Population and Activity of Onroad
Vehicles in MOVES4
£% United States
Environmental Protect
Agency
-------
Population and Activity of Onroad
This technical report does not necessarily represent final EPA decisions
or positions. It is intended to present technical analysis of issues using
data that are currently available. The purpose in the release of such
reports is to facilitate the exchange of technical information and to
inform the public of technical developments.
Vehicles in MOVES4
Assessment and Standards Division
Office of Transportation and Air Quality
U.S. Environmental Protection Agency
NOTICE
4>EPA
United States
Environmental Protection
Agency
EPA-420-R-23-005
August 2023
-------
Table of Contents
1. Introduction 7
2. MOVES Vehicle and Activity Classifications 11
2.1. HPMS Class 11
2.2. Source Use Types 11
2.3. Regulatory Classes 12
2.4. Fuel and Technology Types 13
2.5. Road Types 14
2.6. Source Classification Codes (SCC) 15
2.7. Model Year Groups 15
2.8. Source Bins 16
2.9. Allowable Vehicle Modeling Combinations 17
2.10. Default Inputs and Fleet and Activity Generators 19
3. VMT by Calendar Year and Vehicle Type 21
3.1. Historic Vehicle Miles Traveled (1990 and 1999-2021) 21
3.2. Projected Vehicle Miles Traveled (2022-2060) 24
4. Vehicle Populations by Calendar Year 28
4.1. Historic Source Type Populations (1990, 1999-2021) 28
4.2. Projected Vehicle Populations (2022-2060) 32
5. Fleet Characteristics 36
5.1. Source Type Definitions 36
5.2. Sample Vehicle Population 38
6. Vehicle Age-Related Characteristics 48
6.1. Age Distributions 49
6.2. Relative Mileage Accumulation Rate 54
7. VMT Distribution of Source Type by Road Type 64
8. Average Speed Distributions 66
8.1. Description of Telematics Dataset 66
8.2. Derivation of Default National Average Speed Distributions 67
8.3. Updated average speed distributions 69
9. Driving Schedules and Ramps 71
9.1. Driving Schedules 71
9.2. Modeling of Ramps in MOVES 76
10. Off-Network Idle Activity 77
10.1. Off-Network Idle Calculation Methodology and Definitions 77
10.2. Light-Duty Off-Network Idle 78
10.3. Heavy-Duty Off-Network Idle 85
10.4. Off-network Idling Summary 91
11. Hotelling Activity 92
11.1. Hotelling Activity Distribution 92
11.2. National Default Hotelling Rate 94
12. Engine Start Activity 97
12.1. Light-Duty Start Activity 98
12.2. Heavy-Duty Start Activity 104
12.3. Motorcycle and Motorhome Starts 123
1
-------
13. Temporal Distributions 125
13.1. VMT Distribution by Month of the Year 125
13.2. VMT Distribution by Type of Day 127
13.3. VMT Distribution by Hour of the Day 128
13.4. Parking Activity 130
13.5. Hourly Hotelling Activity 134
14. Geographical Allocation of Activity 139
14.1. Source Hours Operating Allocation to Zones 139
14.2. Parking Hours Allocation to Zones 140
15. Vehicle Mass and Road Load Coefficients 141
15.1. Source Mass and Fixed Mass Factor 142
15.2. Road Load Coefficients 146
16. Air Conditioning Activity Inputs 150
16.1. ACPenetrationFraction 150
16.2. FunctioningACFraction 151
16.3. Air Conditioning Activity Demand 152
17. Conclusion and Areas for Future Research 154
Appendix A Fuel Type and Regulatory Class Fractions from Previous Versions of MOVES 156
Al. Distributions for Model Years 1960-1981 156
A2. Distributions for Model Years 1982-1999 158
Appendix B 1990 Age Distributions 168
Bl. Motorcycles 168
B2. Passenger Cars 168
B3. Trucks 168
B4. Other Buses 169
B5. School Buses and Motor Homes 169
B6. Transit Buses 169
Appendix C Detailed Derivation of Age Distributions 171
CI. Vehicle Survival by Source Type 171
C2. Vehicle Sales by Source Type 173
C3. Base Year Age Distributions 176
C4. Historic Age Distributions 177
C5. Projected Age Distributions 178
Appendix D Driving Schedules 181
Appendix E Total Idle Fraction Regression Coefficients 184
Appendix F Source Masses for Light-Duty Vehicles 187
Fl. Motorcycles 188
F2. Passenger Cars 189
F3. Light-Duty Trucks 189
Appendix G NREL Fleet DNA Preprocessing Steps 191
Appendix H Averaging Methods for Heavy-duty Telematics Activity Data 195
HI. Evaluated Methods 195
H2. Comparison of Evaluated Methods 198
H3. Future W ork 199
Appendix I Road Load Coefficient for Combination Trucks in HD GHG Rule 201
Appendix J MOVES4 SourceUseTypePhysics Table 211
2
-------
Appendix K AVFT Tool 217
Kl. Proportional 217
K2. National Average 218
K3. Known Fractions 218
K4. Constant 218
18. References 219
3
-------
List of Acronyms
AEO
Annual Energy Outlook publication
AMPO
Association of Metropolitan Planning Organizations
APU
auxiliary power unit
ARCADIS
Design & Consultancy firm for natural and built assets
ASD
average speed distribution
AVFT
Alternate Vehicle Fuel and Technology
AVGCK
VIUS broad average weight category
CAN
Controller Area Network
CARB
California Air Resources Board
CBI
confidential business information
CE-CERT
College of Engineering - Center for Environmental Research and
Technology
CFR
Code of Federal Regulations
CNG
Compressed Natural Gas
CO
carbon monoxide
CO2
carbon dioxide
CRC
Coordinating Research Council
DOT
U.S. Department of Transportation
EIA
U.S. Energy Information Administration
EPA
U.S. Environmental Protection Agency
ERG
Eastern Research Group, Inc.
E85
gasoline containing 70-85 percent ethanol by volume
FFV
flexible fuel vehicle
FHWA
Federal Highway Administration
FMCSA
Federal Motor Carrier Safety Administration
FTA
Federal Transit Administration
GEM
Greenhouse Gas Emissions Model
GHG
Greenhouse Gases
g/hr
Grams per hour
GPO
U.S. Government Publishing Office
GPS
Global Positioning System
GVWR
Gross Vehicle Weight Rating
HC
Hydrocarbons
HD
Heavy-Duty
HDDBT
Heavy-Duty Diesel transit buses
HDV
Heavy-Duty Vehicle
HHD
Heavy-Heavy-Duty
HHDDT
Heavy-Heavy-Duty Diesel Truck
HPMS
Highway Performance Monitoring System
ID
Identification
IHS
Information Handling Services (research consulting firm)
I/M
Inspection and Maintenance program
kg/m
kilogram per meter
LD
Light-Duty
4
-------
LDT
Light-Duty Truck
LDV
Light-Duty Vehicle
LHD
Light-Heavy-Duty
MAR
mileage accumulation rate
MC
Motorcycle
MD
Medium-duty
MHD
Medium-Heavy-Duty
M0BILE6
(predecessor to MOVES)
MOVES
Motor Vehicle Emission Simulator
mph
miles per hour
MPO
Metropolitan Planning Organization
MSA
Metropolitan Statistical Area (U.S. Census)
MSOD
Mobile Source Observation Database
NACFE
North American Council for Freight Efficiency
NCHRP
National Cooperative Highway Research Program
NCTCOG
North Central Texas Council of Government
NEI
National Emission Inventory
NHTS
National Household Travel Survey
NHTSA
National Highway Traffic Safety Administration
NOx
nitrogen oxide
NPMRDS
National Performance Management Research Dataset
NREL
National Renewable Energy Laboratory
NTD
National Transit Database
NVPP
National Vehicle Population Profile
OHIM
Office of Highway Information Management
ONI
off-network idle
OPCLASS
operator classification
ORNL
Oak Ridge National Laboratory
PAMS
portable activity measurement systems
PM
Particulate Matter
PM2.5
fine particles of particulate matter
PM10
Particles of particulate matter 10 micrometers and smaller
RMAR
relative mileage accumulation rate
RPM
revolutions per minute
RT
Road Type
SBDG
Source Bin Distribution Generator
see
Source Classification Codes
SCR
selective catalytic reduction
SHI
source hours idle
SHO
source hours operating
SIP
State Implementation Plan
SMOKE
Sparse Matrix Operator Kernel Emissions
ST
Source Type
STP
scaled-tractive power
SUV
sport utility vehicle
SVP
Sample Vehicle Population
5
-------
TDM
Travel Demand Models
TEDB
Transportation Energy Data Book
TIF
total idle fraction
TIUS
Truck Inventory and Use Survey
TRB
Transportation Research Board
TRLHP
tractive road load horsepower
TxDOT
Texas Department of Transportation
VIUS
Vehicle Inventory and Use Survey
VMT
Vehicle Miles Traveled
VSP
vehicle specific power
VTRIS
Vehicle Travel Information System
WMATA
Washington Metropolitan Area Transit Authority
6
-------
1. Introduction
The United States Environmental Protection Agency's Motor Vehicle Emission Simulator—
commonly referred to as MOVES—is a set of modeling tools for estimating air pollution
emissions produced by onroad (highway) and nonroad mobile sources. MOVES estimates the
emissions of greenhouse gases (GHGs), criteria pollutants and selected air toxics. The MOVES
model is currently the official model for use for state implementation plan (SIP) submissions to
EPA and for transportation conformity analyses outside of California. The model is also the
primary modeling tool for estimating the impact of mobile source regulations on emission
inventories.
MOVES calculates emission inventories by multiplying emission rates by the appropriate
emission-related activity, applying correction and adjustment factors as needed to simulate
specific situations and then adding up the emissions from all sources and regions.
Vehicle population and activity data are critical inputs for calculating emission inventories from
emissions processes such as running exhaust, start exhaust and evaporative emissions. In
MOVES, most running emissions are distinguished by operating modes, depending on road type
and vehicle speed. Start emissions are determined based on the time a vehicle has been parked
prior to the engine starting, known as a "soak." Evaporative emission modes are affected by
vehicle operation and the time that vehicles are parked. Emission rates are further categorized by
grouping vehicles with similar fuel type, regulatory classification, and other vehicle
characteristics into "source bins."
This report describes the sources and derivation for onroad vehicle population and activity
information and associated adjustments as stored in the MOVES default database. In particular,
this report describes the data used to fill the default database tables listed below in Table 1-1.
Note that technical details on the default database values for emission rates, correction factors
and other inputs, including information on nonroad equipment, are described in other MOVES
technical reports.1
These data have been updated for MOVES4 from previous versions of MOVES. In MOVES4,
we have updated vehicle activity based on newer data from Annual Energy Outlook, Highway
Statistics, Transportation Energy Data Book, and School Bus Fleet Fact Book. We also updated
vehicle distributions based on IHS2020 and relative mileage accumulations based on FHWA
analysis. In addition, updates have been made for gliders, Class 2b and 3 light heavy-duty
(LHD) vehicles, and electric vehicles for MOVES4.
Properly characterizing emissions from onroad vehicles requires a detailed understanding of the
vehicles that comprise the national fleet and their patterns of operation. The MOVES default
database has a domain that encompasses the entire United States, Puerto Rico and the Virgin
Islands. In MOVES, users may analyze emission inventories in 1990 and every year from 1999
to 2060. The national default activity information in MOVES provides a reasonable basis for
estimating national emissions. As described in this report, the most important of these inputs,
such as vehicle miles travelled (VMT) and population estimates, come from long-term
systematic national measurements.
7
-------
Due to the availability of these national measurements, the most recent year of measured data in
the model and the base year for projected emissions, is 2021.
It is important to note that uncertainties and variability in the default data contribute to the
uncertainty in the resulting emission estimates. Therefore, MOVES has been specifically
designed to accommodate the input of alternate, user-supplied activity data. In particular, when
modellers estimate emissions for specific geographic locations, EPA guidance recommends
replacing many of the MOVES fleet and activity defaults with local data. This is especially true
for inputs where local data is more detailed or up to date than those provided in the MOVES
defaults. EPA's Technical Guidance2 provides more information on customizing MOVES with
local inputs.
Population and activity data are ever changing as new historical data becomes available and new
projections are generated. As part of the MOVES development process, the model undergoes
major updates and review every few years. The significant updates made to MOVES since the
MOVES2014 release were peer-reviewed under EPA's peer review guidance3 in two separate
reviews conducted in 2017, 2019, and 2020. Materials from each peer review, including peer-
review comments and EPA responses are located on the EPA's science inventory webpage.4 5
The development of fleet and activity inputs will continue to be an important area of focus and
improvement for MOVES.
8
-------
Table 1-1 MOVES database elements covered in this report
Database Table Name
Content Summary
Report Sections
AvgSpeedDistribution
Distribution of time among average speed bins
Section 8
DayVMTFraction
Distribution of VMT between weekdays and
weekend days
Section 13
DriveSchedule
Average speed of each drive schedule
Section 9
DriveScheduleAssoc
Mapping of which drive schedules are used for
each combination of source type and road type
Section 9
DriveScheduleSecond
Speed for each second of each drive schedule
Section 9
FuelType
Broad fuel categories that indicate the fuel
vehicles are capable of using
Section 2
HotellingActivityDistribution
Distribution of hotelling activity to the various
operating modes
Section 11
HotellingCalendarY ear
Rate of hotelling hours per total restricted access
VMT
Section 11
HourVMTFraction
Distribution of VMT among hours of the day
Section 13
HPMSVtypeYear
Annual VMT by HPMS vehicle types
Section 3
IdleRegion
Map of idle regions to idle region IDs.
Section 10
ModelY earGroup
A list of years and groups of years corresponding
to vehicles with similar emissions performance
Section 2
MonthGroupHour
Coefficients to calculate air conditioning demand
as a function of heat index
Section 16
MonthVMTFraction
Distribution of annual VMT among months
Section 13
OpModeDistribution
The distribution of engine start soak times for each
source type, day type, hour of the day and
pollutant.
Section 12
PollutantProcessModelYear
Assigns model years to appropriate groupings,
which vary by pollutant and process
Section 2
RegulatoryClass
Categorizes vehicles into weight-rating based
groups used to assign emission rates.
Section 2
RoadType
Distinguishes roadways as urban or rural and by
type of access, particularly the use of ramps for
entrance and exit
Section 2
RoadTypeDistribution
Distribution of VMT among road types
Section 7
SampleVehicleDay
Identifies vehicles in the SampleVehicleTrip table
Section 13
Sample VehiclePopulation
Fuel type and regulatory class distributions by
source type and model year.
Section 5
SampleVehicleTrip
Trip start and end times used to determine parking
times for evaporative emission calculations.
Section 13
see
Source Classification Codes that identify the
vehicle type, fuel type, road type and emission
process in MOVES output
Section 2
StartsHourFraction
The fraction of total starts that occur in each hour
of the day. This allocationFraction varies by
county (zonelD) and day type.
Sectionl2
StartsMonthAdjust
The monthAdjustFactor adjusts the starts per day
to reflect monthly variation in the number of
starts.
Sectionl2
StartsPerDay
StartsPerDay value is the number of starts per
average vehicle (of all source types). This value
varies by county (zonelD) and day type.
Sectionl2
StartsSourceTypeFraction
The allocation of total starts per day for all
vehicles to each of the MOVES source types.
Sectionl2
9
-------
Table 1-1 MOVES database elements covered in this report
Database Table Name
Content Summary
Report Sections
SourceBinDistribution
Distribution of population among different vehicle
sub-types (source bins)
Section 2
SourceTypeAge
Rate of survival to subsequent age, relative
mileage accumulation rates and fraction of
functional air conditioning equipment
Appendix C
Section 6
Section 16
SourccTvpcAgcDistribution
Distribution of vehicle population among ages
Section 6
SourceTypeHour
The distribution of total daily hotelling among
hours of the day
Section 13
SourceTypeModelY ear
Prevalence of air conditioning equipment
Section 16
SourceTypePolProcess
Indicates which source bin discriminators are
relevant for each source type and pollutant/process
Section 2
SourceTypeYear
Source type vehicle counts by year
Section 4
SourceUseType
Mapping from HPMS class to source type,
including source type names
Section 2
SourceU seTypePhy sics
Road load coefficients and vehicle masses for each
source type used to calculate vehicle specific
power (VSP) and scaled tractive power (STP)
Section 15
TotalldleFraction
Fraction of vehicle operating time when speed is
zero.
Section 10
Zone
Allocation of activity to zone (county)
Section 14
ZoneRoadType
Allocation of driving time to zone (county) and
road type
Section 14
10
-------
2. MOTES Vehicle ami Activity Classifications
Fundamentally, onroad mobile source emission inventories are estimated by applying vehicle
populations and activity to appropriate emission rates. We wanted to enter vehicle population
and activity data in a form as close as possible to how this data is collected by highway
departments and vehicle registrars, but we had to map these to existing emission standards and
in-use emission rates. Thus, EPA developed MOVES-specific terminology classifying vehicles
according to how they are operated, such as "source types," and to emission-related
characteristics, such as "regulatory classes" and "fuel types." At the most detailed level, vehicles
are classified into "source bins" which have a direct mapping to emission rates by vehicle
operating mode in the MOVES emission rate tables.
This section provides definitions of the various vehicle classifications used in MOVES. The
MOVES terms introduced in this section will be used throughout the report. Later sections
explain how default vehicle populations and activity are assigned and allocated to these
classifications.
2.1.HPMS Class
In this report, MOVES HPMS class refers to one of five categories derived from the US
Department of Transportation (DOT) Highway Performance Monitoring System (HPMS) based
vehicle classes used by the Federal Highway Administration (FHWA) in the Table VM-1 of their
annual Highway Statistics report.6 The five HPMS classes used in MOVES are as follows:
motorcycles (HPMSVTypelD 10), light-duty vehicles (25), buses (40), single-unit trucks (50)
and combination trucks (60). Please note that the light-duty vehicles class (25) here represents
the combination of the VM-1 categories for long wheelbase and short wheelbase light-duty cars
and trucks. More details on how HPMS classes are used in MOVES may be found in Section 3.
2.2. Source Use Types
The primary vehicle classification in MOVES is source use type, or, more simply, source type.
Source types are groups of vehicles with similar activity and usage patterns and are more specific
than the HPMS vehicle classes described above. In addition, source types have common body
types, and the road load coefficients (rolling load, mechanical rotating friction, aerodynamic
drag) are defined by source type as discussed in Section 15.
Vehicles are classified into source types based on body type as well as other characteristics, such
as whether they are registered to an individual, a commercial business, or a transit agency;
whether they have specific travel routines such as a refuse truck; and whether they typically
travel short- or long-haul routes (greater than 200 miles per day). The MOVES source types are
listed in Table 2-1 along with the associated HPMS classes. More detailed source type
definitions are provided in Section 5.1.
11
-------
Table 2-1 Onroad Source Types in MOVES
sourceTypelD
Source Type Name
HPMSVTypelD
HPMS Description
11
Motorcycles
10
Motorcycles
21
Passenger Cars
25
Light-Duty Vehicles
31
Passenger Trucks (primarily personal use)
25
Light-Duty Vehicles
32
Light Commercial Trucks (primarily non-
personal use)
25
Light-Duty Vehicles
41
Other Buses (non-school, non-transit)
40
Buses
42
Transit Buses
40
Buses
43
School Buses
40
Buses
51
Refuse Trucks
50
Single-Unit Trucks
52
Single Unit Short-Haul Trucks
50
Single-Unit Trucks
53
Single Unit Long-Haul Trucks
50
Single-Unit Trucks
54
Motor Homes
50
Single-Unit Trucks
61
Combination Short-Haul Trucks
60
Combination Trucks
62
Combination Long-Haul Trucks
60
Combination Trucks
2.3.Regulatory Classes
In contrast to source types, regulatory classes are used to group vehicles subject to similar
emission standards. The EPA regulates vehicle emissions based on groupings of technologies
and classifications that do not necessarily correspond to DOT activity and usage patterns. To
properly estimate emissions, it is critical for MOVES to account for these emission standards.
The regulatory classes used in MOVES are summarized in Table 2-2 below. The "doesn't
matter" regulatory class is used internally in the model if the emission rates for a given pollutant
and process are independent of regulatory class. The motorcycle (MC) and light-duty vehicle
(LDV) regulatory classes have a one-to-one correspondence with source type. Other source types
are allocated between regulatory classes based primarily on gross vehicle weight rating (GVWR)
classification, which is a set of eight classes defined by FHWA based on the manufacturer-
defined maximum combined weight of the vehicle and its load. Urban buses have their own
regulatory definition and therefore are an independent regulatory class.
Table 2-2 F
Regulatory Classes in MOVES
regClassID
Regulatory Class Name
Description
0
Doesn't Matter
Doesn't Matter
10
MC
Motorcycles
20
LDV
Light-Duty Vehicles
30
LDT
Light-Duty Trucks
41
LHD2b3
Class 2b and 3 Trucks (8,500 lbs < GVWR <= 14,000 lbs)*
42
LHD45
Class 4 and 5 Trucks (14,000 lbs. < GVWR <= 19,500 lbs.)*
46
MHD
Class 6 and 7 Trucks (19,500 lbs. < GVWR < =33,000 lbs.)
47
HHD
Class 8a and 8b Trucks (GVWR > 33,000 lbs.)
48
Urban Bus
Urban Bus (see CFR Sec. 86.091 2)
49
Gliders
Glider Vehicles7
* Class 2b trucks (GVWR 8,501-10,000 lbs) are only modeled in passenger trucks and light commercial trucks
(source types 31 and 32, respectively). Class 3 trucks (GVWR 10,001-14,000 lbs) are only modeled in heavy-duty
source types (buses and single unit trucks). Model year 2017-and-later engine-certified Class 3 trucks (only
present within source types 52, 53, and 54) are classified as LHD45 (regclassID 42) for modeling purposes.
12
-------
The EPA regulatory distinction between light-duty (LD) and heavy-duty (HD) trucks falls in the
midst of FHWA GVWR Class 2. Trucks of 6,001-8,500 lbs. GVWR are Class 2a; in MOVES,
they are considered light-duty trucks in regulatory class 30. Vehicles of 8,500-14,000 lbs.
GVWR are Class 2b and Class 3 and considered light heavy-duty vehicles (LHD) in regulatory
class 41.
In MOVES4, we reclassified diesel light-heavy-duty Class 3 engine-certified vehicles for model
year 2017 and later years as LHD45 vehicles. The population fraction of diesel light-heavy-duty
vehicles is based on data in IHS2020. The emission rates for LHD2b3 vehicles are based on the
assumption that all vehicles are chassis-certified. Because Class 3 engine-certified vehicles are
subject the same emission standards as Class 4 and 5 engine-certified vehicles, we reclassified
these vehicles as LHD45 vehicles. Model year 2017 was selected because this is the first model
year when the emission rates are different between LHD2b3 and LHD45.11
In the MOVES model, "Gliders" refers to post-2007 heavy-duty diesel vehicles with new chassis
but with older engines that do not meet 2007 or 2010 emissions standards and thus are treated as
a separate regulatory class.
Section 5.2 provides more information on the distribution of vehicles among regulatory classes.
Vehicle weights in MOVES are defined by both regulatory class and source type as discussed in
Section 15.
2.4.Fuel ami Technology Types
MOVES models vehicles powered by following fuel types: gasoline, diesel, E-85 (fuels
containing 70 percent to 85 percent ethanol by volume), compressed natural gas (CNG) and
electricity. Note that in some cases, a single vehicle can use more than one fuel. For example,
flexible fuel vehicles (FFV) are capable of running on either gasoline or E-85. In MOVES, fuel
type refers to the capability of the vehicle rather than the fuel in the tank. The fuel use actually
modeled depends on a number of factors including the location, year and month in which the fuel
was purchased, as explained in the MOVES technical report on fuel supply.8 MOVES also
allows the modeling of technology types, although these are not distinguished in MOVES output.
In MOVES4, technology type is used to distinguish battery and fuel-cell electric vehicles. Table
2-3 below summarizes the fuel types and technology types populated in MOVES4. These are
recorded in the default database FuelType, EngineTech and FuelEngTechAssoc tables.
13
-------
Table 2-3 A List of Allowable Fuel Types to Power Vehicles in MOVES
fuelTypelD
Description
Default Fuel
FormulationlD8
EngTechID
Technology
Description
1
Gasoline
10
1
Conventional Internal
Combustion
2
Diesel Fuel
20
1
Conventional Internal
Combustion
3
Compressed Natural Gas (CNG)
30
1
Conventional Internal
Combustion
5
Ethanol (E-85)
50
1
Conventional Internal
Combustion
9
Electricity
90
30
Electric
40
Fuel Cell
It is important to note that not all fuel type/source type combinations can be modeled in
MOVES. For example, MOVES cannot model gasoline-fueled long-haul combination trucks or
diesel motorcycles. Similarly, flexible fuels (E85-compatible) are only modeled for passenger
cars, passenger trucks and light commercial trucks. In addition, MOVES does not explicitly
model hybrid powertrains, but accounts for these vehicles in calculating fleet-average energy
consumption and CO2 rates.a For more information on how MOVES models the impact of fuels
on emissions, please see the MOVES documentation on fuel effects.9
2.5. Road Types
MOVES calculates onroad emissions separately for each of four road types and for "off-
network" activity when the vehicle is not moving. The road types used in MOVES are listed in
Table 2-4. The four MOVES road types (2-5) are aggregations of FHWA functional facility
types.
Table 2-4 Road Types in MOVES
roadTypelD
Description
FHWA Functional Types
1
Off Network
Off Network
2
Rural Restricted Access
Rural Interstate & Rural Freeway/Expressway
3
Rural Unrestricted Access
Rural Other Principal Arterial, Minor Arterial, Major
Collector, Minor Collector & Local
4
Urban Restricted Access
Urban Interstate & Urban Freeway/Expressway
5
Urban Unrestricted Access
Urban Other Principal Arterial, Minor Arterial, Major
Collector, Minor Collector & Local
a While we have considered creating a separate category for hybrid vehicles, modeling their emissions separately is
not required for regulatory purposes and presents a number of challenges, including obtaining representative detailed
data on hybrid vehicle emissions and usage and accounting for offsetting emissions allowed under the fleet-
averaging provisions of the relevant emissions standards.
14
-------
The MOVES road types are based on two important distinctions in how FWHA classifies roads:
1) urban versus rural roadways are distinguished based on surrounding land use and human
population density and 2) unrestricted versus restricted are distinguished based on roadway
access—restricted roads require the use of ramps. The urban/rural distinction is used primarily
for national level calculations. It allows different default speed distributions in urban and rural
settings. Of course, finer distinctions are possible. Users with more detailed information on
speeds and acceleration patterns may run MOVES at project level where emissions can be
calculated for individual links.
2.6. Source Classification Codes (SCC)
Source Classification Codes (SCC) are used to group and identify emission sources in large-scale
emission inventories. They are often used when post-processing MOVES output to further
allocate emissions temporally and spatially when preparing inputs for air quality modeling. In
MOVES, SCCs are numerical codes that identify the vehicle type, fuel type, road type and
emission process using MOVES identification (ID) values in the following form:
AAAFVVRRPP, where
• AAA indicates mobile source (this has a value of 220 for both onroad and nonroad),
• F indicates the MOVES fuelTypelD value,
• VV indicates the MOVES sourceTypelD value,
• RR indicates the MOVES roadTypelD value and
• PP indicates the MOVES emission processID value.
Building the SCC values in this way allows additional source types, fuel types, road types and
emission processes to be easily added to the list of SCCs as changes are made to future versions
of MOVES. The explicit coding of fuel type, source type, road type and emission process also
allows the new SCCs to indicate aggregations. For example, a zero code (00) for any of the
sourceTypelD, fuelTypelD, roadTypelD and processID strings that make up the SCC indicates
that the reported emissions are an aggregation of all categories of that type. Using the mapping
described above, modelers can also easily identify the sourceTypelD, fuelTypelD, roadTypelD
and processID of emissions reported by SCC. Refer to earlier sections in this document for the
descriptions of the sourceTypelD, fuelTypelD and roadTypelD values currently used by
MOVES. Emission processes are discussed in other MOVES reports on emission rate
development1011 and are not described here. All feasible SCC values are listed in the SCC table
within the default database.
2.7.Model Year Groups
MOVES uses model year groups to avoid unnecessary duplication of emission rates for vehicles
with similar technology and similar expected emission performance. For example, there is a
model year group for "1980 and earlier." In MOVES, model year refers to the year in which the
vehicle was produced, built and certified as compliant with emission standards.
15
-------
The default ModelYearGroup table provides information on the model year group names,
beginning and ending years and a two-digit shorthand identifier (shortModelYrGroupID).
However, the model year groups that are relevant for a given calculation can vary depending on
pollutant and emission process as defined in the PollutantProcessModelYear table. For example,
a 2031 vehicle belongs to the "2031" model year group for estimating running total energy
consumption but belongs to the "2031-2050" group for estimating nitrous oxide running
emissions. Because these groupings are determined based on analysis of the actual or expected
emissions performance, the rationale for each model year grouping is provided in the MOVES
emission rate reports.10'11
2.8. Source Bins
The MOVES default database identifies emission rates by emission-related characteristics such
as the type of fuel that a vehicle uses and the emission standards it is subject to. These
classifications are called "source bins." They are named with a sourceBinID that is a unique 19-
digit identifier in the following form:
1FFEERRMM0000000000, where
• 1 is a placeholder,
• FF is a MOVES fuelTypelD,
• EE is a MOVES engTechID,
• RR is a MOVES regClassID,
• MM is a MOVES shortModYrGroupID and
• 10 trailing zeros for future characteristics.
The model allocates vehicle activity and population to these source bins as described below.
A mapping of model year to model year groups is stored in the PollutantProcessModelYear
table. Distributions of fuel type and regulatory class by source type are stored by model year in
the SampleVehiclePopulation table. MOVES combines information from these two tables (see
Table 2-5) to create a detailed SourceBinDistribution. In general, fuel type is relevant for all
emission calculations, but the relevance of regulatory class and model year group depend on the
pollutant and process being modeled. See Section 2.10 for more information on how MOVES
uses generators to calculate detailed activity information.
16
-------
Table 2-5 Data Tables Used to Allocate Source r
"ype to Source Bin
Table Name
Key Fields*
Additional Fields
Notes
SourceTypePolProcess
sourceTypelD
polProcessID
isRegClassReqd
isMY GroupReqd
Indicates which pollutant-processes the
source bin distributions may be applied
to and indicates which discriminators
are relevant for each sourceTypelD and
polProcessID (pollutant/process
combination)
PollutantProcessModelY ear
polProcessID
modelYearlD
modelY earGroupID
Assigns model years to appropriate
model year groups for each
polProcessID.
SampleVehiclePopulation
sourceTypelD
modelYearlD
fuelTypelD
engTechID
regClassID
stmyFuelEngFraction
stmyFraction
Includes fuel type and regulatory class
fractions for each source type and
model year, even for some source
type/fuel type combinations that do not
currently have any appreciable market
share (i.e. CNG motor homes). This
table provides default fractions for the
Alternative Vehicle Fuel & Technology
(AVFT) importer.
Note:
* In these tables, the sourceTypelD and modelYearlD are combined into a single sourceTypeModelYearlD.
While details of the SourceTypePolProcess and PollutantProcessModel Year tables are discussed
in the reports on the development of the light- and heavy-duty emission rates,10'11 the
SampleVehiclePopulation (SVP) table is a topic for this report and is discussed in Section 5.2
2.9. Allowable Vehicle Modeling Combinations
In theory, the MOVES source bins would allow users to model any combination of source type,
model year, regulatory class and fuel type. However, each combination must have accompanying
emission rates; combinations that lack data from emissions testing or have negligible market
share cannot be directly modeled in MOVES.
Table 2-6 summarizes the allowable source type-fuel type combinations. Most of the gasoline
and diesel combinations exist with a few exceptions, but options for alternative fuels are limited,
as discussed earlier in Section 2.4. MOVES also stores regulatory class distributions by source
type in the SampleVehiclePopulation table. Table 2-7 summarizes the allowable source type-
regulatory class combinations. Table 2-8 shows the full set of allowable source type, fuel type
and regulatory class combinations. Also see the mapping of fuel types and technology types
shown in Table 2-3. Additional discussion about decisions to include and exclude certain types
of vehicles can be found in Section 5.
