an additional MOVES
run to extract the default 5 mph Op-Mode distribution from the MOVES execution
database. By selecting “save data” for the “Operating Mode Distribution Generator
(Running OMDG)” under the MOVES “Advanced Performance Features” panel, the Op-
Mode distributions generated for 5 mph on an urban unrestricted road type are saved in
the MOVES execution database in the MySQL table “opmodedistribution.” The Op-
Mode distribution used in the analysis for Link 3 is partially shown in Exhibit F-9.

Exhibit F-9. Link 3 (Queue Link) Op-Mode Distribution Input Table (Partial)

F-12
Off-Network

As it is assumed that there are no off-network links (such as parking lots or truck stops)
that would have to be considered using the Off-Network Importer (bus idling at the
terminal is captured by the terminal links), there is no need to use this option in this
example. As noted earlier, this assumption is made to simplify the example. Most transit
projects would include rider parking lots and should include these emissions in a PM hot-
spot analysis.

F.5.6 Generating emission factors for use in air quality modeling

After generating the run specification and entering the required information into the
Project Data Manager as described above, MOVES is run 17 times: 16 runs (four hours
of the day for four quarters of the year) plus an initial run to generate the Op-Mode
distribution for 5 mph as discussed earlier. Upon completion of each run, the MOVES
output is located in the MySQL output database table “movesoutput” and sorted by
Month, Hour, LinkID, ProcessID, and PollutantID. An aggregate PM2.5 emission factor
is then calculated by the project sponsor for each Month, Hour, and LinkID combination
using the following equation and the guidance given in Section 4.4.7 of the guidance: 4

PMaggregate total = (PMtotal running) + (PMtotal crankcase running) + (brake wear) + (tire wear)

For each link, the total emissions are divided by the number of vehicles on each link (as
reported in the “movesactivityoutput” table ActivitytypeID = 6) to produce a
grams/vehicle-hour value. This value is then multiplied by the number of buses on each
link, for each of the 24 hours where data are available (see Exhibit F-2).

The emission factor (grams/vehicle-hour) for LinkID 4 (links 4 through 9) is converted

into grams per vehicle-minute, and then multiplied by the total idle time for each unique
hour. For instance, the hour from 5 p.m. to 6 p.m. has a volume of 42 buses per hour (7
buses per bus bay). If each bus is expected to idle for 60 seconds each hour, the total idle
time for each bus bay for that hour would be 7 minutes per hour. If MOVES calculated a
PM emission factor of 2.0 grams per vehicle-minute, the emission factor for each bus bay
link under this scenario would be 14.0 grams/hour.

To account for temperature changes throughout the day, emission factors are evenly
paired with corresponding traffic volumes (six hours per period):
• 6 a.m. results – traffic data from 3 a.m. to 9 a.m.
• 12 p.m. results – traffic data from 9 a.m. to 3 p.m.
• 6 p.m. results – traffic data from 3 p.m. to 9 p.m.
• 12 p.m. results – traffic data from 9 p.m. to 3 a.m.

The emission factor results for each quarter are similarly paired with traffic volumes.

4
EPA is considering creating one or more MOVES scripts that would automate the summing of aggregate
emissions when completing project-level analyses. These scripts would be made available for download on
the MOVES website (www.epa.gov/otaq/models/moves/tools.htm), when available.

F-13
The 96 resulting grams/hour emission factors (24 hours each for four quarters) for each
link are then ready to be used as an input to the AERMOD dispersion model to predict
PM2.5 concentrations.

F.6 ESTIMATE EMISSIONS FROM ROAD DUST, CONSTRUCTION AND

ADDITIONAL SOURCES (STEP 4)

F.6.1 Estimating re-entrained road dust

In this case, this area does not have any adequate or approved SIP budgets for either
PM2.5 NAAQS, and neither the EPA nor the state air agency has made a finding that road
dust emissions are a significant contributor to the air quality problem for either PM2.5
NAAQS. Therefore, PM2.5 emissions from road dust do not need to be considered in this
analysis (see Sections 2.5.3 and 6.2).

F.6.2 Estimating transportation-related construction dust

The construction of this project will not occur during the analysis year. Therefore,
emissions from construction dust are not included in this analysis (see Sections 2.5.5 and
6.4).

F.6.3 Estimating additional sources of emissions in the project area

Through interagency consultation, it is determined that the project area in the analysis
year does not include locomotives or other nearby emissions sources that would have to
be included in the air quality modeling (see Section 6.6).

F.7 SELECT AN AIR QUALITY MODEL, DATA INPUTS, AND RECEPTORS

(STEP 5)

F.7.1 Characterizing emission sources

Because this is a transit terminal project, EPA’s AERMOD model is determined to be the
appropriate dispersion model to use for this analysis (see Section 7.3). AERMOD is run
to estimate PM2.5 concentrations in and around the bus terminal project. Each link is
represented in AERMOD as an “Area Source” with dimensions matching the project
description (see Exhibit F-1). The emission release height is set to three meters, the
approximate exhaust height of most transit buses.

Emission factors generated from the MOVES runs are added to the AERMOD input file
(see Exhibit F-10). For this analysis, emissions vary significantly from hour to hour due
to fluctuating bus volumes as well as from daily and quarterly temperature effects.
Adjustment factors (EMISFACT) are used to model these hourly and quarterly variations

F-14
in emission factors. Refer to Appendix J for additional detail on using hour-by-hour
emission differences in an AERMOD input file.

Exhibit F-10. AERMOD Input File (Partial) with Seasonal (Quarterly) and Hourly
Adjustment Factors (Circled)

F.7.2 Incorporating meteorological data

A representative set of meteorology data, as well as an appropriate surface roughness are

selected (see Section 7.5). The recommended five years of meteorological data is
obtained from a local airport for calendar years 1998-2002. The appropriate surface
roughness is set at 1 meter, consistent with the recommendations made in the AERMOD
Implementation Guide.

F-15
F.7.3 Placing receptors

Using the guidance given in Section 7.6, receptors are placed at appropriate locations
within the area substantially affected by the project (see Exhibit F-11). 5 It is determined
in this instance to locate receptors around the perimeter of the project in increments of
five meters as well as within the passenger loading areas adjacent to the bus bays.
Receptor heights are set at 1.8 meters. A background concentration of “0” is input into
the model. Representative background concentrations are added at a later step (see Step
7).

AERMOD is run using five years of meteorological data and output produced for all
receptors for each of the five years of meteorological data.

Exhibit F-11. Area Source and Receptor Locations for Air Quality Modeling

5
The number and arrangement of receptors used in this example are simplified for ease of explanation;
real-world projects could expect to see more receptors and include the surrounding area.

F-16
F.8 DETERMINE BACKGROUND CONCENTRATIONS FROM NEARBY AND
OTHER EMISSION SOURCES (STEP 6)

Through the interagency consultation process, a nearby upwind PM2.5 monitor that has
been collecting ambient data for both the annual and 24-hour PM2.5 NAAQS is
determined to be representative of the background air quality at the project location (see
Exhibit F-12). The most recent data set is used (in this case, calendar year 2008 through
2010) and average 24-hour PM2.5 values are provided in a four-day/three-day
measurement interval. As previously noted, no nearby sources requiring modeling are
identified.

Note: This is a highly simplified situation for illustrative purposes; refer to Section 8 of
the guidance for additional considerations for how to most accurately reflect background
concentrations in a real-world scenario.

Exhibit F-12. PM2.5 Monitor Data from a Representative Nearby Site (Partial)

F-17
F.9 CALCULATE DESIGN VALUES AND DETERMINE CONFORMITY (STEP 7)

With both MOVES outputs and background concentrations now available, the project
sponsor can calculate the design values. For illustrative purposes, calculations for a
single receptor with the highest modeled concentrations for the build scenario are shown
in this example. 6 In Step 7, the guidance from Sections 9.3.2 and 9.3.3 are used to
calculate design values from the modeled results and the background concentrations for
comparison with the annual and 24-hour PM2.5 NAAQS.

F.9.1 Determining conformity to the annual PM2.5 NAAQS

First, average background concentrations are determined for each year of monitored data
(shown in Exhibit F-13). The three-year average background concentration is then
calculated (see Exhibit F-14).

Exhibit F-13. Annual Average Background Concentration for Each Year

Annual Average
Monitoring Background
Year Concentration
2008 13.348
2009 12.785
2010 13.927
Annual
Average 13.353

Exhibit F-14. Calculation of Annual Design Value (At Highest Receptor)

Annual Average Annual Average

Background Modeled Sum of
Concentration Concentration Background + Annual Design Value
(Three-year (Five-year Project
Average) Average)
13.353 1.423 14.776 14.8

6
In an actual PM hot-spot analysis, design values would be calculated at additional receptors as described
in Section 9.3.

F-18
To determine the annual PM2.5 design value, the annual average background
concentration is added to the five-year annual average modeled concentration (at the
receptor with the highest annual average concentration from the AERMOD output). This
calculation is shown in Exhibit F-14. The sum (project + background) results in a design
value of 14.8 µg/m3. This value at the highest receptor is less than the 1997 annual PM2.5
NAAQS of 15.0 µg/m3. It can be assumed that all other receptors with lower modeled
concentrations will also have design values less than the 1997 annual PM2.5 NAAQS. In
this example it is unnecessary to determine appropriate receptors in the build scenario or
develop a no-build scenario for the annual PM2.5 NAAQS, since the build scenario
demonstrates that the hot-spot analysis requirements in the transportation conformity rule
are met at all receptors.

F.9.2 Determining conformity to the 24-Hour PM2.5 NAAQS

The next step is to calculate a design value to compare with the 2006 24-hour PM2.5
NAAQS through a “Second Tier” analysis as described in Section 9.3.3. For ease of
explanation, this process has been divided into individual steps, consistent with the
guidance.

Step 7.1
The number of background measurements is counted for each year of monitored data
(2008 to 2010). Based on a 4-day/3-day measurement interval, the dataset has 104 values
per year.

Step 7.2
For each year of monitored concentrations, the eight highest daily background
concentrations for each quarter are determined, resulting in 32 values (4 quarters; 8
concentrations/quarter) for each year of data (shown in Exhibit F-15).

Step 7.3
Identify the highest modeled 24-hour concentration in each quarter, averaged across each
year of meteorological data. For illustrative purposes, the highest average concentration
across five years of meteorological data for a single receptor in each quarter is shown in
Exhibit F-16. Note that, in a real-world situation, this process would be repeated for all
receptors in the build scenario.

F-19
Exhibit F-15. Highest Daily Background Concentrations for Each Quarter and
Each Year

2008
Rank Q1 Q2 Q3 Q4
1 20.574 21.262 22.354 20.434
2 20.152 20.823 22.042 20.016
3 19.743 20.398 21.735 19.611
4 19.346 19.985 21.434 19.218
5 18.961 19.584 21.140 18.837
6 18.588 19.196 20.851 18.467
7 18.226 18.819 20.568 18.109
8 17.874 18.454 20.291 17.761
2009
Rank Q1 Q2 Q3 Q4
1 20.195 20.867 21.932 20.058
2 19.784 20.440 21.628 19.651
3 19.386 20.026 21.329 19.257
4 19.000 19.624 21.037 18.875
5 18.625 19.235 20.750 18.504
6 18.262 18.857 20.469 18.145
7 17.910 18.490 20.194 17.796
8 17.568 18.135 19.924 17.457
2010
Rank Q1 Q2 Q3 Q4
1 21.137 21.847 22.980 20.990
2 20.698 21.390 22.655 20.556
3 20.272 20.948 22.336 20.135
4 19.860 20.519 22.023 19.726
5 19.459 20.102 21.717 19.330
6 19.071 19.698 21.417 18.945
7 18.694 19.307 21.123 18.572
8 18.329 18.927 20.834 18.211

Exhibit F-16. Five-year Average of Highest Modeled Concentrations for Each

Quarter (At Example Receptor)

Q1 Q2 Q3 Q4
Five Year Average
Maximum
6.51 6.64 6.71 6.63
Concentration (At
Example Receptor)

F-20
Step 7.4
The highest modeled concentration in each quarter (from Step 7.3) is added to each of the
eight highest monitored concentrations for the same quarter for each year of monitoring
data (from Step 7.2). As shown in Exhibit F-17, this step results in eight concentrations
in each of four quarters for a total of 32 values for each year of monitoring data. As
mentioned, this example analysis shows only a single receptor’s values, but project
sponsors should calculate design values at all receptors in the build scenario.

Exhibit F-17. Sum of Background and Modeled Concentrations at Example

Receptor for Each Quarter

2008
Q1 Q2 Q3 Q4
1 27.088 27.901 29.063 26.948
2 26.667 27.462 28.750 26.530
3 26.258 27.037 28.443 26.125
4 25.861 26.624 28.143 25.732
5 25.476 26.224 27.848 25.351
6 25.102 25.835 27.560 24.982
7 24.740 25.459 27.277 24.623
8 24.389 25.093 27.000 24.275
2009
Q1 Q2 Q3 Q4
1 26.709 27.506 28.641 26.572
2 26.298 27.079 28.336 26.166
3 25.900 26.665 28.038 25.772
4 25.514 26.264 27.745 25.389
5 25.140 25.874 27.459 25.019
6 24.776 25.496 27.178 24.659
7 24.424 25.130 26.903 24.310
8 24.082 24.774 26.633 23.971
2010
Q1 Q2 Q3 Q4
1 27.651 28.486 29.689 27.505
2 27.212 28.030 29.363 27.070
3 26.787 27.587 29.044 26.649
4 26.374 27.158 28.732 26.240
5 25.974 26.742 28.426 25.844
6 25.585 26.338 28.125 25.460
7 25.209 25.946 27.831 25.087
8 24.843 25.566 27.543 24.725

F-21
Step 7.5
As shown in Exhibit F-18, for each year of monitoring data, the 32 values from Step 7.4
are ordered together in a column and assigned a yearly rank for each value, from 1
(highest concentration) to 32 (lowest concentration).

Exhibit F-18. Ranking Sum of Background and Modeled Concentrations at

Example Receptor for Each Year of Background Data

Rank 2008 2009 2010

1 29.063 28.641 29.689
2 28.750 28.336 29.363
3 28.443 28.038 29.044
4 28.143 27.745 28.732
5 27.901 27.506 28.486
6 27.848 27.459 28.426
7 27.560 27.178 28.125
8 27.462 27.079 28.030
9 27.277 26.903 27.831
10 27.088 26.709 27.651
11 27.037 26.665 27.587
12 27.000 26.633 27.543
13 26.948 26.572 27.505
14 26.667 26.298 27.212
15 26.624 26.264 27.158
16 26.530 26.166 27.070
17 26.258 25.900 26.787
18 26.224 25.874 26.742
19 26.125 25.772 26.649
20 25.861 25.514 26.374
21 25.835 25.496 26.338
22 25.732 25.389 26.240
23 25.476 25.140 25.974
24 25.459 25.130 25.946
25 25.351 25.019 25.844
26 25.102 24.776 25.585
27 25.093 24.774 25.566
28 24.982 24.659 25.460
29 24.740 24.424 25.209
30 24.623 24.310 25.087
31 24.389 24.082 24.843
32 24.275 23.971 24.725

F-22
Step 7.6
For each year of monitoring data, the value with a rank that corresponds to the projected
98th percentile concentration is determined. As discussed in Section 9, an analysis
employing 101-150 background values for each year (as noted in Step 7.1, this analysis
uses 104 values per year) uses the 3rd highest rank to represent a 98th percentile. The 3rd
highest concentration (highlighted in Exhibit F-18) is referred to as the “projected 98th
percentile concentration.”

Step 7.7
Steps 7.1 through 7.6 are repeated to calculate a projected 98th percentile concentration at
each receptor based on each year of monitoring data and modeled concentrations.

Step 7.8
For the example receptor, the average of the three projected 98th percentile concentrations
(highlighted in Exhibit F-18) is calculated.

Step 7.9
The resulting value of 28.508 µg/m3 is then rounded to the nearest whole µg/m3 resulting
in a design value at the example receptor of 29 µg/m3. At each receptor this process
should be repeated. However, in the case of this analysis, the example receptor is the
receptor with the highest design value in the build scenario.

Step 7.10
The design values calculated at each receptor are compared to the NAAQS. In the case
of this example, the highest 24-hour design value (29 µg/m3) is less than the 2006 24-
hour PM2.5 NAAQS of 35 µg/m3. Since this is the design value at the highest receptor, it
can be assumed that the conformity requirements are met at all receptors in the build
scenario. Therefore, it is unnecessary for the project sponsor to calculate design values
for the no-build scenario for the 24-hour PM2.5 NAAQS.

F.10 CONSIDER MITIGATION AND CONTROL MEASURES (STEP 8)

In this case, the project is determined to conform. In situations when this is not the case,
it may be necessary to consider additional mitigation or control measures. If measures
are considered, additional air quality modeling would need to be completed and new
design values calculated to ensure that conformity requirements are met. See Section 10
for more information, including some specific measures that might be considered.

F.11 DOCUMENT THE PM HOT-SPOT ANALYSIS (STEP 9)

The final step is to properly document the PM hot-spot analysis in the conformity
determination (see Section 3.10).

F-23
 A ppendix G :
Example of Using EMFAC for a Highway Project

G.1 INTRODUCTION

The purpose of this appendix is to demonstrate the procedures described in Section 5 of

the guidance on using EMFAC2007 to generate emission factors for air quality modeling.
The following example, based on a hypothetical highway project, illustrates the modeling
steps required for users to change EMFAC’s default VMT distribution and to develop
project-specific PM running exhaust emission factors. This example uses the “Emfac”
mode in EMFAC2007 (v2.3) to generate gram per mile (g/mi) emission factors stored in
the “Summary Rate” output file (.rts file) suitable for use in an air quality model. Users
will be able to generate running emission factors in a single EMFAC model run; multiple
calendar years can also be handled within one model run. As described in the main body
of this section, each run will be specific to either PM10 or PM2.5; however, this example is
applicable to both. This example does not include the subsequent air quality modeling;
refer to Appendix E for an example of how to run an air quality model for a highway
project for PM hot-spot analyses.

G.2 PROJECT CHARACTERISTICS

The hypothetical highway project is located in Sacramento County, California. For

illustrative purposes, the project is characterized by a single link with an average link
travel speed for all traffic equal to 65 mph. 1 The project’s first full year of operation is
assumed to be the year 2013. Through the interagency consultation process, it is
determined that 2015 should be the analysis year (based on the project’s emissions and
background concentrations). The build scenario 2015 traffic data for this highway
project shows that 25% of the total project VMT is from trucks and 75% from non-trucks.

1
These are simplified data to illustrate EMFAC’s use; this example does not, for instance, separate data by
peak vs. off-peak periods, divide the project into separate links, or consider additional analysis years, all of
which would likely be required for an actual project.

G-1
G.3 PREPARING EMFAC BASIC INPUTS

Based on the project characteristics, it is first necessary to specify the basic inputs and
default settings in EMFAC (see Exhibit G-1).

Exhibit G-1. Basic Inputs in EMFAC for the Hypothetical Highway Project

Step Input Category Input Data Note

1 Geographic Area County  Sacramento Select from drop-down list

Use Average Default (not shown in the EMFAC
Calculation Method
user interface)
2 Calendar Years 2015 Select from drop-down list
3 Season or Month Annual Select from drop-down list
4 Scenario Title Use default Define default title in the EMFAC
user interface
5 Model Years Use default Include all model years
6 Vehicle Classes Use default Include all vehicle classes
7 I/M Program Schedule Use default Include all pre-defined I/M program
parameters
8 Temperature 60F Delete all default temperature bins
and input 60
9 Relative Humidity 70%RH Delete all default relative humidity
bins and input 70
10 Speed Use default Include all speed bins from 5 mph to
65 mph
11 Emfac Rate Files Summary Rates (RTS) Select from EMFAC user interface
12 Output Particulate PM10 Select from EMFAC user interface

G-2
G.4 EDITING EMFAC DEFAULT VMT DISTRIBUTIONS

The next step is to calculate the EMFAC defaults for trucks and non-trucks. As shown in
Exhibit G-2, EMFAC’s 13 vehicle classes are grouped into trucks and non-trucks to
match the project-specific traffic data. Specifically, Light-Duty Autos, Light-Duty
Trucks (T1 and T2) and Motorcycles are grouped together to represent the “non-truck”
class. All other vehicle classes (Medium-Duty Trucks, Light HD Trucks (T4 and T5),
Medium HD Trucks, Heavy HD Trucks, Other Buses, Urban Buses, School Buses, and
Motor Homes) are classified as “trucks.” The total pre-populated VMT for truck and
non-truck for this highway project are 6,269,545 miles and 26,134,922 miles,
respectively.

Exhibit G-2. Example Highway Project Pre-Populated VMT for 13 Default Vehicle
Classes

EMFAC Vehicle Class EMFAC default VMT

01 - Light-Duty Autos (PC) 15,271,757
02 - Light-Duty Trucks (T1) 3,340,492
03 - Light-Duty Trucks (T2) 7,266,306
04 – Medium-Duty Trucks (T3)* 3,535,454
05 - Light HD Trucks (T4)* 816,278
06 - Light HD Trucks (T5)* 302,809
07 – Medium HD Trucks (T6)* 698,543
08 - Heavy HD Trucks (T7)* 704,156
09 - Other Buses* 49,590
10 - Urban Buses* 40,198
11 – Motorcycles 256,367
12 - School Buses* 31,176
13 - Motor Homes* 91,341
Truck VMT 6,269,545
Non-truck VMT 26,134,922
TOTAL 32,404,467
* Classified as trucks to match project-specific data

The next step is to calculate percentage VMT for trucks and non-trucks and their
respective adjustment factors to match project-specific VMT distributions, as shown in
Exhibit G-3. The default VMT percentages for trucks (19%) and non-trucks (81%) are
much different from what the project traffic data suggest (25% and 75% in the build
scenario). Therefore the EMFAC default VMT for each vehicle class is scaled down for
non-trucks and scaled up for trucks, respectively, based on the calculated adjustment
factors (0.93 and 1.29).

G-3
Exhibit G-3. Calculation of Adjustment Factors for Truck and Non-Truck VMT

Column A Column B
Adjustment Factor
VMT % of total VMT % of total VMT
(AF)*
(EMFAC default) (Project-specific)
Trucks 6,269,545 19% 25% 1.29
Non-trucks 26,134,922 81% 75% 0.93
Sum 32,404,467 100% 100%
* Adjustment factor is equal to the ratio between project-specific % VMT (Column B) and
EMFAC default % VMT (Column A), for trucks and non-trucks, respectively.

Multiplying the EMFAC default VMT by the calculated adjustment factors (AF) for each
vehicle class will produce updated VMT numbers that reflect project-specific information
in terms of truck and non-truck VMT percentage. As shown in Exhibit G-4, when the
adjusted VMT values for the truck group are added up, the sum is equal to 8,101,117
(25% of the total VMT). The non-truck VMT is 24,303,350 (75% of the total VMT).
Note that the overall VMT before and after the adjustment stays constant. Next, the
adjusted VMT values are entered into the EMFAC interface; pressing the “Apply” button
accepts the changes.

Exhibit G-4. Example Adjusted VMT for 13 Default Vehicle Classes

Adjusted
VMT Adjusted %
Default % VMT by (default VMT by
Vehicle Class VMT vehicle class VMT*AF) vehicle class
01 - Light-Duty Autos (PC) 15,271,757 47.1% 14,201,491 43.8%
02 - Light-Duty Trucks (T1) 3,340,492 10.3% 3,106,386 9.6%
03 - Light-Duty Trucks (T2) 7,266,306 22.4% 6,757,073 20.9%
04 - Medium-Duty Trucks (T3)* 3,535,454 10.9% 4,568,294 14.1%
05 - Light HD Trucks (T4)* 816,278 2.5% 1,054,743 3.3%
06 - Light HD Trucks (T5)* 302,809 0.9% 391,271 1.2%
07 - Medium HD Trucks (T6)* 698,543 2.2% 902,614 2.8%
08 - Heavy HD Trucks (T7)* 704,156 2.2% 909,867 2.8%
09 - Other Buses* 49,590 0.2% 64,077 0.2%
10 - Urban Buses* 40,198 0.1% 51,941 0.2%
11 – Motorcycles 256,367 0.8% 238,400 0.7%
12 - School Buses* 31,176 0.1% 40,284 0.1%
13 - Motor Homes* 91,341 0.3% 118,025 0.4%
Truck 6,269,545 19.4% 8,101,117 25.0%
Non-truck 26,134,922 80.7% 24,303,350 75.0%
TOTAL 32,404,467 100.0% 32,404,467 100.0%
* Classified as trucks to match project-specific data

G-4
G.5 GENERATING LINK-SPECIFIC EMISSION FACTORS

After the EMFAC run is completed, the project-specific running exhaust emission factors
are presented in Table 1 of the output Summary Rates file (.rts file). As highlighted in
Exhibit G-5, the PM10 running exhaust emission factor is 0.040 g/mi under the associated
speed bin of 65 mph. Tire wear and brake wear PM10 emission factors are 0.009 g/mi
and 0.013 g/mi, respectively, and do not vary by speed. For the one link in this example,
the total running link emission factor is 0.062 g/mi, which is the sum of these three
emission factors. For comparison, the total running link emission factor (based on
EMFAC default VMT distribution) is equal to 0.056 g/mi. It is lower than the project-
specific emission factor because the EMFAC default includes a smaller proportion of
truck VMT than this hypothetical highway project.

Exhibit G-5. Generating Running Exhaust Emission Factors in EMFAC

This completes the use of EMFAC for determining emissions factors for this project.
The total running link emission factor of 0.062 grams per vehicle-mile can be now be
used in combination with link length and link volume as inputs into the selected air
quality model, as discussed in Section 7 of the guidance.

G-5
 A ppendix H :
Example of Using EMFAC to Develop Emission Factors for a
Transit Project

H.1 INTRODUCTION

The purpose of this appendix is to illustrate the modeling steps required for users to
change EMFAC’s defaults and to develop project-specific PM idling and start exhaust
emission factors for a hypothetical bus terminal project. It also shows how to generate
emission factors from EMFAC for a project that involves a limited selection of vehicle
classes (e.g., urban buses). 1 This example uses the “Emfac” mode in EMFAC2007 (v2.3)
to generate grams per hour (g/hr) and grams per trip start (g/trip) emission factors stored
in the “Summary Rate” output file (.rts file) suitable for use in the AERMOD air quality
model. This example does not include the subsequent air quality modeling; refer to
Appendix F for an example of how to run AERMOD for a transit project for PM hot-spot
analyses.

The assessment of a bus terminal or other non-highway project can involve modeling two
different categories of emissions: (1) the start and idle emissions at the project site, and
(2) the running exhaust emissions on the links approaching and departing the project site.
This example is intended to help project sponsors understand how to create representative
idle and start emission factors based on the best available information supplied by
EMFAC, thus providing an example of how users may have to adapt the information in
EMFAC to their individual project circumstances.

As a preliminary note, the reader should understand that to estimate idle emissions, the
main task will involve modifying the default vehicle populations, by vehicle class,
embedded in EMFAC. When estimating start emissions, users will be modifying the
default vehicle trips, also by vehicle class. This appendix walks through the steps to
model idle and start emissions for this hypothetical project. Users will be able to
generate idle and start emission factors in a single EMFAC model run; multiple calendar
years can also be handled within one model run. As described in the main body of this
section, each run will be specific to either PM10 or PM2.5; however, this example is
applicable to both.

1
This is a highly simplified example showing how to employ EMFAC to calculate idle and start emission
factors for use in air quality modeling. An actual project would be expected to be significantly more
complex.

H-1
H.2 PROJECT CHARACTERISTICS

A PM10 hot-spot analysis is conducted for a planned bus terminal project in Sacramento
County, California. The project’s first full year of operation is assumed to be the year
2013. Through the interagency consultation process, it is determined that 2015 should be
the analysis year (based on the project’s emissions and background concentrations). The
PM analysis focuses on idle and start emissions from buses operated in the terminal. It is
assumed that these buses correspond to the “Urban Buses” vehicle class specified in
EMFAC and their average soak time is 540 minutes (all buses are parked overnight
before trip starts).

H.3 PREPARING EMFAC BASIC INPUTS (APPLICABLE TO BOTH IDLE AND

START EMISSIONS ESTIMATION)

Based on the project characteristics, basic inputs and default settings in EMFAC are first
specified (see Exhibit H-1). These basic inputs are similar to those specified for highway
projects. To generate idle emission factors from EMFAC, a speed bin of 0 mph must be
selected in the EMFAC interface.

Exhibit H-1. Basic Inputs in EMFAC for the Hypothetical Highway Project

Step Input Category Input Data Note

1 Geographic Area County  Sacramento Select from drop-down list
Use Average Default (not visible in the EMFAC
Calculation Method
user interface)
2 Calendar Years 2015 Select from drop-down list
3 Season or Month Annual Select from drop-down list
4 Scenario Title Use default Define default title in the EMFAC
user interface
5 Model Years Use default Include all model years
6 Vehicle Classes Use default Include all vehicle classes
7 I/M Program Schedule Use default Include all pre-defined I/M program
parameters
8 Temperature 60F Delete all default temperature bins
and input 60
9 Relative Humidity 70%RH Delete all default relative humidity
bins and input 70
10 Speed Use default Include speed bin of 0 mph
11 Emfac Rate Files Summary Rates (RTS) Select from EMFAC user interface
12 Output Particulate PM10 Select from EMFAC user interface

H-2
H.4 EDITING EMFAC DEFAULT POPULATION DISTRIBUTIONS TO OBTAIN
IDLE EMISSION FACTORS

To generate idle emission factors that reflect the bus terminal project data, vehicle
population by vehicle class must be modified in the EMFAC user interface. EMFAC has
data limitations regarding idle emissions: among the 13 vehicle classes in EMFAC, idle
emission factors are available only for LHDT1, LHDT2, MHDT, HHDT, School Buses,
and Other Buses. Although EMFAC does not provide idle emission factors for the
“Urban Buses” class (the class most typically associated with transit buses), the idle
emission factors for “Other Buses” may be used to represent transit buses.

Note that only the “Other Buses” vehicle population will affect idle emissions in this
example; however, the “Urban Buses” class also needs to be included at this point to
address idling and starting emission factors in one single run. Thus, except for “Other
Buses” and “Urban Buses,” all other vehicle classes are eliminated in EMFAC by
inputting very low values (such as “1”; entering “0” is not allowed in EMFAC). Exhibit
H-2 shows the EMFAC interface before and after vehicle population by vehicle class is
changed.

Note: In this bus terminal example, start emissions are available for “urban buses”;
however, idle emission factors are only available for “other buses.” Therefore, users
will access emission factor information for both “other” and “urban” buses, and the
population data for these fleets are left intact (see modified version of Exhibit H-2).

H-3
Exhibit H-2. Changing EMFAC Vehicle Population Distributions to Estimate Idle
Emission Factors

Default EMFAC data before modification

Modified EMFAC data

H-4
H.5 EDITING EMFAC DEFAULT TRIP DISTRIBUTIONS TO OBTAIN START
EMISSION FACTORS

After users modify the population distribution in EMFAC, the new population
distribution will be used by EMFAC to create vehicle trip distributions. The new
distribution will affect the EMFAC data displayed during the trip distribution
modification steps described below. Users need to manually update the trip distributions
through the EMFAC user interface to obtain project-specific start emission factors.

Average start emission factors in EMFAC depend on the number of trips made by a
particular vehicle class and the corresponding soak time. To generate project-specific
start emission factors, the number of trips by vehicle class must be modified in the
EMFAC user interface. For this example bus terminal project, a very low value (“1”) is
entered into the interface for all vehicle classes except for “Urban Buses” to represent the
project-specific data. Exhibit H-3 shows the EMFAC interface before and after vehicle
trip distributions by vehicle class are changed.