17
-------
Table 2-6 Matrix of the Allowable Source Type-Fuel Type Combinations in MOVES4
(Allowable combinations are marked with an X)
Source Use Type
Motorcycles
Passenger Cars
Passenger Trucks
Light Commercial Trucks
Other Buses
Transit Buses
School Buses
Refuse Trucks
Short-Haul Single Unit
Trucks
Long-Haul Single Unit
Trucks
Motor Homes
Short-Haul Combination
Trucks
Long-Haul Combination
Trucks
Fuel Type
11
21
31
32
41
42
43
51
52
53
54
61
62
Gasoline
1
X
X
X
X
X
X
X
X
X
X
X
X
Diesel
2
X
X
X
X
X
X
X
X
X
X
X
X
CNG
3
X
X
X
X
X
X
X
X
X
E85-Capable
5
X
X
X
Battery
Electric
9
X
X
X
X
X
X
X
X
X
X
X
X
Electric Fuel
Cell
9
X
X
X
X
X
X
X
X
X
Table 2-7 Matrix of the Allowable Source Type-Regulatory Class Combinations in MOVES4
(Allowable combinations are marked with an X)
Source Use Type
Motorcycles
Passenger Cars
Passenger Trucks
Light Commercial
Trucks
Other Buses
Transit Buses
School Buses
Refuse Trucks
Short-Haul Single
Unit Trucks
Long-Haul Single
Unit Trucks
Motor Homes
Short-Haul
Combination Trucks
Long-Haul
Combination Trucks
Regulatory Class
11
21
31
32
41
42
43
51
52
53
54
61
62
MC
10
X
LDV
20
X
LDT
30
X
X
LHD2b3
41
X
X
X
X
X
X
X
LHD45
42
X
X
X
X
X
X
X
MHD67
46
X
X
X
X
X
X
X
X
X
HHD8
47
X
X
X
X
X
X
X
X
X
Urban Bus
48
X
Gliders*
49
X
X
* This table was updated to fix an error in previous versions of the report. Glider assignment to sourcetypes is unchanged
from MOVES3 to MOVES4.
18
-------
Table 2-8 Summary of source type, fuel type, technology, and regulatory class combinations in
MOVES4
sourceTypelD
fuelTypelD
engTechID
regClassID
11
1
1
10
21
1,2,5
1
20
9
30
20
1,2
1
30,41
31, 32
5
1
30
9
30
30,41
9
40
41
1,2,
1
42, 46, 47
41,42
3
1
47, 48
9
30, 40
42, 46, 47
1,2
1
41, 42, 46, 47
43
3
1
47
9
30, 40
41, 42, 46, 47
1,2
1
41, 42, 46, 47
51, 52, 53, 54
3
1
47
9
30, 40
41, 42, 26, 47
1, 9
1
46, 47
61,62
2
1
46, 47, 49
3
1
47
9
30, 40
46,47
2.10. Default Inputs ami Fleet ami Activity Generators
As explained in the introduction, vehicle population and activity data are critical inputs for
calculating emission inventories and MOVES calculators require information on vehicle
population and activity at a very fine scale. In project-level modeling, this detailed information
may be available and manageable. However, in other cases, the fleet and activity data used in
the MOVES calculators must be generated from inputs in a condensed or more readily available
format. MOVES uses "generators" to create fine-scale information from user inputs and MOVES
defaults.
The MOVES Total Activity Generator estimates hours of vehicle activity using vehicle miles
traveled (VMT) and speed information to transform VMT into source hours operating (SHO).
Other types of vehicle activity are generated by applying appropriate factors to vehicle
populations. Vehicle starts, extended idle hours and source hours (including hours operating and
not-operating) are also generated. The default database for MOVES contains national estimates
for VMT and vehicle population for every possible analysis year (1990 and 1999-2060). For
national inventory runs, annual national activity is distributed temporally and spatially using
allocation factors and age distributions for future years are generated from the base year
distribution.
The Source Bin Distribution Generator (SBDG) uses information on model year groupings and
fuel type and regulatory class distributions to estimate activity fractions of each source bin as a
function of source type, model year, pollutant and process. MOVES maps the activity data (by
source types) to source bins which map directly to the MOVES emission rates.
19
-------
There are a number of MOVES modules that generate operating mode distributions based on
vehicle activity inputs. For running emissions, MOVES uses information on speed distributions
and driving patterns (driving schedules) to develop operating mode fractions for each source
type, road type and time of day and to calculate off-network idling activity. Similarly, other
generators use MOVES inputs to develop operating mode distributions for hotelling activity,
starts and vapor venting.
20
-------
3. VMT by Calendar Year ami Vehicle Type
At the national level, MOVES calculates source operating hours from national vehicle miles
traveled (VMT) by vehicle type. The default database contains national VMT estimates for all
analysis years, which include 1990 and 1999-2060. Years 1991-1998 are excluded because there
is no regulatory requirement to analyze them and including them would increase model
complexity. Calendar year 1990 is available to be modeled in MOVES because of the Clean Air
Act Amendments of 1990.
The national VMT estimates are stored in the HPMSVTypeYear table,b which includes three
data fields: HPMSBaseYearVMT (discussed below), baseYearOffNetVMT and
VMTGrowthFactor. Off network VMT refers to the portion of activity that is not included in
travel demand model networks or any VMT that is not otherwise reflected in the other four road
types. The field baseYearOffNetVMT is provided in case it is useful for modeling local areas.
However, the reported HPMS VMT values, used to calculate the national averages discussed
here, are intended to include all VMT. Thus, for MOVES national defaults, the
baseYearOffNetVMT is zero for all vehicle types. Additionally, the VMTGrowthF actor field is
not used in MOVES and is set to zero for all vehicle types.
3.1. Historic Vehicle Miles Traveled (1990 ami 1999-2021)
In MOVES4, VMT estimates for the historic years 1990 and 1999-2021 come from the VM-1
table of US DOT Federal Highway Administration's (FHWA) Highway Statistics series.6 In
reporting years 2007 and later, the VM-1 data are calculated with an updated methodology,12
which implements state-reported data directly rather than a modeled approach and which has
different vehicle categories. The current HPMS-based VM-1 categories are 1) light-duty short
wheelbase, 2) light-duty long wheelbase, 3) motorcycles, 4) buses, 5) single-unit trucks and 6)
combination trucks. Because MOVES categorizes light-duty source types based on vehicle type
and not wheelbase length, the short and long wheelbase categories are combined into a single
category of light-duty vehicles (HPMSVTypelD 25). Internally, the MOVES Total Activity
Generator allocates this VMT to MOVES source types and ages using vehicle populations, age
distributions and relative mileage accumulation rates.
For years prior to 2007, the VM-1 data with historical vehicle type groupings are inconsistent
with the current VM-1 vehicle categories used in MOVES and cannot be used as they are
currently reported. However, in early 2011, FHWA released revised VMT data for years 2000-
2006 to match the new category definitions. Shortly afterward, the agency replaced these revised
numbers with the previously published VMT data stating, "[FHWA] determined that it is more
reliable to retain the original 2000-2006 estimates because the information available for those
b In MOVES, users can enter VMT estimates using four different input methods: annual miles by HPMS class,
annual miles by source type, annual average daily miles by HPMS class and annual average daily miles by source
type. As in previous versions of MOVES, the national defaults are stored as annual miles by HPMS class and any
discussion in this report on annual VMT estimates will be in this context.
21
-------
years does not fully meet the requirements of the new methodology."c However, needing
continuity of the VM-1 vehicle categories, we used these FHWA-revised values by the new
categories as the VMT for 2000-2006.
This left two years, 1990 and 1999, that needed to be adjusted to be consistent with the new
HPMS vehicle categories. Since the methodology that FHWA used to revise the 2000-2006 data
is undocumented, we adjusted 1990 and 1999 using the average ratio of the change for each
vehicle category. This was found by dividing the FHWA-adjusted VMT for each vehicle
category by the original VMT for each year 2000-2006 and then calculating the average ratio for
each category. This ratio was then applied to the corresponding VMT values reported in VM-1
for 1990 and 1999. Since FHWA's adjustments conserved the original total VMT estimates, we
normalized our adjusted values such that the original total VMT for the years were unchanged.
The resulting values for historic years by HPMS vehicle class are listed in Table 3-1. The VMT
for 1990 and 1999 were EPA-adjusted from VM-1, 2000-2006 were FHWA-revised and 2007
and later were unadjusted, other than the simple combination of the short and long wheelbase
classes into light-duty vehicles. In addition to these adjustments, for some years, the VMT values
were revised by FHWA in subsequent publications. Table 3-2 summarizes the data source and
revision date we used for each historical year.
0 This text appears in a footnote to FHWA's Highway Statistics Table VM-1 for publication years 2000-2009.
22
-------
Table 3-1 Historic year VMT by HPMS vehicle class (millions of miles)
Year
Motorcycles
Light-Duty
Vehicles
Buses
Single Unit
Trucks
Combination
Trucks
1990
11,404
1,943,194
10,279
70,861
108,624
1999
13,619
2,401,408
14,853
100,534
160,921
2000
12,175
2,458,221
14,805
100,486
161,238
2001
11,120
2,499,069
12,982
103,470
168,969
2002
11,171
2,555,468
13,336
107,317
168,217
2003
11,384
2,579,195
13,381
112,723
173,539
2004
14,975
2,652,092
13,523
111,238
172,960
2005
13,773
2,677,641
13,153
109,735
175,128
2006
19,157
2,680,537
14,038
123,318
177,321
2007
21,396
2,691,034
14,516
119,979
184,199
2008
20,811
2,630,213
14,823
126,855
183,826
2009
20,822
2,633,248
14,387
120,207
168,100
2010
18,513
2,648,456
13,770
110,738
175,789
2011
18,542
2,650,458
13,807
103,803
163,791
2012
21,385
2,664,060
14,781
105,605
163,602
2013
20,366
2,677,730
15,167
106,582
168,436
2014
19,970
2,710,556
15,999
109,301
169,830
2015
19,606
2,779,693
16,230
109,597
170,246
2016
20,445
2,849,718
16,350
113,338
174,557
2017
20,149
2,877,378
17,227
116,102
181,490
2018
20,076
2,232,588
18,303
120,699
184,165
2019
19,688
2,254,309
17,980
124,746
175,305
2020
17,632
2,568,745
15,104
124,880
177,261
2021
19,738
2,776,073
16,793
131,869
195,616
23
-------
Year
FHWA Publication Source (Publication/Revision Date)
1990
Highway Statistics 1991 (October 1992)
1999
Highway Statistics 1999 (October 2000)
2000
Highway Statistics 2000 (April 2011)
2001
Highway Statistics 2001 (April 2011)
2002
Highway Statistics 2002 (April 2011)
2003
Highway Statistics 2003 (April 2011)
2004
Highway Statistics 2004 (April 2011)
2005
Highway Statistics 2005 (April 2011)
2006
Highway Statistics 2006 (April 2011)
2007
Highway Statistics 2007 (April 2011)
2008
Highway Statistics 2008 (April 2011)
2009
Highway Statistics 2010 (December 2012)
2010
Highway Statistics 2010 (December 2012)
2011
Highway Statistics 2012 (January 2014)
2012
Highway Statistics 2013 (January 2015)
2013
Highway Statistics 2014 (December 2015)
2014
Highway Statistics 2014 (December 2015)
2015
Highway Statistics 2015 (January 2017)
2016
Highway Statistics 2016 (May 2018)
2017
Highway Statistics 2017 (March 2019)
2018
Highway Statistics 2018 (March 2020)
2019
Highway Statistics 2019 (November 2020)
2020
Highway Statistics 2020 (December 2021)
2021
Highway Statistics 2021 (February 2023)
3.2. Projected Vehicle Miles Traveled (2022-2060)
The Annual Energy Outlook (AEO)13 describes the future energy consumption forecasted by
Department of Energy. Vehicle sales and miles traveled are included in the projections because
they strongly influence fuel consumption. In MOVES4, VMT for years beyond 2021 are based
on the reference case VMT projections from AEO2023. Because AEO vehicle categories are
different from HPMS classes, the AEO projections were not used directly. Instead, year-to-year
percent changes in the projected values were calculated and applied to the 2021 base year HPMS
data. Since AEO2023 only projects out to 2050, VMT for years 2051-2060 were assumed to
continue to grow at the same growth rate as between 2049 and 2050.
Table 3-3 shows the mappings between AEO VMT categories and HPMS categories. Where
multiple AEO categories are listed, their VMT were summed before calculating the year-over-
year growth rates. AEO's light-duty category was mapped to both the combined HPMS light-
duty and the motorcycle categories. Motorcycles were included here because they were not
explicitly accounted for elsewhere in AEO. Since buses span a large range of heavy-duty
vehicles and activity, the combination of AEO's light-medium-, medium- and heavy-heavy-duty
categories was mapped to the HPMS bus category. AEO's light-medium- and medium-heavy-
duty categories were combined for mapping to the HPMS single-unit truck category and AEO's
heavy-heavy-duty category was mapped to the HPMS combination truck category. We
acknowledge that using VMT growth estimates from different vehicle types as surrogates for
motorcycles and buses, in particular, will introduce additional uncertainty into these projections.
24
-------
Table 3-3 Mapping AEO categories to
iPMS classes for projecting VMT
AEO VMT Category Groupings
HPMS Class
Total Light-Duty VMT1
10 - Motorcycles
Total Commercial Light Truck VMT11
25 - Light Duty Vehicles
Total Heavy-Duty VMT111
40 - Buses
Light-Medium Subtotal VMT111
+
Medium Subtotal VMT111
50 - Single Unit Trucks
Heavy Subtotal VMT111
60 - Combination Trucks
Notes:
I From AEO2023 Table 41: Light-Duty VMT by Technology Type
II From AEO2023 Table 46: Transportation Fleet Car and Truck VMT by Type and Technology
III From AEO2023 Table 49: Freight Transportation Energy Use
The percent growth over time was calculated for each of the groups described above and applied
by HPMS category to the 2021 base year VMT from Highway Statistics Table VM-1. The
resulting values are presented in Table 3-4 below.
25
-------
Table 3-4
WIT
projections
for 2022-
2060 by
HPMS
Motorcycles
Light-Duty
Vehicles
Buses
Single Unit
Trucks
Combination
Trucks
class
(millions of
miles)Year
2022
20,633
2,901,945
17,214
135,406
200,278
2023
21,047
2,960,259
17,148
134,608
199,807
2024
21,197
2,981,301
17,171
134,815
200,053
2025
21,279
2,992,798
17,325
136,253
201,606
2026
21,434
3,014,635
17,561
138,444
203,994
2027
21,637
3,043,248
17,768
140,450
205,996
2028
21,825
3,069,676
17,958
142,460
207,665
2029
21,973
3,090,504
18,105
144,361
208,574
2030
22,087
3,106,420
18,240
146,472
209,036
2031
22,183
3,119,997
18,401
148,997
209,566
2032
22,267
3,131,809
18,615
152,082
210,565
2033
22,385
3,148,359
18,795
154,890
211,188
2034
22,533
3,169,199
18,971
157,678
211,738
2035
22,670
3,188,409
19,159
160,568
212,421
2036
22,782
3,204,208
19,322
163,251
212,825
2037
22,912
3,222,473
19,511
166,165
213,503
2038
23,058
3,243,010
19,696
169,004
214,176
2039
23,204
3,263,620
19,879
171,934
214,728
2040
23,372
3,287,177
20,081
175,071
215,428
2041
23,544
3,311,409
20,295
178,349
216,218
2042
23,723
3,336,548
20,505
181,564
216,996
2043
23,896
3,360,929
20,702
184,702
217,607
2044
24,075
3,386,035
20,886
187,773
218,001
2045
24,264
3,412,617
21,057
190,856
218,146
2046
24,483
3,443,472
21,255
194,304
218,439
2047
24,719
3,476,595
21,459
197,881
218,709
2048
24,960
3,510,577
21,635
201,225
218,685
2049
25,205
3,544,972
21,826
204,751
218,745
2050
25,474
3,582,818
22,086
208,966
219,456
2051
25,746
3,621,068
22,349
213,267
220,169
2052
26,021
3,659,726
22,615
217,657
220,884
2053
26,298
3,698,797
22,884
222,137
221,601
2054
26,579
3,738,285
23,156
226,709
222,321
2055
26,863
3,778,195
23,432
231,376
223,043
2056
27,150
3,818,530
23,710
236,138
223,767
2057
27,440
3,859,297
23,993
240,999
224,494
26
-------
Table 3-4
WIT
projections
for 2022-
2060 by
HPMS
class
(millions of
miles)Year
Motorcycles
Light-Duty
Vehicles
Buses
Single Unit
Trucks
Combination
Trucks
2058
27,733
3,900,498
24,278
245,960
225,223
2059
28,029
3,942,139
24,567
251,022
225,954
2060
28,328
3,984,225
24,859
256,189
226,688
27
-------
4. Vehicle Populations by Calendar Year
MOVES uses vehicle populations to characterize emissions activity that is not directly dependent
on VMT. These population data are also used to allocate VMT from HPMS class to source type
and age (for more details, see Section 6). The default database stores historic estimates and
future projections of total US vehicle populations in 1990 and 1999-2060 by source type. The
MOVES database stores this information in the SourceTypeYear table, which has three data
fields: sourceTypePopulation, salesGrowthFactor and migrationRate. However, the
salesGrowthFactor and migrationRate fields are not used in MOVES.
4.1. Historic Source Type Populations (1990, 1999-2021)
MOVES populations for calendar years 1990 and 1999-2021 are derived primarily from
registration data summarized in the Federal Highway Administration's annual Highway Statistics
report. Motorcycle populations are from vehicle registrations reported in Table VM-1,6 and
passenger car populations are from registrations reported in Table MV-1.14 The general
categories for truck and bus registrations presented in Highway Statistics were allocated to
specific MOVES source types as described below.
The numbers of single-unit and combination trucks were determined for each calendar year using
registration data in the Highway Statistics Table VM-1. The remaining MV-1 truck registrations
were allocated to the light-duty trucks. The populations were further allocated from the light-
duty, single-unit and combination truck categories to individual source types using the source
type distribution fractions shown below in Table 4-1.
The source type distribution fractions were calculated from national vehicle registration data
purchased from IHS15'16 for calendar years 1999 and 2020. These fractions were calculated as the
ratio of the total registrations by source type to their corresponding HPMS class totals (see Table
2-1 for this mapping). These fractions were then linearly interpolated to estimate the source type
distribution fractions for all years between 1999 and 2020. However, there are a few nuances to
this analysis:
• Starting with MOVES3, the distinction between passenger light-duty trucks (31) and
commercial light-duty trucks (32) was revised. Now in MOVES, a light-duty truck is
considered a passenger truck if it is registered to an individual and a commercial light-
duty truck if it is registered to an organization or business. Since this is inconsistent with
the source type definitions used by the 1999 IHS data, the ratio of passenger to
commercial light-duty trucks from the 2020 IHS data were used for all calendar years.
• The 2020 IHS data were unable to distinguish between short-haul (52) and long-haul (53)
single-unit trucks and consequentially grouped them together. These vehicles are
differentiated in MOVES4 using an earlier IHS data for 2011 which was able to
differentiate between these vehicles. From the earlier data, it was determined that of
short-haul and long-haul single-unit trucks, 95.8 percent are short-haul. This percentage
fraction was applied for all historic years to differentiate between these two source types.
28
-------
Ta
?le 4-1 Source type distributions used to allocate truck
)opulations in MOVES4
Year
31/30
32/30
51/50
52/50
53/50
54/50
61/60
62/60
1990"
0.913632
0.086368
0.013311
0.767722
0.03386
0.185107
0.625648
0.374352
1999"'
0.913632
0.086368
0.015472
0.791929
0.034927
0.157671
0.574437
0.425563
2000
0.913632
0.086368
0.015006
0.794353
0.035034
0.155606
0.57569
0.42431
2001
0.913632
0.086368
0.01454
0.796777
0.035141
0.153541
0.576943
0.423057
2002
0.913632
0.086368
0.014074
0.799201
0.035248
0.151476
0.578196
0.421804
2003
0.913632
0.086368
0.013608
0.801625
0.035355
0.149411
0.57945
0.42055
2004
0.913632
0.086368
0.013142
0.804049
0.035462
0.147346
0.580703
0.419297
2005
0.913632
0.086368
0.012676
0.806473
0.035569
0.145281
0.581956
0.418044
2006
0.913632
0.086368
0.01221
0.808897
0.035676
0.143216
0.583209
0.416791
2007
0.913632
0.086368
0.011744
0.811321
0.035783
0.141151
0.584462
0.415538
2008
0.913632
0.086368
0.011278
0.813746
0.03589
0.139086
0.585715
0.414285
2009
0.913632
0.086368
0.010813
0.81617
0.035997
0.137021
0.586969
0.413031
2010
0.913632
0.086368
0.010347
0.818594
0.036103
0.134956
0.588222
0.411778
2011
0.913632
0.086368
0.009881
0.821018
0.03621
0.132891
0.589475
0.410525
2012
0.913632
0.086368
0.009415
0.823442
0.036317
0.130826
0.590728
0.409272
2013
0.913632
0.086368
0.008949
0.825866
0.036424
0.128761
0.591981
0.408019
2014
0.913632
0.086368
0.008483
0.82829
0.036531
0.126696
0.593235
0.406765
2015
0.913632
0.086368
0.008017
0.830714
0.036638
0.124631
0.594488
0.405512
2016
0.913632
0.086368
0.007551
0.833138
0.036745
0.122567
0.595741
0.404259
2017
0.913632
0.086368
0.007085
0.835562
0.036852
0.120502
0.596994
0.403006
2018
0.913632
0.086368
0.006619
0.837986
0.036959
0.118437
0.598247
0.401753
2019
0.913632
0.086368
0.006153
0.84041
0.037066
0.116372
0.599501
0.400499
2020*"
0.913632
0.086368
0.005687
0.842834
0.037173
0.114307
0.600754
0.399246
Note:
* Fractions may not sum to one due to rounding.
" Fractions from 1990 were retained from MOVES201417 with the exceptions noted in the text.
"* Fractions from 1999 and 2020 were calculated from IHS registration data with the exceptions noted in the text;
fractions for other years were estimated from these values.
Buses were allocated using different data sources:
• School bus (43) populations for 2002-2021 come from the School Bus Fleet Fact Book18
publication series' School Transportation Statistics tables. Since these values are
presented as totals corresponding to academic years (e.g., 2016-2017) and MOVES
requires national values to be entered for calendar years, the data were taken to
correspond to the year in which the school year ends (2017, in the example). For 1990
and 1999-2001, school buses were assumed to be a constant proportion of the total bus
population in each year based on the 2002 counts. Note that the School Bus Fleet Fact
Book series did not publish the School Transportation Statistics tables for the academic
years 2019-2020 or 2020-2021. Therefore, the school bus populations for calendar years
2020 and 2021 were linearly interpolated from the 2019 and 2022 school bus population
values.
• Transit bus (42) populations were calculated from the Federal Transit Administration's
National Transit Database (NTD)19 data series on Revenue Vehicle Inventory and Rural
Revenue Vehicle Inventory. See Section 5.1.4 for more information on the definition of
transit buses in MOVES. For 1990 and 1999-2001, transit buses were assumed to be a
constant proportion of the total bus population in each year based on the 2002 counts.
• Other bus (41) populations were calculated as the remainder of the MV-1 bus
registrations less the school bus and transit bus populations. Note that the Highway
29
-------
Statistics series on bus populations show large changes in bus registrations for 2011,
2012, and 2016-2020, inconsistent with intermediate years as well as historic populations.
Lacking evidence that these specific data reflect actual changes in the number of buses
operating in the US, the bus registration values for those years were dropped and
estimated instead with linear interpolation/extrapolation. Specifically, 2011 and 2012
values were linearly interpolated from 2010 and 2013 registrations, and 2016-2020 values
were linearly interpolated from 2015 and 2021 registrations.
For all source type populations derived from Table VM-1, note that this registration data has the
same vehicle category differences as the VMT data for reporting years prior to 2007 as described
in Section 3.1. Similar to the VMT analysis, we used the FHWA-revised values for 2000-2006
and adjusted the registration data ourselves for 1990 and 1999 as described in Section 3.1.
30
-------
rable 4-2 H
istoric source type populations for calent
ar years 1990 and 1999-2021
yearlD
Motorcycle
Passenger
Car
Passenger
Truck
Light
Com.
Truck
Other
Bus
Transit
Bus
School
Bus
Refuse
Truck
Single Unit
Short-haul
Single Unit
Long-haul
Motorhome
Combination
Short-haul
Combination
Long-haul
1990
3,657,632
143,549,627
35,233,652
3,330,743
172,025
48,151
318,050
57,229
3,300,770
145,578
795,855
1,010,435
604,587
1999
4,032,581
132,432,044
67,831,975
6,412,359
226,133
63,296
418,087
102,180
5,229,900
230,661
1,041,260
1,320,899
978,569
2000
4,346,068
133,621,420
71,372,854
6,747,089
238,473
66,750
440,902
98,708
5,225,066
230,448
1,023,540
1,387,368
1,022,554
2001
4,903,056
137,633,467
75,468,041
7,134,220
239,567
67,056
442,925
101,383
5,555,547
245,023
1,070,569
1,425,354
1,045,174
2002
5,004,156
135,920,677
76,343,163
7,216,948
243,137
68,055
449,525
98,025
5,566,262
245,496
1,054,998
1,395,588
1,018,105
2003
5,370,035
135,669,897
78,109,534
7,383,928
237,582
68,604
470,364
96,029
5,656,706
249,485
1,054,327
1,386,937
1,006,605
2004
5,780,870
136,430,651
82,631,649
7,811,417
256,107
68,796
470,371
94,274
5,767,659
254,378
1,056,953
1,393,895
1,006,464
2005
6,227,146
136,568,083
85,821,437
8,112,957
264,495
69,514
473,044
94,076
5,985,115
263,969
1,078,182
1,433,416
1,029,685
2006
6,678,958
135,399,945
89,179,459
8,430,401
274,929
70,232
476,798
94,704
6,273,755
276,699
1,110,776
1,503,506
1,074,482
2007
7,138,476
135,932,930
91,130,392
8,614,829
266,607
82,378
485,451
95,326
6,585,230
290,437
1,145,679
1,540,261
1,095,086
2008
7,752,926
137,079,843
90,786,036
8,582,276
270,188
84,739
488,381
93,477
6,744,360
297,455
1,152,754
1,514,209
1,071,021
2009
7,929,724
134,879,600
90,986,822
8,601,256
293,181
86,756
462,056
90,350
6,819,992
300,791
1,144,963
1,536,166
1,080,952
2010
8,009,503
130,892,240
90,954,042
8,598,158
284,828
89,097
472,126
85,019
6,726,539
296,669
1,108,962
1,501,651
1,051,214
2011
8,437,502
125,656,528
98,841,145
9,343,749
291,480
88,076
472,661
77,257
6,419,582
283,131
1,039,085
1,445,179
1,006,459
2012
8,454,939
111,289,906
111,893,065
10,577,586
298,491
91,912
467,980
77,108
6,744,224
297,449
1,071,506
1,458,563
1,010,530
2013
8,404,687
113,676,345
111,768,111
10,565,774
297,426
94,222
472,901
72,716
6,710,991
295,983
1,046,316
1,462,992
1,008,356
2014
8,417,718
113,898,845
115,351,838
10,904,554
289,926
98,060
484,041
70,649
6,898,626
304,259
1,055,224
1,528,882
1,048,315
2015
8,600,936
112,864,228
118,820,506
11,232,458
300,197
103,669
485,041
67,791
7,024,767
309,822
1,053,921
1,632,988
1,113,894
2016
8,679,380
112,961,266
123,051,307
11,632,408
316,227
106,871
474,194
66,042
7,287,056
321,390
1,072,030
1,639,505
1,112,538
2017
8,715,204
111,177,029
127,338,527
12,037,692
326,102
108,115
471,461
66,149
7,801,640
344,086
1,125,123
1,726,637
1,165,581
2018
8,666,185
108,570,167
129,863,323
12,276,368
331,268
106,645
476,150
68,357
8,654,635
381,706
1,223,201
1,738,513
1,167,498
2019
8,596,314
108,547,710
132,720,051
12,546,423
334,921
107,660
479,867
62,514
8,538,930
376,603
1,182,386
1,753,665
1,171,545
2020
8,347,435
105,135,300
135,714,463
12,829,494
340,474
107,199
483,161
56,346
8,351,145
368,321
1,132,597
1,796,832
1,194,130
2021
9,892,706
102,973,881
141,340,004
13,361,293
346,155
106,610
486,454
60,937
9,031,555
398,330
1,224,875
1,888,460
1,255,024
31
-------
Note that the decline in sales seen in the 2008 recession results in a flattening of total population
growth rates and eventually a decline in total population for passenger cars and long-haul
combination trucks as shown in Table 4-2. This suggests that the decline in sales was
accompanied by a delay in the scrappage of older vehicles. The dynamic vehicle survival rates in
MOVES and their impact on age distributions are discussed in Appendix C.
4.2.Projected Vehicle Populations (2022-2060)
Vehicle stock estimates from the reference case of AEO2023 were used to project future
populations, using a methodology similar to the VMT projections as described in Section 3.2.
Because AEO vehicle categories differ from MOVES source types, the AEO projected vehicle
stocks were not used directly. Instead, year-to-year percent changes in the projected values were
calculated and applied to the base year populations. Since AEO2023 only projects out to 2050,
populations for years 2051-2060 were assumed to continue to grow at the same growth rate as
between 2049 and 2050.
Table 4-3 shows the mappings between AEO stock categories and MOVES source types. Where
multiple AEO categories are listed, their stocks were summed before calculating the year-over-
year growth rates. AEO's car category was mapped to both motorcycle and passenger car
categories. Motorcycles were included here because they were not explicitly accounted for
elsewhere in AEO. Since buses span a large range of heavy-duty vehicles and activity, the
combination of AEO's light-medium-, medium- and heavy-heavy-duty categories was mapped to
each source type in the HPMS bus category. AEO's light-medium- and medium-heavy-duty
categories were combined for mapping to each source type in the HPMS single-unit truck
category and AEO's heavy-heavy-duty category was mapped to each source type in the HPMS
combination truck category. We acknowledge that using stock growth estimates from different
vehicle types as surrogates for motorcycles and buses, in particular, will introduce additional
uncertainty into these projections.
32
-------
Table 4-3 Mapping AEO categories to source types for projecting vehicle populations
AEO Stock Category Groupings
MOVES Source Type
Total Car Stock1
11 - Motorcycle
21 - Passenger Car
Total Light Truck Stock1
+
Total Commercial Light Truck Stock11
31 - Passenger Truck
32 - Light Commercial Truck
Total Heavy-Duty Stock111
41 - Other Bus
42 - Transit Bus
43 - School Bus
Light-Medium Subtotal Stock111
+
Medium Subtotal Stock111
51 - Refuse Truck
52 - Single Unit Short-haul Truck
53 - Single Unit Long-haul Truck
54 - Motor Home
Heavy Subtotal Stock111
61 - Combination Short-haul Truck
62 - Combination Long-haul Truck
Notes:
I From AE02020 Table 45: Light-Duty Vehicle Stock by Technology Type
II From AE02020 Table 46: Transportation Fleet Car and Truck Stock by Type and Technology
III From AE02020 Table 49: Freight Transportation Energy Use
The percent growth over time was calculated for each of the groups described above and applied
to the 2021 base year source type populations. The resulting populations are presented in Table
4-4.
33
-------
yearlD
2022
2023
2024
2025
2026
2027
2028
2029
2030
2031
2032
2033
2034
2035
2036
2037
2038
2039
2040
2041
2042
2043
2044
2045
2046
2047
2048
2049
Table 4-4 Projected source type populations for 2022-2060
Motorcycle
Passenger
Car
Passenger
Truck
Light Com.
Truck
Other
Bus
Transit
Bus
School
Bus
Refuse
Truck
Single Unit
Short-haul
Single Unit
Long-haul
Motorhome
Combination
Short-haul
9,960,980
100,879,903
145,568,230
13,761,000
355,038
109,346
498,937
62,881
9,319,612
411,035
1,263,942
9,980,784
98,864,871
148,495,717
14,037,744
363,032
111,808
510,171
64,581
9,571,615
422,149
1,298,119
10,020,715
96,976,695
151,808,977
14,350,956
371,449
114,400
522,000
66,401
9,841,381
434,047
1,334,705
10,068,231
95,019,911
155,406,263
14,691,019
380,201
117,096
534,299
68,309
10,124,111
446,516
1,373,049
10,118,077
93,044,514
159,087,837
15,039,049
389,222
119,874
546,976
70,263
10,413,770
459,292
1,412,333
10,163,889
91,023,562
162,716,059
15,382,035
398,448
122,715
559,941
72,244
10,707,410
472,242
1,452,157
10,202,610
88,929,876
166,241,760
15,715,330
407,504
125,504
572,668
74,200
10,997,330
485,029
1,491,477
10,241,184
86,849,273
169,747,957
16,046,782
415,952
128,106
584,540
76,081
11,276,029
497,321
1,529,274
10,274,461
84,768,348
173,113,252
16,364,913
423,875
130,547
595,675
77,899
11,545,483
509,205
1,565,818
10,302,337
82,703,488
176,315,354
16,667,617
431,222
132,809
605,999
79,616
11,800,003
520,430
1,600,337
10,326,938
80,701,901
179,354,785
16,954,944
438,391
135,017
616,074
81,330
12,054,102
531,637
1,634,798
10,347,718
78,770,062
182,209,219
17,224,782
444,914
137,026
625,240
82,916
12,289,045
541,999
1,666,661
10,365,945
76,944,360
184,869,219
17,476,240
450,953
138,886
633,727
84,443
12,515,385
551,982
1,697,358
10,387,180
75,271,063
187,427,971
17,718,126
457,036
140,760
642,277
85,970
12,741,715
561,964
1,728,053
10,414,307
73,806,497
189,895,322
17,951,372
463,417
142,725
651,243
87,540
12,974,404
572,226
1,759,611
10,446,868
72,549,531
192,260,465
18,174,956
469,912
144,725
660,371
89,138
13,211,315
582,675
1,791,741
10,486,428
71,507,929
194,555,849
18,391,946
476,380
146,717
669,460
90,740
13,448,712
593,145
1,823,937
10,532,381
70,655,368
196,796,636
18,603,774
482,531
148,612
678,104
92,319
13,682,721
603,466
1,855,674
10,585,589
69,980,785
199,019,056
18,813,866
488,362
150,407
686,299
93,854
13,910,230
613,500
1,886,529
10,641,874
69,428,865
201,177,470
19,017,908
493,995
152,142
694,215
95,418
14,142,009
623,723
1,917,964
10,701,018
68,980,482
203,288,861
19,217,504
499,891
153,958
702,500
96,978
14,373,263
633,922
1,949,327
10,761,830
68,587,840
205,378,385
19,415,033
506,440
155,975
711,703
98,681
14,625,625
645,052
1,983,552
10,824,308
68,253,422
207,443,011
19,610,208
513,482
158,144
721,600
100,486
14,893,131
656,850
2,019,832
10,888,569
67,967,034
209,498,009
19,804,473
520,345
160,258
731,244
102,251
15,154,826
668,392
2,055,323
10,954,013
67,765,663
211,483,323
19,992,151
526,737
162,226
740,227
103,950
15,406,596
679,496
2,089,469
11,019,269
67,572,902
213,453,366
20,178,385
533,008
164,158
749,040
105,664
15,660,640
690,701
2,123,923
11,083,706
67,405,457
215,371,446
20,359,706
539,141
166,046
757,658
107,373
15,913,938
701,872
2,158,276
11,148,959
67,255,915
217,289,944
20,541,068
544,712
167,762
765,487
108,989
16,153,368
712,432
2,190,748
34
-------
yearlD
Motorcycle
Passenger
Car
Passenger
Truck
Light Com.
Truck
Other
Bus
Transit
Bus
School
Bus
Refuse
Truck
Single Unit
Short-haul
Single Unit
Long-haul
Motorhome
Combination
Short-haul
Combination
Long-haul
2050
11,217,514
67,150,524
219,243,961
20,725,787
549,881
169,354
772,752
110,555
16,385,455
722,668
2,222,224
2,340,070
1,555,153
2051
11,286,490
67,045,298
221,215,550
20,912,167
555,100
170,962
780,085
112,143
16,620,877
733,051
2,254,152
2,336,793
1,552,975
2052
11,355,891
66,940,237
223,204,868
21,100,223
560,368
172,584
787,489
113,754
16,859,681
743,584
2,286,539
2,333,520
1,550,800
2053
11,425,718
66,835,341
225,212,076
21,289,971
565,686
174,222
794,962
115,389
17,101,916
754,267
2,319,391
2,330,252
1,548,628
2054
11,495,975
66,730,609
227,237,334
21,481,424
571,054
175,875
802,506
117,047
17,347,632
765,104
2,352,716
2,326,988
1,546,459
2055
11,566,664
66,626,041
229,280,804
21,674,600
576,474
177,544
810,122
118,728
17,596,878
776,097
2,386,519
2,323,729
1,544,293
2056
11,637,787
66,521,637
231,342,651
21,869,512
581,944
179,229
817,810
120,434
17,849,705
787,248
2,420,808
2,320,475
1,542,130
2057
11,709,348
66,417,397
233,423,039
22,066,177
587,467
180,930
825,572
122,164
18,106,164
798,559
2,455,589
2,317,225
1,539,971
2058
11,781,349
66,313,320
235,522,135
22,264,611
593,042
182,647
833,406
123,920
18,366,309
810,032
2,490,870
2,313,979
1,537,814
2059
11,853,792
66,209,406
237,640,108
22,464,830
598,670
184,381
841,315
125,700
18,630,190
821,671
2,526,658
2,310,739
1,535,660
2060
11,926,681
66,105,655
239,777,127
22,666,848
604,352
186,130
849,300
127,506
18,897,864
833,476
2,562,961
2,307,502
1,533,509
35
-------
5. Fleet Characteristics
Despite the availability of vehicle registration databases, comprehensive surveys for
characterizing travel pattern and sophisticated sensors and cameras for measuring vehicle
activity, it is still difficult to estimate vehicle populations in the categories needed for emissions
inventory modeling. Differentiating, for example, between passenger car and trucks, or between
light-duty and heavy-duty trucks presents substantial modeling challenges since the
characteristics that are important for emissions are not always readily observable.20 21 To develop
MOVES defaults, we have merged registration and survey data with activity measurements in an
effort to identify key vehicle parameters such as weight, axle and tire configuration and typical
trip range.
MOVES categorizes vehicles into thirteen source types as described in Section 2.1, which are
defined using physical characteristics, such as number of axles and tires and travel behavior
characteristics, such as typical trip lengths. This section describes the defining characteristics of
the source types in greater detail, explains how source type is related to fuel type and regulatory
class through the SampleVehiclePopulation table and how MOVES estimates and projects the
number of vehicles in each category.
5.1.Source Type Definitions
MOVES source types are intended to further divide HPMS vehicle classifications into groups of
vehicles with similar activity patterns. For example, passenger trucks and light commercial
trucks are expected to have different daily trip patterns.
5.1.1. Motorcycles
According to the HPMS vehicle description, motorcycles (sourceTypelD 11) are, "all two- or
three-wheeled motorized vehicles, typically with saddle seats and steered by handlebars rather
than a wheel."22 This category usually includes any registered motorcycles, motor scooters,
mopeds and motor-powered bicycles. Please note that off-road motorcycles are regulated as
nonroad equipment and are not covered in this report.
5.1.2. Passenger Cars
Passenger cars are defined as any coupes, compacts, sedans, or station wagons with the primary
purpose of carrying passengers.22 For consistency with vehicle emission standards, the category
also includes some small crossover vehicles.23 All passenger cars (sourceTypelD 21) are
categorized in the light-duty vehicle regulatory class (regClassID 20).
5.1.3. Light-Duty Trucks
Light-duty trucks include pickups, most sport utility vehicles (SUVs) and vans.22 FHWA's
vehicle classification specifies that light-duty vehicles are those weighing less than 10,000
pounds, specifically vehicles with a GVWR in Class 1 and 2; with the exception of Class 2b
trucks (8,500 to 10,000 lbs) with two axles or more and at least six tires, colloquially known as
"duallies", which FHWA classifies into the single-unit truck category.
36
-------
In MOVES, a light-duty truck is considered a passenger truck (sourceTypelD 31) if it is
registered to an individual, or a light-duty commercial truck (sourceTypelD 32) if it is registered
to an organization or business.
Because the Class 2b trucks with only 2 axles and only 4 tires are classified in the light-duty
source types, sourceTypelDs 31 and 32 contain vehicles in both the light-duty truck regulatory
class (regClassID 30) and the Class 2b and 3 truck regulatory class (regClassID 41) as discussed
in Section 5.2.3.
5.1.4. Buses
MOVES has three bus source types: other (sourceTypelD 41), transit (sourceTypelD 42) and
school buses (sourceTypelD 43).
Transit buses in MOVES are defined as any active vehicle with a bus body type ("bus",
"articulated bus", "over-the-road bus", "double decked bus" and "cutaway") that must be
reported to Federal Transit Administration's (FTA) National Transit Database (NTD). According
to the FTA, these are buses owned by a public transit organization for the primary purpose of
transporting passengers on fixed routes and schedules.24
School buses in MOVES are defined as according to FHWA: vehicles designed to carry more
than ten passengers and are used to transport K-12 students between their home and school.25
Any other buses that do not fit into the transit or school bus categories are modeled in MOVES
as "other" buses.d For example, these may include intercity buses not owned by transit agencies.
Please note that these definitions allow similar vehicle types to be modeled in both the transit and
other bus source types. For example, a shuttle bus operated by a transit agency would be
modeled as a transit bus, but an airport shuttle bus operated by a private company would be
modeled as an "other" bus. Due to the similarities between these source types, they have
identical fuel type and regulatory class distributions. However, they do have different age
distributions and driving schedules as described in subsequent sections.
5.1.5. Single-Unit Trucks
The single-unit HPMS class in MOVES consists of refuse trucks (sourceTypelD 51), short-haul
single-unit trucks (sourceTypelD 52), long-haul single-unit trucks (sourceTypelD 53) and motor
homes (sourceTypelD 54). FHWA's vehicle classification specifies that single-unit trucks are
single-frame trucks with a gross vehicle weight rating of greater than 10,000 pounds or with two
axles and at least six tires—colloquially known as "dualies." The difference between short-haul
and long-haul single-unit trucks is their primary trip length; short-haul trucks travel less than or
equal to 200 miles a day and long-haul trucks travel more than 200 miles a day.
5.1.6. Combination Trucks
d Note, in previous versions of MOVES, "other" buses were called "intercity" buses and defined slightly differently.
37
-------
The combination truck HPMS class in MOVES consists of two source types: short-haul
(sourceTypelD 61) and long-haul combination trucks (sourceTypelD 62). These are heavy-duty
trucks that are not single-frame. Like single-unit trucks, short-haul and long-haul combination
trucks are distinguished by their primary trip length; short-haul trucks travel less than or equal to
200 miles a day and long-haul trucks travel more than 200 miles a day. Generally, short-haul
combination trucks are older than long-haul combination trucks and these short-haul trucks are
often purchased in secondary markets, such as for drayage applications, after being used
primarily for long-haul trips.26
5.2. Sample Vehicle Population
To match source types to emission rates, MOVES must associate each source type with specific
fuel types, technologies (EngTech), and regulatory classes. As vehicle markets shift, these
distributions change with model year. This information is stored in the SampleVehiclePopulation
(SVP) table, which contains two fractions: stmyFraction and stmyFuelEngFraction.
The stmyFraction represents the default national fuel type, EngTech, and regulatory class
allocation for each source type and model year. We define the stmyFraction as shown in
Equation 5-1.
/ (stmy)stmyjtetrc —
Nst,my,ft,et,rc
y rtar Nst,my,rwc Equation S-l
' I of-CPT
'etEET
rcERC
where the number of vehicles JV in a given source type st, model year my, fuel type ft, EngTech
et, and regulatory class rc is divided by the sum of vehicles across the set of all fuel types FT,
EngTechs ET, and regulatory classes RC. That is, the denominator is the total number of vehicles
in a given source type and model year and so the stmyFraction must sum to one for each source
type and model year. For example, model year 2010 passenger trucks have stmyFractions that
indicate the distribution of these vehicles between gasoline, diesel, E85, and electricity fuel types
with their associated EngTechs, and regulatory classes 30 and 41. A value of zero indicates that
the MOVES default population of vehicles of that source type, model year, fuel type, EngTech,
and regulatory class is negligible in the national population or does not exist.
Because a modeler may modify fuel type and EngTech distributions by source type and model
year to simulate local conditions through the Alternative Vehicle Fuel and Technology (AVFT)
table—but is not expected to modify regulatory class distributions—the
SampleVehiclePopulation table also contains the stmyFuelEngFraction. When a modeler
supplies an AVFT table, MOVES will use the StmyFuelEngFraction to apply a default regulatory
class distribution to the user-supplied fuel type and EngTech distributions, regardless of whether
these vehicles exist in the default. Similar to the stmyFraction above, we define
StmyFuelEngFraction as shown in Equation 5-2.
38
-------
f (stTYiyfllclCTig^st :rny,ft ,et ,rc
ZNst,my,ft,
rceRC
Equation 5-2
st,my,ft,et,rc
for number of vehicles JV in a given source type st, model year my, fuel type ft, EngTech et,
regulatory class rc, and the set of all regulatory classes RC. In this case, the denominator is the
total for a given source type, model year, fuel type, and EngTech, and so the
stmyFuelEngFraction must sum to one for each combination of source type, model year, fuel
type, and EngTech.
For a concrete example of how stmyFraction and StmyFuelEngFraction are used in MOVES,
take the example of MY2030 combination long-haul trucks. The stmyFraction assigns the default
fuel, EngTech, and regulatory classes to the population of MY2030 combination long-haul
trucks, which is nearly all HHD class 8 diesel. However, a modeler could create a future scenario
in which there is a high penetration of fuel cell electric trucks. The StmyFuelEngFraction allows
MOVES to assign vehicles to regulatory classes without also requiring the modeler to supply
future weight class distributions.
As noted in Section 2.4, these fuel type fractions indicate the fuel capability of the vehicle, which
is not necessarily the fuel being used by the vehicle. MOVES allocates fuel to specific vehicles
in a two-step process: 1) vehicles are classified by the type of fuel they can use in the fuel type
fraction and then 2) fuels are distributed according to how much of each fuel is used relative to
the vehicles' total fuel consumption in the fuel usage fraction. For example, Figure 5-1 shows the
national default fuel type fractions for all light-duty vehicles among the different MOVES fuel
types. In this report's nomenclature, E85-capable and flexible fuel vehicles are synomous—they
describe vehicles that can accept either gasoline or E-85 fuel. The amount of E-85 versus the
amount of gasoline used out of all the fuel consumed by the vehicle is stored in the
fuelUsageFraction table. Discussion on fuel usage can be found in the MOVES Fuel Supply
Report.8
39
-------
1960 1970 1980 1990 2000 2010 2020 2030 2040 2050 2060
Model Year
Figure 5-1 Default fuel fractions for light-duty source types in MOVES4
In MOVES4, both the stmyFractions and the stmyFuelEngFractions were primarily calculated
using 2014 and 2020 IHS data1516 as explained below. Electric vehicle fleet fraction in MOVES
were based on analysis for EPA rulemaking, considering costs, consumer preferences, and
CARB regulatory programs; see details in each vehicle type section below.
For model years 2000-2013, the 2014 IHS data were used to calculate fuel type, regulatory class,
and EngTech distributions. Values for model years 2014 to 2019 were calculated using the 2020
IHS data. As the 2020 IHS data does not contain complete information on model year 2020 and
later vehicles, we held regulatory class distributions for these vehicles constant at the model year
2019 values, except where noted below.
Before the fuel type and regulatory class distributions could be calculated from the 2014 and
2020 IHS data, the data needed to be cleaned. For the source type field, there were many Class 3
trucks that were classified as light-duty; as MOVES requires Class 3 trucks to be modeled in a
heavy-duty source type, these were all re-classified as "other single-unit trucks" (see Section
5.2.5 for an explanation of this source categorization). Additionally, some compact SUVs were
40
-------
originally classified as light trucks where EPA emission certification data showed that those
makes and models were regulated as cars;23 we re-classified these vehicles as passenger cars. For
the fuel type field, electric hybrids with gasoline or diesel were grouped with fully gasoline or
diesel vehicles since MOVES does not model hybrids separately. Vehicles categorized as
"ethanol" or "flexible" were considered to be in the MOVES E-85 fuel category. If the fuel type
was unknown, it was set to be the most common fuel type for the vehicle's source type and
model year. Any remaining vehicles with alternative fuels (including hydrogen fuel cell,
methanol and "convertible"), or vehicles with source type/fuel type combinations that MOVES
cannot model (such as CNG light commercial trucks) were dropped from the data.
Each of the subsections below describe in more detail which data sources were applied to which
model years for each source type.
5.2.1. Motorcycles
All motorcycles fall into the motorcycle regulatory class (regClassID 10) and must be fueled by
gasoline. Although some alternative fuel motorcycles may exist, they account for a negligible
fraction of total US motorcycle sales and cannot be modeled in MOVES.
5.2.2. Passenger Cars
All passenger cars fall into the light-duty vehicle regulatory class (regClassID 20). IHS data
provided the split between gasoline, diesel, electricity, and E-85 capable cars in the
SampleVehiclePopulation table. For model years 2000 through 2013, the 2014 IHS data were
used, while for model years 2014 to 2019, the 2020 IHS data were used.
In MOVES4 national defaults, for model years prior to 2023, electric passenger cars were
modeled with market shares from the 2022 EPA Trends report.27 For model years 2023 and later,
the electric passenger car sales fractions are based on light-duty electric vehicle cost and
consumer preferences as modeled in EPA OMEGA (Optimization Model for reducing Emissions
of Greenhouse Gases from Automobiles) model.28
The fuel type distributions for gasoline, diesel, and E-85 capapable passenger cars for model
years 2022 and later were derived from Department of Energy car sales projections from
AEO2023's table "Light-Duty Vehicle Sales by Technology Type".13 Fuel type distributions for
MY2020-MY2021 were not available in AEO2023, so we used AEO2021 and AEO2022 for
those model years, respectively.
Note that MOVES may be run at the county or project scale with local information to accurately
capture EV market penetration and other fuel type variation by geographic region. As explained
in the MOVES Technical Guidance,2 this can be done through the AVFT importer in the
MOVES interface. MOVES cannot model CNG or fuel cell electric passenger cars.
41
-------
5.2.3. Light-Duty Trucks
Since passenger and light commercial trucks are defined as light-duty vehicles, they are
constrained to regulatory class 30 and 41. Light-duty trucks in the 2020 IHS data with a GVWR
class of 1, 2, or 2a were classified as regulatory class 30, and Class 2b trucks were classified as
regulatory class 41. IHS data provided the split between gasoline, diesel, electricity, and E-85
capable trucks. Note that all E-85 light-duty trucks are modeled as regulatory class 30.
For model years 2000 through 2013, the 2014 IHS data were used to calculate fuel type and
regulatory class distributions; for model years 2014 to 2019, the 2020 IHS data were used.
In MOVES4 national defaults, for model years prior to 2023, electric passenger trucks were
modeled with market shares from the 2022 EPA Trends report.29 For model years 2023 and later,
light-duty (regClassID 30) electric passenger truck and light commercial truck sales fractions are
based on light-duty electric vehicle cost and consumer preferences as modeled in EPA OMEGA
(Optimization Model for reducing Emissions of Greenhouse Gases from Automobiles) model.30
Projections for battery electric Class 2b trucks (regClassID 41) were based on OMEGA outputs
and an analysis of the national impact of CARB's Advanced Clean Trucks regulation.31
The fuel type distributions for light-duty gasoline, diesel, and E-85 capable trucks for model
years 2022 and later were derived from Department of Energy light truck and light commercial
truck sales projections from AEO2023's tables "Light-Duty Vehicle Sales by Technology Type"
and "Transportation Fleet Car and Truck Sales by Type and Technology".13 Fuel type
distributions for MY2020-MY2021 were not available in AEO2023, so we used AEO2021 and
AEO2022 for those model years, respectively.
Note that MOVES may be run at the county or project scale with local information to accurately
capture fuel distribution and EV market penetration variation by geographic region. As explained
in the MOVES Technical Guidance,2 this can be done through the AVFT importer in the
MOVES interface. MOVES cannot model CNG or fuel cell electric passenger trucks or light
commercial trucks.
5.2.4. Buses
MOVES4 can model diesel, gasoline, CNG and electric buses, but cannot model E-85 buses.
Since school buses have a distinguishing characteristic in their VIN and they are well represented
in the 2014 and 2020 IHS data, we were able to calculate their fuel type and regulatory class
distributions, with model years 2000-2013 based on the 2014 IHS data and model years 2014-
2019 based on the 2020 IHS data. All CNG school buses are assigned to regulatory class 47.
On the other hand, transit buses and "other buses" are not distinguished from each other in the
2014 and 2020 IHS datas. The National Transit Database is a potential alternate data source for
transit buses, but since it lacks weight class information, it could not be used to calculate
regulatory class distributions. Instead, considering that the vehicles in the transit and "other bus"
categories may overlap, we grouped these categories together when determining fuel type and
regulatory class distributions. The only difference between the transit and other bus distributions
42
-------
is in the categorization of Class 8 buses, since urban transit buses are regulated separately from
other heavy-duty vehicles under 40 CFR 86.091-2.32 For this reason, Class 8 CNG and diesel
transit buses were classified in regulatory class 48, whereas Class 8 gasoline transit buses and all
Class 8 other buses were classified as regulatory class 47. Additionally, MOVES can only model
CNG other buses in regulatory class 47. For model years 2013 and earlier, the 2014 IHS data
were used to calculate fuel type and regulatory class distributions; for model years 2014 to 2019,
the 2020 IHS data were used.
For all bus source types, for model years 2022 and later, we used Department of Energy heavy-
duty sales projections from AEO2023's "Freight Transportation Energy Use"13 table to derive
year-over-year growth for all heavy-duty gasoline, diesel, CNG, battery electric, and fuel cell
electric vehicles. Data for MY2020-MY2021 were not available in AEO2023, so we used
AEO2021 and AEO2022 for those model years, respectively. We applied the year-over-year
growth in vehicle sales to the model year 2019 bus counts in the 2020 IHS data in order to derive
future year fuel type and regulatory class distributions. Battery electric bus projections in
AEO2023 were adjusted for model years 2024 and later based on an analysis of the national
impact of CARB's Advanced Clean Trucks regulation.31
5.2.5. Single-Unit Trucks
Single-unit vehicles are distributed among the heavy-duty regulatory classes (regClassIDs 41,
42, 46 and 47) and between fuels based on IHS data. The IHS data categorized single-unit trucks
into refuse trucks (based on ownership), motor homes and "other single-unit trucks." Lacking a
way to differentiate these trucks into short-haul and long-haul, we used the fuel type and
regulatory class distributions for "other single-unit trucks" identically for both short-haul and
long-haul single-unit trucks. As with the other heavy-duty vehicles, MOVES can only model
CNG single-unit trucks in regulatory class 47. MOVES cannot model E-85 single-unit trucks.
The ability to model electric single unit trucks was added in MOVES4.
For single unit short-haul and long-haul trucks (sourceTypelDs 52 and 53), the 2014 IHS data
were used to calculate fuel type and regulatory class distributions for model years 2000-2013,
and the 2020 IHS data were used for model years 2014-2019.
For refuse trucks (sourceTypelD 51), we use the 2014 and 2020 IHS data for model years 2000-
2013 and 2014-2019, respectively. However, we found that electric refuse trucks were not well
represented in the IHS data, so we used electric refuse truck counts reported to EPA from the
2019 Annual Production Volume Reports into Engine and Vehicle Compliance Information
System33 instead of the electric refuse truck counts in the IHS data.
For motor homes (sourceTypelD 54), we used the 2014 IHS data for all model years 2013 and
earlier, and the 2020 IHS data for model years 2014-2019.
For all single unit trucks model years 2020 and later (including refuse trucks and motorhomes),
we used Department of Energy heavy-duty sales projections from AEO2023's "Freight
Transportation Energy Use"13 table to derive year-over-year growth for light heavy-duty and
medium heavy-duty gasoline, diesel, CNG, battery electric, and fuel cell electric vehicles. Data
43
-------
for MY2020-MY2021 were not available in AEO2023, so we used AEO2021 and AEO2022 for
those model years, respectively. We applied the year-over-year growth in vehicle sales to the
model year 2019 single unit truck counts (from the sources described above) to derive future
year fuel type and regulatory class distributions. Battery electric and fuel cell electric vehicle
projections in AEO2023 were adjusted for model years 2024 and later based on an analysis of
the national impact of CARB's Advanced Clean Trucks regulation.31
5.2.6. Combination Trucks
Combination trucks consist mostly of Class 8 trucks in the MOVES HHD regulatory class
(regClassID 47) but also include Class 7 trucks in the MHD regulatory class (regClassID 46) and
glider trucks (regClassID 49).
Almost all combination trucks are diesel-fueled, but MOVES4 also can model CNG and electric
combination trucks, as well as gasoline short-haul combination trucks. Combination trucks were
split between long-haul and short-haul by IHS using vehicle registration characteristics. As with
the other heavy-duty vehicles, MOVES does not model E-85 combination trucks.
For combination short-haul trucks (sourceTypelD 61), the 2014 IHS data were used to calculate
fuel type and regulatory class distributions for model years 2000-2013, and the 2020 IHS data
were used for model years 2014-2019. However, we found that battery electric combination
trucks were not well represented in the IHS data, so we used electric combination truck counts
reported to EPA from the 2019 Annual Production Volume Reports into Engine and Vehicle
Compliance Information System33 instead of the battery electric combination truck counts in the
IHS data.
For combination long-haul trucks (sourceTypelD 62), the 2014 IHS data were used to calculate
fuel type and regulatory class distributions for model years 2000-2013, and the 2020 IHS data
were used for model years 2014-2019.
For model years 2020 and later, we used Department of Energy heavy-duty sales projections
from AEO2023's "Freight Transportation Energy Use"13 table to derive year-over-year growth
for heavy heavy-duty gasoline, diesel, CNG, battery electric, and fuel cell electric vehicles. Data
for MY2020-MY2021 were not available in AEO2023, so we used AEO2021 and AEO2022 for
those model years, respectively. We applied the year-over-year growth in vehicle sales to the
model year 2019 combination truck counts in the 2020 IHS data in order to derive future year
fuel type and regulatory class distributions. Battery electric and fuel cell electric vehicle
44
-------
projections in AEO2023 were adjusted for model years 2024 and later based on an analysis of
the national impact of CARB's Advanced Clean Trucks regulation.31 e
5.2.6.1. Glider Truck Populations
"Glider trucks" in MOVES refers to vehicles with new chassis but with older engines that do not
meet MY 2007 or 2010 emissions standards (Section 2.3). Most glider trucks are Class 8
vehicles that use diesel heavy heavy-duty engines. For simplicity, in MOVES, we assume that all
glider vehicles are HHD but modeled as a separate regulatory class (regClassID 49) and are only
populated within the combination short- and long-haul truck source types (sourceTypelD 61 and
62, respectively).
We used sales data from both glider kit manufacturers and glider assembler manufacturers to
estimate glider truck populations in MOVES. The glider kits contain the vehicle chassis and cab,
but lack the engine and transmission. Glider assembler manufacturers (referred heareafter as
"glider assemblers") assemble the glider vehicle by installing the engine and transmission into
the glider kit produced by the glider kit manufacturer. Most glider assemblers are small
businesses that sell less than 10 glider vehicles per year. However, most of the glider vehicles
made from 2016 to 2020 have been produced by a handful of large glider assemblers.
We estimated the glider population based on annual glider production volume (sales) data for
model years 2010 to 2016 shared as claimed confidential business information (CBI) from the
two major glider kit manufacturers.34 For use in MOVES, we assumed annual sales of 500 for
glider vehicles for years prior to 2010 and rounded the reported production volumes in the years
2010 to 2016 to the nearest thousand, as shown in Table 5-1. The rounded values reflect the
uncertainty regarding the number of gliders in the fleet, including the contribution of small
volume glider manufacturers and the number used in single-unit vehicles/
Table 5-1: Annual Glider Vehicle Sales Estimates Applied in MOVES Based on Claimed CBI Data
Shared by Manufacturers
MY
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
Glider
Population
500
500
1000
3000
4000
5000
8000
12000
7000
7500
3500
1500
0
0
e We made the simplifying assumption that all electric short-haul trucks are battery EVs and all electric long-haul
trucks are fuel cell EVs. However, there were no fuel cell EV combination trucks in production in our base year of
2020, so we could not directly apply year-over-year growth in AEO to project future distributions. Instead, we
calculated the ratio of AEO2023's HHD fuel cell EV sales to HHD battery EV sales and applied this ratio as a
scaling factor to the battery EV short-haul truck projections to estimate future year fuel cell EV long-haul truck
counts. We then derived future year fuel type and regulatory class distributions from these counts.
f In 2017, glider manufacturers are limited to producing their maximum production between MYs 2010 and 2014.
See 81 FR 73478 for more information.
45
-------
For estimating the glider sales for 2017 and 2018, we did not have data from the two glider kit
manufacturers, but we have data from the glider assemblers. As part of EPA's Greenhouse Gas
Emissions and Fuel Efficiency Standards for Medium- and Heavy-Duty Engines and Vehicles -
Phase 2 rulemaking (Phase 2)91, the agency adopted new rules for glider kits, glider vehicles and
glider engines. Starting in model year 2018, a glider assembler could continue to sell glider
vehicles, without limit, if the glider engine was from a 2010 or later model year. If a glider
assembler wishes to sell glider vehicles with earlier model year engines, they are limited to the
lesser of 300 per year or the number of glider vehicles they sold in calendar year 2014. The
regulation requires glider assemblers to report their sales data to EPA, including their 2014 sales
to identify their individual sales allowances. The number of manufacturers who reported glider
sales data to EPA for 2014, 2017, 2018, and 2019 are shown in Table 5-2, and the reported sales
are displayed in Figure 5-2. Prior to the 2018 model year, more than 260 glider assemblers
reported their sales data, including five manufacturers which produced more than 300 gliders and
whose sales were capped starting in 2018. The reported sales from the glider assemblers in 2014
was close to 9,000 vehicles, which compares well to the 8,000 glider kit sales reported for 2014
shown in Table 5-1.
Table 5-2. Number of Glider Assemblers that reported to EPA, grouped by glider production in
2014
Glider Production
in 2014
Manufacturers
reporting in
2014
Manufacturers
reporting in
2017
Manufacturers
reporting in
2018
Manufacturers
reporting in
2019
<=10
208
70
73
21
10-50
53
23
23
6
51-300
8
5
5
0
300 +
5
2
2
0
Total
274
100
103
27
46
-------
12017 Manufacturer Reported Sales
12018 Manufacturer Reported Sales
9 Manufacturer Reported Sales
<=10
10-SO
51-300
300 +
Figure 5-2. Reported Glider Sales by Glider Assemblers for Calendar Year 2014, 2017,2018, and
2019.
The total number of glider sales by glider assemblers is significantly reduced in 2017, 2018, and
2019 as shown in Figure 5-2. One reason for the decrease is that we only received reported sales
from a subset of the assemblers who reported their sales in 2014. For the assemblers who
reported sales in 2017, we calculated the ratio of sales to the number of sales these assemblers
reported in 2014. For the assemblers who reported sales in 2018 and 2019, we calculated the
ratio of the sales to maximum allowable (the smaller of their sales in 2014 or 300 glider
vehicles). As shown in Table 5-3, the larger glider assemblers tended to produce more gliders in
comparison to the 2014 sales and their maximum allowable sales.
Table 5-3. Ratio of 2017, 2018, and 2019 Glider Sales to 2014 Sales or the Maximum Allowable
A
B
C
Glider Production
in 2014
Ratio of 2017
sales to 2014
sales for
assemblers that
reported in 2017
Ratio of 2018
sales to
maximum
allowable for
assemblers that
reported in 2018
Ratio of 2019 to
2014 sales to
maximum
allowable for
assemblers that
reported in 2019
<=10
0.30
0.46
0.62
10-50
0.57
0.63
0.70
51-300
0.68
0.71
300 +
1.02
1.00
To estimate the number of glider sales in 2017, we multiplied the ratios in Column (A) of Table
5-3 by total number of gliders sales by glider assembler size reported in 2014. This is under the
assumption that the sale growth/reduction rate from 2014 to 2017 is the same between reported
glider assemblers and the overall glider assemblers. And similarly, to estimate the number of
glider sales in 2018, we multiplied the ratios in Column (B) by the maximum allowable of all
47
-------
glider assembler manufacturers who reported in 2014. When rounded to the nearest 500, this
yielded total glider sales estimates of 7,500 in 2017 and 3,500 in 2018 as shown in Table 5-1.
To estimate 2019 sales, we had limited data from the glider assembler manufacturers (only 27
assembler manufacturers reported 2019 sales at the time of the analysis, all of which sold less
than 50 gliders per manufacturer). The two major glider kit manufacturers informed EPA that
they had stopped production of glider kits in 2018. As such, we assumed that the 2019 glider
vehicles would be 1500 vehicle sales, which is calculated from 50 gliders per manufacturer and
27 reported assembler manufacturers, rounded to the nearest 500. We assumed zero sales for
2020 and later model years. Assuming an insignificant number of gliders in future years is
appropriate due to decreasing availability of pre-2010 engines and requirements for 2021 and
later model year glider vehicles to meet the Medium and Heavy-duty Greenhouse Gas Phase 2
emissions and fuel economy standards.83
We calculated the fraction of gliders (stmyFraction) by dividing the estimated glider production
by the number of age-0 combination trucks for each model year using Equation 5-3. We applied
this fraction to both short-haul combination (sourceTypelD 61) and long-haul combination
trucks (sourceTypelD 62). Note all gliders are regclass 49 diesel trucks.
/ (,Stmy)regciass 49 model year i,
_ Glidersi Equation 5-3
Combination Tvuckssoucetype 61+62,i
5.2.7. Older Moid ¥ ears
For pre-2000 model years, most SampleVehiclePopulation values are based on combining 1999
and 2011 IHS vehicle registration data with data from the 1997 and 2002 Vehicle Inventory and
Use Survey (VIUS).g The documentation of the pre-2000 model years may be found in
Appendix A. Note that there are two exceptions to our reliance on the VlUS-based analysis for
model years before 2000:
• For passenger trucks and light commercial trucks, we used the 2014 IHS data for model
years 1981-2000 because the MOVES definition of these vehicle types is no longer
consistent with the VIUS definition. Unfortunately, the data are too scarce in 2014 IHS
and later for pre-1981 model years, so we continue to rely on the previous analysis as
described in Appendix A analysis for those model years.