H-5
Exhibit H-3. Changing EMFAC Trip Distributions to Estimate Start Emission
Factors

Default EMFAC data before modification

Modified EMFAC data

H-6
H.6 GENERATING IDLE AND START EMISSION FACTORS

“Urban Buses” is the vehicle class best representing transit buses in this hypothetical bus
terminal project. After the EMFAC run is completed, the project-specific idle exhaust
emission factors are presented in Table 1 of the output Summary Rates file (.rts file). As
shown in Exhibit H-4, the PM10 idle exhaust emission factor for the example bus terminal
project (0.734 grams/idle-hour) can be found under the 0 mph speed bin for the HDT
vehicle class (associated with “Other Buses” because EMFAC does not provide “Urban
Buses” idle emission factors). The start emission factor for vehicle class “Urban Buses”
(0.011 g/trip) is presented in Table 2 under the 540-min time bin in the column “All” or
“UBUS” (see Exhibit H-5).

In order to produce a grams/hour emission factor for use in AERMOD, several post-
processing calculations are necessary. First, the idle emission factor (0.734 grams/idle-
hour) is multiplied by the number of vehicle idle-hours. Next, the start emissions can be
calculated by multiplying the start emission factor (0.011 grams/trip) by the number of
starts expected in a given hour. If the area being modeled has both idling and starts, these
values can be summed to produce an aggregate grams/hour value.

Exhibit H-4. Generating Idling Emission Factors in EMFAC

H-7
Exhibit H-5. Generating Start Emission Factors in EMFAC 2

This completes the use of EMFAC for determining start and idle emission factors for this
project. The aggregate grams/hour value for starts and idle can now be input into
AERMOD, as discussed in Section 7 of the guidance.

2
Note that the start emission factors for UBUS and ALL are identical in this exhibit because the user
modified the number of trips by vehicle class to include activity from only “Urban Buses”. EMFAC
collapsed the 13 vehicle classes to six vehicle groups in the output file. The collapsed output provides start
emission factors for the “Urban Buses” in the UBUS category and because fleet activity was composed
entirely from this vehicle class, the start emission factors for UBUS and ALL are essentially the same.

H-8
 A ppendix I :
Estimating Locomotive Emissions

I.1 INTRODUCTION

This appendix describes how to quantify locomotive emissions when they are a
component of a transit or freight terminal or otherwise a source in the project area being
modeled. Note that state or local air quality agencies may have experience modeling
locomotive emissions and therefore could be of assistance when quantifying these
emissions for a PM hot-spot analysis.

Generally speaking, locomotive emissions can be estimated in the following manner:

1. Determine where in the project area locomotive emissions should be estimated.

2. Determine when to analyze emissions.

3. Describe the locomotive activity within the project area, including:

• The locomotives present in the project area (the “locomotive roster”); and
• The percentage of time each locomotive spends in various throttle settings
(the “duty cycle”).

4. Calculate locomotive emissions using either:

• Horsepower rating and load factors, or
• Fuel consumption data. 1

The estimated locomotive emission rates that result from this process would then be used
for air quality modeling. The interagency consultation process must be used to evaluate
and choose the model and associated method and assumptions used for quantifying
locomotive emissions for PM hot-spot analyses (40 CFR 93.105(c)(1)(i)).

I.2 DETERMINING WHERE IN THE PROJECT AREA LOCOMOTIVE

EMISSIONS SHOULD BE ESTIMATED

Under certain circumstances, it is appropriate to model different locations within the

project area as separate sources to characterize differences in locomotive type and/or
activity appropriately. This step is analogous to dividing a highway project into links (as
described in Sections 4.2 and 5.2 of the guidance) and improves the accuracy of
emissions modeling and subsequent air quality modeling. For example, in an intermodal
terminal, emissions from a mainline track (which will have a large percentage of higher

1
These are the two methods described in this appendix; others may be possible. See Appendix I.5 for
details.

I-1
speed operations with little idling) should be estimated separately from the associated
passenger or freight terminal (which would be expected to experience low speed
operations and significant idling).

The following activities are among those typically undertaken by locomotives and are
candidates for being modeled as separate sources if they occur at different locations
within the project area:
• Idling within the project area;
• Trains arriving into, or departing from, the project area (e.g., terminal arrival and
departure operations);
• Testing, idling, and service movements in maintenance areas or sheds;
• Switching operations;
• Movement of trains passing through, but not stopping in, the project area.

The project area may also be divided into separate sources if it includes several different
locomotive rosters (see Appendix I.4.1, below)

I.3 DETERMINING WHEN TO ANALYZE EMISSIONS

The number of hours and days that have to be analyzed depends on the range of activity
expected to occur within the project area. For rail projects where activity varies from
hour to hour, day to day, and possibly month to month, it is recommended that, at a
minimum, project sponsors calculate emissions based on 24 hours of activity for both a
typical weekday and weekend day and for four representative quarters of the analysis
year when comparing emissions to all PM2.5 NAAQS. 2 For projects in areas that violate
only the 24-hour PM10 or PM2.5 NAAQS, the project sponsor may choose to model only
one quarter, in appropriate cases. See Section 3.3.4 of the guidance for further
information.

These resulting emission rates should be applied to AERMOD and used to calculate
design values to compare with the applicable PM NAAQS as described in Sections 7
through 9 of the guidance.

I.4 DESCRIBING THE LOCOMOTIVE ROSTERS AND DUTY CYCLES

Before calculating locomotive emission rates, it is necessary to know what locomotives

are present in the locations being analyzed in the project area (see Appendix I.2, above)
and what activities these locomotives are undertaking at these locations. This data will
impact how emissions are calculated.

2
If there is no difference in activity between weekday and weekend activity, it may not be necessary to
examine weekend day activity separately. Similarly, if there is no difference in activity between quarters,
emission rates can be determined for one quarter, which can then be used to represent every quarter of the
analysis year.

I-2
I.4.1 Locomotive rosters

Because emissions can vary significantly depending on a locomotive’s make, model,

engine, and year of engine manufacture (or re-manufacture), it is important to know what
locomotives are expected to be operating within the project area. Project sponsors should
develop a “locomotive roster” (i.e., a list of each locomotive’s make, model, engine, and
year) for the locomotives that will be operating within the specific project area being
analyzed. The more detailed the locomotive roster, the more accurate the estimated
emissions will be.

In some cases, it will be necessary to develop more than one locomotive roster to reflect
the operations in the project area accurately (for example, switcher locomotives may be
confined to one portion of a facility and therefore may be represented by their own
roster). In these situations, users should model areas with different rosters as separate
sources to account for the variability in emissions (see Appendix I.2).

I.4.2 Locomotive duty cycles

Diesel locomotive engine power is controlled by “notched” throttles; idling, braking, and
moving the locomotive is conducted by placing the throttle in one of several available
“notch settings.” 3 A locomotive’s “duty cycle” is a description of how much time, on
average, the locomotive spends in each notch setting when operating. Project sponsors
should use the latest locally-generated or project-specific duty cycles whenever possible;
this information may be available from local railway authorities or the state or local air
agency. 4 The default duty cycles for line-haul and switch locomotives, found in Tables 1
and 2 of 40 CFR 1033.530 (EPA’s regulations on controlling emissions from
locomotives), should be used only if they adequately represent the locomotives that will
be present in the project area and no local or project-specific duty cycles are available.

I.5 CALCULATING LOCOMOTIVE EMISSIONS

Once a project’s locomotive rosters and respective duty cycles have been determined,
locomotive emissions can then be calculated for each part of the project area using either
(1) horsepower rating and load factors, or (2) fuel consumption data. These two methods
are summarized below. Unless otherwise determined through consultation, only one
method should be used for a given project.

3
A diesel locomotive typically has eight notch settings for movement (run notches), in addition to one or
more idle or dynamic brake notch settings. Dynamic braking is when the locomotive engine, rather than
the brake, is used to control speed.
4
The state or local air agency may have previously developed locally-appropriate duty cycles for emissions
inventory purposes.

I-3
I.5.1 Finding emission factors

Regardless of method chosen, locomotive emissions factors will be needed for the
analysis. Locomotive emission factors depend on the type of engine, the power rating of
the locomotive (engine horsepower), and the year of engine manufacture (or re-
manufacture). Default PM10 emission factors for line-haul and switch locomotives can be
obtained from Tables 1 and 2 of EPA’s “Emission Factors for Locomotives,” EPA-420-
F-09-025 (April 2009). 5 These PM10 emission factors are in grams/horsepower-hour and
can easily be converted to PM2.5 emission factors. However, these are simply default
values; locomotive-specific data may be available from manufacturers and should be
used whenever possible. In addition, see Appendix I.5.4 for other variables that must be
considered when determining the appropriate locomotive emission factors.

Note that the default locomotive emission factors promulgated by EPA may change over
time as new information becomes available. The April 2009 guidance cited above
contains the latest emission factors as of this writing. Project sponsors should consult the
EPA’s website at: www.epa.gov/otaq/locomotives.htm for the latest locomotive default
emission factors and related guidance.

I.5.2 Calculating emissions using horsepower rating and load factors

One way locomotive emissions can be calculated is to use PM2.5 or PM10 locomotive
emission factors, the horsepower rating of the engines found on the locomotive roster,
and engine load factors (which are calculated from the duty cycle).

Calculating Engine Load Factors

The horsepower of the locomotive engines, including the horsepower used in each notch
setting, should be available from the rail operator or locomotive manufacturer.
Locomotive duty cycle data (see Appendix I.4.2) can then be used to determine how
much time each locomotive spends in each notch setting, including braking and idling.
An engine’s “load factor” is the percent of maximum available horsepower it uses over
the course of its duty cycle. In other words, a load factor is the weighted average power
used by the locomotive divided by the engine’s maximum rated power. 6 Load factors
can be calculated by summing the actual horsepower-hours of work generated by the
engine in a given period of time and dividing it by the engine’s maximum horsepower
and the hours during which the engine was being used, with the result expressed as a
percentage. For example, if a 4000 hp engine spends one hour at full power (generating
4000 hp-hrs) and one hour at 50 percent power (generating 2000 hp-hrs), its load factor
5
Table 1 of EPA’s April 2009 document includes default emission factors for higher power cycles
representative of general line-haul operation; Table 2 includes emission factors for lower power cycles used
for switching operations. The April 2009 document also includes information on how to convert PM10
emission factors for PM2.5 purposes. Note that Table 6 (PM10 Emission Factors) should not be used for PM
hot-spot analyses, since these factors are national fleet averages rather than emission factors for any
specific project.
6
“Weighted average power” in this case is the average power used by the locomotive weighted by the time
spent in each notch, as explained further below.

I-4
would be 75 percent (6000 hp-hrs ÷ 4000 hp ÷ 2 hrs). Note that, in this example, it
would be equivalent to calculate the load factor using the percent power values instead:
((100% * 1 hr) + (50% * 1 hr) ÷ 2 hrs = 75%). To simplify emission factor calculations,
it is recommended that locomotive activity be generalized into the operational categories
of “moving” and “idling,” with separate load factors calculated for each.

An engine’s load factor is calculated by completing the following steps:

Step 1. Determine the number of notch settings the engine being analyzed has and the
horsepower used by the engine in each notch setting. 7 Alternatively, as described above,
the percent of maximum power available in each notch could instead be used.

Step 2. Identify the percentage of time the locomotive being analyzed spends in each
notch setting based on its duty cycle (see Appendix I.4.2).

Step 3. To make emission rate calculations easier, it is useful to calculate two separate
load factors for an engine: one for when the locomotive is idling and one for when it is
moving. 8 Therefore, the percentage of time the locomotive spends in each notch (from
Step 2) needs to be adjusted so that all idling and all moving notches are considered
separately. For example, if a locomotive has just one idle notch setting, it spends 100%
of its idling time in that setting, even if it only idles during part of its duty cycle. While
calculating the time spent idling will usually be simple, for the non-idle (moving) notch
settings some additional adjustment to the locomotive’s duty cycle percentages will be
required to determine the time spent in each moving notch as a fraction of total time spent
moving, disregarding any time spent idling.

For example, say a locomotive spends 30% of its time idling and 70% of its time moving
over the course of its duty cycle and that 15% of this total time (idling and moving
together) is spent in notch 2. When calculating the moving load factor, this percentage
needs to be adjusted to determine what fraction of just the 70% of time spent moving is
spent in notch 2. In this example, 15% of the total duty cycle spent in notch 2 would
equal 21.4% (15% * 100% ÷ 70%) of the locomotive’s time when it is not at idle; that is,
whenever it is moving, this locomotive spends 21.4% of its time in notch 2. This
calculation is repeated for each moving notch setting. The result will be the fraction of
time spent in each notch when considering idle and moving modes of operation
separately.

Step 4. The next step is to calculate what fraction of maximum available horsepower is
being used based on the time spent in each notch setting as was calculated in Step 3. This
is determined by summing the product of the percentage of time spent in each notch
(calculated in Step 3) by the horsepower generated by the engine at that notch setting
(determined in Step 1). For example, if the locomotive with a rated engine power of

7
For locomotives that are equipped with multiple dynamic braking notches and/or multiple idle notches, it
may be necessary to assume a single dynamic braking notch and a single idle notch, depending on what
information is available about the particular engine.
8
In this case, “moving” refers to all non-idle notch settings: that is, dynamic braking and all run notches.

I-5
3000 hp spends 21.4% of its moving time in notch 2 and 78.6% of its moving time in
notch 6, and is known to generate 500 hp while in notch 2 and 2000 hp while in notch 6,
then its weighted average power would be 1679 hp (107 hp (500 hp * 0.214) + 1572 hp
(2000 hp * 0.786) = 1679 hp).

Step 5. The final step is to determine the load factors. This is done by dividing the
weighted average horsepower (calculated in Step 4) by the maximum engine horsepower.
For idling, this should be relatively simple. For example, if there is one idle notch setting
and it is known that a 4000 hp engine uses 20 hp when in its idle notch, then its idle load
factor will be 0.5% (20 hp ÷ 4000 hp). To determine the load factor for all power
notches, the weighted horsepower calculated in Step 4 should be divided by the total
engine horsepower. For example, if the same 4000 hp engine is determined to use an
average of 1800 hp while in motion (as determined by adjusting the horsepower by the
time spent in each “moving” notch setting in Step 4), then the moving load factor would
be 45% (1800 hp ÷ 4000 hp).

The resulting idling and moving load factors represent the average amount of the total
engine horsepower the locomotive is using when idling and moving, respectfully. These
load factors can then be used to modify PM emission factors and generate emission rates
as described below.

Generating Emission Rates Based on Load Factors

As noted above, EPA’s “Emission Factors for Locomotives” provides emission factors in
grams/brake horsepower-hour. This will also likely be the case with any specific
emission factors obtained from manufacturer’s specifications. These units can be
converted into grams/second (g/s) emission rates by using the load factor on the engines
and the time spent in each operating mode, as described below.

The first step is to adjust the PM emission factors to reflect how the engine will actually
be operating. 9 This is done by multiplying the appropriate PM emission factor by the
idling and moving load factors calculated for that particular engine. 10 Next, to determine
the emission rate, this adjusted emission factor is further multiplied by the amount of
time the locomotive spends idling and moving while in the project area. 11

For example, if the PM emission factor known to be 0.18 g/bhp-hr, the engine being
analyzed has an idling load factor of 0.5%, and the locomotive is anticipated to idle 24
minutes per hour in the project area, then the resulting emission rate would be 0.035
grams/hour (0.18 g/bhp-hr * 0.5% * 0.4 hours).

9
Because combustion characteristics of an engine vary by throttle notch position, it is appropriate to adjust
the emission factor to reflect the average horsepower actually being used by the engine.
10
Project sponsors are reminded to check www.epa.gov/otaq/locomotives.htm to ensure the latest default
emission factors for idle and moving emissions are being used.
11
Note that this may or may not match up with the idle and moving time as described by the duty cycle
used to calculate the load factors, depending on how project-specific that duty cycle is.

I-6
Emission rates need to be converted into g/s for use by AERMOD, as described further in
Sections 7 through 9 of the guidance. These calculations should be repeated until the
entire locomotive roster is represented in each part of the project area being analyzed.