• We also relied exclusively on the 2014 and 2020 IHS data for all model years of transit
buses, other buses, and motor homes.
6. Vehicle Age-Related Characteristics
gAt this writing, VIUS 2002 is the latest VIUS available. DOT has begun collecting data for the 2021 VIUS and we
hope to incorporate data from this survey in future versions of MOVES.
48
-------
Age is an important factor in calculating vehicle emission inventories. MOVES employs a
number of different age dependent factors, including deterioration of engine and emission after-
treatment technology due to tampering and mal-maintenance, vehicle scrappage and fleet
turnover and mileage accumulation over the lifetime of the vehicle. Deterioration effects are
detailed in the MOVES reports on the development of light-duty and heavy-duty emission
rates.10'11 This section describes vehicle age distributions and relative mileage accumulation rates
by source type.
6.1. Age Distributions
Vehicle age is defined in MOVES as the difference between a vehicle's model year and the year
of analysis. Age distributions in MOVES vary by source type and range from 0 to 30+ years, so
that all vehicles 30 years and older are modeled together. Therefore, an age distribution is
comprised of 31 fractions, where each fraction represents the number of vehicles present at a
certain age divided by the vehicle population for all ages. Since sales and scrappage rates are not
constant, these distributions vary by calendar year. Ideally, all historic age distributions could be
derived from registration data sources. However, acquiring such data is prohibitively costly, so
MOVES4 only contains registration-based age distributions for two analysis years: 1990 and
2020. The age distributions for all other analysis years in MOVES4 were projected forwards or
backwards from the 2020 base age distribution. All default age distributions are available in the
SourceTypeAgeDistribution table in MOVES database.
The rest of this section details the derivation of the base 2020 age distribution and the forwards
and backwards projections for all years other than 1990. The 1990 age distributions are discussed
in Appendix B.
6.1.1. Base Age Distributions
The 2020 base age distributions for cars and trucks were primarily derived from the 2020 IHS
data and the 2017 National Transit Database (NTD). The 2020 IHS data had vehicle counts by
age for motorcycles (11), passenger cars (21), passenger trucks (31), light commercial trucks
(32), school buses (43), refuse trucks (51), motor homes (54), combination short-haul trucks (61)
and combination long-haul trucks (62), as well as other single-unit trucks and non-school buses.
The age distribution for the other single-unit trucks was applied to both short-haul (52) and long-
haul (53) single-unit trucks and the age distribution for non-school buses was applied to the other
bus source type (41). Transit bus (42) age distributions were calculated from the NTD active
fleet vehicles using the definition of a transit bus in Section 5.1.4.
Since the age distributions in MOVES represent the full calendar year, additional calculations
were necessary for determining the fraction of age 0 vehicles in the fleet because the 2020 IHS
data did not capture all vehicles sold in 2020. Vehicle sales by source type in 2020 were
calculated from a variety of sources as described in Appendix C.2. The source type sales were
divided by the 2020 source type populations (see Section 4.1) to determine the age 0 fractions.
The other fractions for ages 1-30 were renormalized so that each source type's age distribution
summed to 1. This was done instead of directly using the sales numbers to calculate the age
49
-------
distributions (i.e., using the sales values as age 0 counts) because the IHS data is only used in
MOVES to determine vehicle distributions, not for vehicle populations.
Figure 6-1 shows the fraction of vehicles by age and source type for calendar year 2020, which
formed the basis for forecasting and back-casting age distributions as described in the following
sections. Please note that since all vehicles age 30 and older are grouped together, there is an
uptick in this age bin for most source types.
50
-------
16%-r
12%-
c
o
• f—4
o
s
m
<
16%-
12% ¦
8% ¦
4% ¦
0% ¦
10
20
30
Source Type
Motorcycle
Passenger Car
Passenger Truck
•*"= Light Commercial Truck
Source Type
Intercity Bus
Transit Bus
— School Bus
Source Type
Refuse Truck
Single Unit (Short- & Long-haul)
Motor Home
Source Type
Combination Short-haul Truck
Combination Long-haul Truck
Figure 6-1 2020 age distributions by source type in MOVES4
51
-------
6.1.2. ilistoric Age Distributions
The 2000-2019 age distributions were backcast from the 2020 base age distribution using
historic population and sales estimates. Age distributions are calculated from population counts,
if the populations are known by age:
f Equation
^ Py 6-1
In Equation 6-1, fay is the age fraction, pa is the population of vehicles at age a and Py is the
total population in calendar year y. In this section, arrow notation will be used if the operations
are to be performed for all ages. For example, fy is used to represent all age fractions in calendar
year y. Another example is Py; it represents an array of pa values at each permissible age in
calendar year y. In contrast, Py represents the total population in year y.
Intuitively, backcasting an age distribution one year involves removing the new vehicles sold in
the base year and adding the vehicles scrapped in the previous year, as shown in Equation 6-2:
p * — — N ' 4- 7? * Equation
y-i ry Ivy """ Ky-1 g_2
where Py-\ is the population (known at each age) of the previous year, Py is the population in the
base year, Ny is new vehicles sold in the base year and Ry-i is the population of vehicles
removed in the previous year. Please note that the sales term only includes new vehicles at age 0.
This can be represented algorithmically as follows:
1. Calculate the base population distribution (Py) by multiplying the base age distribution
(fy) and base population (Py).
2. Remove the age 0 vehicles (Ny).
3. Decrease the population age index by one (for example, 3-year-old vehicles are
reclassified as 2-year-old vehicles).
4. Add the vehicles that were removed in the previous year (Ry_i).
5. Convert the resulting population distribution into an age distribution using Equation 6-1.
6. Replace the new age 29 and 30+ fractions with the base year age 29 and 30+ fractions
and renormalize the new age distribution to sum to 1 while retaining the original age 29
and 30+ fractions.
7. This results in the previous year age distribution (fy_ i). If this algorithm is to be
repeated, /y_x becomes fy for the next iteration.
The fraction of age 30+ vehicles is kept constant because most source types have a sizeable
fraction in this age bin in the base age distributions. If left unconstrained, the algorithm can
either grow this age bin unreasonably large or shrink it unreasonably small, depending on the
source type. This indicates that the base survival rates for the oldest age bins may be
inappropriate. However, lacking better data, we decided to keep the age 30+ bin at a constant
fraction for all historic age distributions.
52
-------
Age 29 is additionally retained because when the number of scrapped vehicles is calculated, a
large proportion of them come from the age 30 bin. In reality, these scrapped vehicles have a
distribution well beyond age 30, but they are all grouped together in this analysis. When the
scrapped vehicles are added to the index-shifted population distribution, this results in a large
addition to the age 29 bin. To prevent this from happening, the base year age 29 fractions are
also retained in each backcasted year.
Please see Appendix C, Detailed Derivation of Age Distributions, for more information on how
this algorithm was applied to derive the historic national default age distributions in MOVES.
6.1.3. Projected Age Distributions
The method used to forecast the 2021-2060 age distributions from the 2020 distribution is similar
to the backcasting method described above. To forecast an age distribution one year, Equation
6-2 of the previous section can be rewritten as Equation 6-3:
Py+1 — Py Ry + Ny+1
Essentially, this is done by taking the base year's population distribution, removing the vehicles
scrapped in the base year and adding the new vehicles sold in the next year. This can be
represented algorithmically as follows:
1. Calculate the base population distribution (Py) by multiplying the base age distribution
(fy) and base population (Py ).
2. Remove the vehicles that did not survive (Ry) at each age level.
3. Increase the population age index by one (for example, 3-year-old vehicles are
reclassified as 4-year-old vehicles).
4. Add new vehicle sales (Ny+1) as the age 0 cohort.
5. Convert the resulting population distribution into an age distribution using Equation 6-1.
6. Replace the new age 30+ fraction with the base year age 30+ fraction and renormalize the
new age distribution to sum to 1 while retaining the original age 0 and age 30+ fractions.
7. This results in the next year age distribution (fy+1). If this algorithm is to be repeated,
fy+1 becomes fy for the next iteration.
The fraction of age 30+ vehicles is kept constant in the projection algorithm for the same reasons
given for the backcasting algorithm. However, there is no issue with an artificially growing
population of age 29 vehicles when projecting forward. Therefore, the age 29 bin is calculated as
the others are instead of being retained from the base age distribution.
Please see Appendix C, Detailed Derivation of Age Distributions, for more information on how
this algorithm was applied to derive the historic national default age distributions in MOVES.
In addition to producing the default projected age distributions, this algorithm was implemented
in the spreadsheet-based Age Distribution Projection Tool.35 This tool can be used to project
future local age distributions from user-supplied baseline distributions, provided that the baseline
Equation
6-3
53
-------
year is 2011 or later. This requirement ensures that the 2008-2009 recession is fully accounted
for in the baseline. The sales rates and scrappage assumptions are the same in the tool as they are
in the national default. This is because local projections of sales and scrappage are generally
unavailable and the national trends are the best available data. Thus, projections made with the
tool tend to converge with the national age distributions for far future years.
6.2.Relative Mileage Accumulation Rate
For emission calculations, MOVES needs to estimate the miles travelled by each age and source
type. MOVES uses a relative mileage accumulation rate (RMAR) in combination with source
type populations (see Section 4) and age distributions described in Section 6.1 to distribute the
total annual miles driven by each HPMS vehicle type (see Section 3) to each source type and age
group. Using this approach, the vehicle population and the total annual vehicle miles traveled
(VMT) can vary from calendar year to calendar year, but the proportional travel by an individual
vehicle of each age will not vary.
The RMAR is determined from the mileage accumulation rate (MAR) within each HPMS
vehicle classification such that the annual mileage accumulation for a single vehicle of each age
of a source type is relative to the mileage accumulation of all of the source types and ages within
the HPMS vehicle classification. For example, passenger cars, passenger trucks and light
commercial trucks are all within the same HPMS vehicle classification (Light-duty vehicles,
HPMSVTypelD 25). As described below in Section 6.2.1.1, new (age 0) passenger trucks and
light commercial trucks are defined to have a RMAR of one (1.0)h and new passenger cars have
a RMAR of 0.95. This means that when MOVES allocates the VMT assigned to the light-duty
vehicle HPMS class to passenger cars, passenger trucks and light commercial trucks, a passenger
car of age 0 will be assigned only 95 percent of the annual VMT assigned to a passenger truck or
light commercial truck of age 0. The RMAR values used in MOVES4 are shown in Figure 6-2.
h Within each HPMS vehicle class, an RMAR value of one is assigned to the source type and age with the highest
annual VMT accumulation. Because we use the same mileage accumulation data for passenger trucks and light
commercial trucks, they both have a value of one.
54
-------
Motorcycle
Light Duty
Light Duty
Light Duty
1.00
0.75
0.50
0.25
1.00-¦
0.75-
0.50-
0.25-
0.00 -¦
Single Unit Trucks
Combination Trucks
Figure 6-2. Relative Mileage Accumlation Rates (RMAR) by HPMS Class and SourceTypelD
Buses
Single Unit Trucks
Buses
Single Unit Trucks
Combination Trucks
1.00
0.75
0.50
0.25
0.00
1.00
0.75
0.50
0.25
The derivation of the RMAR values for each sourcetype and HPMS class are discussed in the
following subsections.
6.2.1. Motorcycles
The RMAR values were calculated from mileage accumulations for motorcycles (sourceTypelD
11) based on the model years and odometer readings listed in motorcycle advertisements. A
stratified sample of about 1,500 ads were examined. A modified Wei bull curve was fit to the data
to develop the relative mileage accumulation rates used in MOVES.36
55
-------
6.2.1.1. Passenger Cars, Passenger Trucks and Light-Commercial
Trucks
In earlier versions of MOVES, the RMAR values for passenger cars, passenger trucks and light
commercial trucks (sourceTypelD 21, 31 & 32) were taken from a NHTSA report on
survivability and mileage schedules.112 In that NHTSA analysis, annual mileage by age was
determined for cars and for trucks using data from the 2001 National Household Travel Survey.
For MOVES4, we updated the RMAR values for passenger cars, passenger trucks and light
commercial trucks (sourcetypelD 21,31, and 32). We leveraged the vehicle miles traveled
analysis done by NHTSA for their CAFE standards for MY 2024-2026 presented in their
Technical Support Documentation.37 NHTSA used a random national sample of one million
vehicles based on data from IHS-Polk for 2016, which provides a longitudinal dataset with
information on individual vehicles over time, providing data on vehicle mileage as they age. The
smoothed annual mileage schedules used by NHTSA are presented in Table 6-1:
56
-------
Table 6-1 VMT Annual Mileage Schedules Derived by NHTSA and Weights Used to Generate LDT
Mileage for MOVES
Vehicle
Age
Cars
Vans/SUVs
Pickups
Weight 31
Weight 32
0
15,922
16,234
18,964
0.831
0.169
1
15,379
15,805
17,986
0.884
0.116
2
14,864
15,383
17,076
0.896
0.104
3
14,378
14,966
16,231
0.900
0.100
4
13,917
14,557
15,449
0.905
0.095
5
13,481
14,153
14,726
0.917
0.083
6
13,068
13,756
14,060
0.924
0.076
7
12,677
13,366
13,448
0.918
0.082
8
12,305
12,982
12,886
0.915
0.085
9
11,952
12,605
12,372
0.930
0.070
10
11,615
12,234
11,903
0.928
0.072
11
11,294
11,870
11,476
0.940
0.060
12
10,986
11,512
11,088
0.945
0.055
13
10,690
11,161
10,737
0.945
0.055
14
10,405
10,816
10,418
0.947
0.053
15
10,129
10,477
10,131
0.944
0.056
16
9,860
10,146
9,871
0.942
0.058
17
9,597
9,820
9,635
0.943
0.057
18
9,338
9,501
9,421
0.946
0.054
19
9,081
9,189
9,226
0.944
0.056
20
8,826
8,883
9,047
0.946
0.054
21
8,570
8,583
8,882
0.946
0.054
22
8,313
8,290
8,726
0.950
0.050
23
8,051
8,004
8,577
0.957
0.043
24
7,785
7,724
8,433
0.962
0.038
25
7,511
7,450
8,290
0.964
0.036
26
7,229
7,183
8,146
0.965
0.035
27
6,938
6,923
7,998
0.968
0.032
28
6,635
6,669
7,842
0.971
0.029
29
6,319
6,421
7,676
0.984
0.016
30
5,988
6,180
7,497
0.831
0.169
We derived RMAR values using the same methodology used for previous MOVES versions.38
Passenger cars, passenger trucks and light commercial trucks were grouped together as light-duty
vehicles (HPMSVTypelD 25). The NHTSA data for light-duty trucks were used for both the
passenger truck and commercial truck source types. Since the trucks had a higher MAR than
passenger cars, each source type's mileage by age was divided by truck mileage at age 1 to
determine a relative MAR. Analysis of the data determined that new passenger cars (age 0)
accumulate 95 percent of the annual miles accumulated by new light-duty trucks.
We chose to continue using the same RMAR for all light trucks rather than deriving individual
mileage accumulation rates for sourceTypelD 31 and 32 because the NHTSA results did not
distinguish passenger and commercial trucks. However, rather than computing a simple
57
-------
arithmetic average of the annual mileage for the SUV/Van and Pickup categories, we weighted
these values using factors from a separate IHS-Polk sample obtained by EPA for 2017 in five
states: Colorado, Georgia, New Jersey, California, and Illinois. We originally purchased this
sample to update the RMAR for light-duty vehicles but given the availability of a national
sample with similar longitudinal characteristics, we opted to use the NHTSA mileage schedules
as our main data source. EPA's five-state sample did, however, classify vehicles to MOVES
sourcetypes based on body shape, GVWR and personal vs commercial registration type. We
calculated weights for SUV/Vans and Pickups based on the five-state samples for sourcetype 31
and 32, respectively, and applied them to the NHTSA schedules (see Table 6-1), assuming that
the SUV/Van category loosely mapped to passenger vehicles while the Pickup category mapped
to light commercial trucks and assuming no significant difference between the populations in
2016 and 2017. The resulting weighted mileages for LDV and LDT are shown in Figure 6-3.
20000
18000
c 18000
0J
>
5 14000
tfi
= 12000
2
1 10000
c
< 8000
0)
Ui
S 6000
>
< 4000
2000
0
Figure 6-3 Annual Mileage for LDV and LDT
The updated RMARs are presented in Figure 6-4 and shown in comparison to the relative MAR
for light-duty vehicles used in MOVES3. For cars, the new curves suggest an increase in miles
driven across the age range in comparison to MOVES3, but particularly for vehicles between 15-
25 years old. For LDT, there is a small reduction in miles driven for younger trucks up to 10
years old and a reduction for the oldest trucks (27 years old and later), while an increase in miles
driven is observed between the age range 12-27 years. Furthermore, the relationship between
LDV and LDT in MOVES4 shows how both groups have increasingly similar driving patterns
compared to earlier data.
Emissions sensitivity analysis at the national level indicates that the updated RMARs for light-
duty vehicles results in an increase in the light-duty inventory of one to three percent depending
on the pollutant and calendar year, with smaller increases further in the future. Because vehicle
populations and the total miles for HPMS vehicle class 25 were held constant in this testing, the
emissions change is caused by the shift in VMT to older ages.
Mileage_LDV
Mileage_LDT
Vehicle Age
58
-------
1.00-
0.75-
cc
<
2
| 0.50-
ra
a
0.25-
0.00-
Figure 6-4 LDV and LDT relative MAR for MOVES4 (solid lines) and MOVES3 (dashed lines)
6.2.2. Buses
In MOVES4, the RMARs for Buses are unchanged from MOVES3.
The transit bus (sourceTypelD 42) annual mileage accumulation rate are taken from the
MOBILE6 values for diesel transit buses (HDDBT). This mileage data was obtained from the
1994 Federal Transportation Administration survey of transit agencies as shown in Table 6-3 and
a smoothing function applied to remove the variability in the data.39 The MOBILE6 results were
extended to calculate values for ages 26 through 30.
For MOVES3, we redefined source type 41 as "other bus" (sourceTypelD 41) and assigned the
same RMAR as the transit bus (sourceTypelD 42).
The school bus (sourceTypelD 43) annual mileage accumulation rate (9,939 miles per year) is
derived from the 1997 School Bus Fleet Fact Book.18 For MOVES3, we updated the RMAR for
school buses such that it has the same shape as the transit bus RMAR, but adjusted down such
that year 0 is based on the 9,939 miles per year from the School Bus Fleet Fact Book. The same
relative shape is evident in of the Bus RMAR in Figure 6-2.
Oo<
ss
s
—
LDV
LDT
VS^
-
t 1 1 1 1 1 r
0 5 10 15 20 25 30
Vehicle Age
59
-------
Table 6-2 Annual mileage accumulation of transit buses from 1994 Federal Transit Administration
data
Age
Miles
Age
Miles
Age
Miles
1
*
11
32,540
21
19,588
2
*
12
32,605
22
22,939
3
46,791
13
27,722
23
26,413
4
41,262
14
28,429
24
23,366
5
42,206
15
32,140
25
11,259
6
39,160
16
28,100
26
23,228
7
38,266
17
24,626
27
21,515
8
36,358
18
23,428
28
25,939
9
34,935
19
22,575
29
20,117
10
33,021
20
23,220
30
17,515
* Insufficient data
6.2.3. Other Heavy-Duty Vehicles
In MOVES4, the RMARs for other HD vehicles are unchanged from MOVES3.
The RMAR values for source types 51 (refuse trucks), 52 (short-haul single-unit trucks), 53
(long-haul single-unit trucks), 61 (short-haul combination trucks) and 62 (long-haul combination
trucks) use the data from the 2002 Vehicle Inventory and Use Survey (VIUS).40 The total
reported annual miles traveled by truck in each source type by age, as shown in Table 6-3, was
divided by the vehicle population by age to determine the average annual miles traveled per
truck by source type.
60
-------
Table 6-3 VIUS2002 annual mileage by vehicle age
Age
Model
Year
Single-Unit Trucks
Combination Trucks
Refuse
(51)
Short-Haul
(52)
Long-Haul
(53)
Short-Haul
(61)
Long-Haul
(62)
0
2002
26,703
21,926
40,538
60,654
109,418
1
2001
32,391
22,755
28,168
59,790
128,287
2
2000
31,210
24,446
30,139
61,651
117,945
3
1999
31,444
23,874
49,428
62,865
110,713
4
1998
31,815
21,074
33,266
55,113
99,925
5
1997
28,450
21,444
23,784
54,263
94,326
6
1996
25,462
16,901
21,238
40,678
85,225
7
1995
30,182
15,453
27,562
38,797
85,406
8
1994
20,722
13,930
21,052
33,485
71,834
9
1993
25,199
13,303
11,273
30,072
71,160
10
1992
23,366
11,749
18,599
27,496
67,760
11
1991
18,818
13,675
15,140
24,175
80,207
12
1990
12,533
11,332
13,311
22,126
48,562
13
1989
15,891
9,795
9,796
21,225
64,473
14
1988
19,618
9,309
12,067
21,163
48,242
15
1987
12,480
9,379
16,606
20,772
58,951
16
1986
12,577
4,830
8,941
11,814
35,897
0-3
1999-2002
Average
30,437
23,250
37,069
61,240
116,591
For each source type, in the first few years, the data showed only small differences in the annual
miles per vehicle and no trend. After that, the average annual miles per vehicle declined in a
fairly linear manner, at least until the vehicles reach age 16 (the limit of the data). MOVES,
however, requires mileage accumulation rates for all ages to age 30. The relative mileage
accumulation rate at age 30 were derived from the 1992 Truck Inventory and Use Survey (TIUS)
as documented in the ARCADIS report.41
Mileage accumulation rates for these vehicles were determined for each age from 0 to 30 using
the following method:
1) Ages 0 through 3 use the same average annual mileage accumulation rate for age 0-3
vehicles of that source type.
2) Ages 4 through 16 use mileage accumulation rates calculated using a linear regression
of the VIUS data. The average mileage accumulation rate of ages 0 to 3 were used for
age 3 in the regression. The resulting coefficients are summarized in Table 6-4,
3) Age 30 uses the 1992 TIUS relative mileage accumulation rate for age 30. These
relative mileage accumulation rates were allocated to the MOVES source types from
the MOBILE6 mileage accumulation rates, they were converted to mileage based on
the mileage data used in MOVES, then converted back to an RMAR consistent with
the other ages.
4) Ages 17 through 29 use values from interpolation between the values in age 16 and
age 30.
61
-------
Table 6-4 Regression coefficients for heavy-duty truck average annual mileage accumulation rates
(ages 4-16)
Measurement
Refuse
Truck (51)
Single-Unit
Short-Haul (52)
Single-Unit
Long-Haul (53)
Combination
Short-Haul (61)
Combination
Long-Haul (62)
Average 0-3a
30,437
23,250
37,069
61,240
116,591
Intercept13
30,437
23,250
37,069
61,240
116,591
Slopeb
-1,361
-1,368
-2,476
-4,092
-6,418
Age 30 RMAR
0.027
0.0115
0.086
0.015
0.052
a Average sample annual miles traveled for ages 0 through 3.
b Intercept at age 3; slope from ages 4 through 16.
The RMAR values for heavy-duty were updated for MOVES3. The resulting relative mileage
accumulation rates are shown in Table 6-5 below and Figure 6-2 above. As in previous versions
of MOVES, the first four ages (age 0 to 3) are identical and then decline linearly to age 16 and
then linearly to age 30 with a different slope.
6.2.4. Motor Homes
For MOVES3, we updated the RMAR values and added a decreasing trend with age. Data from
the 2017 National Household Travel Survey42 was used for the motor home RMAR calculation.
The calculation methodology is different from the other heavy-duty trucks. The same average
annual mileage accumulation rate was used for age 0-3 motor homes. Age 4 through 30 used
mileage accumulation rates that were calculated using a linear regression of the National
Household Travel Survey data.
Based on this data, the average annual vehicle miles of travel per vehicle for age 0 to 3 is 6003.
In the regression analysis, this value was used as intercept at age 3. The slope from age 4
through 30 was calculated at -83 miles/year. The motor home mileage accumulation values were
then converted to RMARs by dividing by the average mileage for age 0-3 long-haul single-unit
trucks (37,069).
The resulting relative mileage accumulation rates of motor homes are shown in Table 6-5 below
and Figure 6-2 above. Note that first four ages are identical and then decline linearly to age 30
since the 2017 National Household Travel Survey has data available from age 0 to 30.
62
-------
Table 6-5 Relative mileage accumulation rates for heavy-duty trucks in MOVES
agelD
Refuse (51)
Short-Haul
Single-Unit (52)
Long-Haul
Single-Unit
(53)
Motor Home
(54)
Short-Haul
Combination
(61)
Long-Haul
Combination
(62)
0
0.8211
0.6272
1.0000
0.1620
0.5253
1.0000
1
0.8211
0.6272
1.0000
0.1620
0.5253
1.0000
2
0.8211
0.6272
1.0000
0.1620
0.5253
1.0000
3
0.8211
0.6272
1.0000
0.1620
0.5253
1.0000
4
0.7844
0.5903
0.9332
0.1597
0.4902
0.9473
5
0.7477
0.5534
0.8664
0.1575
0.4551
0.8945
6
0.7110
0.5165
0.7996
0.1552
0.4200
0.8418
7
0.6743
0.4796
0.7328
0.1529
0.3849
0.7891
8
0.6376
0.4427
0.6660
0.1507
0.3498
0.7363
9
0.6009
0.4058
0.5992
0.1484
0.3147
0.6836
10
0.5642
0.3689
0.5323
0.1462
0.2796
0.6309
11
0.5275
0.3320
0.4655
0.1439
0.2445
0.5781
12
0.4908
0.2950
0.3987
0.1417
0.2094
0.5254
13
0.4541
0.2581
0.3319
0.1394
0.1743
0.4727
14
0.4174
0.2212
0.2651
0.1372
0.1392
0.4199
15
0.3807
0.1843
0.1983
0.1349
0.1041
0.3672
16
0.3440
0.1474
0.1315
0.1327
0.0690
0.3145
17
0.3214
0.1380
0.1282
0.1304
0.0652
0.2957
18
0.2987
0.1285
0.1249
0.1282
0.0613
0.2769
19
0.2761
0.1191
0.1216
0.1259
0.0575
0.2581
20
0.2535
0.1097
0.1184
0.1236
0.0536
0.2394
21
0.2309
0.1002
0.1151
0.1214
0.0498
0.2206
22
0.2083
0.0908
0.1118
0.1191
0.0460
0.2018
23
0.1857
0.0814
0.1085
0.1169
0.0421
0.1830
24
0.1631
0.0719
0.1052
0.1146
0.0383
0.1642
25
0.1405
0.0625
0.1019
0.1124
0.0344
0.1454
26
0.1179
0.0530
0.0986
0.1101
0.0306
0.1267
27
0.0953
0.0436
0.0954
0.1079
0.0267
0.1079
28
0.0727
0.0342
0.0921
0.1056
0.0229
0.0891
29
0.0500
0.0247
0.0888
0.1034
0.0191
0.0703
30
0.0274
0.0153
0.0855
0.1011
0.0152
0.0515
63
-------
7. VMT Distribution of Source Type by Road Type
For each source type, the RoadTypeVMTFraction field in the RoadTypeDistribution table stores
the fraction of total VMT for each source type that is traveled on each of the MOVES five road
types nationally. Users may supply the VMT distribution by vehicle class for each road type for
individual counties when using County Scale. For National Scale, the default distribution is
allocated to individual counties using the SHOAllocFactor found in the ZoneRoadType table.
The national default distribution of VMT to source type for each road type in MOVES4 were
derived to reflect the VMT data included in the 2017 National Emission Inventory (NEI) Version
2.43 This data is provided by states every three years as part of the NEI project and is
supplemented by EPA estimates based on data provided by FHWA Highway Statistics44 when
state supplied estimates are not available. The FHWA road types mapped to the MOVES road
type ID values (the eighth and ninth digits of the 10-digit onroad SCC) are shown below in Table
7-1.
Table 7-1 Mapping of FHWA road types to IV
OVES road types
FHWA Road Type
MOVES
Road Type ID
MOVES Road Type
Rural Interstate
2
Rural Restricted Access
Rural Other Freeways and Expressways
2
Rural Restricted Access
Rural Other Principal Arterial
3
Rural Unrestricted Access
Rural Minor Arterial
3
Rural Unrestricted Access
Rural Major Collector
3
Rural Unrestricted Access
Rural Minor Collector
3
Rural Unrestricted Access
Rural Local
3
Rural Unrestricted Access
Urban Interstate
4
Urban Restricted Access
Urban Other Freeways & Expressways
4
Urban Restricted Access
Urban Other Principal Arterial
5
Urban Unrestricted Access
Urban Minor Arterial
5
Urban Unrestricted Access
Urban Major Collector
5
Urban Unrestricted Access
Urban Minor Collector
5
Urban Unrestricted Access
Urban Local
5
Urban Unrestricted Access
The national distribution of road type VMT by source type is calculated from the NEI VMT
estimates and is summarized in Table 7-2. The off-network road type (roadTypelD 1) is
allocated no VMT.
Note that because it is difficult to distinguish single unit short-haul and long-haul trucks in
roadway VMT measurements, the distributions for single-unit short-haul trucks are virtually the
same as those for single-unit long-haul trucks.
64
-------
Table 7-2 MOV
S4 road type distribution by source type
Road Type3
Source
Type
Description
Rural
Restricted
Rural
Unrestricted
Urban
Restricted
Urban
Unrestricted
2
3
4
5
All
11
Motorcycle
0.0825631
0.267313
0.198403
0.451721
1.000
21
Passenger Car
0.08177
0.204595
0.259544
0.454091
1.000
31
Passenger Truck
0.0958223
0.265213
0.222866
0.416098
1.000
32
Light Commercial Truck
0.0839972
0.217512
0.262385
0.436105
1.000
41
Other Bus
0.131819
0.246451
0.222309
0.399421
1.000
42
Transit Bus
0.122177
0.232623
0.259237
0.385963
1.000
43
School Bus
0.133622
0.290446
0.202762
0.37317
1.000
51
Refuse Truck
0.133744
0.281628
0.244409
0.340218
1.000
52
Single-Unit Short-Haul Truck
0.133827
0.290565
0.233264
0.342345
1.000
53
Single-Unit Long-Haul Truck
0.124627
0.288468
0.224945
0.36196
1.000
54
Motor Home
0.146173
0.297276
0.211836
0.344715
1.000
61
Combination Short-Haul Truck
0.172224
0.327849
0.244772
0.255155
1.000
62
Combination Long-Haul Truck
0.338174
0.240709
0.256685
0.164432
1.000
' RoadTypelD = 1 (Off Network) is assigned no VMT.
65
-------
8. Average Speed Distributions
Average speed is used in MOVES to convert VMT inputs into the source hours operating (SHO)
units that MOVES uses for internal calculations. It is also used to select appropriate driving
cycles, which are then used to calculate exhaust running operating mode distributions at the
national, county and sometimes project level. Instead of using a single average speed in these
tasks, MOVES uses a distribution of average speeds by bin. The AvgSpeedDistribution table
lists the default fraction of driving time for each source type, road type, day and hour in each
average speed bin. The fractions sum to one for each combination of source type, road type, day
and hour. The MOVES average speed bins are defined in Table 8-1.
Table 8-1 MOVES s
peed bin categories
Bin
Average Speed (mph)
Average Speed Range (mph)
1
2.5
speed <2.5 mph
2
5
2.5 mph <= speed < 7.5 mph
3
10
7.5 mph <= speed < 12.5 mph
4
15
12.5 mph <= speed < 17.5 mph
5
20
17.5 mph <= speed < 22.5 mph
6
25
22.5 mph <= speed <27.5 mph
7
30
27.5 mph <= speed < 32.5 mph
8
35
32.5 mph <= speed < 37.5 mph
9
40
37.5 mph <= speed < 42.5 mph
10
45
42.5 mph <= speed <47.5 mph
11
50
47.5 mph <= speed < 52.5 mph
12
55
52.5 mph <= speed < 57.5 mph
13
60
57.5 mph <= speed < 62.5 mph
14
65
62.5 mph <= speed <67.5 mph
15
70
67.5 mph <= speed < 72.5 mph
16
75
72.5 mph <= speed
As described below, the default average speed distributions for all sourcetypes were updated in
MOVES3 using the telematics data. There were no updates for MOVES4.
8.1.Description of Telematics Dataset
In a study done by the Coordinating Research Council (CRC A-100)45, the GPS data collected by
Streetlight Data were used to develop inputs for the 2014 National Emissions Inventory. The
dataset consists of data from billions of trips derived from smart phone applications, in-
dashboard car navigation systems and commercial fleet management systems on vehicles
operating over a period of 12 consecutive months between September 2015 and August 2016 at a
high temporal and spatial resolution.