Appendix I.7 provides an example of calculating g/s locomotive emission rates using this
methodology.

I.5.3 Calculating emissions using fuel consumption data

Another method to calculate locomotive emissions involves using fuel consumption data.
Chapter 6.3 of EPA’s “Procedure for Emission Inventory Preparation -- Volume IV:
Mobile Sources” (reference information provided in Appendix I.6, below) is a useful
reference and should be consulted when using this method.

Note that, for this method, it may be useful to scale down data already available to the
project sponsor. For example, if rail car miles/fuel consumption is known for trains
operating in situations identical to those being estimated in the project area, this data can
be used to estimate fuel consumption rates for a defined track length within the project
area.

Calculating Average Fuel Consumption

Locomotive fuel consumption is specific to a particular locomotive engine and the

throttle (notch) setting it is using. Data on the fuel consumption of various engines at
different notch settings can often be obtained from the locomotive or engine
manufacturer’s specifications. When only partial data is available (e.g., only data for the
lowest and highest notch settings are known), interpolation combined with best available
engineering judgment can be used to determine fuel consumption at the intermediate
notch settings.

A locomotive’s average fuel consumption can be calculated by determining how long

each locomotive is expected to spend in each notch setting based on its duty cycle (see
Appendix I.4.2). This data can be aggregated to generate an average fuel consumption
rate for each locomotive type. See Chapter 6.3 of Volume IV for details on how to
generate this data based on a specific locomotive roster and duty cycle.

Once the average fuel consumption rates have been determined, they should be
multiplied by the appropriate emission factors to determine a composite average hourly
emission rate for each engine in the roster. Since the objective is to determine an average
fuel consumption rate for the entire locomotive roster, this calculation should be repeated
for each engine on the roster at each location analyzed.

If several individual sources will be modeled at different sections of the project area as
described in Appendix I.2, train schedule data should be consulted to determine the hours
of operation of each locomotive within each section of the project area. Hourly emission
rates per locomotive should then be multiplied by the number of hours the locomotive is

I-7
operating, for each hour of the day in each section of the project area to provide average
hourly emission rates for each section of the project. These should then be converted to
grams/second for use in AERMOD, as described further in Sections 7 through 9 of the
guidance.

Examples of calculating locomotive emissions using this method can be found in Chapter
6 of Volume IV.

I.5.4 Factors influencing locomotive emissions and emission factors

The following considerations will influence locomotive emissions regardless of the

method used and should be examined when determining how to characterize locomotives
for emissions modeling or when choosing the appropriate emission factors:

• Project sponsors should be aware of the emission reductions that would result
from remanufacturing existing locomotives (or replacing existing locomotives
with new locomotives) that meet EPA’s Tier 3 or Tier 4 emission standards when
they become available. The requirements that apply to existing and new
locomotives were addressed in EPA’s 2008 rulemaking entitled “Control of
Emissions of Air Pollution from Locomotive Engines and Marine Compression-
Ignition Engines Less Than 30 liters Per Cylinder” (73 FR 37095). Beginning in
2012 all locomotives will be required to use ultra-low sulfur diesel fuel (69 FR
38958). Additionally, when existing locomotives are remanufactured, certified
remanufacture systems will have to be installed to reduce emissions. Beginning
in 2011, new locomotives must meet tighter Tier 3 emission standards. Finally,
beginning in 2015 even more stringent Tier 4 emission standards for new
locomotives will begin to be phased in.

• For locomotives manufactured before 2005, a given locomotive may be in one of

three possible configurations, depending on when it was last remanufactured: (1)
uncertified; (2) certified to the standards in 40 CFR Part 92; or (3) certified to the
standards in 40 CFR Part 1033. Each of these configurations should be treated as
a separate locomotive type when conducting a PM hot-spot analysis.

• Emissions from locomotives certified to meet Family Emission Limits (FELs)

may differ from the emission standard identified on the engine’s Emission
Control Information label. Rail operators will know if their locomotives
participate in this program. Any locomotives in the project area participating in
this program should be identified so that the actual emissions from the particular
locomotives being analyzed are considered in the analysis, rather than the family
emissions level listed on their FEL labels.

I-8
I.6 AVAILABLE RESOURCES

These resources and websites should be checked prior to beginning any PM hot-spot
analysis to ensure that the latest data (such as emission factors) are being used:

• “Emission Factors for Locomotives,” EPA-420-F-09-025 (April 2009). Available

online at: www.epa.gov/otaq/locomotives.htm.

• Chapter 6 of “Procedure for Emission Inventory Preparation - Volume IV: Mobile

Sources.” Available online at: www.epa.gov/OMS/invntory/r92009.pdf. Note
that, as of this writing, the emission factors listed in Volume IV have been
superseded by the April 2009 publication listed above for locomotives certified to
meet current EPA standards. 12

• “Control of Emissions from Idling Locomotives,” EPA-420-F-08-014, March

2008. Available online at:
www.epa.gov/otaq/regs/nonroad/locomotv/420f08014.htm.

• See Section 10 of the guidance for additional information regarding potential

locomotive emission control measures.

I.7 EXAMPLE OF CALCULATING LOCOMOTIVE EMISSION RATES USING

HORSEPOWER RATING AND LOAD FACTOR ESTIMATES

The following example demonstrates how to estimate locomotive emissions using the
engine horsepower rating/load factor method described in Appendix I.5.2.

The hypothetical proposed project in this example includes the construction of an

intermodal terminal in an area that is designated as nonattainment for both the 1997
annual PM2.5 NAAQS and the 2006 24-hour PM2.5 NAAQS. The terminal in this
example is to be completed and operational in 2013. The hot-spot analysis is performed
for 2015, because it is determined through interagency consultation that this will be the
year of peak emissions, when considering the project’s emissions and the other emissions
in the project area.

In this example, the operational schedule anticipates that 32 locomotives will be in the
project area over a 24-hour period, with 16 locomotives in the project area during the
peak hour. Based on the schedule, it is further determined that while in the project area
each train will spend 540 seconds idling and 76 seconds moving.

12
Although the emission factors have been superseded, the remainder of the Volume IV guidance remains
in effect.

I-9
The locomotive PM2.5 emissions are calculated based on horsepower rating and load
factors.

I.7.1 Calculate idle and moving load factors

As described in I.5.2, the project sponsor uses a series of steps to calculate load factors.
These steps are described below and the results from each step are shown in table form in
Exhibit I-1.

Step 1: The project sponsor first needs some information about the locomotives expected
to be operating at the terminal in the analysis year.

For each locomotive, the horsepower used by the locomotive in each notch setting as well
as under dynamic braking and at idle must be determined. For the purpose of this
example it is assumed that all of the locomotives that will serve this terminal are very
similar: all use the same horsepower under each of operating conditions, and all have
only one idle and dynamic braking notch setting. The horsepower generated at each
notch setting is obtained from the engine specifications (see second column of Exhibit I-
1). In this case, the rated engine horsepower is 4000 hp (generated at notch 8).

Step 2: The next step is to determine the average amount of time that the locomotives
spend in each notch and expressing the results as a percentage of the locomotive’s total
operating time. In this example, it is determined that, based on their duty cycle, the
locomotives that will service this terminal spend 38% of their time idling and 62% of
their time in motion in one of the eight run notch settings or under dynamic braking. The
percentage of time spent in each notch is shown in the third column of Exhibit I-1.

Step 3: To make emission factor calculations easier, it is decided to calculate separate

idling and moving load factors. The next step, then, is for the project sponsor to calculate
the actual percentage of time that the locomotives spend in each notch, treating idling and
moving time separately. This is done by excluding the time spent idling and
recalculating the percentage of time spent in the other notches (i.e., dynamic braking and
each of the eight notch settings) so that the total time spent in non-idle notches adds to
100%. The results are shown in the fourth column of Exhibit I-1.

Step 4: The next step is to calculate the weighted average horsepower for this engine
using the horsepower generated in each notch and the percentage of time spent in each
notch as adjusted in Step 3. For locomotives that are idling, this is simply the horsepower
used at idle. For the other notches, the actual horsepower for each notch is determined by
multiplying the horsepower generated in a given notch (determined in Step 1) by the
actual percentage of time that the locomotive is in that notch, as adjusted (calculated in
Step 3). The results are shown in the fifth column of Exhibit I-1.

Step 5: The final step in this part of the analysis is to determine the idle and moving load
factors. The idle load factor is just the horsepower generated at idle divided by the
maximum engine horsepower, with the result expressed as a percentage. To determine

I-10
the moving load factor, the weighted average horsepower for all non-idle notches
(calculated in Step 4) is divided by the maximum engine horsepower, with the result
expressed as a percentage. The final column of Exhibit I-1 shows the results of these
calculations, with the idling and moving load factors highlighted.

Exhibit I-1. Calculating Locomotive Load Factors

Step 3: Step 4:
Reweighted Time-
Step 1: Step 5:
Step 2: time spent in weighted
Horsepower Load
Notch Average % each notch hp used,
(hp) factors
Setting time spent (adjusted so based on
used in (idle and
in notch that non-idle time
notch moving)
notches add to spent in
100%) notch
Idling load factor:
Idle 14 38.0% 100.0% 14.0 0.4%
Moving load factor:
Dynamic 136 12.5% 20.2% 27.5
Brake
1 224 6.5% 10.5% 23.5
2 484 6.5% 10.5% 50.8
3 984 5.2% 8.4% 82.7
4 1149 4.4% 7.1% 81.6
5 1766 3.8% 6.1% 107.8
6 2518 3.9% 6.3% 158.6
7 3373 3.0% 4.8% 161.9
8 4,000 16.2% 26.1% 1,044.0
Total 62.0% 100.0% 1,752.4 43.8%

I.7.2 Using the load factors to calculate idle and moving emission rates

Now that the idle and moving load factors have been determined, the gram/second (g/s)
emission rates can be calculated for the idling and moving locomotives.

First, the project sponsor would determine how many locomotives are projected to be
idling and how many are projected to be in motion during the peak hour of operation and
over a 24-hour period. As previously noted, it is anticipated that 32 locomotives will be
in the project area over a 24-hour period, with 16 locomotives in the project area during
the peak hour. It was further determined that, while in the project area, each train will
spend 540 seconds idling and 76 seconds moving.

For the purpose of this example, it has been assumed that each locomotive idles for the
same amount of time and is in motion for the same amount of time. Note that, in this
case, the number of locomotives considered “moving” will be double the actual number

I-11
of locomotives present in order to account for the fact that each locomotive moves twice
through the project area (as it arrives and departs the terminal).

Next, the project sponsor would determine the PM2.5 emission factor to be used in this
analysis for 2015. These emission factors can be determined from the EPA guidance
titled “Emission Factors for Locomotives.”

Table 1 of “Emission Factors for Locomotives” presents PM10 emission factors in terms
of grams/brake horsepower-hour (g/bhp-hr) for line haul locomotives that are typically
used by commuter railroads. Emission factors are presented for uncontrolled
locomotives, locomotives manufactured to meet Tier 0 through Tier 4 emission
standards, and locomotives remanufactured to meet more stringent emission standards.
It’s important to determine the composition of the fleet of locomotives that will use the
terminal in the year that is being analyzed so that the emission factors in Table 1 can be
used in the calculations. This information would be available from the railway operator.

In this example, we are assuming that all of the locomotives meet the Tier 2 emission
standard. However, an actual PM hot-spot analysis would likely have a fleet of
locomotives that meets a combination of these emission standards. The calculations
shown below would have to be repeated for each different standard that applies to the
locomotives in the fleet.

The final step in these calculations is to use the information shown in Exhibit I-1 and the
other project data collected to calculate the PM2.5 emission rates for idling and moving
locomotives during both the peak hour and over a 24-hour basis. 13

Calculating Peak Hour Idling Emissions

The following calculation would be used to determine the idling emission rate during the
peak hour of operation: 14

PM2.5 Emission Rate = (16 trains/hr) * (1 hr/3,600 s) * (540 s/train) * (4,000 hp) *
(0.004) * (0.18 g/bhp-hr) * (1 hr/3,600 s) * (0.97)
PM2.5 Emission Rate = 0.0019 g/s

Where:
• Trains per hour = 16 (number of trains present in peak hour)
• Idle time per train = 540 s (from anticipated schedule)
• Locomotive horsepower = 4,000 hp (from engine specifications)
• Idle load factor = 0.004 (0.4%, calculated in Exhibit I-1)

13
Peak hour emission rates will not be necessary for all analyses; however, for certain projects that involve
very detailed air quality modeling analyses, peak hour emission rates may be necessary to more accurately
reflect the contribution of locomotive emissions to air quality concentrations in the project area.
14
Note that, for the calculations shown here, any units expressed in hours or days need to be converted to
seconds since a g/s emission rate is required for AERMOD.

I-12
 • Tier 2 Locomotive Emission Factor = 0.18 g/bhp-hr (from “Emission
Factors for Locomotives”)
• Ratio of PM2.5 to PM10 = 0.97 (from “Emission Factors for Locomotives”)

Calculating 24-hour Moving Emissions

Similarly, the following equation would be used to calculate the moving emission rate for
the 24-hour period:

PM2.5 Emission Rate = (64 trains/day) * (76 s/train) * (1 day/86,400 s) * (4,000 hp) *
(0.438) * (0.18 g/bhp-hr) * (1hr/3,600 s) * (0.97)
PM2.5 Emission Rate = 0.0048 g/s

Where:
• Trains per day = 64 (double the actual number of trains present over 24
hours to account for each train moving twice through the project area)
• Moving time per train = 76 s (from anticipated schedule)
• Locomotive horsepower = 4,000 hp (from engine specifications)
• Moving load factor = 0.438 (43.8%, calculated in Exhibit I-1)
• Tier 2 Locomotive Emission Factor = 0.18 g/bhp-hr (from “Emission
Factors for Locomotives”)
• Ratio of PM2.5 to PM10 = 0.97 (from “Emission Factors for Locomotives”)

A summary of the variables used in the above equations and the resulting emission rates
can be found in Exhibit I-2, below.

Exhibit I-2. PM2.5 Locomotive Emission Rates

Calculated
PM2.5 Calculated
Operational Number of Time/ 24-hour
Emission Peak Hour
Mode Locomotives Train Emission
Factor Emission Rate
Rate
Peak 24
(s) (g/bhp-hr) (g/s) (g/s)
hour hours
Idle 16 32 540 0.18 0.0019 0.00016
Moving 32 64 76 0.18 0.057 0.0048

These peak and 24-hour emission rates can now be used in air quality modeling for the
project area, as described in Sections 7 through 9 of the guidance.

Note that, since this area is designated as nonattainment for both the 1997 annual PM2.5
NAAQS and the 2006 24-hour PM2.5 NAAQS, the results of the analysis will be
compared to both NAAQS (see Section 3.3.4 of the guidance). Since the area is in
nonattainment of the annual PM2.5 NAAQS, all four quarters will need to be included in

I-13
the analysis to estimate a year’s worth of emissions. If there is no change in locomotive
activity across quarters, the emission rates calculated here could be used for each quarter
of the year (see Appendix I.3).

I-14
 A ppendix J :
Additional Reference Information on Air Quality Models and
Data Inputs

J.1 INTRODUCTION

This appendix supplements Section 7’s discussion of air quality models. Specifically,
this appendix describes how to configure AERMOD and CAL3QHCR for PM hot-spot
analysis modeling, as well as additional information on handling the data required to run
the models for these analyses. This appendix is not intended to replace the user guides
for air quality models, but discuss specific model inputs, keywords, and formats for PM
hot-spot modeling. This appendix is organized so that it references the appropriate
discussions in Section 7 of the guidance.

J.2 SELECTING AN APPROPRIATE AIR QUALITY MODEL

The following discussion supplements Section 7.3 of the guidance and describes how to
appropriately configure AERMOD and CAL3QHCR when completing a PM hot-spot
analysis. Users should also refer to the model user guides, as appropriate.

J.2.1 Using AERMOD for PM hot-spot analyses

There are no specific commands unique to transportation projects that are necessary when
using AERMOD. By default, AERMOD produces output for particulate matter in units
of micrograms per cubic meter of air (µg/m3). All source types in AERMOD require that
emissions are specified in terms of emissions per unit time, although AREA-type sources
also require specification of emissions per unit time per unit area. AERMOD has no
specific traffic queuing mechanisms. Emissions output from MOVES, EMFAC, AP-42,
and other types of methods should be formatted as described in the AERMOD User
Guide. 1

J.2.2 Using CAL3QHCR for PM hot-spot analyses

CAL3QHCR is an extension of the CAL3QHC model that allows the processing of a full
year of hourly meteorological data, the varying of traffic-related inputs by hour of the
week, and calculation of long-term average concentrations. It also will display the five
highest concentration days for the time period being modeled. Emissions output from
MOVES, EMFAC, AP-42, and other emission methods should be formatted as described

1
Extensive documentation is available describing the various components of AERMOD, including user
guides, model formulation, and evaluation papers. See EPA’s SCRAM website for AERMOD
documentation: www.epa.gov/scram001/dispersion_prefrec.htm#aermod.

J-1
in the CAL3QHCR User Guide. 2 In addition, the following guidance is provided when
using CAL3QHCR for a PM hot-spot analysis:

Specifying the Right Pollutant

When using CAL3QHCR for PM hot-spot analyses, the MODE keyword must be used to
specify analyses for PM so that concentrations are described in micrograms per cubic
meter of air (µg/m3) rather than parts per million (ppm).

Entering Emission Rates

MOVES emission rates for individual roadway links are based on the Op-Mode
distribution associated with each link and are able to include emissions resulting from
idling. MOVES-based emission factors that incorporate relevant idling time and other
delays should be entered in CAL3QHCR using the EFL keyword. Therefore, within
CAL3QHCR, the IDLFAC keyword’s emission rates should be set to zero, because the
effects of idling are already included within running emissions. (Note that if a non-zero
emission rate is used in CAL3QHCR, the model will treat idling emission rates separately
from running emission rates.) The same recommendation applies when using emission
rates calculated by EMFAC.

Assigning Speeds

Although the user guide for CAL3QHCR specifies that the non-queuing links should be
assigned speeds in the absence of delay caused by traffic signals, the user should use
speeds that reflect delay when using CAL3QHCR for a hot-spot analysis. Since MOVES
emission factors already include the effects of delay (i.e., Op-Mode distributions that are
user-specified or internally calculated include the effects of delay), the speeds used in
CAL3QHCR links will already reflect the relevant delay on the link over the appropriate
averaging time. The same recommendation applies when using EMFAC.

Using the Queuing Algorithm

When applying CAL3QHCR for the analysis of highway and intersection projects, its
queuing algorithm should not be used. 3 This includes the CAL3QHCR keywords
NLANE, CAVG, RAVG, YFAC, IV, and IDLFAC. As discussed in Sections 4 and 5,
idling vehicle emissions should instead be accounted for by properly specifying links for
emission analysis and reflecting idling activity in the activity patterns used for MOVES
or EMFAC modeling.

2
The CAL3QHCR user guide and other model documentation can be found on EPA’s SCRAM website:
www.epa.gov/scram001/dispersion_prefrec.htm#cal3qhc.
3
CAL3QHCR’s algorithm for estimating the length of vehicle queues associated with intersections is based
on the 1985 Highway Capacity Manual, which is no longer current. Furthermore, a number of other
techniques are now available that can be used to estimate vehicle queuing around intersections.

J-2
J.3 CHARACTERIZING EMISSION SOURCES
The following discussion supplements Section 7.4 of the guidance and describes in more
detail how to characterize sources in CAL3QHCR and AERMOD, including the physical
characteristics, location, and timing of sources. This discussion assumes the user is
familiar with handling data in these models, including the use of specific keywords. For
additional information, refer to the CAL3QHCR and AERMOD user guides.

J.3.1 Physical characteristics and locations of sources in CAL3QHCR

CAL3QHCR characterizes highway and intersection projects as line sources. The

geometry and operational patterns of each roadway link are described using the following
variables, which in general may be obtained from engineering diagrams and design plans
of the project: 4
• The coordinates (X, Y) of the endpoints of each link; 5
• The width of the “highway mixing zone” (see below);
• The type of link (“at grade,” “fill,” “bridge,” or “depressed”);
• The height of the roadway relative to the surrounding ground (not to exceed ±10
meters); 6 and
• The hourly flow of traffic (vehicles per hour).

CAL3QHCR treats the area over each roadway link as a “mixing zone” that accounts for
the area of turbulent air around the roadway resulting from vehicle-induced turbulence.
The width of the mixing zone is an input to the model. Users should specify the width of
a link in CAL3QHCR as the width of the traveled way (traffic lanes, not including
shoulders) plus three meters on either side. Users should treat divided highways as two
separate links. See Section 7.6 of the guidance for more information on placing
receptors.

J.3.2 Timing of emissions in CAL3QHCR

The CAL3QHCR User Guide describes two methods for accepting time-varying
emissions and traffic data; these are labeled the “Tier I” and “Tier II” approaches. 7

4
Traffic engineering plans and diagrams may include information such as the number, width, and
configuration of lanes, turning channels, intersection dimensions, and ramp curvature, as well as
operational estimates such as locations of weave and merge sections and other descriptions of roadway
geometry that may be useful for specifying sources.
5
In CAL3QHCR, the Y-axis is aligned due north.
6
The CALINE3 dispersion algorithm in CAL3QHCR is sensitive to the height of the road. In particular,
the model treats bridges and above-grade “fill” roadways differently. It also handles below-grade roadways
with height of less than zero (0) meters as “cut” sections. Information on the topological features of the
project site is needed to make such a determination. Note that in the unusual circumstance that a roadway
is more than ten meters below grade, CALINE3 has not been evaluated, so CAL3QHCR is not
recommended for application. In this case, the relevant EPA Regional Office should be consulted for
determination of the most appropriate model.
7
This nomenclature is unrelated to EPA’s motor vehicle emission standards and the design value
calculation options described in Section 9 of this guidance.

J-3
Project-level PM hot-spot modeling should use the Tier II method, which can
accommodate different hourly emission patterns for each day of the week. Most
emissions data will not be so detailed, but the Tier II approach can accommodate
emissions data similar to that described in Sections 4 and 5 of the guidance. The
CAL3QHCR Tier I approach should not be used, as it employs only one hour of
emissions and traffic data and therefore cannot accommodate the emissions data required
in a PM hot-spot analysis.

Through the IPATRY keyword, CAL3QHCR allows up to seven 24-hour profiles

representing hour-specific emission, traffic, and signalization (ETS) data for each day of
the week. Depending on the number of MOVES runs, the emission factors should be
mapped to the appropriate hours of the day. For example, peak traffic emissions data for
each day would be mapped to the CAL3QHCR entry hours corresponding to the relevant
times of day (in this case, the morning and afternoon peak traffic periods). If there are
more MOVES runs than the minimum specified in the Section 4, they should be modeled
and linked to the correct days and hours using IPATRY.

As described in Section 7 of the guidance, the number of CAL3QHCR runs required for a
given PM hot-spot analysis will vary based on the amount of meteorological data
available.

J.3.3 Physical characteristics and locations of sources in AERMOD

The following discussion gives guidance on how to best characterize a source.

AERMOD includes different commands (keywords) for volume, area, and point sources.

Modeling Volume Sources

Many different sources in a project undergoing a PM hot-spot analysis might be modeled

as volume sources. Examples include areas designated for truck or bus queuing or idling
(e.g., off-network links in MOVES), driveways and pass-throughs in transit or freight
terminals, and locomotive emissions. 8 AERMOD can also approximate a highway “line
source” using a series of adjacent volume sources (see the AERMOD User Guide for
suggestions). Certain nearby sources that have been selected to be modeled may also be
appropriately treated as a volume source (see Section 8 of the guidance for more
information on considering background concentrations from other sources).

Volume source parameters are entered using the source parameter (SRCPARAM)
keyword in the AERMOD input file. This requires the user to provide the following
information:
• The emission rate (mass per unit time, such as g/s);
• The initial lateral dispersion coefficient determined from the initial lateral
dimension (width) of the volume;

8
See Section 6 and Appendix I for information regarding calculating locomotive emissions.

J-4
 • The initial vertical dispersion coefficient determined from the initial vertical
dimension (height) of the volume; and
• The source release height of the volume source center, (i.e., meters above the
ground).

Within AERMOD, the volume source algorithms are applicable to line sources with some
initial plume depth (e.g., highways, rail lines). 9 There are three inputs needed to
characterize the initial size of a roadway plume:

1. Initial lateral dispersion coefficient (σyo, Syinit). First, estimate the initial lateral
dimension (or width) of the volume source. One of the following options can be
used:
a) The average vehicle width plus 6 meters, when modeling a single lane of
traffic;
b) The road width multiplied by 2; or
c) A set width, such as 10 meters per lane of traffic.

To specify the initial lateral dispersion coefficient (σyo), referred to as Syinit in

AERMOD, the AERMOD User Guide recommends dividing the initial width by 2.15.
This is to ensure that the overlapping distributions from adjacent volume sources
simulate a line source of emissions.

2. Initial vertical dispersion coefficient (σzo, Szinit). First, estimate the initial vertical
dimension (height) of the plume for volume sources. A typical approach is to assume
it is about 1.7 times the average vehicle height, to account for the effects of vehicle-
induced turbulence. For light-duty vehicles, this is about 2.6 meters, using an average
vehicle height of 1.53 meters or 5 feet. For heavy-duty vehicles, this is about 6.8
meters, using an average vehicle height of 4.0 meters. Since most road links will
consist of a combination of light-duty and heavy-duty traffic, the initial vertical
dimension should be a combination of their respective values. There are two options
available to estimate initial vertical dimension:
a) Estimate the initial vertical dimension using an emissions-weighted average.
For example, if light-duty and heavy-duty vehicles contribute 40% and 60%
of the emissions of a given volume source, respectively, the initial vertical
dimension would be (0.4 * 2.6) + (0.6 * 6.8) = 5.1 meters.
b) Alternatively, the initial vertical dimension may be estimated using a traffic
volume weighted approach based on light-duty and heavy-duty vehicle
fractions.

The AERMOD User Guide recommends that the initial vertical dispersion coefficient
(σzo), termed Szinit in AERMOD, be estimated for a surface-based volume source by
dividing the initial vertical dimension by 2.15. For typical light-duty vehicles, this

9
The vehicle-induced turbulence around roadways with moving traffic suggests that prior to transport
downwind, a roadway plume has an initial size; that is, the emissions from the tailpipe are stirred because
the vehicle is moving and therefore the plume “begins” from a three-dimensional volume, rather than from
a point source (the tailpipe).

J-5
 corresponds to a Szinit (σzo) of 1.2 meters. For typical heavy-duty vehicles, the initial
value of Szinit (σzo) is 3.2 meters.

3. Source release height. The source release height (Relhgt in AERMOD), which is
the height at which wind effectively begins to affect the plume, may be estimated
from the midpoint of the initial vertical dimension. For moving light-duty vehicles,
this is about 1.3 meters. For moving heavy-duty vehicles, it is 3.4 meters. Since
most road links will consist of a combination of light-duty and heavy-duty traffic, the
source release height should be a combination of their respective values. There are
two options available to estimate source release height:
a) Estimate using an emissions-weighted average. For a 40% light-duty and
60% heavy-duty emissions share, the source release height would be (0.4 *
1.3) + (0.6 * 3.4) = 2.6 meters.
b) Alternatively, the source release height may be estimated using a traffic
volume weighted approach based on light-duty and heavy-duty YHKLFOHIUDFWLRQV.

Another way of dealing with Syinit, Szinit, and/or Relhgt parameters that change as a
result of different fractions of light-duty and heavy-duty vehicles is to create two
overlapping versions of each roadway source, corresponding to either light-duty and
heavy-duty traffic. These two sources could be superimposed in space, but have
emission rates and Syinit, Szinit, and Relhgt parameters that are specific to light-duty or
heavy-duty traffic.

Also, AERMOD (version dated 09292) allows Syinit, Szinit, and Relhgt to change by
hour of the day, which may be considered if the fraction of heavy-duty vehicles is
expected to significantly change throughout a day. Users should consult the latest
information on AERMOD when starting a PM hot-spot analysis.

Groups of idling vehicles may also be modeled as one or more volume sources. In those
cases, the initial dimensions of the source, dispersion coefficients, and release heights
should be calculated assuming that the vehicles themselves are inducing no turbulence.

When using adjacent volume sources to represent emissions from a source such as a
roadway, a sufficient number of volume sources should be employed to represent a
consistent density of emissions for a single link in a MOVES or EMFAC analysis. In
addition, when the source-receptor spacing in AERMOD is shorter than the distance
between adjacent volume sources, AERMOD may produce aberrant results. In the
present version of the model, receptors within a volume source in AERMOD are assigned
concentrations of zero. When volume sources are used and publicly-accessible locations
are closer to a source than the distance between adjacent volume sources, it is
recommended that smaller volume sources be used with shorter spacing between them.

For example, for such a segment along a highway segment, individual lanes might be
modeled discretely, rather than using a single volume source for all lanes. This will
reduce the spacing between volume sources and increase the quality of results closest to a
source. Receptors near area and point sources are not affected by this concern.

J-6
Consult the AERMOD User Guide and AERMOD Implementation Guide for details in
applying AERMOD to roadway sources.

Modeling Area Sources

AERMOD can represent rectangular, polygon-shaped, and circular area sources using the
AREA, AREAPOLY, or AREACIRC keywords. Sources that may be modeled as area
sources may include areas within which emissions occur relatively evenly, such as a
single link modeled using MOVES or EMFAC. 10 Evenly-distributed ground-level
sources might also be modeled as area sources.

AERMOD requires the following information when modeling an area source:

• The emission rate per unit area (mass per unit area per unit time);
• The release height above the ground;
• The length of the north-south side of the area;
• The length of the east-west side of the area (if the area is not a square);
• The orientation of the rectangular area in degrees relative to north; and
• The initial height (vertical dimension) of the area source plume.

In using a series of area sources to represent emissions of a roadway, the release height
and initial vertical dimension of the plume should be calculated as described above for
volume sources.

Modeling Point Sources

It may be appropriate to model some emission sources as fixed point sources, such as
exhaust fans or stacks on a bus garage or terminal building. If a source is modeled with
the POINT keyword in AERMOD, the model requires:
• The emission rate (mass per unit time);
• The release height above the ground;
• The exhaust gas exit temperature;
• The stack gas exit velocity; and,
• The stack inside diameter in meters.

These parameters can often be estimated using the plans and engineering diagrams for
ventilation systems.

For projects with emissions on or near rooftops, such as bus terminals or garages,
building downwash should also be modeled for the relevant sources. The potential for
building downwash should also be addressed for nearby sources whose emissions are on
or near rooftops in the project area. Building downwash occurs when air moving over a

10
At present, the area sources in AERMOD do not include AERMOD’s “plume meander approach.”
Consult the latest version of the AERMOD Implementation Guide for the most current information on
when volume sources or area sources are most appropriate.

J-7
building mixes to the ground on the “downwind” side of the building. AERMOD
includes algorithms to model the effects of building downwash on plumes from nearby or
adjacent point sources. Consult the AERMOD User Guide for additional detail on how
to enter building information.

J.3.4 Placement and sizing of sources within AERMOD

There are several general considerations with regard to placing and sizing sources within
AERMOD.

First, volume, area, and point sources should be placed in the locations where emissions
are most likely to occur. For example: if buses enter and exit a bus terminal from a single
driveway, the driveway should be modeled using one or more discrete volume or area
sources in the location of that driveway, rather than spreading the emissions from that
driveway across the entire terminal yard.

Second, for emissions from the sides or tops of buildings (as may be found from a bus
garage exhaust fan), it may be necessary to use the BPIPPRIME utility in AERMOD to
appropriately capture the characteristics of these emissions (such as downwash).

Third, the initial dimensions and other parameters of each source should be as realistic as
is feasible. Chapter 3 of the AERMOD User Guide includes recommendations for how
to appropriately characterize the shape of area and volume sources.