The data included latitude, longitude and timestamps corresponding to the instantaneous position
that each vehicle sends to a central server. Streetlight overlays the coordinates on their roadway
network to determine distance traveled between consecutive points. From the distance and time
66
-------
between points, average speeds were calculated and further classified by month, day of the week
and hour. The dataset also was able to discriminate between personal vehicles, medium-duty
commercial trucks (Class 6 and lower) and heavy-duty commercial trucks (Class 7 and 8). The
personal dataset was available at high resolution (1 Hz) and low resolution (one point every 10 or
30 seconds) while the commercial dataset was available at a lower resolution with one point
every 60 or 180 seconds. The data included a GIS shapefile containing road information
classified into the four MOVES road types and a second shapefile containing county boundaries
to generate data with the appropriate mapping.
Note that since the CRC A-100 project was developed to improve inputs used in the NEI, the
definitions of urban and rural applied to the CRC study were consistent with the requirements of
EPA's platform modeling for the NEI and regulatory impact analyses46, which follow the
definitions established by the U.S. Census Bureau. This is inconsistent with the urban-rural
roadtype definitions used in MOVES, which follow those established by FHWA. The main
difference in the definitions established by the U.S. Census Bureau and FHWA is the population
threshold used to distinguish between urban and rural. The U.S. Census Bureau defines an urban
area as areas with a population of 2500 or more, whereas the FHWA defines an urban area as
areas with a population of 5000 or more. Therefore, telematics speed data gathered by
Streetlight Data in some areas that are considered rural by FHWA and MOVES may have been
assigned to "urban" roadtypes. For MOVES modeling purposes, this discrepancy implies that the
average speed distributions derived from this dataset could be biased high by some degree, since
vehicles on rural roads generally spend more time traveling at faster speeds than those on urban
roads.
Due to restrictions in time and resources, the final dataset consisted of only 1/16th of the
information available to StreetLight Data. This aggregated subset totaled 250 million records
classified into 3 vehicle categories:
- Personal Passenger vehicles
- Medium-Duty commercial trucks (under 26,000 lbs of GVWR)
- Heavy-Duty commercial trucks (over 26,000 lbs of GVWR)
The final dataset contains information for the three vehicle categories mentioned above across
3,109 counties in the mainland US. The dataset was classified into MOVES roadtypes and
MOVES speedbins, for 12 months of the year, seven days of the week and 24 hours of the day.
For further details, see the CRC A-100 report.45
A single set of default national average speed distributions for the MOVES default database were
developed using the national database which contains average speed distributions for each
county and hour of the day, for weekday/weekend, varying by road type and source type.
Additionally, we used activity (VMT and average speed by county, fuel, source type and road
type) from the beta version of the NEI collaborative 2016 modeling platform47. The following
section describes the procedure to generate the average speed distributions included in MOVES.
8.2.Derivation of Default National Average Speed Distributions
67
-------
The general steps for the derivation of default average speed distributions were:
1. Calculation of source hours operating (SHO) for each source type on each road type
aggregated over all counties to represent the entire U.S.
2. Calculation of average speed distributions for each hour of the day, day of the week, road
type and source type, weighted by the fraction of SHO in each county in reference to the
national SHO for a given source type and road type combination.
For the first step, we used county-specific annual VMT classified by fuel, source type and road
type as well as county-specific annual average speed values classified by source type and road
type. Both data files were used in the development of activity for the NEI collaborative 2016
beta modeling platform and are based on FHWA and CRC A-100 information (where available),
respectively. We calculated a county-specific annual value of source-hours operating (SHO) for
each source type - road type combination, as shown in Equation 8-1, by adding all the VMT
assigned to different fuels (z) for each source type (ST) - road type (RT) combination in each
county (Co) and dividing by the corresponding annual average speed:
. , „,m Y,i=fuei Annual VMTSTiRTi Co
Annual SHOST RT Co =
Annual Average SpeedSTRT Co
miles
miles/hour.
Equation 8-1
Then, we aggregate over all counties z to obtain a national annual SHO for each source type (ST)
- road type (RT) combination following Equation 8-2:
National Annual SHOst rt = y Annual SHO^st rt^ [hours] Equation 8-2
' li=Co
In the second step, we used a data file from the CRC A-100 project containing average speed
distributions by hour of the day and day typefor each source type - road type combination for
each county. These values were weighted togheter using the SHO for each county developed in
Equation 8-1 divided by the national annual SHO determined in Equation 8-2. This results in
average speed distributions (ASD) weighted by the national activity for a given source type -
road type combination for each hour (h) of each weekday/weekend (d). This is summarized in
Equation 8-3:
ASD
h.d.ST.RT
I
AverageSpeedFractioni h d ST RT Co x Annual SHOSTRT Co Equation 8-3
i=16 National Annual SHOst rt
68
-------
Note that the sum over all 16 speed bins should be equal to 1 for each hour and type of day for a
given source type and road type combination.
For the default national average speed distributions used in MOVES4, we used the same
mapping of telematics data to MOVES source type used in the NEI to maintain consistency. For
buses, refuse trucks, and motor homes for which no direct mapping was provided, we assigned
the medium-duty commercial profile. The final mapping is detailed in Table 8-2:
Table 8-2 Map of MOVES Source Types to telematics data vehicle type
MOVES Source Type ID
MOVES Source Type Name
Telematics Vehicle Type
11
Motorcycle
Personal
21
Passenger Car
Personal
31
Passenger Truck
Personal
32
Light Commercial Truck
Medium-Duty Commercial
41
Intercity Bus
Medium-Duty Commercial
42
Transit Bus
Medium-Duty Commercial
43
School Bus
Medium-Duty Commercial
51
Refuse Truck
Medium-Duty Commercial
52
Single Unit Short-haul Truck
Medium-Duty Commercial
53
Single Unit Long-haul Truck
Heavy-Duty Commercial
54
Motor home
Medium-Duty Commercial
61
Combination Unit Short-haul Truck
Heavy-Duty Commercial
62
Combination Unit Long-haul Truck
Heavy-Duty Commercial
8.3.1 pdated average speed distributions
As an example, the resulting default average speed distributions for different vehicle types are
shown in Figure 8-1 for all road types and day types at 5 pm.
• Differences between Personal, Medium-Duty and Heavy-Duty commercial are most
noticeable on rural restricted roads, where the Personal category (mapped to Passenger
Cars, Passenger Trucks and Motorcycles) shows notably more time traveling at speeds
above 75 mph.
• For all vehicle types, weekday-weekend differences between average speed profiles are
generally small; the exception is for urban restricted access roads, reflecting the expected
difference between weekend and weekday traffic volumes at 5 pm on urban freeways.
69
-------
Weekday
• Weekerd
0 20 40 60 80 0 20 40 60 80 0 20 40 60 80
Average Speed (mph)
Figure 8-1 Average speed distributions for 5pm (hourlD 17) on the different MOVES road types.
For mapping between MOVES source types and telematics vehicle types, see Table 8-2.
The Streetlight data improves on previous estimates, with more data and enough detail to
differentiate between Personal, Medium-Duty and Heavy-Duty vehicle types. However, it does
not provide information to differentiate between vocation-specific trucks or buses. As new
datasets beome available, we will continue to update and improve these inputs.
70
-------
9. Driving Schedules ami Ramps
Drive schedule refers to a second-by-second vehicle speed trajectory. The drive schedules in
MOVES are intended to include all vehicle operation from the time the engine starts until the
engine is keyed off, both driving (travel) and idling time.1 Drive schedules are used in MOVES
to determine the operating mode distribution for MOVES running processes for calculation of
emissions and energy consumption. The drive schedules in MOVES4 are unchanged from
MOVES3 and are mostly the same as those in MOVES2014, with the exception of drive
schedules for transit and school buses, as described below, and the handling of ramps as
described in Section 9.2.
More specifically, each second of vehicle operation is assigned to an operating mode as a
function of vehicle velocity in each second and the specific power (VSP) for light-duty vehicles
or scaled tractive power (STP) for heavy-duty vehicles. The distinction between VSP and STP is
discussed in Section 15. Each operating mode is associated with an emission rate (in grams per
hour of vehicle operation). The average speed distribution is used to weight the operating mode
distributions determined from driving schedules with different average speeds into a composite
operating mode distribution that represents overall travel by vehicles. The distribution of
operating modes is used by MOVES to weight the emission rates to account for the vehicle
operation.
9.1.Driving Schedules
A key feature of MOVES is the capability to accommodate many schedules to represent driving
patterns across source type, road type and average speed. For the national default case, MOVES
uses 49 drive schedules with various average speeds, mapped to specific source types and road
types.
MOVES stores all drive schedule information in three database tables. The DriveSchedule table
provides the drive schedule name, identification number and the average speed of the drive
schedule. The DriveScheduleSecond table contains the second-by-second vehicle trajectories for
each schedule. In some cases, the vehicle trajectories are not contiguous; as detailed below, they
may be formed from several unconnected microtrips that overall represent driving behavior. The
Drive Schedule Assoc table defines the set of schedules which are available for each combination
of source use type and road type.
Table 9-1 through Table 9-6 below list the driving schedules used in MOVES. Some driving
schedules are used for both restricted access (freeway) and unrestricted access (non-freeway)
driving. In these cases, for example, at extreme congestion or unimpeded high speeds, we
assume that the road type itself has little impact on the expected driving behavior (driving
1 However, as described in Section 10, recent data suggests that drive schedules miss a substantial fraction of real-
world idling. MOVES has been updated to better account for the idling that was not captured in previous versions
of the model.
71
-------
schedule). Similarly, some driving schedules are used for multiple source types where vehicle
specific information was not available.
Table 9-1 MOVES driving cycles for motorcycles, passenger cars, passenger trucks and light
commercial trucks (11, 21, 31, 32)
ID
Cycle Name
Average
Speed
Unrestricted Access
Restricted access
Rural
Urban
Rural
Urban
101
LD Low Speed 1
2.5
X
X
X
X
1033
Final FC14LOSF
8.7
X
X
1043
Final FC19LOSAC
15.7
X
X
1041
Final FC17LOSD
18.6
X
X
1021
Final FC11LOSF
20.6
X
X
1030
Final FC14LOSC
25.4
X
X
153
LD LOS E Freeway
30.5
X
X
1029
Final FC14LOSB
31.0
X
X
1026
Final FC12LOSE
43.3
X
1020
Final FC11LOSE
46.1
X
X
1011
Final FC02LOSDF
49.1
X
1025
Final FC12LOSD
52.8
X
1019
Final FC11LOSD
58.8
X
X
1024
Final FC12LOSC
63.7
X
X
1018
Final FC11LOSC
64.4
X
X
1017
Final FC11LOSB
66.4
X
X
1009
Final FC01LOSAF
73.8
X
X
X
X
158
LD High Speed Freeway 3
76.0
X
X
X
X
Table 9-2 MOVES driving cycles for other buses (41
ID
Cycle Name
Average
Speed
Unrestricted access
Restricted access
Rural
Urban
Rural
Urban
398
CRC E55 HHDDT Creep
1.8
X
X
X
X
404
New York City Bus
3.7
X
X
201
MD 5mph Non-Freeway
4.6
X
X
X
X
405
WMATA Transit Bus
8.3
X
X
202
MD lOmph Non-Freeway
10.7
X
X
X
X
203
MD 15mph Non-Freeway
15.6
X
X
X
X
204
MD 20mph Non-Freeway
20.8
X
X
X
X
205
MD 25mph Non-Freeway
24.5
X
X
X
X
206
MD 30mph Non-Freeway
31.5
X
X
X
X
251
MD 30mph Freeway
34.4
X
X
X
X
252
MD 40mph Freeway
44.5
X
X
X
X
253
MD 50mph Freeway
55.4
X
X
X
X
254
MD 60mph Freeway
60.4
X
X
X
X
255
MD High Speed Freeway
72.8
X
X
X
X
397
MD High Speed Freeway Plus 5mph
77.8
X
X
X
X
72
-------
ID
398
401
404
201
405
202
402
203
204
403
205
206
251
252
253
254
255
397
ID
398
501
301
302
303
304
305
306
351
352
353
354
355
396
Table 9-3 MOVES driving cycles for transit and school buses (42, 43)
Cycle Name
Average
Speed
Unrestricted access
Rural
Urban
Restricted access
Rural
CRC E55 HHDDT Creep
U
X
X
X
Bus Low Speed Urban
3.1
X
X
New York City Bus
3.7
X
X
MD 5mph Non-Freeway
4.6
X
WMATA Transit Bus
1.3
X
X
MD lOmph Non-Freeway
10.7
X
Bus 12mph Non-Freeway
11.5
X
X
MD 15mph Non-Freeway
15.6
X
MD 20mph Non-Freeway
201
X
Bus 30mph Non-Freeway
21.9
X
X
MD 25mph Non-Freeway
24.5
X
MD 30mph Non-Freeway
31.5
X
MD 30mph Freeway
34.4
X
MD 40mph Freeway
44.5
X
MD 50mph Freeway
55.4
X
X
X
MD 60mph Freeway
60.4
X
X
X
MD High Speed Freeway
72.8
X
X
X
MD High Speed Freeway Plus 5mph
111
X
X
X
Table 9-4 MOVES driving cycles for refuse trucks (51)
Cycle Name
Average
Speed
Unrestricted access
Restricted access
Rural
Urban
Rural
CRC E55 HHDDT Creep
11
X
Refuse Truck Urban
2.2
X
X
HD 5mph Non-Freeway
5.S
X
HD lOmph Non-Freeway
11.2
X
X
X
HD 15mph Non-Freeway
15.6
X
X
X
HD 20mph Non-Freeway
19.4
X
X
X
HD 25mph Non-Freeway
25.6
X
X
X
HD 30mph Non-Freeway
32.5
X
X
X
HD 30mph Freeway
34.3
X
X
X
HD 40mph Freeway
47.1
X
X
X
HD 50mph Freeway
54.2
X
X
X
HD 60mph Freeway
59.4
X
X
X
HD High Speed Freeway
71.7
X
X
X
HD High Speed Freeway Plus 5mph
111
X
X
X
73
-------
Table 9-5 MOVES driving cycles for single-unit trucks and motor homes (52, 53, 54)
ID
Cycle Name
Average
Speed
Unrestricted access
Restricted access
Rural
Urban
Rural
Urban
398
CRC E55 HHDDT Creep
1.8
X
X
X
X
201
MD 5mph Non-Freeway
4.6
X
X
X
X
202
MD lOmph Non-Freeway
10.7
X
X
X
X
203
MD 15mph Non-Freeway
15.6
X
X
X
X
204
MD 20mph Non-Freeway
20.8
X
X
X
X
205
MD 25mph Non-Freeway
24.5
X
X
X
X
206
MD 30mph Non-Freeway
31.5
X
X
X
X
251
MD 30mph Freeway
34.4
X
X
X
X
252
MD 40mph Freeway
44.5
X
X
X
X
253
MD 50mph Freeway
55.4
X
X
X
X
254
MD 60mph Freeway
60.4
X
X
X
X
255
MD High Speed Freeway
72.8
X
X
X
X
397
MD High Speed Freeway Plus 5mph
77.8
X
X
X
X
Table 9-6 MOVES driving cycles for combination trucks (61, 62)
ID
Cycle Name
Average
Speed
Unrestricted access
Restricted access
Rural
Urban
Rural
Urban
398
CRC E55 HHDDT Creep
1.8
X
X
X
X
301
HD 5mph Non-Freeway
5.8
X
X
X
X
302
HD lOmph Non-Freeway
11.2
X
X
X
X
303
HD 15mph Non-Freeway
15.6
X
X
X
X
304
HD 20mph Non-Freeway
19.4
X
X
X
X
305
HD 25mph Non-Freeway
25.6
X
X
X
X
306
HD 30mph Non-Freeway
32.5
X
X
X
X
351
HD 30mph Freeway
34.3
X
X
X
X
352
HD 40mph Freeway
47.1
X
X
X
X
353
HD 50mph Freeway
54.2
X
X
X
X
354
HD 60mph Freeway
59.4
X
X
X
X
355
HD High Speed Freeway
71.7
X
X
X
X
396
HD High Speed Freeway Plus 5mph
77.8
X
X
X
X
The default drive schedules for light-duty vehicles listed in the tables above were developed
from several sources. "LD LOS E Freeway" and "HD High Speed Freeway" were retained from
MOBILE6 and are documented in report M6.SPD.001.48 "LD Low Speed 1" is a historic cycle
used in the development of speed corrections for MOBILE5 and is meant to represent extreme
stop-and-go "creep" driving. "LD High Speed Freeway 3" was developed for MOVES to
represent very high-speed restricted access driving. It is a 580-second segment of restricted
access driving from an in-use vehicle instrumented as part of EPA's On-Board Emission
Measurement Shootout program,49 with an average speed of 76 mph and a maximum speed of 90
mph. Fifteen additional light-duty "final" cycles were developed for MOVES based on urban
and rural data collected in California in 2000 and 2004.111 These cycles were selected to best
cover the range of road types and average speeds modeled in MOVES.
74
-------
The driving schedules (ID 201-206, 251-255, 397, and 398) used for all buses (41,42,43) are
borrowed directly from driving schedules used for single-unit trucks. The "New York City
Bus"50 and "WMATA Transit Bus"51 drive schedules are included for urban driving that
includes transit-type bus driving behavior. The "CRC E55 HHDDT Creep"52 cycle was included
to cover extremely low speeds for heavy-duty trucks. The "Bus 12 mph Non-Freeway" (ID 402)
and the "Bus 30 mph Non-Freeway" (ID 403) cycles used for transit and school buses were
based on Ann Arbor Transit Authority buses instrumented in Ann Arbor, Michigan.53 The bus
"flow" cycles were developed using selected non-contiguous snippets of driving from one stop to
the next stop, including bus-stop idling, to create cycles with the desired average driving speeds.
The "Bus Low Speed Urban" bus cycle (ID 401) is the last 450 seconds of the standard New
York City Bus cycle.
For MOVES3 and later versions of MOVES, we revised the handling of bus speeds; we changed
the driving cycle mapping in the DriveSchedule table to be the actual speed for all bus drive
cycles. Consistent with our changes, users should input the actual average speed distribution for
transit buses.
The "Refuse Truck Urban" cycle represents refuse truck driving with many stops and a
maximum speed of 20 mph but an average speed of 2.2 mph. This cycle was developed by West
Virginia University for the State of New York. For restricted access driving of refuse trucks at
extremely low speeds, the CRC E55 HHDDT Creep cycle is used instead. All of the other
driving cycles used for refuse trucks are the same as the driving cycles developed for heavy-duty
combination trucks, described below.
Single-unit and combination trucks use driving cycles developed specifically for MOVES, based
on data from 150 medium- and heavy-duty vehicles instrumented to gather instantaneous speed
and GPS measurements.54 The drive cycle data were segregated into restricted access and
unrestricted access driving for medium- and heavy-duty vehicles and then further stratified by
vehicles trips according the pre-defined ranges of average speed covering the range of vehicle
operation. The medium-duty cycles are used with single-unit trucks and heavy-duty cycles are
used with combination trucks.
The developed schedules are not contiguous schedules which could be run on a chassis
dynamometer but are made up of non-contiguous "snippets" of driving (microtrips) meant to
represent target distributions. For use with MOVES, we modified the schedules' time field in
order to signify when one microtrip ended and one began. The time field of the driving schedule
table increments two seconds (instead of one) when each new microtrip begins. This two-second
increment signifies that MOVES should not regard the microtrips as contiguous operation when
calculating accelerations.
Both single-unit and combination trucks use the CRC E55 HHDDT Creep cycle for all driving at
extremely low speeds. At the other end of the distribution, none of the existing driving cycles
for heavy-duty trucks included average speeds sufficiently high to cover the highest speed bin
used by MOVES. To construct such cycles, EPA started with the highest speed driving cycle
and added 5 mph to each point, effectively increasing the average speed of the driving cycle
without increasing the acceleration rate at any point. We have checked the feasibility of these
75
-------
new driving cycles (396 and 397) using simulations with the EPA's Greenhouse Gas Emissions
Model (GEM)55 for medium- and heavy-duty vehicle compliance. GEM is a forward-looking full
vehicle simulation tool that calculates fuel economy and GHG emissions from an input drive
trace and series of vehicle parameters. One of the aspects of forward-looking models is that the
driver model is designed to demand torque until the vehicle drive trace is met. Our results
indicate that the simulated vehicles could follow the speed demands of the proposed driving
cycles without exceeding maximum torque or power.
We compared the operating mode distrition estimated for a national scale run in MOVES to the
operating mode distribution measured from the Heavy-Duty In-Use Testing (HDIUT) program in
the Appendix G of the heavy-duty exhaust report. Overall, the operating mode distributions
compare well. One notable differene is, for a national scale run, MOVES estimates a higher
percentage of activity in the highest power, high speed operating mode bins.11 This may be
reasonable because the manufactur-run testing for the HDIUT data are expected to under-
represent high power operation due to steep grades, high speeds, and heavy-pay loads (e.g.,
multiple trailers, over-weight trailers) compared to the in-use fleet. Or perhaps, the discrepancy
could be due in part to the high-speed driving cycle being overly aggressive compared to in-use
driving. As mentioned in the Conclusions section, we suggest that a further evaluation of the in-
use operating mode distributions and heavy-duty driving cycles be considered for future work for
MOVES.
9.2. Modeling of Ramps in MOVES
For MOVES3 and later versions of MOVES, we simplified the modeling of emissions on
restricted access roadways by modeling ramps as part of highway driving. The MOVES3
Population and Activity Report has a detailed discussion of this change.38 For future versions of
MOVES, we hope to investigate whether drive cycles can be further improved by incorporating a
representative mix of ramp and highway driving.
At the project-scale, it is important to model ramps separately to identify localized areas where
high acceleration and deceleration events cause increases in exhaust emissions56 and brake
emissions. Users can continue to estimate ramps as individual links in project-scale. Preferably,
project-level users can characterize the specific operating mode or driving cycle of the ramps
they are evaluating.
76
-------
10. Off-Network Idle Activity
To better account for observed levels of idling, we added a new emission calculation to
MOVES3 for County and National Scale runs' allowing the model to estimate idle emissions that
occur off the road network (i.e., on roadTypeID=l) for all soucetypes. We have made no updates
for MOVES4. This section summarizes the new calculation methodology employed by MOVES
and then provides information on the idling data available for both light-duty and heavy-duty
vehicles.
10.1. Off-Network Idle Calculation Methodology and Definitions
We are defining the total idle fraction (TIF) as the ratio of the total source hours idling and total
source hours operating. This value can be derived from instrumented vehicles as explained
below. MOVES defines "idle" as any seconds in the driving schedules where the speed is less
than one mile per hour (opModeID=l) during engine operation. Using the fraction of vehicle
operation hours that are opModeID=l, the source hours idle (SHI2-5) during normal daily vehicle
operation for each of the four onroad road types (roadTypelDs 2, 3, 4, & 5) can be determined
from the driving schedules used for vehicle operation on roadways. We exclude any extended
engine idle that occurs during the mandated rest period for combination long-haul truck
(sourceTypelD 62), which we call hotelling (see Section 11). Total idle fractions are stored in
the new TotalldleFraction table in the MOVES default database.
Since the estimates of TIF are greater than the idle time accounted for in the MOVES driving
schedules (SHI2-5), we also need to increase MOVES' estimate of total source hours operating
(SHO). In particular, the off-network idle (ONI) time is defined as the additional idle hours that
need to be added to the on-network source hours operating (SHO2-5) in order to account for the
additional idle time. The on-network SHO2-5 is derived from the VMT and speed distribution.
Starting with MOVES3, the additional ONI hours are assigned to the running exhaust process
(processID=l) for the off-network road type (roadTypeID=l).
In MOVES2014, total SHO was calculated from vehicle miles traveled (VMT) and average
speed for all onroad roadTypelDs 2, 3, 4 and 5. Starting with MOVES3, we renamed this value
as on-network SHO2-5 to indicate that additional time needs to be added to account for off-
network idle time. The SHO for all road types will now include the "extra" operating time (ONI)
implied by the larger total idle fraction value:
SHO = (Y"5 SHOt) + ONI
^—>i=2
Where i = roadTypelD
Equation 10-1
J In Project Scale, MOVES does not adjust activity to account for off-network idling. Instead, the user can provide
location-specific idling activity as appropriate.
77
-------
Source hours idle (SHI) then is the total hours of idle, excluding diesel long-haul combination
truck hotelling idle:
SHI = (Y SHIi ) + ONI
t—>i=2
Equation 10-2
Where i = roadTypelD
All running exhaust activity for roadTypeID=l is idle, so SHOi=SHIi and represent ONI. Since
the TIF values are the measured fraction of idle time during vehicle operation, the SHI is also the
result of applying the TIF to the SHO:
SHI = TIF x SHO Equation 10-3
Thus, from Equation 10-1, Equation 10-2 and Equation 10-3:
tif = (J$=2 SHlJ+ONI Equation 10-4
ffi=2SHOt)+ONI
And, by by re-arranging Equation 10-4 and using the TIF, on-network source hours operating
(SHO2-5) and on-network source hours idling (SHI2-5) from the four network road types, MOVES
can calculate the hours for off-network idle (ONI):
(I?=2 SHOj ) X TIF - Ei=2 SHI, Equation 10-5
(1 - TIF)
Where i = roadTypelD
As an example, the default values of TIF for light-duty vehicles in idleRegionID=101 (New
Jersey) are presented in Table E-2 in Appendix E.
In cases where the ONI is calculated to be less than zero, the ONI will be set to zero. This is
currently true for motorcycles and motorhomes.
Off-network idle emissions are calculated for each hour by using the corresponding emission rate
(grams per hour) for opModeID=l for that hour. All of the adjustments (e.g., fuel effects, air
condition effects) made to the emission rates for opModeID=l for other road types apply to off-
network idle emissions as well. MOVES3 separately reports the emissions from the off-network
idle hours in the movesOutput table as exhaust running process (processID=l) for road type "off-
network" (roadTypeID=l).
10.2. Light-Duty Off-Network Idle
10.2.1. Verizon Telematics Data
78
-------
MOVES defaults for light-duty off network idling continues to rely on our MOVES3 analysis. In
developing MOVES3, Verizon Telematics data for light-duty vehicles was purchased only for
the following five states due to costs - California, New Jersey, Illinois, Georgia and Colorado.
These states were selected for a variety of reasons, including geographic coverage, urban and
rural mix, use of inspection and maintenance programs, and number of vehicles participating in
the program. The data were collected August 2015 through August 2016 using on-board
diagnostic data loggers under contracts with State Farm insurance, Mercedes-Benz and
Volkswagen. The data includes vehicles from model year 2017 back to model year 1996, which
is also the first year manufacturers were required to equip all vehicles with on-board diagnostic
(OBD) systems.57 Vehicle owners allowed their vehicles to be measured for a variety of reasons
and the data cannot be considered a random sample. The Verizon Telematics data were used as a
primary data source for the light-duty off-network idle defaults described in this section and also
for the soak and start defaults described in Section 12.1 The data characteristics and pre-
processing steps for both analyses are described here.
The Verizon data includes activity information gathered on vehicles for all or some subset of the
entire year. The information collected was summarized and processed into individual trips for
analysis. The analysis summary database includes trip start time and date, trip end time and date,
total trip time, total idle time, trip average speed, trip maximum speed and trip distance. Trips
were defined as the time period from key-on to key-off Engine idle was defined as any time
during the trip where the recorded engine RPM was greater than zero and the vehicle speed was
less than one mile per hour. Total idle time is a fraction defined as the ratio of the sum of the idle
time periods in a trip and the total time of the trip from key-on to key-off In addition to the trip
data, each trip was associated with a vehicle ID. For each vehicle ID, the model year and vehicle
registration postal ZIP code was provided. All vehicles were light-duty, either passenger car or
light-duty truck. No information about where the trips occurred was provided in the samples.
Using the provided data, all of the activity by vehicles was assumed to occur within the county in
which they were registered. The counties were categorized as urban or rural based on the U.S.
Census Metropolitan Statistical Area (MSA) classifications. Counties were also grouped as either
having a State Inspection and Maintenance (I/M) program or not.
10.2.2. QA/QC of the Verizon Telematics Data
Table 10-1 shows a high-level summary of Verizon Telematics data. The original dataset
provided by Verizon included around 41 million trip summary records from the five states. Such
large datasets pose several challenges related to data quality and sampling. For example, for
some trips, data were found to be missing or incomplete. Such trips were removed from the
original dataset and the remainder were used to analyze the idle fraction as summarized in the
"Total Trips (Idle)" column of Table 10-1.
79
-------
Table 10-1 Verizon Telematics data sample summary
State
Total Trips
(Original)
Total Trips
(Idle)*
Total Trips
(Soak Time & Starts)"
%Trips*"
California
1,958,858
1,886,947
1,761,184
90%
Colorado
5,644,374
5,390,417
4,977,334
88%
Georgia
15,457,392
14,654,336
13,465,865
87%
Illinois
12,955,252
12,318,387
11,448,257
88%
New Jersey
5,139,506
4,947,792
4,615,346
90%
Only valid trips included in idle analysis.
Only valid trips with previous recorded valid trips included in start and soak
analysis.
*" Percent of total trips remaining after all screening (starts divided by original
total).
In addition, not all vehicles in the sample had 12 complete months of data, due to termination of
subscriptions, instrumentation failures, etc. during the sampling period. To distinguish
infrequently used vehicles from those that had left the program, we developed an algorithm to
extract only those vehicles and their associated monthly data for which there was at least one trip
in the current month, the preceding and succeeding months. In addition, for a given vehicle, the
first and last month of the data for each vehicle was kept in the sampling frame only if there was
at least one trip in the first week and the last week for the month, respectively. Figure 10-1 shows
the Verizon Telematics sample vehicle population by state and month derived using this
sampling approach. Appropriate weighting was then applied to the monthly results to generate
annual averages.
80
-------
12015-8 >2015-9 B2015 - 10 12015 - 11 B2015 - 12 B2016 - 1
12016-3 ¦ 2016 - 4 B2016-5 B2016-6 B2016-7 B2016-8
12016-2
7000
6000
VI
pi
5000
u
2
>
4000
Ct-x
O
N
Q*>
3000
—
=
s
2000
1000
0
J
21 31
California
21 31 21 31 21 31
Colorado Georgia Illinois
sourceTypelD/State
21 31
New Jersey
Figure 10-1 Sample vehicle population in the Verizon Telematics data by month, state and
sourceType. Note: the legend indicates the "year-month" of the data collection.
There were a few instances where the trip time was less than 1 second, or the soak time was less
than two seconds, for example, when a vehicle crossed into a different time zone or when the
data logger recorded erroneous trip starts at midnight for trips that included midnight driving.
Such trips represented less than 1 percent of the total trips for any given state and were removed
from the idle and starts/soak analysis. The remaining trips were used to analyze engine starts and
soaks (see the "Total Trips (Soak Time & Starts)" column in Table 10-1 for the total trip counts).
The erroneous trip starts removed from the start/soak analysis do not affect the results for the
analysis of total idle time.
10.2.3. Estimating MOVES National Defaults from Verizon
Telematics Data
The Verizon Telematics data covered only five states, but MOVES must model the entire U.S.
Thus, we associated each state with nearby states to create vehicle-population weighted national
averages for starts and soaks and regional-specific values for idle time. Table 10-2 lists the
vehicle populations used for computing national averages. Figure 10-2 shows how we mapped
individual states to the Verizon data. We grouped the states qualitatively, considering proximity
and climate. Climate was considered because the monthly patterns varied between areas with
large temperature shifts between seasons (Colorado, Illinois) and states with moderate seasonal
changes (California and Georgia). The weighted average results for the light-duty passenger
trucks (indicated in the data as sourceTypelD 31) were used for light-duty commercial trucks
81
-------
(sourceTypelD 32) as well. Due to lack of data, motorcycle idle fractions were set to zero. This
results in the same roadway (drivecvcle-based) idling as before and no off-network idle.
Table 10-2 2014
Vehicle populations of the idle regions
Verizon data source state
sourceTypelD
Vehicle Population
idleRegionID
California
21
23,114,006
105
California
31
19,917,792
105
Colorado
21
6,902,041
104
Colorado
31
8,823,105
104
Georgia
21
38,269,101
102
Georgia
31
39,358,137
102
Illinois
21
26,768,198
103
Illinois
31
25,510,186
103
New Jersey
21
27,625,575
101
New Jersey
31
23,077,050
101
Calgary
o
.Vancouver
o
Colorado
California
Toronto!
S:hi . ago Detroit,
Illinois
¦^BgBoston
New Jersey
^Philadelphia
fuigton
Georgia
'
_M5%rey
MEXICO
Guadalajara
o
Mi am«
Ha^; ana
CUBA
Esn HERE, Ga..:
Figure 10-2 Default Regions for Weighting Light-Duty Activity14
In addtion to region, the Verizon Telematics data analysis suggested that the following factors
are important when estimating total idling fraction:
k Note, Alaska is associated with Colorado. Hawaii. Puerto Rico and the Virgin Islands are associated with
California.
82
-------
• Month of the year (which depends on the region)
• County type, i.e., whether registered in an urban (MSA) or rural county
• Passenger car or light truck
• Day type, i.e., weekend vs. weekday variation
The analysis showed no significant variation with age or hour of the day. A simplified linear
regression model was built to capture the variability of the total idle fraction (TIF) across
different variables (daylD, sourceTypelD, countyTypelD, idleRegionID and monthID).
MOVES3 default values for TIF were calculated based on the equation below:
TIF = daylDi + sourceTypelDj + countyTypeIDk + idleRegionlDi Equation
+ monthlDm + idleRegionlDi x rnonthlDm + n 10-6
where, i,j, k, l,m are coefficient values for the combinations of daylD (2=Weekend,5=Weekday),
sourceTypelD, countyTypelD, idleRegionID and monthID and n is the intercept (a constant) for
Equation 10-6 above. The regression model handled ordinal categorical variables as independent
variables. The full set of coefficients are available in Appendix E.
As one might expect, idling activity is more common in winter months in colder states and urban
areas have more idling activity than rural areas. There is less idling activity on weekends versus
weekdays. Idling activity is similar for passenger cars and light trucks, but separate idle
fractions were developed for each of the source types.
In MOVES, we use the model fit TIF values from the multi-variable linear model (Equation
10-6, rather than using the averages from the Verizon Telematics data, mainly to smooth the
variation in the Verizon Telematics data. Figure 10-3 below illustrates the model fit against
actual values. TIF model results are represented by solid lines versus average values from the
Verizon Telematics data, shown as dashed lines. As expected, Region 105 (California) which has
the smallest sample size also shows the most variation and deviation from the regression results.
For example, for Region 105 (California), passenger trucks (sourceTypelD 31), weekdays
(daylD 5), the model fit smooths out the abnormally high idle fraction measured for July
(monthID 7).
We also use the model estimated TIF values to estimate values that were not measured by
Verizon. Note that there was no data available for New Jersey from Verizon Telematics for rural
counties (i.e., countyTypeID=0) as shown in Figure 10-3. However, the regression model applies
the rural/urban effect without regard to region. Appendix E shows a sample calculation using
MOVES3 default values for passenger cars in rural counties in idleRegionID=101 (New Jersey).
The model fit TIF values apply to all calendar years in MOVES. Note that idleRegionID and
countyTypelD vary depending on the county location. Each state is assigned an idleRegionID in
the MOVES State table as shown in Figure 10-2. Each county is assigned an "urban" or "rural"
countyTypelD in the MOVES County table based on the MSA designation. As discussed earlier,
the results for the light-duty passenger trucks (indicated in the data as sourceTypelD 31) are also
used for light-duty commercial trucks (sourceTypelD 32).
83
-------
—•—TIF (Modeled) - 0 —•—TIF (Modeled) - 1 0 represents courilyTypcJD-0
-.-TIF (Actual)-0 - — TIF(Actual)- 1 1 comtvT»pelI)-l
sourceTypelD=21, idleRegionID=101
source'fypelD=21, id IeRegionlD=l 03
sourccTypelD=31, id I c Region I1)-101
0.3
E 0.28
'I 0.26
2
't 0.24
-5 0.22
5
0.2
0.18
0.16
monthlD/day'ID
soureeTypeTD=31, idIcRegionTD=l 03
0.34
0.32
^ 0.3
E 0.28
5
0.26
jt 0.24
I0'22
0.2
0.18
0.16
source'i'ypelD=21, idleRegionlD=l 02
1 2 3 4 5 6 7 8 9 101112| 1 2 3 4 5 6 7 8 9 1011 12
montblDMayll)
soureeTypeID=21, idleRegionID=l04
0.34
0.32
0.3
E- 0.28
1
'§ 0.26
2
U 0.24
^ 0.22
|
0.2
0.18
0.16
12345678? 1011
monthlD/daylD
sourceTypeTD=31, idleRegionTD=l 02
sourecTypeID=31, idleRegionID=l04
monlhTD'ilayTT)
sourceTypeID=21, idleRegionlD=105 source l'ypelD=31, idleRegionlD=105
Figure 10-3: TIF model results compared to the values from the Verizon Telematics data
84
-------
10.3. Heavy-Duty Off-Network Idle
The Verizon Telematics data exclusively covered light-duty vehicles. Heavy-duty vehicles are
spread across a wide range of vocations and have activity patterns that are distinctly different
from light-duty. Currently, the idling captured in the MOVES driving cycles represents the idling
at intersections and on congested highways, but do not include a full estimate of "workday idle"
that many commercial heavy-duty trucks experience in their daily operation, such as queuing at
distribution centers, or loading and unloading payload. Off-network idle is also intended to
address these gaps in idle activity modeling.
The heavy-duty off-network idle defaults were derived from the National Renewable Energy
Laboratory (NREL) Fleet DNA clearinghouse of commercial fleet vehicle operating data. The
data processing applied to the Fleet DNA dataset is described in this section. This analysis has
not been updated since MOVES3. However, the University of California Riverside's Bourns
College of Engineering Center for Environmental Research and Technology (CE-CERT) has
concluded their data collection for a study to evaluate the selective catalytic reduction (SCR)
behavior of heavy-duty vehicles. We hope to apply the same processing steps to the latest CE-
CERT dataset and expect to combine the results the with Fleet DNA data in a future MOVES
update.
The same Fleet DNA dataset and pre-processing steps described in this section were used for the
soak and start defaults described in Section 12.2
10.3.1. NREL Fleet DNA Database
We partnered with NREL to make use of their expansive Fleet DNA database58 of heavy-duty
vehicles to develop idle activity estimates for heavy-duty vehicles. NREL's Fleet DNA database
is developed from vehicles operating in the field with devices to record 1-Hz telematics and
CAN (controller area network59) data.
While the Fleet DNA database includes a wide range of fuels, vehicle drivetrains and propulsion
mechanisms, only diesel-powered conventional vehicles were included in the analysis to ensure
the selected drive cycles are representative of traditional operation and not modified to
accommodate the vehicle architecture. This analysis used data from 415 conventional heavy-duty
vehicles with over 120,000 hours of operation, providing a diverse data encompassing 23 vehicle
vocations in 36 states. The number of conventional vehicles in the Fleet DNA database by
MOVES source type are shown in Table 10-3. The table also includes the number of states with
activity in each Fleet DNA sample. The geographic distribution could influence average idle
85
-------
emission rates, due to differences in congestion, topography and regional policies.1 60 However,
as presented in the NREL project report,61 truck idling and start activity was observed to be
largely a function of the truck vocation, rather than the US state of operation. Likely a larger
sample size of vehicles across vocations and states would be needed to elucidate geographic
differences in truck activity.
Table 10-3. Sample size of conventional vehicles in the Fleet DNA database by MOVES source type
sourceTypelD
Source Type Name
Number of Vehicles
in Fleet DNA
Number of States
with Recorded
Activity
41
Other Buses (non-school, non-transit)
0
0
42
Transit Buses
16
3
43
School Buses
7
1
51
Refuse Trucks
37
4
52
Single-Unit Short-Haul Trucks
119
8
53
Single-Unit Long-Haul Trucks
0
0
54
Motor Homes
0
0
61
Combination Short-Haul Trucks
105
8
62
Combination Long-Haul Trucks
131
32
Total
415
Note: The number of trucks operating in each US state is listed in the NREL project report61
Table 10-4 shows the vocational distribution of the short-haul source types (single unit and
combination short-haul trucks) and the sample size of each vocation category. A complete
description of the Fleet DNA dataset, additional pre-processing performed, and analyses not
discussed in this report can be found in the NREL report.61
1 For example, California has a regulation prohibiting idling for more than five minutes for vehicles that
are not California clean idle certified.1 However, other states, counties and cities also have idling
regulations. In addition, most recent heavy-duty vehicles are California clean-idle certified. For example,
all fourteen of the MY 2008 and later heavy heavy-duty tractors tested for extended idling emission rates
(produced from four major engine manufacturers) were all clean idle certified.5
86
-------
Table 10-4. Vocation types of the Combination Short-Haul and Single-Unit Short-Haul vehicles within the
Fleet DNA database
Combination Short-Haul
Vehicle Vocation
Number of
Vehicles in
Fleet DNA
Single-Unit Short-Haul
Vehicle Vocation
Number of
Vehicles in
Fleet DNA
Beverage Delivery
10
Warehouse Delivery
9
Food Delivery
13
Parcel Delivery
39
Dray age
28
Linen Delivery
17
Transfer Truck
28
Food Delivery
30
Local Delivery
7
Snow Plow
11
Regional Haul
7
Towing
4
Dump Truck
4
Concrete
3
Parcel Delivery
5
Delivery
1
Dry Van
3
Shredder
1
Propane Tank
1
Dump Truck
3
87
-------
10.3.2. CE-CERT Study
The California Air Resources Board (CARB) contracted with CE-CERT to conduct a large-scale
study in which vehicle and engine activity data were collected from 90 heavy-duty vehicles that
are mapped to 19 different groups defined by a combination of vocational use, gross vehicle
weight rating and geographic region within California. EPA supported the test program by
providing data loggers and data quality analysis through a Cooperative Research and
Development Agreement with CE-CERT. Most of these vehicles were registered in California
and traveled a majority of their miles in-state. The study did include some out-of-state vehicles in
the line-haul and pick-up/delivery categories. Almost all the vehicles were of model year 2010 or
newer and most were equipped with SCR technology. One drayage truck was model year 2008
(with no SCR) and all the buses were CNG fueled. In addition, some of the vehicles in the study
were hybrids. We hope to incorporate data from the CE-CERT study in future versions of
MOVES.
10.3.3. Heavy-Duty Off-network Idle Data Processing
The NREL Fleet DNA data were preprocessed to identify starts and idle periods for the analysis.
Two key parameters are engine speed (revolutions per minute [rpm]) and wheel speed (miles per
hour [mph]). An engine speed greater than zero indicates that the vehicle engine is running and a
wheel speed greater than zero mph signifies that the vehicle is in motion. In this analysis, vehicle
starts are calculated by identifying the transition from an engine speed of zero to greater than
zero. Vehicle soak is defined as the length of time between engine off (engine speed of zero) and
the next time it is started (engine speed greater than zero). A vehicle is considered to be idling
when its wheel speed is less than one mph and the engine speed is greater than zero. The total
operating time (engine RPM > 0) occurring within each daylD is also calculated.
Periods of contiguous idle are identified by length and the daylD corresponding to the start of the
idle. If an idle period started during one daylD and ended on another, the idle time was only
counted for the daylD in which the trip started. Idle periods longer than an hour were categorized
separately as "extended idle" for long-haul combination trucks (sourceTypelD 62) and not
included in the average idle time of the off-network idle fraction calculation below.
Vehicle activity values in MOVES represent average activity at a national scale. MOVES uses
"total idle fraction" to quantify off-network idle. In this analysis, total idle fraction was
calculated by first summing the daily average idle time for each individual vehicle across all
vehicles within the same vehicle (sourceType) and day type (daylD) classification. Those
summed idle times were then divided by the sum of the daily average operating time for each
individual vehicle across all vehicles within the same vehicle and day type. This sum-over-sum
approach normalizes the recorded activity by the amount of time each vehicle was instrumented
and weights the average idle fraction towards the vehicles with the most daily-average activity.™
m We evaluated several approaches for calculating average idle fraction using the Fleet DNA data. The approach
presented here (Equation 10-7) is equivalent to Equation 17-23 (Method 3 "normalized sum over sum") in Appendix
I. Appendix I includes an overview of each approach and a comparison between calculation approaches.
88
-------
Equation 10-7 shows the calculation of the total idle fraction for each source type and specific
day type (weekday or weekend).
v, fidle hours; /
M Uays)
Idle fractions d = r -
' Y /operating hoursi / \
V ! day Si) Equation 10-7
Where:
i = individual vehicle ID
s = source type ID
d = day type ID
10.3.4. Heavy-duty Off-network Idle Results
As seen in Table 10-3, several heavy-duty source types were not available in the Fleet DNA
database at the time of this report. Additionally, none of the school buses instrumented for this
dataset operated on the weekend, so there is no data for day ID 2. We hope to have more of the
source types and day ID's covered when we process the CE-CERT dataset and combine it with
the Fleet DNA dataset in a future version of MOVES. In the interim, we assumed the idle
behavior of the missing vehicles closely matched others. We chose to use the transit bus
(sourceTypelD 42) to represent other buses (sourceTypelD 41), applied the weekday data from
the school bus (sourceTypelD 43) for the missing weekend data, used the single-unit short-haul
data (sourceTypelD 52) to represent the single-unit long-haul trucks (sourceTypelD 53).
Lacking data for motorhomes (sourceTypelD 54), we set their total idle fraction to zero. This
will result in the same roadway (drivecycle-based) idling as in MOVES2014 and no off-network
idle. While this is an area that would benefit from more research, we think it is unlikely for
motorhomes to idle significantly when they are not on roadways since they are equipped with
APUs and often park where auxiliary power is available.
Figure 10-4 and Figure 10-5 show the idle fraction values for weekends and weekdays,
respectively. In both figures, the solid blue bars represent the off-network idle for each heavy-
duty vehicle sourceType. The hashed bars represent the extended idle portion, which is only
available to the long-haul combination trucks (sourceTypelD 62). The specific values added to
the MOVES TotalldleFraction database table for this update are shown in Table 10-5.
89
-------
Idle Fraction, Weekends
0.6
0.5
e
o
0.4
S3
m
u_
0.3
-------
Table 10-5 Idle fraction values for heavy-duty sourceTypes based on data from NREL's Fleet DNA
database
SourceType
Vehicle Description
Weekend It
le Fractions
Weekday It
le Fractions
Off-
Network
Extended
Off-
Network
Extended
41
Other Bus
0.388
0.000
0.390
0.000
42
Transit Bus
0.388
0.000
0.390
0.000
43
School Bus
0.314
0.000
0.314
0.000
51
Refuse Truck
0.503
0.000
0.469
0.000
52
Single Unit, Short
0.420
0.000
0.348
0.000
53
Single Unit, Long
0.420
0.000
0.348
0.000
61
Combo, Short
0.312
0.000
0.332
0.000
62
Combo, Long
0.130
0.127
0.145
0.138
10.4. Off-network Idling Summary
Figure 10-6 displays the off-network idling fraction and the on-network idling fraction for an
urban county in the midwestern idle region. The off-network idling accounts for most of the
idling for most source types. Note that the idle fraction, and subsequently, the off-network idling
fraction changes significantly between January and July for the light-duty vehicles. However, it
is unchanged for the heavy-duty vehicles. Also, note that the idling fraction for long-haul
combination trucks is lower than for other vehicles because long-duration idling (> 1 hour) for
long-haul combination trucks is modeled as hotelling activity discussed in the Section 11.
91
-------
daylD: 2
0.5
0.4 i
0.3
0.2
c 0-1
o
S0.0H
0.5
0.41
0.3
0.2
0.1
0.01
daylD: 5
T3
idle_fraction
| off_network
I on network
21 31 32 41 42 43 51 52 53 61 62 21 31 32 41 42 43 51 52 53 61 62
sourceTypelD
Figure 10-6. On-network idle and Off-network idle fractions for an Urban County in the
Midwestern Region
ll.Hotelling Activity
MOVES defines "hotelling" as any long period of time (e.g., > 1 hour) that drivers spend in their
vehicles during mandated rest times during long distance deliveries by tractor/trailer combination
heavy-duty trucks. During the mandatory rest time, drivers can stay in motels or other
accommodations, but most of these trucks have sleeping berths built into the cab of the truck and
drivers stay in their vehicles.
Hotelling hours are included in MOVES to account for the energy used and pollutants generated
to power air conditioning, heat and other amenities. These amenities require power for operation,
which can be obtained by running the main truck engine (extended idle) or by use of smaller on-
board power generators (auxiliary power units, APU). Some truck stop locations include power
hookups (truck stop electrification or "shore power") to allow use of amenities without running
either the truck engines or APUs. Some of the rest time may occur without the use of amenities
at all.
In MOVES, only the long-haul combination truck source use type (sourceTypelD 62) is assumed
to have any hotelling activity. All source use types other than long-haul combination trucks have
hotelling activity fractions set to zero.
11.1. Hotelling Activity Distribution
92
-------
In MOVES, hotelling hours are divided into operating modes which define the emissions
associated with the type of hotelling activity. As explained above, long-haul trucks are often
equipped with sleeping berths and other amenities to make the drive rest periods more
comfortable. Table 11-1 shows the hotelling operating modes available in MOVES.
Table 11-1 Hotelling activity operating modes in MOVES
OpModelD
Description
200
Extended Idling of Main Engine
201
Hotelling Diesel Auxiliary Power Unit (APU)
203
Hotelling Shore Power (plug in)
204
Hotelling Battery or All Engines and Accessories Off
The hotelling activity distributions in MOVES are consistent with the hotelling assumptions used
in EPA's Heavy-Duty Greenhouse Gas Phase 2 rulemaking, which included increasing adoption
of battery or electric supplemental power.62 Additionally, we updated the model to include a
fraction of hotelling time when the driver did not require any supplemental power. Starting in
2011, the hours-of-service regulations from the Federal Motor Carrier Safety Administration
(FMCSA) were updated to encourage longer periods of rest.63 Drivers could split their 10 hours
of mandated off-duty time between the sleeper berth for at least 8 hours and another location for
the remaining 2 hours. We assume the drivers do not require power when not in the sleeper berth
and applied a constant 20 percent of hotelling time to represent the 2 hours off-duty time not in
the sleeper berth for all years.
The HotellingActivityDistribution table, shown in Table 11-2, contains the MOVES default
values for the distribution of hotelling activity to the operating modes. For model years before
2010, we assume 80 percent of time is extended idling and 20 percent does not require
supplemental power, as mentioned previously. Starting with the 2010 model year, an increased
number of trucks equipped with APUs are expected as a result of the Phase 1 Heavy Duty
Greenhouse Gas Standards64 and a fraction of the time that previously was assigned to extended
idle is now assigned to opModelD 201 (the use of APUs). In later model years, we continue to
assume a constant fraction of time with no supplemental power and distribute the remaining time
among extended idle, APU use and increasing battery use based on EPA's assessment of
technologies expected to be used by tractor manufacturers to comply with the Heavy-Duty
Greenhouse Gas standards Phase 2, with stepwise increases in model years 2021, 2024 and
2027.65 Similar to pre-2010 model years, we assumed drivers would not require supplemental
power 20 percent of the time for model years 2010 and later.
Alternative fueled long-haul trucks are assumed to have the same hotelling activity distribution
as diesel trucks as described above with the following adjustments:
• CNG trucks are assumed to not have diesel APUs and instead rely on the main engine
idling.
• Fuel Cell EVs are assumed to use shore power for the 80 percent time in sleeper berth.
93
-------
Table 11-2 Default hotelling activity distributions
Fuel Type
beginModelYearlD
endModelYearlD
opModeFraction for given opModelD
200
201
203
204
Idle
APU
Shore Power
Battery/Off
Diesel
1960
2009
0.80
0.00
0.00
0.20
2010
2020
0.73
0.07
0.00
0.20
2021
2023
0.48
0.24
0.00
0.28
2024
2026
0.40
0.32
0.00
0.28
2027
2060
0.36
0.32
0.00
0.32
CNG
1960
2020
0.80
0.00
0.00
0.20
2021
2026
0.72
0.00
0.00
0.28
2027
2060
0.68
0.00
0.00
0.32
Fuel Cell
EV
1960
2060
0.00
0.00
0.80
0.20
Based on peer-review comments on the above analysis for diesel trucks in 2017, we reevaluated
our assumptions about APU and hotelling battery penetration rates. The diesel APU usage
assumptions for model year 2010 through 2020 in Table 11-2 are qualitatively consistent with
two fleet surveys: NACFE 2018 Annual Fleet Fuel Study6611 and Shoettle et al. (2016).67 0 On
the other hand, both surveys suggested a higher (non-zero) penetration of hotelling battery units
in 2010-2020, as well as projecting a higher penetration in future years. However, given
concerns about the representativeness of the surveys, we have decided to retain the current
assumption regarding fleet-average APU and battery usage in MOVES4 and recognize that the
current hotelling battery usage may be a low estimate. Future MOVES updates could utilize
instrumented truck and APU measurements to replace these projections.
11.2. National Default Hotelling Rate
To estimate hotelling activity, MOVES uses a hotelling rate. As shown in Equation 11-1, the
default hotelling rate is the national total hours of hotelling divided by the national total miles
driven by long-haul combination trucks on all restricted access roads (both urban and rural).
Hotelling Hours ^ „
Hotelling Rate = - —- Equation 11-1
Total Restricted Miles Traveled
Where: Total Restricted Miles Traveled is the total miles traveled by diesel long-haul
combination trucks on rural and urban restricted access roads (freeways) in MOVES.
n NACFE (2018) reported increasing diesel APU and and battery penetration rates for model year 2010-2016
vehicles. The diesel APU values span the MOVES values for model year 2010-2020. The NACFE 2013 Annual
Fleet Fuel Study reports survey values from 20 participating fleets, which are likely earlier adopters and may not be
representative of the entire fleet.
° Shoettle et al. reports that 38.7 percent of fleets use auxiliary power sets and 30.1 percent use battery packs based
on a survey of 96 heavy-duty fleet managers. However, information regarding the percentage of vehicles within a
fleet is not provided.
94
-------
The hotelling rate is used to estimate hotelling in different calendar years and to spatially allocate
hotelling to counties across the US. The hotelling rate is based on travel on rural and urban
restricted access roads (freeways), because this is where long-haul trucks are most frequently
operated and most hotelling occurs at locations near those roadways (i.e., rest stops or truck
stops).
For MOVES3 and later versions of MOVES, the national default hotelling rate is based on data
collected and analyzed by the National Renewable Energy Laboratory (NREL) Fleet DNA61 as
discussed in Section 10.3.1. For the hotelling analysis, NREL analyzed data collected from 131
long-haul combination diesel trucks operating in the United States. The 131 trucks had broad
coverage across the United States, with home bases in 32 states.
Because the NREL data did not include information on all operating modes of hotelling activity,
we back-calculated the hours of hotelling from the data on extended idling using Equation 11-2.
First, we estimated the extended idle hours per mile from the NREL data. Vehicles were
assumed to be extended idling (hotelling with the main engine running in idle), if the vehicle
speed = 0 and the duration of the idling was > 1 hour. For the 131 long-haul trucks, the trucks
averaged 3.45 extended idle hours for every 1,000 miles driven. Then, we calculated a ratio of
total miles traveled to restricted access miles using the MOVES national default values presented
in Table 7-2 (the rural restricted VMT fraction = 0.34 and urban restricted VMT fraction = 0.26).
This allows better spatial allocation of hoteling activity to counties with freeways. Finally, we
multiply the extended idle hours by the ratio of hotelling hours to the extended idle hours. We
did not have information from NREL about use of auxiliary power units from any of the trucks
in the Fleet DNA data, so we used the 80 percent extended idling value for pre-2010 model year
trucks which assumes no APU usage as presented in Table 11-2.
Hotelling Rate =
(Extended Idle Hours\( Total Miles Traveled \ / Hotelling Hours \
\Total Miles Traveled) \Restricted Access Miles Traveled) \Extended Idle Hours)
= ( ) (—)
V1000/ V0.34+ 0.26/ V0.8/
= (—) (—)
Viooo/ V0.6/V0.8/
7.2 Hotelling Hours
1000 Restriced Access Miles Traveled
Figure 11-1 compares the hotelling rate in MOVES3 derived from NREL Fleet DNA, with the
default value used in MOVES2014 for the 2014 NEI version 268 and two other studies. Lutsey et
Equation
11-2
95
-------
al.69 presented data from a nationwide truck surveyp and NCHRP 08-10170 conducted an analysis
of an instrumented truck dataset with 300 trucks.q The hotelling rate was unchanged between
MOVES3 and MOVES4.
30.00 i
25.00 :
20.00 :
15.00 :
W \
x 10.00 : ¦¦¦¦¦I
5.00 :
0.00 4 , , , .
MOVES2014 MOVES3 Lutsey (UC Davis) et NCHRP 08-101
(2014 NEI v2) (NREL Fleet DNA) al. (2004)
Figure 11-1. Hotelling hours per 1000 miles driven on freeways compared across different datasets.
In MOVES, the national rate of hotelling hours per mile of restricted access roadway VMT is
stored in the HotellingCalendarYear table for each calendar year. When the hotelling rate is
applied, it is multiplied by the rural and urban restricted access VMT by long-haul combination
trucks to estimate the default hotelling hours for any location, month or day. In MOVES, the
national rate of hotelling hours per mile of restricted access roadway VMT is stored in the
HotellingCalendarYear table for each calendar year. When the hotelling rate is applied, it is
multiplied by the rural and urban restricted access VMT by long-haul combination trucks to
estimate the default hotelling hours for any location, month or day.
The County Data Manager includes the HotellingHours table which provides the opportunity for
states and other users to provide their own estimates of hotelling hours specific to their location
and time. Whenever possible, states and local areas should obtain and use more accurate local
estimates of hotelling hours when modeling local areas.
The allocation of hotelling to specific hours of the day is described below in Section 13.5.
p Lutsey reported average idling hours and driving hours per day. Using default national hours driving and restricted
access miles driven reported in Table 11-1 of the MOVES2014 Population and Activity Report, we derived an
estimate of extended idle hours per restricted access miles. We also used the ratio of hotelling hours to extended idle
hours as was done in Equation 11-2.
q Equation 11-2 was also used to calculate the hotelling rates from the data reported from NCHRP 08-101. The
definition of hotelling for the NCHRP 08-101 data was idling between 8 and 16 hours of duration, which is different
than used by the NREL analysis.
96
-------
12.Engine Start Activity
Immediately following the start of an internal combustion engine, fuel is inefficiently burned due
to the relatively cool temperature of the engine and the need to provide excess fuel to promote
combustion. During this time, the quantity and profile of the pollutants generated by the engine
are significantly different than when the running engine is fully warm. Additionally, the after-
treatment technology employed on modern vehicles often requires time to become fully
functional. For these reasons, MOVES accounts for the effects of engine starts separately from
the estimates for hot running emissions.
The temperature of the engine and after-treatment systems depend not only on ambient
temperature, but the time since the last engine operation (soak time) as discussed in the light-
duty10 and heavy-duty11 emission rate reports. MOVES accounts for the soak time using "soak
time operating modes." The distribution of the soak times for engine starts can have a significant
effect on the emissions estimated for trips.
MOVES uses the following set of tables in the default database to determine the default number
of starts, soak times and their temporal distributions:
• StartsPerDayPerVehicle
• Starts Age Adjustment
• StartsHourFraction
• StartsMonthAdjust
• StartsOpModeDistribution
The StartsPerDayPerVehicle table contains a factor (StartsPerDayPerVehicle) which, when
multiplied by the total number of vehicles of a given source type calculates the number of starts
in a day. The StartsPerDayPerVehicle factor represents the average starts per day for each
sourcetype and day type (weekday/weekend)
Starting with MOVES3, starts vary by source type, day type and vehicle age to account for the
lower average start activity that is expected to occur as vehicles age (see Section 12.1.1 for light-
duty and Section 12.2.3 for heavy-duty). Figure 12-1 shows the calculation of starts per day per
vehicle by vehicle age for light-duty vehicles. Note that the age 0 starts per day are greater than
the fleet-average starts per day, and the starts at age 30 are lower than the fleet-average starts per
day.
MOVES accounts for the effect of age using the ageAjustment factors stored in the
Starts Age Adjustment table. This table stores the number of starts by vehicle age within each
sourcetype, relative to the number of starts at age 0. All of the ageAdjustment factors in MOVES
are based on the mileage accumulation rates (discussed in Section 6.2). By using the mileage
accumulation rates to derive the start ageAdjustement factors, we are assuming that the starts per
mile is constant over the life of the vehicle. In other words, as vehicles travel fewer miles per day
as they age, they similarly conduct fewer starts. The ageAdjustment factors for each source type
are set equal to one at age zero, and decrease from one as the age increases, reflecting relatively
lower starts as the vehicles age. MOVES does not use the absolute values in this table, but scales
97
-------
the ageAdjustment factors in conjunction with the source type age distributions of the MOVES
run (Section 6.1) such that the average starts reported in in the StartsPerDayPerVehicle table is
conserved. Using this method, MOVES estimates starts by vehicle age without having the
default or input age distribution impact the estimated number of starts. However, the
StartsPerDayPerVehicle factor value stored in the StartsPerDayPerVehicle is intended to be
representative of the fleet-average starts, and we consider the age distributions when estimating
these fleet-average starts as discussed in the following subsections.
The StartsMonthAdjust table contains the monthAdjust factor which adjusts the starts per day to
reflect monthly variation in the number of engine starts (see Section 12.1.2.2 for light-duty and
Section 12.2.3.2 for heavy-duty). The month Adjustment is used as a raw multiplicative factor,
with values greater and less than one. Unlike the startsageadjustment table, MOVES does not
scale the month Adjustment factors to conserve starts for each model year. The average
monthAdjust values across all 12 months is one, so the annual number of starts estimated by
MOVES is consistent with the values in the StartsPerDayPerVehicle table. However, the
numbers of starts for a given month vary from the values in the StartsPerDayPerVehicle table
according to the month Adjustment factors.
The StartsHourFraction distributes the starts in a day to the hours of the day. The
allocationFraction value varies by source type, day type and hour of the day (see Section 12.1.2.1
for light-duty and Section 12.2.3.1 for heavy-duty).
The StartsOpModeDistribution table contains the distribution of engine start soak times for each
source type, age, day type and hour of the day (see Section 12.1.3 for light-duty and Section
12.2.4 for heavy-duty).
MOVES allows users to update the starts table if they have more representative data for their
purposes. MOVES provides additional start input tables and flexibilities for entering starts as
described in the Technical Guidance2.
The data inputs for motorcycles and motorhomes for the four start tables are discussed in Section
12.3.
As discussed in Section 13.4, the MOVES2014 SampleVehicleTrip table is still used in
MOVES4 for estimating evaporative emission activity. Thus, the number and time of starts used
to estimate start emissions is inconsistent with the trips and parking time used for evaporative
emissions in MOVES3 and later versions. While we think the impact of these inconsistencies is
small, we plan to address this conflict in future versions of MOVES.
12.1. Light-Duty Start Activity
For MOVES4, light-duty start activity are calculated from the same sample of vehicles from the
Verizon Telematics data discussed in Section 10.2.1.
12.1.1. Starts Per Day Per Vehicle
98
-------
The vehicle starts input format was substantially updated for MOVES3 to better allow inputs
based on the summary activity from large telematic datasets. In addition, the start inputs have
been updated to account for differences in start activity by month, day type, hour of day, and
vehicle age. To calculate the national average light-duty starts per day for MOVES, we
calculated the average starts from a set of telematics data obtained from Verizon (discussed in
Section 10.2.1) and adjusted this average to account for vehicle age. There were no additional
updates for MOVES4.
Table 12-1 below shows the starts per day per vehicle derived from the Verizon telematics
dataset for passenger cars (sourceTypelD 21) and passenger trucks (sourceTypelDs 31) and by
weekend days (daylD 2) and weekdays (daylD 5). We calculated a weighted-average starts per
day per vehicle from the Verizon dataset using the regional populations from each state sampled
(California New Jersey, Illinois, Georgia and Colorado) as documented in Table 10-2. The
values shown for passenger trucks (sourceTypelD 31) are also being used for light commercial
trucks (sourceTypelD 32).
Next, we calculated the average age of the vehicles in the Verizon dataset, using the model year
for each vehicle stored in the the vehicle metadata file from all the vehicles in the Verizon
dataset. We assumed the base year = 2015.6 (5 months of the Verizon dataset were in 2015 and 6
months occurred in 2016). We calculated the average vehicle age for each vehicle using
Equation 12-1.
Age = Base Year (2015.6) — Average Model Year
Equation 12-1
We then calculated the average age for each state included in the dataset and then calculated a
Verizon weighted-average shown in Table 12-1 using the regional populations used previously
(Table 10-2)
Table 12-1 National Average Starts per Day per Vehicle for Light-duty Vehicles based on Verizon
Telematics data per Vehicle
Source Type
Source-
TypelD
Verizon
weighted
average age
(years)
MOVES3
CY 2016
average age
(years)
Day of the
Week
Verizon
weighted
average starts
per vehicle per
day
Calculated
national
average starts
per day per
vehicle
Passenger
Cars
21
7.3
9.55
Weekend
3.36
3.13
Weekday
3.96
3.68
Passenger
Trucks
31
8.54
10.1
Weekend
3.49
3.32
Weekday
4.09
3.89
Light-
Commercial
Trucks
32
8.54
8.47
Weekend
3.49
3.52
Weekday
4.09
4.13
Next, we adjusted the starts for each vehicle age. We could not use the Verizon data directly
because it did not include a full range of vehicle ages. Instead, we used factors derived from the
mileage accumulation rates as discussed in the beginning of Section 12. We scaled the age
adjustment factors, such that at the average age (e.g., 7.3 years for passenger cars), the starts per
99
-------
day is equal to the average estimated from Verizon (e.g., 3.96 per day for weekdays for
passenger cars). The resulting starts per day by age for light-duty vehicles are presented in Figure
12-1. The starts per day for age 0 are higher than the Verizon weighted average starts per day,
while the starts at age 30 are substantially lower.
vehicle age
dayName — Weekdays ~ ' Weekend
Figure 12-1. Starts per day per vehicle by vehicle age calculated from the Verizon dataset and
MOVES ageAdjustment factors
We then used Equation 12-2 to calculate the MOVES age-weighted average starts per vehicle per
day using the starts per day per vehicle by age calculated in Figure 12-1 and the 2016 default age
distributions in MOVES. The purpose of this calculation is to adjust the average starts per day
from the Verizon sample to represent the nation, given that the national age distribution is
different than the age distribution of vehicles sampled in the Verizon datasets. As shown in Table
12-lFigure 12-1, the age in MOVES for CY 2016 passenger cars and passenger trucks is older
than in the Verizon dataset, while the average age of light commercial trucks is slightly older in
MOVES than in the Verizon dataset. We adjusted the average starts using the 2016 default age
distribution because the Verizon dataset was conducted in 2015-2016.