Finally, if nearby sources are to be included in air quality modeling (see discussion in
Section 8 of the guidance), a combination of all these source types may be needed to
appropriately represent their emissions within AERMOD. For instance, evenly-
distributed ground-level sources might also be modeled as area sources, while a nearby
power plant stack might be modeled as a point source.

J.3.5 Timing of emissions in AERMOD

Within AERMOD, emissions that vary across a year should be described with the
EMISFACT keyword (see Section 3.3.5 of the AERMOD User Guide). The number of
quarters that need to be analyzed may vary based on a particular PM hot-spot analysis.
See Section 2.5 of the guidance for more information on when PM emissions need to be
evaluated, and Sections 4 and 5 of the guidance on determining the number of MOVES
and EMFAC runs.

The Qflag parameter under EMISFACT may be used with a secondary keyword to
describe different patterns of emission variations throughout a year. Note that AERMOD
defines seasons in the following manner: winter (December, January, February), spring
(March, April, May), summer (June, July, August), and fall (September, October,
November). Emission data obtained from MOVES or EMFAC should be appropriately
matched with the relevant time periods in AERMOD. For example, if four MOVES or
EMFAC runs are completed (one for each quarter of a year), there are emission estimates

J-8
corresponding to four months of the year (January, April, July, October) and peak and
average periods within each day. In such a circumstance, January runs should be used to
represent all AERMOD winter months (December, January, February), April runs for all
spring months (March, April, May), July runs for all summer months (June, July,
August), and October for all fall months (September, October, November).

If separate weekend emission rates are available, season-specific weekday runs should be
used for the Monday-Friday entries; weekend runs would be assigned to the Saturday and
Sunday entries. The peak/average runs for each day should be mapped to the AERMOD
entry hours corresponding to the relevant time of day from the traffic analysis. Qflag can
be used to represent emission rates that vary by season, hour of day, and day of the week.
Consult the AERMOD User Guide for details.

J.4 INCORPORATING METEOROLOGICAL DATA

This discussion supplements Section 7.5 of the guidance and describes in more detail
how to handle meteorological data in AERMOD and CAL3QHCR. Section 7.2.3 of
Appendix W to 40 CFR Part 51 provides the basis for determining the urban/rural status
of a source. Consult the AERMOD Implementation Guide for instructions on what type
of population data should be used in making urban/rural determinations.

J.4.1 Specifying urban or rural sources in AERMOD

As described in Section 7 of the guidance, AERMOD employs nearby population as a

surrogate for the magnitude of differential urban-rural heating (i.e., the urban heat island
effect). When modeling urban sources in AERMOD, users should use the URBANOPT
keyword to enter this data.

When considering urban roughness lengths, users should consult the AERMOD
Implementation Guide. Any application of AERMOD that utilizes a value other than 1
meter for the urban roughness length should be considered a non-regulatory application
and would require appropriate documentation and justification as an alternate model (see
Section 7.3.3 of the guidance).

For urban applications using representative National Weather Service (NWS)

meteorological data, consult the AERMOD Implementation Guide. For urban
applications using NWS data, the URBANOPT keyword should be selected, regardless of
whether the NWS site is located in a nearby rural or urban setting. When using site-
specific meteorological data in urban applications, consult the AERMOD Implementation
Guide.

J-9
J.4.2 Specifying urban or rural sources in CAL3QHCR

CAL3QHCR requires that users specify the run as being rural or urban using the “RU”
keyword. 11 Users should make the appropriate entry depending if the source is
considered urban or rural as described in Section 7.5.5 of the guidance.

J.5 MODELING COMPLEX TERRAIN

This discussion supplements Section 7.5 of the guidance and describes in more detail
how to address complex terrain in AERMOD and CAL3QHCR. In most situations, the
project area should be modeled as having flat terrain. Additional detail on how this
should be accomplished in each model is found below. However, in some situations a
project area may include complex terrain, such that sources and receptors included in the
model are found at different heights.

J.5.1 AERMOD

This guidance reflects the AERMOD Implementation Guide as of March 19, 2009.
Analysts should consult the most recent AERMOD Implementation Guide for the latest
guidance on modeling complex terrain.

For most highway and transit projects, the analyst should apply the non-DFAULT option
in AERMOD and assume flat, level terrain. In the AERMOD input file, the FLAT option
should be used in the MODELOPT keyword. This recommendation is made to avoid
underestimating concentrations in two circumstances likely to occur with the low-
elevation, non-buoyant emissions from transportation projects. First, in DFAULT mode,
AERMOD will tend to underestimate concentrations from low-level, non-buoyant
sources where there is up-sloping terrain with downwind receptors uphill since the
DFAULT downwind horizontal plume will pass below the actual receptor elevation.
Second, in DFAULT mode, AERMOD will tend to underestimate concentrations when a
plume is terrain-following. Therefore, the FLAT option should be selected in most cases.

There may be some cases where significant concentrations result from nearby elevated
sources. In these cases, interagency consultation should be used on a case-by-case basis
to determine whether to include terrain effects and use the DFAULT option. In those
cases, AERMAP should be used to prepare input files for AERMOD; consult the
AERMOD and AERMAP user guides and the latest AERMOD Implementation Guide
for information on obtaining and processing relevant terrain data.

11
Specifying urban modeling with the “RU” keyword converts stability classes E and F to D.

J-10
J.5.2 CAL3QHCR

CAL3QHCR does not handle complex terrain. No action is therefore required.

J.6 RUNNING THE MODEL AND OBTAINING RESULTS

This discussion supplements Section 7.7 of the guidance and describes in more detail
how to handle data outputs in AERMOD and CAL3QHCR. AERMOD and CAL3QHCR
produce different output file formats, which must be post-processed in different ways to
enable calculation of design values as described in Section 9.3 of the guidance. This
guidance is applicable regardless of how many quarters are being modeled.

J.6.1 AERMOD output

AERMOD requires that users specify the type and format of output files in the main input
file for each run. See Section 3.7 of the AERMOD User Guide for details on the various
output options. Output options should be specified to enable the relevant design value
calculations required in Section 9.3. Note that many users will have multiple years of
meteorological data, so multiple output files may be required (unless the meteorological
files have been joined prior to running AERMOD).

For the annual PM2.5 design value calculations described in Section 9.3.2, averaging
times should be specified that allow calculation of the annual average concentrations at
each receptor. For example, when using five years of meteorological data, the PERIOD
averaging time could be specified using the CO AVERTIME keyword.

For the 24-hour PM2.5 design value calculations described in Section 9.3.3, the
DAYTABLE option provides output files with 24-hour concentrations at each receptor
for each day processed. Users should flag the quarter and year for each day listed in the
DAYTABLE that AERMOD generates. Note users should also specify a 24-hour
averaging time with the CO AVERTIME command as well.

Another option for calculating 24-hour PM2.5 design values is with a POSTFILE, a file of
results at each receptor for each day processed. By specifying a POSTFILE with a 24-
hour averaging time, a user can generate a file of daily concentrations for each day of
meteorological data. When using this option, users should specify a POSTFILE with a
24-hour averaging time to generate the outputs needed to calculate design values and flag
the quarter and year for each day listed in the POSTFILE that AERMOD generates. Note
that POSTFILE output files can be very large.

For the 24-hour PM10 calculations described in Section 9.3.4, the RECTABLE keyword
may be used to obtain the six highest 24-hour concentrations over the entire modeling
period. A RECTABLE is a file summarizing the highest concentrations at each receptor
over an averaging period (e.g., 24 hours) across a modeling period (e.g., 5 years).

J-11
EPA is actively working towards a post-processing tool for AERMOD that will provide
the appropriate modeling metrics that may then be combined with background
concentrations for comparisons to the PM NAAQS. EPA will announce these new
options as they become available on EPA’s SCRAM website at:
www.epa.gov/scram001/.

J.6.2 CAL3QHCR output

For each year of meteorological data and quarterly emission inputs, CAL3QHCR reports
the five highest 24-hour concentrations and the quarterly average concentrations in its
output file.

For calculating annual PM2.5 design values using CAL3QHCR output, some post-
processing is required. CAL3QHCR’s output file refers to certain data under the display:
“THE HIGHEST ANNUAL AVERAGE CONCENTRATIONS.” If four quarters of
emission data are separately run in CAL3QHCR, each quarter’s outputs listed under
“THE HIGHEST ANNUAL AVERAGE CONCENTRATIONS” are actually quarterly-
average concentrations. As described in Section 7, per year of meteorological data,
CAL3QHCR should be run for as many quarters as analyzed using MOVES and
EMFAC, as CAL3QHCR accepts only a single quarter’s emission factors per input file.

Calculating 24-hour PM2.5 design values under a first or second tier analysis is described
in Section 9.3.3. To get annual average modeled concentrations for a first tier analysis
(Step 1), the highest 24-hour concentrations in each quarter and year of meteorological
data should be identified. Within each year of meteorological data, the highest 24-hour
concentration at each receptor should be identified. For a first tier analysis, at each
receptor, the highest concentrations from each year of meteorological data should be
averaged together. Under a second tier analysis, at each receptor, the highest modeled
concentration in each quarter, from each year of meteorological data, should be averaged
together. These average highest 24-hour concentrations in each quarter, across multiple
years of meteorological data, are used in second tier PM2.5 design value calculations.

In calculating 24-hour PM10 design values, it is necessary to estimate the sixth-highest

concentration in each year if using five years of meteorological data. For each period of
meteorological data, CAL3QHCR outputs the five highest 24-hour concentrations. To
estimate the sixth-highest concentration at a receptor, the five highest 24-hour
concentrations from each quarter and year of meteorological data should be arrayed
together and ranked. From all quarters and years of meteorological data, the sixth-
highest concentration should be identified. This concentration, at each receptor, is used
in calculations of the PM10 design value described in Section 9.3.4.

J-12
 A ppendix K :
Examples of Design Value Calculations for PM Hot-spot
Analyses

K.1 INTRODUCTION
This appendix supplements Section 9’s discussion of calculating and applying design
values for PM hot-spot analyses. While this guidance can apply to any PM NAAQS, this
appendix provides examples of how to calculate design values for the PM NAAQS in
effect at the time the guidance was issued (the 1997 annual PM2.5 NAAQS, the 2006 and
1997 24-hour PM2.5 NAAQS, and the 1987 24-hour PM10 NAAQS). The design values
in this appendix are calculated using the steps described in Section 9.3. Readers should
reference the appropriate sections of the guidance as needed for more detail on how to
complete each step of these analyses.

These illustrative example calculations demonstrate the basic procedures described in the
guidance and therefore are simplified in the number of receptors considered and other
details that would occur in an actual PM hot-spot analysis. Where users would have to
repeat steps for additional receptors, it is noted. These examples are organized according
to the build/no-build analysis steps that are described in Sections 2 and 9 of this guidance.

The final part of this appendix provides mathematical formulas that describe the design
value calculations discussed in Section 9 and this appendix.

K.2 P ROJECT DESCRIPTION AND CONTE XT FOR ALL EXAMPLES

For the following examples, a PM hot-spot analysis is being done for an expansion of an
existing highway with a significant increase in the number of diesel vehicles (40 CFR
93.123(b)(1)(i)). The highway expansion will serve an expanded freight terminal. The
traffic at the terminal will increase as a result of the expanded highway project’s increase
in truck traffic, and therefore the freight terminal is projected to have higher emissions
under the build scenario than under the no-build scenario. The freight terminal is not part
of the project; however, it is a nearby source that will be included in the air quality
modeling, as described further below.

The air quality monitor selected to represent background concentrations from other
sources is a Federal Equivalent Method (FEM) monitor that is 300 meters upwind of the
project. The monitor is on a 1-in-3 day sampling schedule. In this example, the three
most recent years of monitoring data are from 2008, 2009, and 2010. Since 2008 is a
leap year (366 days), for this example, there are 122 monitored values in that year and
121 values for both 2009 and 2010 (365 days each). 1

1
Note that the number of air quality monitoring measurements may vary by year. For example, with 1-in-
3 measurements, there could be 122 or 121 measurements in a year with 365 days. Or, there may be fewer

K-1
However, through interagency consultation, it is determined that the freight terminal’s
emissions are not already captured by this air quality monitor. AERMOD has been
selected as the air quality model to estimate PM concentrations produced by the project
(the highway expansion) and the nearby source (the freight terminal). 2 There are five
years of representative off-site meteorological data being used in this analysis.

As discussed in Section 2.4, a project sponsor could consider mitigation and control
measures at any point in the process. However, since the purpose of these examples is to
show the design value calculations, in this appendix such measures are not considered
until after the calculations are done.

K.3 EXAMPLE: ANNUAL PM2.5 NAAQS

K.3.1 General

This example illustrates the approach to calculating design values for comparison to the
annual PM2.5 NAAQS, as described in Section 9.3.2. The annual PM2.5 design value is
the average of three consecutive years’ annual averages. The design value for
comparison is rounded to the nearest tenth of a μg/m3 (nearest 0.1 μg/m3). For example,
15.049 rounds to 15.0, and 15.050 rounds to 15.1. 3

Each year’s annual average concentrations include contributions from the project, any
nearby sources modeled, and background concentrations. For air quality monitoring
purposes, the annual PM2.5 NAAQS is met when the three-year average concentration is
less than or equal to the current annual PM2.5 NAAQS (i.e., 15.0 μg/m3):

Annual PM2.5 design value = ([Y1] average + [Y2] average + [Y3] average) ÷ 3

Where:
[Y1] = Average annual PM2.5 concentration for the first year of air quality
monitoring data
[Y2] = Average annual PM2.5 concentration for the second year of air quality
monitoring data
[Y3] = Average annual PM2.5 concentration for the third year of air quality
monitoring data

actual monitored values if sampling was not conducted on some scheduled days or the measured value was
invalidated due to quality assurance concerns. The actual number of samples with valid data should be
used.
2
EPA notes that CAL3QHCR could not be used in this particular PM hot-spot analysis, since air quality
modeling included the project and a nearby source. See Section 7.3 of the guidance for further information.
3
A sufficient number of decimal places (3-4) should be retained during intermediate calculations for design
values, so that there is no possibility of intermediate rounding or truncation affecting the final result.
Rounding to the tenths place should only occur during final design value calculations, pursuant to
Appendix N to 40 CFR Part 50.

K-2
For this example, the project described in Appendix K.2 is located in an annual PM2.5
NAAQS nonattainment area. This example illustrates how an annual PM2.5 design value
could be calculated at the same receptor in the build and no-build scenarios, based on air
quality modeling results and air quality monitoring data. In an actual PM hot-spot
analysis, design values would be calculated at additional receptors, as described further in
Section 9.3.2.

K.3.2 Build scenario

For the build scenario, the PM2.5 impacts from the project and from the nearby source are
estimated with AERMOD at all receptors. 4

Steps 1-2. Because AERMOD is used for this project, Step 1 is skipped. The receptor
with the highest average annual concentration, using five years of meteorological data, is
identified directly from the AERMOD output. This receptor’s average annual
concentration is 3.603 μg/m3.

Step 3. Based on the three years of measurements at the background air quality monitor,
the average monitored background concentrations in each quarter is determined. Then,
for each year of background data, the four quarters are averaged to get an average annual
background concentration (last column of Exhibit K-1). These three average annual
background concentrations are averaged, and the resulting value is 11.582 μg/m3, as
shown in Exhibit K-1:

Exhibit K-1. Background Concentrations

Background Average
Q1 Q2 Q3 Q4
Concentrations Annual
2008 13.013 17.037 8.795 8.145 11.748
2009 14.214 14.872 7.912 7.639 11.159
2010 11.890 16.752 9.421 9.287 11.838
3-year average: 11.582

Step 4. The 3-year average annual background concentration (from Step 3) is added to
the average annual modeled concentration from the project and nearby source (from Step
2):
11.582 + 3.603 = 15.185

Step 5. Rounding to the nearest 0.1 µg/m3 produces a design value of 15.2 µg/m3.

4
As noted above, there is a single nearby source that is projected to have higher emissions under the build
scenario than the no-build scenario as a result of the project and its impacts are not expected to be captured
by the monitor chosen to provide background concentrations. Therefore, emissions from the project and
this nearby source are both included in the AERMOD output.

K-3
In this example, the concentration at the highest receptor is estimated to exceed the
current annual PM2.5 NAAQS of 15.0 µg/m3.

Steps 6-8: Since the design value in Step 5 is greater than the NAAQS, design value
calculations are then completed for all receptors in the build scenario, and receptors with
design values above the NAAQS are identified. After this is done, the no-build scenario
is modeled for comparison.

K.3.3 No-build scenario

The no-build scenario (i.e., the existing highway and freight terminal without the
proposed highway and freight terminal expansion), is modeled at all of the receptors in
the build scenario, but design values are only calculated in the no-build scenario at
receptors where the design value for the build scenario is above the annual PM2.5 NAAQS
(from Steps 6-8 above).

Step 9. For this example, the receptor with the highest average annual concentration in
the build scenario is used to illustrate the no-build scenario design value calculation. The
average annual concentration modeled at this receptor in the no-build scenario is 3.521
µg/m3.

Step 10. The background concentrations from the representative monitor are unchanged
from the build scenario, so the average annual modeled concentration of 3.521 is added to
the 3-year average annual background concentrations of 11.528 µg/m3 from Step 3:
11.582 + 3.521 = 15.103

Step 11. Rounding to the nearest 0.1 µg/m3 produces a design value of 15.1 µg/m3.

In this example, the design value at the receptor in the build scenario (15.2 μg/m3) is
greater than the design value at the same receptor in the no-build scenario (15.1 μg/m3). 5
In an actual PM hot-spot analysis, design values would also be compared between build
and no-build scenarios at all receptors in the build scenario that exceeded the annual
PM2.5 NAAQS. The interagency consultation process would then be used to discuss next
steps, e.g., appropriateness of receptors. Refer to Sections 9.2 and 9.4 for additional
details.

If it is determined that conformity requirements are not met at all appropriate receptors,
the project sponsor should then consider additional mitigation or control measures, as
discussed in Section 10. After measures are selected, a new build scenario that includes
the controls should be modeled and new design values calculated. Design values for the
no-build scenario shown above would not need to be recalculated since the no-build
scenario would not change.

5
Values are compared after rounding. As long as the build design value is no greater than the no-build
design value after rounding, the project would meet conformity requirements at a given receptor, even if
the pre-rounding build design value is greater than the pre-rounding no-build design value.

K-4
K.4 EXAMPLE: 24-HOUR PM2.5 NAAQS
K.4.1 General

This example illustrates the two-tiered approach to calculating design values for
comparison with the 24-hour PM2.5 NAAQS, as described in Section 9.3.3. The 24-hour
design value is the average of three consecutive years’ 98th percentile PM2.5 concentration
of 24-hour values for each of those years. For air quality monitoring purposes, the
NAAQS is met when that three-year average concentration is less than or equal to the
currently applicable 24-hour PM2.5 NAAQS for a given area’s nonattainment designation
(35 µg/m3 for nonattainment areas for the 2006 PM2.5 NAAQS and 65 µg/m3 for
nonattainment areas for the 1997 PM2.5 NAAQS). 6 The design value for comparison to
any 24-hour PM2.5 NAAQS is rounded to the nearest 1 µg/m3 (i.e., decimals 0.5 and
greater are rounded up to the nearest whole number, and any decimal lower than 0.5 is
rounded down to the nearest whole number). For example, 35.499 rounds to 35 µg/m3,
while 35.500 rounds to 36. 7

For this example, the project described in Appendix K.2 is located in a nonattainment
area for the 2006 24-hour PM2.5 NAAQS. This example presents first tier and second tier
build scenario results for a single receptor to illustrate how the calculations should be
made based on air quality modeling results and air quality monitoring data. It also shows
second tier no-build scenario results for this same receptor. In an actual PM hot-spot
analysis, design values would be calculated at additional receptors, as described further in
Section 9.3.3.

As explained in Section 9.3.3, project sponsors can start with either a first or second tier
analysis. This example begins with a first tier analysis. However, it would also be
acceptable to begin with the second tier analysis and skip the first tier altogether.

K.4.2 Build scenario

PM2.5 contributions from the project and the nearby source are estimated together with
AERMOD in each of four quarters using meteorological data from five consecutive
years, using a 24-hour averaging time. As discussed in Appendix K.2 above, the one
nearby source (the freight terminal) was included in air quality modeling.

6
There are only two PM2.5 areas where conformity currently applies for both the 1997 and 2006 24-hour
NAAQS. While both 24-hour NAAQS must be considered in these areas, in practice if the more stringent
2006 24-hour PM2.5 NAAQS is met, then the 1997 24-hour PM2.5 NAAQS is met as well.
7
A sufficient number of decimal places (3-4) should be retained during intermediate calculations for design
values, so that there is no possibility of intermediate rounding or truncation affecting the final result.
Rounding should only occur during final design value calculations, pursuant to Appendix N to 40 CFR Part
50.

K-5
First Tier Analysis

Under a first tier analysis, the average highest modeled 24-hour concentrations at a given
receptor are added to the average 98th percentile 24-hour background concentrations,
regardless of the quarter in which they occur. The average highest modeled 24-hour
concentrations are produced by AERMOD, using five years of meteorological data in one
run.

Step 1. The receptor with the highest average modeled 24-hour concentration is
identified. This was obtained directly from the AERMOD output. 8 For this example, the
data from this receptor is shown in Exhibit K-2. Exhibit K-2 shows the highest 24-hour
concentration for each year of meteorological data used, regardless of the quarter in
which they were modeled. The average concentration of these outcomes, 6.710 µg/m3
(highlighted in Exhibit K-2), is the highest, compared to the averages at all of the other
receptors.

Exhibit K-2. Modeled PM2.5 Concentrations from Project and Nearby Source

Highest PM2.5
Year Concentration
Met Year 1 6.413
Met Year 2 5.846
Met Year 3 6.671
Met Year 4 7.951
Met Year 5 6.667
Average 6.710

Step 2. The average 98th percentile 24-hour background concentration for a first tier
analysis is calculated using the 98th percentile 24-hour concentrations of the three most
recent years of monitoring data from the representative air quality monitor selected (see
Appendix K.2). Since the background monitor is on a 1-in-3 day sampling schedule, it
made either 122 or 121 measurements per year during 2008 - 2010. According to Exhibit
9-5, with this number of monitored values per year, the 98th percentile is the third highest
concentration. Exhibit K-3 depicts the top eight monitored concentrations (in µg/m3) of
the monitor throughout the years employed for estimating background concentrations.

8
If CAL3QHCR were being used, some additional processing of model output would be needed. Refer to
Section 9.3.3.

K-6
Exhibit K-3. Top Eight Monitored Concentrations in Years 2008 – 2010

Rank 2008 2009 2010

1 34.123 33.537 35.417
2 31.749 32.405 31.579
3 31.443 31.126 31.173
4 30.809 30.819 31.095
5 30.219 30.487 30.425
6 30.134 29.998 30.329
7 30.099 29.872 30.193
8 28.481 28.937 28.751

The third-ranked concentration of each year (highlighted in Exhibit K-3) is the 98th
percentile value. These are averaged:
(31.443 + 31.126 + 31.173) ÷ 3 = 31.247 µg/m3.

Step 3. Then, the highest average 24-hour modeled concentration for this receptor (from
Step 1) is added to the average 98th percentile 24-hour background concentration (from
Step 2):
6.710 + 31.247 = 37.957 µg/m3.

Rounding to the nearest whole number results in a 24-hour PM2.5 design value of 38
µg/m3.

Because this concentration is greater than the 2006 24-hour PM2.5 NAAQS (35 µg/m3),
this first tier analysis does not demonstrate that conformity is met. As described in
Section 9.3.3, the project sponsor has two options:
• Repeat the first tier analysis for the no-build scenario at all receptors that
exceeded the NAAQS in the build scenario. If the calculated design value for the
build scenario is less than or equal to the design value for the no-build scenario at
all of these receptors, then the project conforms; 9 or
• Conduct a second tier analysis.

In this example, the next step chosen is to conduct a second tier analysis.

Second Tier Analysis

In a second tier analysis, the highest modeled concentrations are not added to the 98th
percentile background concentrations on a yearly basis. Instead, a second tier analysis
uses the average of the highest modeled 24-hour concentration within each quarter of
each year of meteorological data. Impacts from the project, nearby sources, and other
background concentrations are calculated on a quarterly basis before determining the 98th
9
In certain cases, project sponsors can also decide to calculate the design values for all receptors in the
build and no-build scenarios and use the interagency consultation process to determine whether a “new”
violation has been relocated (see Section 9.2).

K-7
percentile concentration resulting from these inputs. The steps presented below follow
the steps described in Section 9.3.3.

Step 1. The first step is to count the number of measurements for each year of
monitoring data used for background concentrations. As described in Appendix K.2 and
in Step 2 of the first tier analysis above, there are 122 monitored values during 2008, 121
values during 2009, and 121 values during 2010.

Step 2. For each year of monitoring data, the eight highest 24-hour background
concentrations in each quarter are determined. The eight highest concentrations in each
quarter of 2008, 2009, and 2010 are shown in Exhibit K-4.

Exhibit K-4. Eight Highest 24-hour Background Concentrations By Quarter for

Each Year

Rank of
Year Background Q1 Q2 Q3 Q4
Concentration
1 27.611 31.749 34.123 30.099
2 25.974 30.219 31.443 28.096
3 25.760 30.134 30.809 26.990
4 25.493 28.368 28.481 25.649
2008
5 25.099 27.319 27.372 25.526
6 24.902 25.788 25.748 25.509
7 24.780 25.564 25.288 25.207
8 23.287 24.794 24.631 24.525
1 26.962 32.405 33.537 31.126
2 24.820 30.487 30.819 28.553
3 24.330 28.937 29.998 25.920
4 23.768 27.035 29.872 25.856
2009
5 23.685 25.880 25.596 25.565
6 23.287 25.867 25.148 24.746
7 23.226 25.254 24.744 24.147
8 22.698 24.268 24.267 23.142
1 27.493 31.579 35.417 30.425
2 24.637 31.173 31.095 26.927
3 24.637 30.193 30.329 26.263
4 24.392 27.994 28.751 25.684
2010
5 24.050 25.439 26.084 25.170
6 23.413 24.253 24.890 24.254
7 22.453 23.006 24.749 23.425
8 22.061 21.790 22.538 22.891

K-8
Step 3. The highest modeled 24-hour concentrations in each quarter are identified at each
receptor. Exhibit K-5 presents the highest 24-hour concentrations within each quarter at
one receptor (for each of the five years of meteorological data used in air quality
modeling) as well as the average of these quarterly concentrations. This step would be
repeated for each receptor in an actual PM hot-spot analysis.

Exhibit K-5. Highest Modeled 24-hour Concentrations Within Each Quarter (Build
Scenario)

Q1 Q2 Q3 Q4
Met Year 1 6.413 3.332 6.201 6.193
Met Year 2 3.229 3.481 5.846 4.521
Met Year 3 6.671 3.330 5.696 6.554
Met Year 4 7.095 3.584 7.722 7.951
Met Year 5 6.664 4.193 4.916 6.667
Average 6.014 3.584 6.076 6.377

The average highest concentrations on a quarterly basis (i.e., the values highlighted in
Exhibit K-5) constitute the contributions of the project and nearby source to the projected
24-hour PM2.5 design value, and are used in subsequent calculations.

Step 4. For each receptor, the highest modeled 24-hour concentration in each quarter
(from Step 3) is added to each of the eight highest monitored concentrations for the same
quarter for each year of monitoring data (from Step 2). To obtain this result, the average
highest modeled concentration for each quarter, found in the last row of Exhibit K-5, is
added to each of the eight highest background concentrations in each quarter in Exhibit
K-4. The results are shown in Exhibit K-6.

K-9
Exhibit K-6. Sum of Modeled and Monitored Concentrations (Build Scenario)

Rank of
Year Background Q1 Q2 Q3 Q4
Concentration
1 33.625 35.333 40.200 36.476
2 31.989 33.803 37.520 34.474
3 31.774 33.718 36.886 33.368
4 31.507 31.952 34.557 32.026
2008
5 31.113 30.903 33.448 31.903
6 30.916 29.372 31.824 31.886
7 30.794 29.148 31.365 31.584
8 29.301 28.378 30.707 30.902
1 32.976 35.989 39.613 37.503
2 30.835 34.071 36.895 34.931
3 30.344 32.521 36.074 32.297
4 29.782 30.619 35.948 32.233
2009
5 29.700 29.464 31.672 31.942
6 29.301 29.451 31.225 31.124
7 29.240 28.838 30.820 30.524
8 28.712 27.852 30.343 29.520
1 33.507 35.163 41.493 36.802
2 30.651 34.757 37.172 33.304
3 30.651 33.777 36.405 32.640
4 30.406 31.578 34.827 32.062
2010
5 30.064 29.022 32.160 31.547
6 29.428 27.837 30.966 30.631
7 28.468 26.590 30.825 29.803
8 28.075 25.374 28.614 29.269

Step 5. The 32 values from each year in Exhibit K-6 are then ranked from highest to
lowest, regardless of the quarter from which each value comes. This step is shown in
Exhibit K-7. Note that only the top eight values are shown for each year instead of the
entire set of 32. Exhibit K-7 also displays the quarter from which each concentration
comes and the value’s rank within its quarter.

K-10
Exhibit K-7. Eight Highest Concentrations in Each Year, Ranked from Highest to
Lowest (Build Scenario)

Yearly Quarterly
Year μg/m3 Quarter
Rank Rank
40.200 1 Q3 1
37.520 2 Q3 2
36.886 3 Q3 3
36.476 4 Q4 1
2008
35.333 5 Q2 1
34.557 6 Q3 4
34.474 7 Q4 2
33.803 8 Q2 2
39.613 1 Q3 1
37.503 2 Q4 1
36.895 3 Q3 2
36.074 4 Q3 3
2009
35.989 5 Q2 1
35.948 6 Q3 4
34.931 7 Q4 2
34.071 8 Q2 2
41.493 1 Q3 1
37.172 2 Q3 2
36.802 3 Q4 1
36.405 4 Q3 3
2010
35.163 5 Q2 1
34.827 6 Q3 4
34.757 7 Q2 2
33.777 8 Q2 3

Steps 6-7. The value that represents the 98th percentile 24-hour concentration is
determined, based on the number of background concentration values there are. As
described in Step 1, there are 122 monitored values for the year 2008 and 121 values for
both 2009 and 2010. According to Exhibit 9-7 in Section 9.3.3, for a year with 101-150
samples per year, the 98th percentile is the 3rd highest concentration for that year.
Therefore, for this example, the 3rd highest 24-hour concentration of each year,
highlighted in Exhibit K-7, represents the 98th percentile value for that year.

Step 8. At each receptor, the average of the three 24-hour 98th percentile concentrations
is calculated. For the receptor in this example, the average is:
(36.886 + 36.895 + 36.802) ÷ 3 = 36.861

K-11
Step 9. The average for the receptor in this example from Step 8 (36.861 µg/m3) is then
rounded to the nearest whole number (37 µg/m3) and compared to the 2006 24-hour
PM2.5 NAAQS (35 µg/m3).

The design value at the receptor in this example is higher than the relevant 24-hour PM2.5
NAAQS. In an actual PM2.5 hot-spot analysis, the design value calculations need to be
repeated for all receptors, and compared to the NAAQS. Since one (and possibly more)
receptors have design values greater than the 24-hour PM2.5 NAAQS, the project will
only conform if the design value in the build scenario is less than or equal to the design
value in the no-build scenario for all receptors that exceeded the NAAQS in the build
scenario. Therefore, the no-build scenario needs to be modeled for comparison, as
described further below. Because the build scenario was modeled with a second tier
analysis, the no-build scenario must also be modeled with a second tier analysis.

K.4.3 No-build scenario

The no-build scenario is described in Section 9.3.3 as Step 10:

• Step 10. Using modeling results for the no-build scenario, repeat Steps 3 through
9 for all receptors with a design value that exceeded the PM2.5 NAAQS in the
build scenario. The result will be a 24-hour PM2.5 design value at such receptors
for the no-build scenario.