National Average Starts per Vehicle per Day
30
= ^ (Starts per Day Per Vehicle)age X ageFractionage Equation
age=0 12-2
Table 12-2 demonstrates the calculation of Equation 12-2 for passenger cars on weekdays. Table
12-1 shows the calculated national average starts per vehicle per day which are used in MOVES.
100
-------
The national average starts per vehicle day in 12-1 are used to estimate the average starts for
these source types and day types for all calendar years in MOVES.
Table 12-2. Calculation of the National Average Starts per Vehicle per Day for Passenger Cars
Vehicle
age
(agelD)
Starts per Day Per
Vehicle by Age
CY 2016 Age Distribution
(ageFraction)
Starts per Day
per Vehicle x
ageFraction
0
4.76
0.061
0.29
1
4.67
0.066
0.31
2
4.57
0.067
0.31
3
4.47
0.062
0.28
4
4.36
0.056
0.24
5
4.24
0.043
0.18
6
4.12
0.044
0.18
7
4.00
0.040
0.16
8
3.87
0.050
0.19
9
3.74
0.055
0.21
10
3.61
0.051
0.19
11
3.48
0.050
0.17
12
3.35
0.046
0.15
13
3.22
0.045
0.15
14
3.09
0.040
0.12
15
2.97
0.034
0.10
16
2.84
0.033
0.09
17
2.72
0.025
0.07
18
2.61
0.021
0.05
19
2.50
0.017
0.04
20
2.39
0.013
0.03
21
2.30
0.012
0.03
22
2.21
0.009
0.02
23
2.12
0.007
0.01
24
2.05
0.006
0.01
25
1.99
0.005
0.01
26
1.93
0.004
0.01
27
1.89
0.003
0.01
28
1.86
0.003
0.00
29
1.84
0.002
0.00
30
1.84
0.030
0.05
National Age-Weighted Average Starts per Vehicle per Day =
3.68
101
-------
12.1.2. Temporal Distributions
There were no updates to default temporal distibutions in MOVES4.
12.1.2.1. Hourly Distribution
The number of starts varies by hour of day. National values for the distribution of starts per day
by hour for passenger cars and light-duty trucks were calculated from the five-state Verizon
sample data described above in Section 10.2.1. The resulting national defaults for start
distribution in MOVES3 and later are illustrated in Figure 12-2. The start fraction values for
hourlDs 1 through 24 sum to 1.0 for a given sourceTypelD and daylD combination. The start
distribution curve in MOVES3 is much smoother than the start distribution based on the
SampleVehicleTrip table in MOVES2014 owing to the larger sample size of the Verizon data.
However, the overall trends are similar.
Jj 0.06
I
in
0.04
0 1 1 I r i 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Horn' ID
MOVE S3
Weekday
0.12
MOVE S3
Weekend
- - MOVES2014
Weekday
- - MOVES2014
Weekend
Figure 12-2 Start distribution for source type 21: MOVES3 derived from Verizon data vs.
MOVES2014
12.1.2.2. Monthly Distribution
For MOVES, we assume that the starts/mile is the same across months. We use the same
monthly distribution for starts in the MonthAdjust table as for VMT in the MonthVMTFraction
102
-------
table discussed in Section 13.1. Light-duty vehicles and all other source types (except
motorcycles) follow the same monthly variation, with slightly elevated starts during the summer
months, and corresponding decrease in starts in the winter.
12.1.3. Start Soak Distributions
As discussed in the beginning of Section 12, soak times are binned into different operating
modes, shown in Table 12-3. The fraction of starts assigned to each soak bin is the "soak
distribution." The light-duty soak distributions derived from Verizon differ by source type, day
type and hour of the day.
The engine soak time distributions for all source types are available in the OpModeDistribution
table of the default database (see Section 10.2) for the national default value calculation method
from the Verizon sample data and Table 10-1 for the number of sample trips used. Figure 12-3
illustrates the MOVES national default soak distribution for a weekday for passenger cars
(sourceTypelD 21).
Table 12-3 MOVES engine soak operating modes
opModelD
Description
101
Soak Time < 6 minutes
102
6 minutes <= Soak Time < 30 minutes
103
30 minutes <= Soak Time < 60 minutes
104
60 minutes <= Soak Time < 90 minutes
105
90 minutes <= Soak Time < 120 minutes
106
120 minutes <= Soak Time < 360 minutes
107
360 minutes <= Soak Time < 720 minutes
108
720 minutes <= Soak Time
103
-------
¦ 101 ¦ 102 ¦ 103 ¦ 104 ¦ 105 ¦ 106 ¦ 107 ¦ 108
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
HourlD
Figure 12-3 MOVES national average engine soak distribution for source type 21 and weekday
MOVES has the capability to model different soak distributions by vehicle age, but we are
currently using the same soak distribution across all vehicle ages. In general, as vehicles age, we
would expect less vehicle starts on average and a soak distribution to shift towards longer soak
times. Access to a large data covering a wider range of ages would help us better quantify this.
12.2. Heavy-Duty Start Activity
Like light-duty vehicles, starts from heavy-duty vehicles can be an important contributor to
emission inventories (e.g., THC and NOx). Additionally, heavy-duty diesel aftertreatment
technologies such as selective catalytic reduction (SCR) systems are also not fully active at
controlling NOx emissions below the catalyst light-off temperature.
Compared to light-duty vehicles, less data are available on heavy-duty vehicle start activity and
there are more subgroups of vehicles with potentially unique activity patterns. For example,
delivery vehicles have different start and soak patterns than long-haul trucks. For MOVES3 and
later, data that covers a wider range of heavy-duty vocations was available. The engine start
analysis below was applied to the same NREL Fleet DNA dataset used in the off-network idle
analysis discussed in Section 10. We are aware of the additional heavy-duty activity data
104
-------
collected by CE-CERT (also described in Section 10) and we expect to incorporate the data in a
future version of MOVES.
12.2.1. Heavy-Duty Engine Start Activity Data Processing
Starts were identified in the data using the data channel for engine speed, measured in
revolutions per minute (RPM). All the instances when the engine speed transitioned from zero to
greater than zero were considered new starts. If the data logger was installed but did not record
any activity, the start fraction is zero; however, if the data logger was not installed on a specific
day type, those values were denoted as "nan" (not-a-number) and were removed from the
analysis. The sum of the hourly fractions across all hours of the day is one.
The number of starts per day was calculated on a per vehicle basis and averaged equally across
all vehicles as shown in Equation 12-3. Thus, vehicles that start frequently or infrequently are
equally weighted in the average starts per day.
Similar to the off-network idle discussion in Section 10, we applied a sum-over-sum approach to
our hourly start fractions. Using Equation 12-4, a start fraction for each hour was calculated by
dividing the daily average starts-per-hour by the average starts-per-day for each combination of
sourceType and daylD. This sum-over-sum approach normalizes the recorded start activity by
the amount of time each vehicle was instrumented and weights the average start fraction towards
the vehicles with the most daily-average starts/
£ ^startSi
Starts Per Daysd =
n
Equation 12-3
i = Vehicle ID within a given sourceType, s
daysi = days within a given daylD, d, when vehicle, is instrumented
n = number ofVehiclelDs withing a given sourceType, s
/day Si)
2 ^startsh i^
Start fraction^ = ,starts I
v ' days J Equation 12-4
h = hour of the day
i = Vehicle ID within a given sourceType, s
r We evaluated several approaches for calculating average start and soak fractions using the Fleet DNA data. The
equations presented in this section are equivalent to the equations labeled "Method 3 'normalized sum over sum'" in
Appendix I. Appendix I includes an overview and a comparison of the calculation approaches.
105
-------
daysi = days within a given daylD, d, when vehicle, is instrumented
Vehicle soak is defined as the time difference between when an engine stops and the next time
the engine starts, as shown in Equation 12-5. The engine stop is defined as the time when engine
speed transitions from greater than zero to zero and engine start is defined as the time when
engine speed transitions from zero to greater than zero.
Every start was assigned a soak opModelD based on the definitions in Table 12-3 .s We then
calculated the average soak fraction, using a normalized sum-over-sum approach like we did for
the start fraction/ For each vehicle, hour and daytype, an average number of starts by soak length
was calculated by summing the number of starts matching each soak opModelD for each hourlD
and daylD and dividing by the number of unique days of measurement for that vehicle. The
hourly soak fraction distribution for each opModelD, sourceType and daylD was then calculated
using Equation 12-6. The sum of the eight opModelD soak fractions will equal 1.0 for each
combination of daylD, hourlD and sourceTypelD.
soak time = engine stop time — engine start time
Equation 12-5
Equation 12-6
h = hour of the day
i = Vehicle ID within a given sourceType, s
o = operating mode/soak length
daysi = days within a given daylD, d, vehicle, is instrumented
12.2.2. Starts Per Vehicle Per Day
s The first start identified for each vehicle was not considered when calculating soak time due to lack of a previous
recorded stop time.
106
-------
As seen in Table 10-3, several heavy-duty source types were not available in the Fleet DNA
database at the time of this analysis. In the future, we hope to update the analysis using both
Fleet DNA and CE-CERT datasets. In the interim, we assumed the start behavior of the missing
vehicles closely matched others. We chose to use the transit bus (sourceTypelD 42) to represent
other bus (sourceTypelD 41), used the single-unit short-haul data from the weekend
(sourceTypelD 52) to represent both the weekday and weekend data of the single-unit long-haul
trucks (sourceTypelD 53) and continued to use the same starts per day for motorhomes
(sourceTypelD 54) as in MOVES2014 (See Table 13-8).
None of the school buses (sourceTypelD 43) instrumented in the Fleet DNA dataset operated on
the weekend, so there is no data for that day ID. In Section 10, we applied the weekday school
bus off-network idle data for the weekend data, assuming the idle behavior of buses was similar
regardless of day type. We opted to retain the zero starts-per-day value for weekends (daylD 2),
assuming the frequency of school bus starts differed between weekends and weekdays.
Figure 12-4 and Figure 12-5 show the starts-per-day values for weekends and weekdays,
respectively.
Daily Starts, Weekends
Other Bus
Transit Bus
School Bus
Refuse
Single Unit,
Single Unit,
Combo,
Combo,
Truck
Short
Long
Short
Long
41
42
43
51
52
53
61
62
Vehicle Description / MOVES SourceType
Figure 12-4 Weekend starts per day for heavy-duty source types based on data from NREL's Fleet
DNA database
107
-------
Daily Starts, Weekdays
Other Bus
41
Transit Bus
School Bus
Refuse
Single Unit,
Single Unit,
Combo,
Combo,
Truck
Short
Long
Short
Long
42
43
51
52
53
61
62
Vehicle Description / MOVES SourceType
Figure 12-5 Weekday starts per day for heavy-duty source typesbased on data from NREL's Fleet
DNA database
As shown in Figure 12-5, the single-unit short-haul trucks (sourceTypelD 52) have significantly
more starts on weekdays. To understand this, we evaluated the impact of the vehicles' vocations
on their start behavior. Figure 12-6 shows that the parcel delivery vocation contributes many
more starts than the other vocations. While we did see differences in starts activity due to vehicle
vocation, we did not account for vocation differences when calculating starts for MOVES
because we could not identify a means to map the vocations represented in this dataset to a
nationally-representative population of vocations. We plan to revisit these estimates in future
versions of MOVES.
108
-------
0
o
"0
o
03
CL
Daily Starts by Vocation, Weekdays
Single-Unit Short-Haul Trucks
&
0
0
>
>
"0
"0
Q
Q
c
~o
0
0
c
0
—I
LL
CD
C
0
0
o
c
o
o
0
>
0
Q
CD
"O
"O
0
_c
CO
c
03
I-
0
C
03
Q_
O
o
Q_
E
Q
Vehicle Vocation
Figure 12-6 Vocation impacts on weekday starts-per-day for heavy-duty, single-unit short-haul
vehicles (sourceType 52) based on data from NREL's Fleet DNA database
As discussed in the beginning of Section 12, the startsPerDayPerVehicle factor stored in the
MOVES startsPerDayPerVehicle table represents the national average starts per day by
sourcetype and day of the week. We developed adjusted the starts measured from Fleet DNA to
be consistent with the national average age distribution in MOVES.
First, we adjusted the heavy-duty start values obtained from Fleet DNA using the ageAdjustment
factors (derived from the mileage accumulation rates in Section 6.2), by assuming that the Fleet
DNA starts are representative of vehicles at age 0. We assumed the Fleet DNA starts are
representative of activity at age 0 because:
• Of the vehicles with a recorded age (112 out of 415 vehicles in the Fleet DNA database),
most are younger than 3 years of age.
• NREL has informed the US EPA that vehicles chosen to be instrumented in the Fleet
DNA database tend to be active vehicles.
Figure 12-7 displays the resulting starts per day per vehicle across all ages for heavy-duty
vehicles calculated using these assumptions. Note that the starts per day per vehicle at age 0 are
equivalent to the average values reported from the Fleet DNA database, while the starts per day
for age 30 source types are significantly lower.
109
-------
— Weekdays
¦ ' Weekend
20
vehicle age
Figure 12-7. Starts per day per vehicle by vehicle age calculated from the Fleet DNA dataset and
MOVES ageAdjustment factors
Next, we calculated the national average starts per day for heavy-duty vehicles using Equation
12-2 with the starts per day per vehicle by age in Figure 12-7 and the 2014 heavy-duty default
age distributions in MOVES. We used the 2014 age distributions because it was the calendar
year with the most vehicle measurements in the FleetDNA dataset; the average age from the
MOVES 2014 age distributions are shown in Table 12-4. The resulting national average starts
per day per vehicle are also displayed in Table 12-4, which are significantly lower than the
average starts per day as measured from the FleetDNA database.
110
-------
Table 12-4 National Average Starts Per Day Per Vehicle for Heavy-duty Vehicles based on data
from NREL's Fleet DNA database
Source Type
SourceTypelD
MOVES3 CY
2014 Average
Age (years)
Day of the
Week
FleetDNA
Starts per
day per
vehicle
Calculated
national
average starts
per day per
vehicle
Other Bus
41
10.4
Weekend
2.64
1.93
Weekday
8.70
6.38
Transit Bus
42
6.5
Weekend
2.64
2.16
Weekday
8.70
7.13
School Bus
43
10.3
Weekend
0.00
0.00
Weekday
5.41
3.98
Refuse Truck
51
11.7
Weekend
0.16
0.10
Weekday
2.74
1.71
Single Unit Short-haul
52
11.8
Weekend
2.58
1.41
Weekday
36.26
19.86
Single Unit Long-haul
53
11.8
Weekend
2.58
1.33
Weekday
2.58
1.33
Combination Short-haul
61
12.0
Weekend
2.82
1.35
Weekday
12.25
5.87
Combination Long-haul
62
10.5
Weekend
0.60
0.37
Weekday
0.81
0.51
12.2.3. Temporal Distribution
12.2.3.1. Hourly Distribution
This section describes the temporal distribution of starts (also referred to as the start fractions)
for heavy-duty vehicles in MOVES based on data from NREL's Fleet DNA. As seen in Table
10-3, several heavy-duty source types were not available in Fleet DNA at the time of the
analysis. We expect data collected by CE-CERT to cover more of the source types and daylDs
and we hope to update the analysis using both Fleet DNA and CE-CERT datasets. In the
interim, we assumed the start behavior of the missing vehicles closely matched others, as
described when the figures are presented below.
The Fleet DNA dataset did not conain any information from buses meeting the MOVES
definition of "other buses." We assumed the start distributions from transit bus (sourceTypelD
42) represented other buses (sourceTypelDs 41) for both weekends and weekdays. Figure 12-8
shows the resulting starts distribution for these two bus types.
Ill
-------
Transit
0.12 -I
0.10 -
c 0.08 -
o
| 0.08 -
•e
m
W 0.04 -
0.02 -
0.00
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Hour
Figure 12-8 Start fraction temporal distribution for transit buses (sourceType 42) and other buses
(sourceType 41) based on data from NREL's Fleet DNA database
The school buses and refuse vehicles in the Fleet DNA dataset did not operate during certain
hours of the day. To avoid tables with zero values for those hours, we averaged adjacent blocks
of time, so those zeros were replaced with very small, nonzero values. The school buses in this
dataset did not operate from the hours of 8:00 PM to 6:00 AM on weekdays. We replaced the
zeros in those hours with 0.0008, which was the average start fraction from 6:00 PM through
6:00 AM, as depicted by the red boxes in Figure 12-9. The refuse trucks did not operate on
weekends from 7:00 PM to 4:00 AM and no data were collected in the hour between 6:00 AM
and 7:00 AM. The missing night hours' data were replaced with 0.01 (the average of the data for
7:00 PM to 4:00 AM) and the missing 6:00 AM data point was replaced with 0.07 (the average
of the 4:00 AM to 7:00 PM data). For each case, we renormalized the results once the zeros were
replaced, so the start fractions across all 24 hours of the day continued to sum to 1.0.
Buses (sourceType 42) & Other Buses (sourceType 41)
n I ~ Weekends
j ¦ Weekdays
112
-------
0.25
0 20
0.15
a 0.10
0 05
0.00
School Bus, Weekdays
1 2 3 4 5 6
7 8 9 10 11 12 13 14 15 18 17
Hour
18 19 20 21 22 23 24
0.30
0.25
0 20
0.15
0.10
0 05
0.00
Refuse Trucks, Weekends
1 2 3
4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Hour
Figure 12-9 Approach for renormalizing the start fraction results to avoid zeros in hours when no
data were collected. The zero-value start fraction in the red boxes were replaced with the average
start fractions from the range of hours in the red boxes. For refuse trucks, the 6:00 AM missing
datapoint was replaced with an average of the hours not outlined in red boxes.
Figure 12-10 and Figure 12-11 show the resulting start fractions by hour for school buses and
refuse trucks, respectively. Note that, in MOVES, school buses (sourceTypelD 43) have zero
starts per day on weekends, and none of the school buses instrumented in the Fleet DNA dataset
operated on the weekend, so we applied the start fractions from the weekday school bus data to
weekends.
113
-------
School Buses (sourceType 43)
0.25
0.20 -
° 0.15
O
m
i o.io
¦*-»
CO
0.05
0.00
Weekends
Weekdays
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Hour
Figure 12-10 Start fractions by hour for school buses (sourceType 43) based on data from NREL's
Fleet DNA database
Refuse Trucks (sourceType 51)
0.25
0.20
5 0.15
o
ra
e
m
0.10
0.05 -
0.00
Weekends
Weekdays
1 2 3 4 5 8 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Hour
Figure 12-11 Start fractions by hour for refuse trucks (sourceType 51) based on data from NREL's
Fleet DNA database
The Fleet DNA dataset did not contain any single-unit long-haul trucks (sourceType 53) so we
assumed their start distribution was similar to the single-unit short-haul trucks (sourceType 52).
114
-------
Figure 12-12 shows the start distribution applied to both single-unit truck types for weekends and
weekdays.
0.12
0,10
c 0.08
.2
I 0.06
t
w 0.04
0.02
0.00
Single-Unit Trucks (sourceTypes 52 & 53)
i Weekends
I Weekdays
1 2 3 4 5 8 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Hour
Figure 12-12 Start fraction by hour for single-unit short-haul trucks (sourceType 52) and single-
unit long-haul trucks (sourceType 53) based on data from NREL's Fleet DNA database
Additional consideration was given before using the FleetDNA data to populate the hourly
fraction tables. The Fleet DNA dataset contained many combination long-haul trucks
(sourceType 62) and NREL staff are confident that the average idle data described in Section 10
and average starts-per-day data described earlier in this section represent the activity of
combination long-haul trucks. However, NREL believes there was a time zone-related logging
error when the data were reported to NREL. Most of the data from the 131 combination long-
haul trucks in the Fleet DNA dataset were collected by an industry partner and NREL was unable
to accurately confirm which time zone the activity was recorded in. The data consistently
showed that the trucks operated more at night with hotelling during the day, which conflicted
with other data sources as discussed in Section 13.5. Figure 12-13 shows the original Fleet DNA
data for long-haul combination trucks that was not applied in MOVES due to the possible time
misalignment. Instead, we assumed the start distribution from short-haul combination trucks was
a better representation. Figure 12-14 shows the starts distribution that was applied in MOVES for
both short- (sourceType 61) and long-haul combination trucks (sourceType 62).
115
-------
Suspected Time Misalignment for
Combination Long-Haui Trucks (sourceType 62)
Hour
Figure 12-13 Start fractions from on NREL's Fleet DNA database that were not applied for
combination long-haul trucks (sourceType 62); we suspect a time misalignment in the the data
Combination Trucks (sourceTypes 61 & 62)
Weekends
Weekdays
1 2 3 4 5 8 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Hour
Figure 12-14 Start fraction temporal distribution for combination short-haul trucks (sourceType
61) and combination long-haul trucks (sourceType 62) based on data from NREL's Fleet DNA
database
116
-------
12.2.3.2. Monthly Distribution
In MOVES, we assume that the starts/mile is the same across all months. We use the same
monthly distribution for starts in the MonthAdjust table as for the VMT in the
MonthVMTFraction table discussed in Section 13.1. Heavy-duty vehicles follow the same
monthly variation as light-duty vehicles, with slightly elevated starts during the summer months,
and corresponding decrease in starts in the winter.
12.2.4. Start Soak Distributions
This section describes the heavy-duty vehicles' soak distributions in MOVES. As seen in Table
10-3, several heavy-duty source types were not available in the Fleet DNA database at the time
of this analysis. We hope to update the analysis using both Fleet DNA and CE-CERT datasets. In
the interim, we applied several assumptions for the soak behavior as described below.
The Fleet DNA dataset did not contain any information from buses meeting the MOVES
definition of "other buses". We assumed the start distributions from transit bus (sourceTypelD
42) represented other buses (sourceTypelDs 41) for both weekends and weekdays. Figure 12-15
and Figure 12-16 show the resulting starts distributions for these two bus types on weekends and
weekdays, respectively.
Transit & Other Buses (sourceTypes 42 & 41), Weekends
100%
90%
80%
c
70%
o
o
60%
0!
i—
LL
50%
(0
o
40%
CO
30%
20%
10%
0%
1 2 3 4 5 6 7 8
9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Hour
Figure 12-15 Weekend soak distributions transit buses (sourceType 42) and other buses
(sourceType 41) based on data from NREL's Fleet DNA database
117
-------
Transit & Other Buses (sourceTypes 42 & 41), Weekdays
100%
90%
80%
c 70%
% 60%
CO
it 50%
§ 40%
m 30%
20%
10%
0%
OpMode
¦ 108
¦ 107
«106
¦ 105
¦ 104
¦i 103
¦ 102
¦ 101
1 2 3 4 5 8 7 8
9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Hour
Figure 12-16 Weekday soak distributions transit buses (sourceType 42) and other buses
(sourceType 41) based on data from NREL's Fleet DNA database
As mentioned previously in the start distribution discussion, school buses (sourceTypelD 43) in
this dataset did not operate from 8:00 PM to 6:00 AM on weekdays. For soak distribution, we
replaced these hours with the average hourly soaks over the period from 6:00 PM through 6:00
AM. The school buses in the Fleet DNA dataset did not operate on the weekends, so we applied
the weekday school bus soak distribution to weekends. Figure 12-17 shows the soak distribution
applied to school buses for both weekends and weekdays.
School Buses (sourceType 43), Weekdays & Weekends
100% hhhhhhhh—bhhi
90% ¦¦¦¦¦¦¦¦¦¦¦¦
HHIlllHl OpMode
80% ¦¦¦¦¦¦¦¦¦¦¦¦
¦¦¦¦¦¦¦¦¦¦¦¦ ¦ 108
c
70% ¦¦¦¦¦¦¦¦¦¦¦¦
llllllllllll "107
a
~*»
Q
60% ¦¦¦¦¦¦¦¦¦¦¦¦
¦hIIIhhihIH 11106
(B
¦¦¦¦¦¦¦¦¦¦¦¦ *105
LL
50% ¦¦¦¦¦¦¦¦¦¦¦¦
a
o
40% ¦¦¦¦¦¦¦¦¦¦¦¦
llllllllllll ¦ 104
w
30% ¦¦¦¦¦¦¦¦¦¦¦¦
llllllllllll >103
20% ¦¦¦¦¦¦¦¦¦¦¦¦
llllllllllll B1°2
10% ¦¦¦¦¦¦¦¦¦¦¦¦
llllllllllll "101
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 18 17 18 19 20 21 22 23 24
Hour
Figure 12-17 Weekend and weekday soak distributions for school buses (sourceType 43) based on
data from NREL's Fleet DNA database
118
-------
Refuse trucks (sourceType 51) in the Fleet DNA dataset did not operate on weekends from 7:00
PM to 4:00 AM and no data were collected in the hour between 6:00 AM and 7:00 AM. The
missing night hours' data were replaced with the average hourly soaks over the period from 7:00
PM to 4:00 AM and the missing 6:00 AM data point was replaced with the average hourly soaks
from 4:00 AM to 7:00 PM. Figure 12-18 and Figure 12-19 show the soak distributions for refuse
trucks on weekends and weekdays, respectively.
100% -
90%
80% I
70%
60%
50%
Q
a
LL
| 40%
m
30%
20%
10%
0%
I
Refuse Trucks (sourceType 51), Weekends
OpMode
¦ 108
¦ 107
¦ 106
¦ 105
¦ 104
¦ 103
¦ 102
¦ 101
1 2 3 4 5 8
7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Hour
Figure 12-18 Weekend soak distributions for refuse trucks (sourceType 51) based on data from
NREL's Fleet DNA database
Refuse Trucks (sourceType 51), Weekdays
100%
90%
OpMode
80%
¦108
c
7°% ¦¦¦HHHHbIHHI
¦111 1,107
.2
o
60% ¦¦¦¦¦¦¦¦¦¦¦¦
1,106
E
UL
5°%
¦HHHHhBbHHHH * 1 os
m
o
40% HHHHHHlfllHHI
bbhbhBBhhhhB ¦104
co
30% HhHHB
"103
20%
¦¦¦¦¦¦¦¦¦¦¦¦ "102
io% BIIbhb
¦¦¦¦¦¦¦¦¦¦¦¦ "101
0% ¦¦¦¦¦¦¦¦¦¦¦¦»
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Hour
Figure 12-19 Weekday soak distributions for refuse trucks (sourceType 51) based on data from
NREL's Fleet DNA database
The Fleet DNA dataset did not contain any single-unit long-haul trucks (sourceType 53) and we
assumed their soak distribution was similar to the single-unit short-haul trucks (sourceType 52).
119
-------
Figure 12-20 and Figure 12-21 show the soak distributions applied to both single-unit truck types
for weekends and weekdays, respectively.
Single-Unit Trucks (sourceTypes 52 & 53), Weekends
90% OpMode
80% HSllflHimfllllllHIIIHI H08
c 70% HBHHHHbHHbHHHHHHHhHHHHHH "107
o
60% *108
I 50% "105
§ 4o% BUlBimHMHHBlBHH 1104
m 30% "103
20% ¦¦hhBBhBhBBBHBBB ¦102
10%
0%
1101
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Hour
Figure 12-20 Weekend soak distributions for single-unit short-haul trucks (sourceType 52) and
single-unit long-haul trucks (sourceType 53) based on data from NREL's Fleet DNA database
Single-Unit Trucks (sourceTypes 52 & 53), Weekdays
100%
90%
80%
70%
60%
50%
40%
30%
20%
10%
0%
1 2 3 4 5 6 7 8
9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Hour
OpMode
¦ 108
¦ 107
i»106
¦ 105
¦ 104
¦ 103
¦ 102
¦ 101
Figure 12-21 Weekday soak distributions for single-unit short-haul trucks (sourceType 52) and
single-unit long-haul trucks (sourceType 53) based on data from NREL's Fleet DNA database
Figure 12-22 shows the weekend soak distribution that was applied in MOVES for combination
short-haul trucks (sourceType 61). Figure 12-23 shows the weekday soak distribution for the
same vehicles.
120
-------
Combination Short-Haul Trucks (sourceType 61), Weekends
100%
90%
80%
70%
60%
50%
c
,9
"3
2
y.
g 40%
m
30%
20%
10%
0%
OpMode
¦ 108
¦ 107
¦ 108
¦ 105
¦ 104
¦ 103
¦ 102
¦ 101
1 2 3 4 5 0 7
8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Hour
Figure 12-22 Weekend soak distributions for combination short-haul trucks (sourceType 61) based
on data from NREL's Fleet DNA database
Combination Short-Haul Trucks (sourceType 61), Weekdays
100%
90%
OpMode
80% HHHH
¦ 108
e
70% HHHHHHlfllHHIHHBHHHHHHHHI
¦ 107
o
¦J3
a
60% ¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦
¦ 106
s
LL
50% ¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦
105
a
o
40% ¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦
¦ 104
m
30% ¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦
«103
20% ¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦
:::: 102
10% ¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦
101
0% ¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦
1 2 3 4 5 6 7 8
9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Hour
Figure 12-23 Weekday soak distributions for combination short-haul trucks (sourceType 61) based
on data from NREL's Fleet DNA database
As mentioned in the start distribution section, we believe there was a time zone-related logging
error for many of the combination long-haul trucks (sourceType 62) in the Fleet DNA dataset.
Consequently, we opted to apply the same average hourly soak distribution from each day type
across all hours of the day. The soak distributions applied for combination long-haul trucks on
weekends are shown in Figure 12-24. The weekday soak distrbituions are in Figure 12-25.
121
-------
Combination Long-Haul Trucks (sourceType 82), Weekends
100%
90% ¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦
OpMode
80% ||||||||||||||||||||||||
¦ 108
c
70% ||||||||||||||||||||||||
¦ 107
o
"¦+-«
o
60% ||||||||||||||||||||||||
it 106
2
UL
50% ||||||||||||||||||||||||
¦ 105
(1
o
40% ||||||||||||||||||||||||
¦ 104
(/)
30% ¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦
¦ 103
20% ¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦
= 102
10% ¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦
101
0% ¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦¦
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Hour
Figure 12-24 Weekend soak distributions for combination long-haul trucks (sourceType 62)
applying the average hourly soak distribution from NREL's Fleet DNA database across all hours of
the day
Combination Long-Hau! Trucks (sourceType 62), Weekdays
100% hhhhhhhhhhhhi
¦¦¦¦¦¦¦¦¦¦¦¦
90% ¦¦¦¦¦¦¦¦¦¦¦¦
llllllllllll OpMode
80% ¦¦¦¦¦¦¦¦¦¦¦¦
¦¦¦¦¦¦¦¦¦¦¦I "108
c
70% ¦¦¦¦¦¦¦¦¦¦¦¦
"107
.o
o
60% ¦¦¦¦¦¦¦¦¦¦¦¦
llllllllllll >106
5
u_
50% ¦¦¦¦¦¦¦¦¦¦¦¦
¦105
a
o
40% ¦¦¦¦¦¦¦¦¦¦¦¦
¦¦llllllllll ¦ 104
m
30% ¦¦¦¦¦¦¦¦¦¦¦¦
llllllllllll 11103
20% ¦¦¦¦¦¦¦¦¦¦¦¦
llllllllllll "102
10% HHHHHHHfllHHIj
llllllllllll ¦ 101
o%
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Hour
Figure 12-25 Weekday soak distributions for combination long-haul trucks (sourceType 62)
applying the average hourly soak distribution from NREL's Fleet DNA database across all hours of
the day
As mentioned for light-duty vehicles, MOVES has the capability to model different soak
distributions by vehicle age. We are currently using the heavy-duty soak distribution estimated
from Fleet DNA across all vehicle ages. In general, as vehicles age, we would expect fewer
vehicle starts and a soak distribution shift towards longer soak times. However, the available data
122
-------
on heavy-duty vehicles at older ages is much more limited. Future work could evaluate the
dependency of soak distributions on vehicle age.
12.3. Motorcycle and Motorhome Starts
Motorcycle and motorhome data are not captured in the Verizon and Fleet DNA datasets used to
update the other source types. The data we used to model starts from motorcycles and
motorhomes is outlined in Table 12-5.