For this part of the example, air quality modeling is completed for the no-build scenario
for the same receptor as the build scenario. Steps 1 and 2 for the build scenario do not
need to be repeated, since the background concentrations in the no-build scenario are
identical to those in the build scenario. Exhibit K-4, which shows the eight highest
monitored concentrations in each quarter over three years, therefore can also be used for
the no-build scenario.

Step 3. For the same receptor examined above in the build scenario, the highest modeled
24-hour concentrations for the no-build scenario are calculated for each quarter, using
each year of meteorological data used for air quality modeling. Exhibit K-8 provides
these concentrations, as well as the quarterly averages (highlighted).

Exhibit K-8. Highest Modeled 24-hour Concentrations Within Each Quarter (No-
Build Scenario)

Q1 Q2 Q3 Q4
Met Year 1 6.757 3.383 6.725 6.269
Met Year 2 3.402 3.535 6.340 4.577
Met Year 3 7.029 3.381 6.177 6.635
Met Year 4 7.476 3.639 8.374 8.048
Met Year 5 7.022 4.258 5.331 6.748
Average 6.337 3.639 6.589 6.455

K-12
Step 4. The highest modeled 24-hour concentration in each quarter (i.e., the values in the
last row of Exhibit K-8) are added to each of the eight highest concentrations for the
same quarter for each year of monitoring data (found in Exhibit K-4), and the resulting
values are shown in Exhibit K-9.

Exhibit K-9. Sum of Modeled and Monitored Concentrations (No-Build Scenario)

Rank of
Year Background Q1 Q2 Q3 Q4
Concentration
1 33.948 35.389 40.713 36.555
2 32.312 33.858 38.033 34.552
3 32.097 33.774 37.399 33.446
4 31.830 32.007 35.070 32.104
2008
5 31.436 30.959 33.961 31.981
6 31.239 29.428 32.337 31.964
7 31.117 29.204 31.878 31.662
8 29.624 28.433 31.220 30.980
1 33.299 36.044 40.127 37.581
2 31.158 34.126 37.408 35.009
3 30.667 32.576 36.587 32.375
4 30.105 30.674 36.461 32.311
2009
5 30.023 29.520 32.185 32.020
6 29.624 29.506 31.738 31.202
7 29.563 28.894 31.333 30.602
8 29.035 27.907 30.856 29.598
1 33.830 35.218 42.007 36.880
2 30.974 34.812 37.685 33.382
3 30.974 33.832 36.918 32.719
4 30.729 31.633 35.340 32.140
2010
5 30.387 29.078 32.674 31.625
6 29.751 27.893 31.479 30.709
7 28.791 26.645 31.338 29.881
8 28.398 25.429 29.127 29.347

Step 5. The 32 values from each year in Exhibit K-9 are ranked from highest to lowest,
regardless of the quarter from which each value comes. This step is shown in Exhibit K-
10. Note that only the top eight values are shown for each year instead of the entire set of
32.

K-13
Exhibit K-10. Eight Highest Concentrations in Each Year, Ranked from Highest to
Lowest (No-Build Scenario)

Yearly Quarterly
Year μg/m3 Quarter
Rank Rank
40.713 1 Q3 1
38.033 2 Q3 2
37.399 3 Q3 3
36.555 4 Q4 1
2008
35.389 5 Q2 1
35.070 6 Q3 4
34.552 7 Q4 2
33.961 8 Q3 5
40.127 1 Q3 1
37.581 2 Q4 1
37.408 3 Q3 2
36.587 4 Q3 3
2009
36.461 5 Q3 4
36.044 6 Q2 1
35.009 7 Q4 2
34.126 8 Q2 2
42.007 1 Q3 7
37.685 2 Q1 3
36.918 3 Q1 2
36.880 4 Q4 8
2010
35.340 5 Q4 6
35.218 6 Q1 1
34.812 7 Q4 2
33.832 8 Q4 3

Steps 6-7. Based on the number of background measurements available per year in this
example (122 for 2008 and 121 for both 2009 and 2010, as discussed in the analysis of
the build scenario), Exhibit 9-7 in Section 9.3.3 indicates that the 3rd highest 24-hour
concentration in each year represents the 98th percentile concentration for that year. The
third highest concentrations are highlighted in Exhibit K-10.

Step 8. For this receptor, the average of the Rank 3 concentrations in 2008, 2009, and
2010 is calculated:
(37.399 + 37.408 + 36.918) ÷ 3 = 37.242

Step 9. The average for the receptor in this example from Step 8 (37.242 µg/m3) is
rounded to the nearest whole µg/m3 (37 µg/m3).

K-14
In this example, the design value at this receptor for both the build and no-build scenarios
is 37 µg/m3, which is greater than the 2006 24-hour NAAQS (35 mg/m3). However, the
build scenario’s design value is equal to the design value in the no-build scenario. 10 For
the project to conform, the build design values must be less than or equal to the no-build
value for all the receptors that exceeded the NAAQS in the build scenario. Assuming
that this is the case at all other receptors, the proposed project in this example would
therefore demonstrate conformity.

K.5 EXAMPLE: 24-HOUR PM10 NAAQS

K.5.1 General

This example illustrates calculating design values for comparison with the 24-hour PM10
NAAQS, as described in Section 9.3.4. The 24-hour PM10 design value is based on the
expected number of 24-hour exceedances of 150 µg/m3, averaged over three consecutive
years. For air quality monitoring purposes, the NAAQS is met when the number of
exceedances is less than or equal to 1.0. The 24-hour PM10 design value is rounded to the
nearest 10 µg/m3. For example, 155.511 rounds to 160, and 154.999 rounds to 150. 11

The 24-hour PM10 design value is calculated at each air quality modeling receptor by
directly adding the sixth-highest modeled 24-hour concentration (if using five years of
meteorological data) to the highest 24-hour background concentration (from three years
of monitored data).

For this example, the project described in Appendix K.2 is located in a nonattainment
area for the 24-hour PM10 NAAQS. This example presents build scenario results for a
single receptor to illustrate how the calculations should be made based on air quality
modeling results and air quality monitoring data.

10
Values are compared after rounding. As long as the build design value is no greater than the no-build
design value after rounding, the project would meet conformity requirements at a given receptor, even if
the pre-rounding build design value is greater than the pre-rounding no-build design value.
11
A sufficient number of decimal places (3-4) in modeling results should be retained during intermediate
calculations for design values, so that there is no possibility of intermediate rounding or truncation
affecting the final result. Rounding to the nearest 10 ug/m3 should only occur during final design value
calculations, pursuant to Appendix K to 40 CFR Part 50. Monitoring values typically are reported with
only one decimal place.

K-15
K.5.2 Build Scenario

Step 1. From the air quality modeling results from the build scenario, the sixth-highest
24-hour concentration is identified at each receptor. These sixth-highest concentrations
are the sixth highest that are modeled at each receptor, regardless of year of
meteorological data used. 12 AERMOD was configured to produce these values.

Step 2. The sixth-highest modeled concentrations (i.e., the concentrations at Rank 6) are
compared across receptors, and the receptor with the highest value at Rank 6 is identified.
For this example, the highest sixth-highest 24-hour concentration at any receptor is
15.218 µg/m3. (That is, at all other receptors, the sixth-highest concentration is less than
15.218 µg/m3.) Exhibit K-11 shows the six highest 24-hour concentrations at this
receptor.

Exhibit K-11. Receptor with the Highest Sixth-Highest 24-Hour Concentration

(Build Scenario)

Highest 24-Hour
Rank
Concentrations
1 17.012
2 16.709
3 15.880
4 15.491
5 15.400
6 15.218

Step 3. The highest 24-hour background concentration from the three most recent years
of monitoring data (2008, 2009, and 2010) is identified. In this example, the highest 24-
hour background concentration from these three years is 86.251 µg/m3.

Step 4. The sixth-highest 24-hour modeled concentration of 15.218 µg/m3 from the
highest receptor (from Step 2) is added to the highest 24-hour background concentration
of 86.251 µg/m3 (from Step 3):
15.218 + 86.251 = 101.469

Step 5. This sum is rounded to the nearest 10 µg/m3, which results in a design value of
100 µg/m3.

12
The six highest concentrations could occur anytime during the five years of meteorological data. They
may be clustered in one or two years, or they may be spread out over several, or even all five, years of the
meteorological data.

K-16
This result is then compared to the 24-hour PM10 NAAQS. In this case, the concentration
calculated at all receptors is less than the 24-hour PM10 NAAQS of 150 µg/m3, therefore
the analysis shows that the project conforms. However, if the design value for this
receptor had been greater than 150 µg/m3, the remainder of the steps in Section 9.3.4
would be completed. That is, build scenario design values for each receptor would be
calculated (Steps 6-7 in Section 9.3.4) and, for all those that exceed the NAAQS, the no-
build design values would also be calculated (Steps 8-10 in Section 9.3.4). The build and
no-build design values would then be compared. 13

K.6 MATHEMATICAL FORMULAS FOR DESIGN VALUE CALCULATIONS

K.6.1 Introduction

This part of the appendix includes mathematical formulas to represent the calculations
described narratively in Section 9.3. This information is intended to supplement Section
9, which may be helpful for certain users.

Appendix K.6 relies on conventions of mathematical and logical notation that are
described after the formulas are presented. Several symbols are used that may be useful
to review prior to reading the individual formulas.

Notation symbols

• x - a single bar over variable x represents a single arithmetic mean of that

variable
• x - double bars over variable x represents an “average of averages”
• x̂ - a “hat” over variable x represents the arithmetic of multiple high
concentration values from different years, either from monitoring data or from
modeling results

Logical symbols

• ∀x - an upside down A before variable x means “for all” values of x

• ∈ x - an “ ∈ ” before variable x means “in x”
• ∀x ∈ y - means “for all x in y”

The following information present equations for calculating design values for the PM2.5
annual NAAQS, 24-hour PM2.5 NAAQS, and 24-hour PM10 NAAQS. The equations are
organized into the sets that are referenced in Section 9.3.

13
Values are compared after rounding. As long as the build design value is no greater than the no-build
design value after rounding, the project would meet conformity requirements at a given receptor, even if
the pre-rounding build design value is greater than the pre-rounding no-build design value.

K-17
K.6.2 Equation Set 1: Annual PM2.5 design value

Formulas

ci = bi + pi

3
bim
bi = ∑
m =1 3
4 bijm
bim = ∑
j =1 4
l
pik
pi = ∑
k =1 l
4 pijk
When using CAL3QHCR, pik = ∑
j =1 4

Definitions

bi = average of three consecutive years’ average annual background concentrations at

receptor i
bim = quarterly-weighted average annual background concentrations at receptor i during
monitoring year m
bijm = quarterly average background concentration at receptor i, during quarter j in
monitoring year m
ci = annual PM2.5 design value at receptor i
i = receptor
j = quarter
k = year of meteorological data
l = length in years of meteorological data record
m = year of background monitoring data
pik = average modeled quarterly average concentrations at receptor i for meteorological
year k. When using AERMOD, it is presumed that AERMOD’s input file is used
to specify this averaging time. When using CAL3QHCR with a single quarter of
meteorological data, pik must be calculated using each pijk for each quarter of
meteorological year k.
pijk = quarterly average concentration at receptor i for quarter j, in meteorological data
year k. This variable is the product of CAL3QHCR when run with a single
quarter of meteorological data. pik can be calculated directly using AERMOD
without explicitly calculating pijk .

K-18
K.6.3 Equation Set 2: 24-Hour PM2.5 design value (First Tier Analysis)

Formulas

cˆi = bˆi + pˆ i
bim = ∀bijm ∈ m
3 bim•rm
bˆi = ∑
m =1 3
l max [max ( p )]
pˆ i = ∑ k jk ijk
(when using CAL3QHCR), which compresses to:
k =1 l
l
max k ( pik )
pˆ i = ∑ (when using AERMOD with maximum concentration by year)
k =1 l

Definitions

b̂i = the average of 98th percentile 24-hour concentrations from three consecutive years of
monitoring data
bijm = daily 24-hour background concentration at receptor i, during quarter j in monitoring
year m
bim = ∀bijm ∈ m = All 24-hour background concentration measurements in year m
bim•rm = The 24-hour period within year m whose concentration rank among all 24-hour
measurements in year m is rm (this represents the 98th percentile of 24-hour
background concentrations within one year.)
ĉi = 24-hour PM2.5 design value at receptor i
i = receptor
j = quarter
k = year of meteorological data
l = length in years of meteorological data record
m = year of background monitoring data
max k = maximum predicted 24-hour concentration within meteorological year k
max jk = maximum predicted 24-hour concentration within quarter j within
meteorological year k
p̂i = average of highest predicted concentrations from each year modeled with the l years
from which meteorological data are used (≥5 years for off-site data, ≥1 year for
on-site data)
pijk = modeled daily 24-hour concentration at receptor i, in quarter j and meteorological
year k
pik = modeled daily 24-hour concentration at receptor i, in meteorological year k

K-19
rm = concentration rank of bim corresponding to 98th percentile of all bim in year m, based
on number of background concentration measurements per year (nm). rm is given
by the following table:

nm rm
1-50 1
51-100 2
101-150 3
151-200 4
201-250 5
251-300 6
301-350 7
351-366 8

K-20
K.6.4 Equation Set 3: 24-Hour PM2.5 design value (Second Tier Analysis)

Formulas
3 cim•rm
cˆi = ∑
m =1 3
cim = ∀cijm ∈ m
cijm = bijm + pˆ ij , for the eight (8) highest bijm in quarter j in monitoring year m
l max jk ( pijk )
pˆ ij = ∑
k =1 l

Definitions

bijm = daily 24-hour background concentration at receptor i, during quarter j in monitoring

year m
ĉi = 24-hour PM2.5 design value at receptor i
cijm = The set of all sums of modeled concentrations ( p̂ij ) with background
concentrations from quarter j and monitoring year m, using the eight highest
background concentrations ( bijm ) for the corresponding receptor, quarter, and
monitoring year.
cim = ∀cijm ∈ m = the set of all cimj corresponding to monitoring year m
cim•rm = predicted 98th percentile total concentration from the project, nearby sources, and
background measurements from year m. Given by the value of cim whose
concentration rank in year m is rm, using background measurements from year m.
i = receptor
j = quarter
k = year of meteorological data
l = length in years of meteorological data record
m = year of background monitoring data
max jk = maximum predicted 24-hour concentration within quarter j within
meteorological year k
pijk = Predicted daily 24-hour concentration at receptor i, during quarter j, based on data
from meteorological year k
p̂ij = Average highest 24-hour modeled concentration ( pijk ) using l years of
meteorological data

K-21
rm = concentration rank of cim corresponding to 98th percentile of all cim in year m, based
on number of background concentration measurements per year (nm). rm is given
by the following table:

nm rm
1-50 1
51-100 2
101-150 3
151-200 4
201-250 5
251-300 6
301-350 7
351-366 8

K-22
K.6.5 Equation Set 5: 24-Hour PM10 design value

Formulas
~
c~i = bi + ~
pi
~
bi = max in (bin )
3
bin =  bim
m =1
~
pi = pil •rl
l
pil =  pik
k =1

Definitions

c~i = 24-hour PM10 design value

~
bi = maximum monitored 24-hour PM10 background concentration at within bin
bim = the set of all monitored 24-hour PM10 background concentrations at receptor i
within monitoring year m
bin = the set of all bim within monitoring years n
i = receptor
k = year of meteorological data
l = length in years of meteorological data record.
maxin = the maximum monitored 24-hour background concentration at receptor i within
monitoring years n
n = the set of all years of monitoring data, m = {1,2,3}
~
pi = pil •rl = modeled 24-hour PM10 concentration with concentration rank of rl among all
concentrations modeled using l years of meteorological data
pil = set of all modeled 24-hour concentrations at receptor i across l years of
meteorological data
rl = l + 1 (for example, rl = 6 when using 5 years of meteorological data)
z

c a = the set (finite union) of all ca with integer values of a = {1,…,z}

a =1

K-23