Table 12-5. Motorcycle and Motorhome Start Data
MOVES Table
Motorcycles (SourceTypelD
11)
Motorhomes
(SourceTypelD 54)
startsPerDayPerVehicle
Starts from Table 13-8
adjusted to represent CY
2014 age distribution
Table 13-8 adjusted to
represent CY 2014 age
distribution
startsHourFraction
Passenger Cars (21)
Passenger Trucks (31)
startsOpmodeDistribution
(soaks)
Passenger Cars (21)
Passenger Trucks (31)
startsMonth Ad] ust
Table 13-2
Table 13-1
For national average starts per day per vehicle, we used the starts per day estimated in
MOVES2014 as presented in Table 13-8. Because these start rates were calculated from
instrumented vehicle data, we assume these start rates are respresentative of active, age 0
vehicles. We thus followed similar steps to calculate national average starts per day per vehicle
as was conducted for heavy-duty vehicles above which used the same assumptions. We
calculated starts per day by vehicle age by applying the ageAdjustment factors to the start data as
shown in Figure 12-26.
(D
o 1-5
JZ
>
«S 1.0
CL
>>
TO
Q
£0.5
CL
en
r
ra ^ „
m 0.0
Motorcycle
1
1
1
1
%
\
\
\s
V
- ^ ^
i
i i
0.6
0.4
0.2
Motor Home
10
20
0.0-
30 0
vehicle age
10
20
30
dayName
Weekdays
Weekend
Figure 12-26. Starts per day per vehicle by vehicle age for motorcycles and motorhomes calculating
using MOVES ageAdjustment factors
We then calculated the national average starts per day for motorcycles and motorhomes using
Equation 12-2 with the starts per day per vehicle by age in Figure 12-26, Figure 12-7 and the
123
-------
2014 heavy-duty default age distributions in MOVES3. We used the 2014 age distributions
because it is the year from which the source type age distributions are based in MOVES3
(Section 6.1). The resulting national average starts per day per vehicle for motorcycles and
motorhomes are displayed in Table 12-4, which are significantly lower than the age zero start
rate.
Table 12-6. National Average Starts Per Day Per Vehicle for Motorcycles and Motor
tiomes
Source
Type
SourceTypelD
MOVES3
CY 2014
Average
Age
(years)
Day of
the
Week
Starts per
day per
vehicle at
age 0
Calculated
national average
starts per day per
vehicle
Motorcycle
11
10.5
Weekend
1.52
0.37
Weekday
0.45
0.11
Motorhome
54
15.0
Weekend
0.57
0.48
Weekday
0.57
0.47
The hourly distribution of starts (stored in the startsHourFraction table) for motorcycles is
assumed to be the same as for passenger cars. For motorhomes, the hourly distribution of starts is
assumed to be the same as for passenger trucks, both of which are estimated from the Verizon
database.Motorcyles soak distributions are the same as passenger cars and motorhomes are the
same as passenger trucks. We assume that the montly pattern of starts (stored in the
startsMonthAdjust table) follows the same pattern as VMT as described in in Section 13.1. Thus,
motorcycle have a pronounced increase in starts during summer months. Motorhomes starts
follow the monthly variation of all other source types, which are only slightly elevated during the
summer months.
124
-------
13.Temporal Distributions
MOVES is designed to estimate emissions for every hour of every day type in every month of
the year. This section describes how VMT is allocated to months of the year, the two day types
and to hours of the day. This section also addresses how sample vehicle trip data is used to
determine and allocate evaporative soak periods to hours of the day. Finally, this section
discusses the derivation of the allocation of hotelling activity for long-haul combination trucks.
See also the discussion of temporal allocations for off-network idle in Section 10 and for engine
starts in Section 12.
In MOVES, VMT are provided in terms of annual miles. These miles are allocated to months,
days and hours using allocation factors, either using default values or values provided by users.
Default values for most temporal VMT allocations are derived from a 1996 report from the
Office of Highway Information Management (OHIM).71 The report describes analysis of a
sample of 5,000 continuous traffic counters distributed throughout the United States. EPA
obtained the data from the report and used it to generate the VMT temporal distribution inputs in
the form needed for MOVES. This information has not been updated for MOVES3 or
MOVES4.
The OHIM report does not specify VMT by vehicle type, so MOVES uses the same values for
all source types, except motorcycles, as described below.
In MOVES, daily truck hotelling hours are calculated as proportional to VMT on restricted
access road types for long-haul combination trucks. However, the hours of hotelling activity in
each hour of the day are not proportional to VMT, as described in Section 13.5.
The temporal distributions for engine start are described in Section 12.1.2. These values are
stored in the StartsMonthAdjust and StartsHourFraction Tables. However, we have not yet
updated the data used to estimate vehicle parking time and associated evaporative emissions. As
in MOVES2014, the engine soak (parked) distributions for evaporative emissions are calculated
from vehicle activity data stored in the SampleVehicleDay and SampleVehicleTrip tables of the
MOVES database. The inconsistency between the updated activity defaults now being used to
calculate engine starts and soaks and the older defaults that MOVES will continue using for
evaporative emissions is not ideal. We plan to resolve this inconsistency in future versions of
MOVES when the code used for the calculation of evaporative emissions is updated.
The temporal allocation of vehicle activity will vary from location to location and EPA guidance
encourages states and local areas to determine their own local vehicle activity parameters for use
with MOVES. When EPA runs MOVES for air quality modeling purposes, we use local activity
data including temporal allocations as explained in EPA technical support documents.46 EPA
plans to update the temporal allocations currently in MOVES using more recent data sources,
such as telematics data, as they become available.
13.1. VMT Distribution by Month of the Year
125
-------
In MOVES, when VMT is entered as an annual value, it is allocated to months of the year using
the factors in the MonthVMTFraction table. For MOVES, we modified the data from the OHIM
report to fit MOVES specifications. Table 13-1 shows VMT/day taken from the OHIM report
(Figure 2.2.1 "Travel by Month, 1970-1995"), normalized to one for January. The VMT per day
in Table 13-1 were used to calculate the fraction of total annual VMT in each month using the
number of days in each month, assuming a non-leap year (365 days). These monthly VMT
allocations are used for all source types, except motorcycles, as described below.
Table 13-1 MonthVMTFraction
Month
Normalized
VMT/day
MOVES
Distribution
January
1.0000
0.0731
February
1.0560
0.0697
March
1.1183
0.0817
April
1.1636
0.0823
May
1.1973
0.0875
June
1.2480
0.0883
July
1.2632
0.0923
August
1.2784
0.0934
September
1.1973
0.0847
October
1.1838
0.0865
November
1.1343
0.0802
December
1.0975
0.0802
Sum
1.0000
FHWA does not report monthly VMT information by vehicle classification. However, it is clear
that in many regions of the United States, motorcycles are driven much less frequently in the
winter months. For MOVES, an allocation for motorcycles was derived using monthly national
counts of fatal motorcycle crashes from the National Highway Traffic Safety Administration
Fatality Analysis System for 2010.72 This allocation increases motorcycle activity (and
emissions) in the summer months and decreases them in the winter compared to the other source
types. These default values in Table 13-2 for motorcycles are only a national average and do not
reflect the strong regional differences that would be expected due to climate.
126
-------
Table 13-2 Mont
lVMTFraction for motorcycles
Month
Month ID
Distribution
January
1
0.0262
February
2
0.0237
March
3
0.0583
April
4
0.1007
May
5
0.1194
June
6
0.1269
July
7
0.1333
August
8
0.1349
September
9
0.1132
October
10
0.0950
November
11
0.0442
December
12
0.0242
Sum
1.0000
The monthly allocation of VMT will vary from location to location and EPA guidance
encourages states and local areas to determine their own monthly VMT allocation factors for use
with MOVES.
13.2. VMT Distribution by Type of Day
The distributions in the DayVMTFraction table divide the weekly VMT estimates into the two
MOVES day types. The OHIM report provides VMT percentage values for each day and hour
of a typical week for urban and rural roadway types for various regions of the United States.
Since the day-of-the-week data obtained from the OHIM report is not disaggregated by month or
source type, the same values were used for every month and for every source type. MOVES uses
the 1995 data displayed in Figure 2.3.2 of the OHIM report.71
The DayVMTFraction needed for MOVES has only two categories: weekdays (Monday,
Tuesday, Wednesday, Thursday and Friday) and weekend (Saturday and Sunday) days. The
OHIM reported percentages for each day of the week were summed in their respective categories
and converted to fractions, as shown in Table 13-3. The OHIM report explains that data for
"3am" refers to data collected from 3am to 4am. Thus, the data labeled "midnight" was summed
with the upcoming day.
Table 13-3 DayVMTFractions
Fraction
Rural
Urban
Weekday
0.72118
0.762365
Weekend
0.27882
0.237635
Sum
1.00000
1.000000
127
-------
We assigned the "rural" fractions to the rural road types (roadTypelDs 2 and 3) and the "urban"
fractions to the urban road types (roadTypelDs 4 and 5). The fraction of weekly VMT reported
for a single weekday in MOVES will be one-fifth of the weekday fraction and the fraction of
weekly VMT for a single weekend day will be one-half the weekend fraction.
The day type allocation of VMT will vary from location to location and EPA guidance
encourages states and local areas to determine their own VMT allocation factors for use with
MOVES.
13.3. VMT Distribution by Hour of the Day
HourVMTFraction uses the same data as for DayVMTFraction. We converted the OHIM
report's VMT data by hour of the day in each day type to percent of day by dividing by the total
VMT for each day type, as described for the DayVMTFraction. There are separate sets of
HourVMTFractions for "urban" and "rural" road types, but unrestricted and unrestricted roads
use the same HourVMTFraction distributions. All source types use the same HourVMTFraction
distributions and Table 13-4 and Figure 13-1 summarize these default values.
128
-------
Table 13-4 MOV]
LS distribution of VMT by hour of the day
hourlD
Description
Urban
Rural
Weekday
Weekend
Weekday
Weekend
1
Hour beginning at 12:00 midnight
0.00986
0.02147
0.01077
0.01642
2
Hour beginning at 1:00 AM
0.00627
0.01444
0.00764
0.01119
3
Hour beginning at 2:00 AM
0.00506
0.01097
0.00655
0.00854
4
Hour beginning at 3:00 AM
0.00467
0.00749
0.00663
0.00679
5
Hour beginning at 4:00 AM
0.00699
0.00684
0.00954
0.00722
6
Hour beginning at 5:00 AM
0.01849
0.01036
0.02006
0.01076
7
Hour beginning at 6:00 AM
0.04596
0.01843
0.04103
0.01768
8
Hour beginning at 7:00 AM
0.06964
0.02681
0.05797
0.02688
9
Hour beginning at 8:00 AM
0.06083
0.03639
0.05347
0.03866
10
Hour beginning at 9:00 AM
0.05029
0.04754
0.05255
0.05224
11
Hour beginning at 10:00 AM
0.04994
0.05747
0.05506
0.06317
12
Hour beginning at 11:00 AM
0.05437
0.06508
0.05767
0.06994
13
Hour beginning at 12:00 Noon
0.05765
0.07132
0.05914
0.07293
14
Hour beginning at 1:00 PM
0.05803
0.07149
0.06080
0.07312
15
Hour beginning at 2:00 PM
0.06226
0.07172
0.06530
0.07362
16
Hour beginning at 3:00 PM
0.07100
0.07201
0.07261
0.07446
17
Hour beginning at 4:00 PM
0.07697
0.07115
0.07738
0.07422
18
Hour beginning at 5:00 PM
0.07743
0.06789
0.07548
0.07001
19
Hour beginning at 6:00 PM
0.05978
0.06177
0.05871
0.06140
20
Hour beginning at 7:00 PM
0.04439
0.05169
0.04399
0.05050
21
Hour beginning at 8:00 PM
0.03545
0.04287
0.03573
0.04121
22
Hour beginning at 9:00 PM
0.03182
0.03803
0.03074
0.03364
23
Hour beginning at 10:00 PM
0.02494
0.03221
0.02385
0.02622
24
Hour beginning at 11:00 PM
0.01791
0.02457
0.01732
0.01917
Sum of All Fractions
1.00000
1.00000
1.00000
1.00000
129
-------
Hour of the Day
Figure 13-1 Hourly VMT fractions by day type and road type
The allocation of VMT to the hours of the day will vary from location to location and EPA
guidance encourages states and local areas to determine their own VMT allocation factors for use
with MOVES. For example, an analysis by CRC has made county specific hourly VMT
distributions available for calendar year 2014.46
13.4. Parking Activity
To properly estimate evaporative fuel vapor losses, it is important to estimate the number of
starts by time of day and the duration of time between vehicle trips. The time between trips with
the engine off is referred to as "soak time". To determine typical patterns of trip starts and ends,
MOVES uses information from instrumented vehicles. This data is stored in two tables in the
MOVES default database, as discussed below. Unlike the information used to determine exhaust
start emissions (see Section 12.1.2), these tables are unchanged from MOVES2014, since
130
-------
updating the activity data in these tables was beyond the scope of more recent MOVES updates.1
Note that the activity described below is applied only to gasoline vehicles since diesel
evaporative emissions (other than refueling spillage) are expected to be negligible and are not
calculated by MOVES.
The first table, SampleVehicleDay, lists a sample population of vehicles, each with an identifier
(vehID), an indication of vehicle type (sourceTypelD) and an indication (daylD) of whether the
vehicle is part of the weekend or weekday vehicle population. Some vehicles were added to this
table to increase the number of vehicles in each day which do not take any trips to better match a
more representative study of vehicle activity in Georgia.73 This analysis is described in greater
detail in the report describing evaporative emissions in MOVES.74
The second table, SampleVehicleTrip, lists the trips in a day made by each of the vehicles in the
SampleVehicleDay table. It records the vehID, daylD, a trip number (tripID), the hour of the trip
(hourlD), the trip number of the prior trip (priorTripID) and the times at which the engine was
turned on and off for the trip. The keyOnTime and keyOffTime are recorded in minutes since
midnight of the day of the trip. 439 trips (about 1.1 percent) were added to this table to assure
that at least one trip is done by a vehicle from each source type in each hour of the day to assure
that emission rates will be calculated in each hour. Table 13-5 shows the resulting number of
vehicles in the SampleVehicleDay table with trip information.
Table 13-5 SampleVehicleDay table
Source Type
Number of Records
sourceTypelD
Description
Weekday (daylD 5)
Weekend (daylD 2)
11
Motorcycle
2214
983
21
Passenger Car
821
347
31
Passenger Truck
834
371
32
Light Commercial Truck
773
345
41
Other Bus
190
73
42
Transit Bus
110
14
43
School Bus
136
59
51
Refuse Truck
205
65
52
Single-Unit Short-Haul Truck
112
58
53
Single-Unit Long-Haul Truck
123
50
54
Motor Home
5431
2170
61
Combination Short-Haul Truck
130
52
62
Combination Long-Haul Truck
122
49
4 Updating the sampleVehicleTrip table to use the data also used to update starts is not straightforward. For example,
the current SampleVehicleTrip table used for evaporative emissions currently contains 37,216 vehicle trips, whereas
the Verizon light-duty database used for starts contains millions of trips. Another approach would be to change the
MOVES algorithm to calculate evaporative emissions based on summarized trip information (as was done for start
emissions) but creating this new algorithm would be a significant programing effort.
131
-------
To account for overnight soaks, many first trips reference a prior trip with a null value for
keyOnTime and a negative value for keyOffTime. The SampleVehicleDay table also includes
some vehicles that have no trips in the SampleVehicleTrip table to account for vehicles that sit
for one or more days without any driving.
The data and processing algorithms used to populate these tables are detailed in two contractor
reports.75'76 The data come from a variety of instrumented vehicle studies, summarized in Table
13-6. These data were cleaned, adjusted, sampled and weighted to develop a distribution
intended to represent average urban vehicle activity.
Table 13-6 Source data for sample vehicle trip information
Study
Study Area
Study
Years
Vehicle Types
Vehicle
Count
3-City FTP
Study
Atlanta, GA; Baltimore, MD;
Spokane, WA
1992
Passenger cars & trucks
321
Minneapolis
Minneapolis/St. Paul, MN
2004-
2005
Passenger cars & trucks
133
Knoxville
Knoxville, TN
2000-
2001
Passenger cars & trucks
377
Las Vegas
Las Vegas, NV
2004-
2005
Passenger cars & trucks
350
Battelle
California, statewide
1997-
1998
Heavy-duty trucks
120
TxDOT
Houston, TX
2002
Diesel dump trucks
4
For vehicle classes that were not represented in the available data, the contractor synthesized
trips using trip-per-operating hour information from the EPA MOBILE677 model and soak time
and time-of-day information from source types that did have data. The application of synthetic
trips is summarized in Table 13-7.
Table 13-7 Synthesis of sample vehicles for source types lacking data
Source Type
Based on
Direct Data?
Synthesized From
Motorcycles
No
Passenger Cars
Passenger Cars
Yes
n/a
Passenger Trucks
Yes
n/a
Light Commercial Trucks
No
Passenger Trucks
Other Buses
No
Combination Long-Haul Trucks
Transit Buses
No
Single-Unit Short-Haul Trucks
School Buses
No
Single-Unit Short-Haul Trucks
Refuse Trucks
No
Combination Short-Haul Trucks
Single-Unit Short-Haul Trucks
Yes
n/a
Single-Unit Long-Haul Trucks
No
Combination Long-Haul Trucks
Motor Homes
No
Passenger Cars
Combination Short-Haul trucks
Yes
n/a
Combination Long-Haul trucks
Yes
n/a
132
-------
The resulting trip-per-day estimates are summarized in Table 13-8. The same estimate for trips
per day is used for all ages of vehicles in any calendar year.
Table 13-8 Trip per day by source type used for evaporative emissions activity
Source Type
Weekday
Weekend
Motorcycles
0.45
1.52
Passenger Cars
5.38
4.99
Passenger Trucks
5.58
4.7
Light Commercial Trucks
6.02
5.06
Other Buses
2.88
1.19
Transit Buses
4.75
4.93
School Buses
5.88
1.64
Refuse Trucks
3.85
1.28
Single-Unit Short-Haul Trucks
7.14
1.67
Single-Unit Long-Haul Trucks
4.45
1.74
Motor Homes
0.57
0.57
Combination Short-Haul trucks
6.07
1.6
Combination Long-Haul trucks
4.29
1.29
The trip activity used for determination of emissions resulting from parked vehicles differs from
the activity used to determine engine start emissions, described in Section 12. Ideally, trips,
engine soak periods, and parking hours would be consistent. However, since both approaches,
although different, are describing the same vehicle activity, the differences are not expected to
have a negative impact on total emission estimates.
Knowing the sequence of starts for each vehicle in the sampleVehicleTrip table allows MOVES
to calculate the length and time of day when each soak occurs. Using this information, the
distribution of soak times in each hour of the day can be calculated for use in the determination
of evaporative emissions from parked vehicles.
The evaporative vapor losses from gasoline vehicle fuel tanks are affected by many factors,
including the number of hours a vehicle is parked without an engine start, referred to as engine
soak time. Most modern gasoline vehicles are equipped with emission control systems designed
to capture most evaporative vapor losses and store them. These stored vapors are then burned in
the engine once the vehicle is operated. However, the vehicle storage capacity for evaporative
vapors is limited and multiple days of parking (diurnals) will overload the storage capacity of
these systems, resulting in larger losses of evaporative vapors in subsequent days.
The detailed description of the calculation for the number of vehicles that have been soaking for
more than a day and the amount of time that the vehicles have been soaking can be found in the
MOVES technical report on evaporative emissions.76
Note, the MOVES County Data Manager allows users to specify the number of engine starts in
each month, day type and hour of the day, as well as by source type and vehicle age. These user
133
-------
inputs override the default start activity values provided by MOVES (see Section 12). However,
these user inputs will not update the soak times used in the calculations for evaporative
emissions, which rely solely on the sample trip data.
13.5. Hourly Hotelling Activity
In Section 11, we updated the hotelling activity rate based on instrumented truck data from the
NREL Fleet DNA database. However, this dataset was not deemed appropriate for updating the
hourly hotelling activity." As discussed below, we found that the hotelling hourly distribution
assumed in MOVES2014 compared well to other datasets. Thus, the hourly hotelling activity is
unchanged from MOVES2014.
To derive the hotelling hour distribution in MOVES, we used the assumption that
the hotelling hours in each day should not directly correlate with the miles traveled in each hour,
since hotelling occurs only when drivers are not driving. Instead, the fraction of hours spent
hotelling by time of day can be derived from other sources. In particular, the report, Roadway-
Specific Driving Schedules for Heavy-Duty Vehicles54 combines data from several instrumented
truck studies and contains detailed information about truck driver behavior. While none of the
trucks in that study were involved in long-haul interstate activity, for lack of better data, we have
assumed that long-haul truck trips have the same hourly truck trip distribution as the heavy
heavy-duty trucks that were studied.
For each hour of the day, we estimated the number of trips that would end in that hour, based on
the number of trips that started 10 hours earlier. The hours of hotelling in that hour is the number
that begin in that hour, plus the number that began in the previous hour, plus the number that
began in the hour before that and so on, up to the required eight hours of rest time. Table 13-9
shows the number of trip starts and inferred trip ends over the hours of the day in the sample of
trucks assuming all trips are 10 hours long. For example, the number of trip ends in hour 1 is the
same as the number of trip starts 10 hours earlier in hour 15 of the previous day.
u The NREL long-haul dataset yielded an hourly hotelling distribution with most of the activity occuring during the
daytime hours, while the MOVES2014, NCHRP 08-101 and truck survey data suggests that most occurs during the
nightime. As discussed in Section 12, NREL could not confirm the time stamp of the data for the long-haul trucks in
the FleetDNA was the local time, or a reference time because the long-haul truck was provided by an industry
partner, not collected by NREL. For these reasons, we decided not to use the FleetDNA data to update the hotelling
hourly distributions.
134
-------
Table 13-9 Hourly distribution of truck trips used to calculate hotelling hours
hourlD
Hour of the Day
Trip Starts
Trip Ends
1
Hour beginning at 12:00 midnight
78
171
2
Hour beginning at 1:00 AM
76
167
3
Hour beginning at 2:00 AM
65
144
4
Hour beginning at 3:00 AM
94
98
5
Hour beginning at 4:00 AM
107
71
6
Hour beginning at 5:00 AM
131
73
7
Hour beginning at 6:00 AM
194
71
8
Hour beginning at 7:00 AM
230
52
9
Hour beginning at 8:00 AM
279
85
10
Hour beginning at 9:00 AM
267
48
11
Hour beginning at 10:00 AM
275
78
12
Hour beginning at 11:00 AM
240
76
13
Hour beginning at 12:00 Noon
201
65
14
Hour beginning at 1:00 PM
211
94
15
Hour beginning at 2:00 PM
171
107
16
Hour beginning at 3:00 PM
167
131
17
Hour beginning at 4:00 PM
144
194
18
Hour beginning at 5:00 PM
98
230
19
Hour beginning at 6:00 PM
71
279
20
Hour beginning at 7:00 PM
73
267
21
Hour beginning at 8:00 PM
71
275
22
Hour beginning at 9:00 PM
52
240
23
Hour beginning at 10:00 PM
85
201
24
Hour beginning at 11:00 PM
48
211
An estimate of the distribution of truck hotelling duration times is derived from a 2004 CRC
paper78 based on a survey of 365 truck drivers at six different locations. Table 13-10 lists the
fraction of trucks in each duration bin. Some trucks are hotelling for more than the required
eight hours, but some are hotelling for less than eight hours.
Table 13-10 Distribution of truck hotelling activity duration
Hotelling Duration
(hours)
Fraction of Trucks
2
0.227
4
0.135
6
0.199
8
0.191
10
0.156
12
0.057
14
0.014
16
0.021
Total
1.000
We assume that all hotelling activity begins at the trip ends shown in Table 13-9. However, not
all trip ends have the same number of hotelling hours. The distribution of hotelling durations
from Table 13-10 is applied to the hotelling that occurs at each of these trip ends.
135
-------
Table 13-11 illustrates the hotelling activity calculations based on the number of trip starts and
trip ends. The hours of hotelling in any hour of the day is the number of trip ends in the current
hour plus the trip ends from the previous hours that are still hotelling. However, since not all
trips begin and end precisely on the hour, we have discounted the oldest hour included in the
calculation by 60 percent to account for those unsynchronized trips.
For example, there are 171 trip ends in hourlD 1. If all trip ends idle for two hours, the number
of hours is 171 (for hourlD 1) and 40 percent of 211 (for hourlD 24) and thus 171 + (0.4*211) =
255.4 hours of hotelling. Similarly, the number of hours can be calculated for other hotelling
time periods. For four-hour hotelling periods, the hotelling hours would be 171 +211+ 201 +
(0.4*240) = 679. Only the oldest hour of the hotelling time period is discounted.
This calculation accounts for the time in the current hour of the day which is a result of hotelling
from trips that ended in the current hour and trips that ended in previous hours. This approach
assumes that all hotelling begins at the trip end. For example, in the hour of the day 1 for the
four hours hotelling bin, the trip ends in hourlD 22 contribute to the hours of hotelling in hourlD
1, since these trip ends are still hotelling (four hours) after the trip end. The trip ends in hourlD
21 do not contribute to the four hours hotelling bin, since it has been more than four hours since
the trip ends occurred.
The initial calculated hours assume that all trucks idle the same amount of time, indicated by the
hotelling hours bin. The distribution (weight) from Table 13-10 is applied to the hour estimate in
each hotelling hours bin to calculate the weighted total idle hours for each hour of the day.
136
-------
Table 13-11 Calculation of hourly distributions of
hourlD
Trip
Starts
Trip
Ends*
2
hours
4
hours
6
hours
8
hours
10
hours
12
hours
14
hours
16
hours
Weighted Total
Idle Hours
Distribution
1
78
171
255.4
679
1204.8
1736
2120.4
2343.6
2495.4
2638.2
1276
0.0628
2
76
167
235.4
629.4
1100
1643.6
2118.6
2408.8
2593
2739.2
1234
0.0611
3
65
144
210.8
566.4
990
1515.8
2047
2431.4
2654.6
2806.4
1166
0.0577
4
94
98
155.6
477.4
871.4
1342
1885.6
2360.6
2650.8
2835
1056
0.0526
5
107
71
110.2
379.8
735.4
1159
1684.8
2216
2600.4
2823.6
930
0.0458
6
131
73
101.4
299.6
621.4
1015.4
1486
2029.6
2504.6
2794.8
823
0.0407
7
194
71
100.2
254.2
523.8
879.4
1303
1828.8
2360
2744.4
728
0.0357
8
230
52
80.4
224.4
422.6
744.4
1138.4
1609
2152.6
2627.6
630
0.0306
9
279
85
105.8
237.2
391.2
660.8
1016.4
1440
1965.8
2497
581
0.0289
10
267
48
82
213.4
357.4
555.6
877.4
1271.4
1742
2285.6
507
0.0255
11
275
78
97.2
231.8
363.2
517.2
786.8
1142.4
1566
2091.8
479
0.0238
12
240
76
107.2
236
367.4
511.4
709.6
1031.4
1425.4
1896
457
0.0221
13
201
65
95.4
238.2
372.8
504.2
658.2
927.8
1283.4
1707
434
0.0221
14
211
94
120
266.2
395
526.4
670.4
868.6
1190.4
1584.4
447
0.0221
15
171
107
144.6
296.4
439.2
573.8
705.2
859.2
1128.8
1484.4
476
0.0238
16
167
131
173.8
358
504.2
633
764.4
908.4
1106.6
1428.4
526
0.0255
17
144
194
246.4
469.6
621.4
764.2
898.8
1030.2
1184.2
1453.8
635
0.0323
18
98
230
307.6
597.8
782
928.2
1057
1188.4
1332.4
1530.6
767
0.0374
19
71
279
371
755.4
978.6
1130.4
1273.2
1407.8
1539.2
1693.2
933
0.0458
20
73
267
378.6
853.6
1143.8
1328
1474.2
1603
1734.4
1878.4
1068
0.0526
21
71
275
381.8
913
1297.4
1520.6
1672.4
1815.2
1949.8
2081.2
1194
0.0594
22
52
240
350
893.6
1368.6
1658.8
1843
1989.2
2118
2249.4
1268
0.0628
23
85
201
297
822.8
1354
1738.4
1961.6
2113.4
2256.2
2390.8
1289
0.0645
24
48
211
291.4
762
1305.6
1780.6
2070.8
2255
2401.2
2530
1308
0.0645
Totals
3428
3428
4799
11655
18511
25367
32223
39079
45935
52791
20213
1.0000
Weight
0.227
0.135
0.199
0.191
0.156
0.057
0.014
0.021
lotelling
activity
Note:
* Assumes every trip ends 10 hours after it starts, such that all trips are 10 hours long. For the first hour of hotelling in each hour
bin, the column sum is reduced by 60 percent to account for trip ends in a column that are not a full hour.
137
-------
The distribution calculated using this method is similar to the behavior observed in a
dissertation79 at the University of Tennessee, Knoxville. This study observed the trucks parking
at the Petro truck travel center located at the 140/175 and Watt Road interchange between mid-
December 2003 and August 2004. Rather than using results from a single study at a specific
location, MOVES uses the more generic simulated values to determine the diurnal distribution of
hotelling behavior. The distribution of total hotelling hours to hours of the day is calculated from
the total hotelling hours and stored in the SourceTypeHour table in MOVES.
MOVES uses this same default hourly distribution from Table 13-11 for all days and locations,
as shown below in Figure 13-2. Note this distribution of hotelling by hour of the day is similar
to the inverse of the VMT distribution used for these trucks by hour of the day.
0.08
0.07
.1 0.06
+J
.g
t 0.05
—
b
tT 0.04
3
0
1
m 0.03
"ai
o 0.02
x
0.01
0
0 2 4 6 8 10 12 14 16 18 20 22 24
Hour of the Day
Figure 13-2 Truck hotelling distribution by hour of the day in MOVES
In Figure 13-2, we also compare the hotelling distribution to hotelling activity derived from the
Vnomics data analyzed by the NCHRP 08-101 project.70 As shown, it provides a constent
diurnal pattern, with most of the hotelling activity occuring during the nighttime hours. This is a
consistent pattern displayed by truck parking results at unofficial locations and truck stops
reported by the Federal Highway Administration.80 The data from the NCHRP 08-101 project
was used only for comparison purposes and not used dirctly to update the hotelling hourly
distribution, because NCHRP 08-101 utilized different definitions of hotelling activity than
MOVES.v
v The definition of hotelling used in the draft NCHRP 08-101 project estimates idling activity with duration > 8
hour, whereas in Section 11 we used an idle duration of > 1 hour.
138
-------
14. Geographical Allocation of Activity
The vehicle miles traveled (VMT) and vehicle populations contained in the default database (see
Section 3 and Section 4, respectively) are for all 50 states and Washington D.C. However,
MOVES can estimate activity in individual states or counties (including Puerto Rico and the
Virgin Islands) when running at Default Scale using geographic allocation factors stored in two
tables: Zone and ZoneRoadType. The geographic allocations in MOVES4 are based on the 2020
NEI81 distribution of VMT by county.
In MOVES4, hotelling hours (including extended idling and auxiliary power unit usage) are
calculated from combination long-haul combination truck VMT in each location and have their
own allocation factors (see Section 11). Similarly, engine starts are calculated based on county
vehicle populations and are not allocated from national estimates, although county vehicle
populations themselves are calculated using allocation factors (SHO and SHP as explained in the
sections below).
Note that the MOVES design only allows for one set of geographic allocations to be stored in the
default database. While real-world geographic allocations change over time, the MOVES
defaults are used for all calendar years. Thus, it is often more accurate to use information other
than the default values. National-level emissions can be generated with calendar year specific
geographical information by running each year separately, with different user-input allocations
for each run. County- and Project-level calculations do not use the default geographical
allocation factors at all. Instead, County and Project scales require that the user input local total
activity for each individual year being modeled.
For MOVES4, we updated the default list of counties (and therefore zones) to incorporate
changes in the state of Alaska that went into effect in January 2019 when the Valdez-Cordova
census area (FIP 02261) was divided into two new census areas: Chugach Census Area (FIP
02263) and Copper Census Area (FIP 02066). For the default database, we split the activity
existing in the zone and zoneroadtype tables for the original Valdez-Cordova Census Area and
allocated it to the new census areas using the population estimate reported in the 2020 US
Census82 for these new areas.
14.1. Source Hours Operating Allocation to Zones
The national total source hours of operation (SHO) are calculated from the estimates of VMT as
described in sections above. This total VMT for each road type is allocated to county using the
SHOAllocFactor field in the ZoneRoadType table. Although the field is named "source hours
operating", it is used only for allocating VMT and not hours of operation.
The 2020 NEI VMT was aggregated into the annual sum for the four MOVES road types in each
county and nationally and used to calculate the SHO AllocFactor using Equation 14-1.
139
-------
SHOAllocF actor,
C ountyV MTRoadTypeID
Equation
RoadTypeiD NationalVMT,
RoadTypelD
14-1
The county allocation values for each roadway type sum to one (1.0) over all 50 states and
Washington D.C. The same SHOAllocFactor set is the default for all calendar years at the
National scale. County- and Project-level calculations do not use the default SHOAllocFactor
allocations at all. Instead, County and Project scales require that the user input all local activity.
14.2. Parking Hours Allocation to Zones
The allocation of the domain-wide hours of parking (time when vehicles are not operating but
continue to have evaporative emissions) to zones is stored in the SHPAllocFactor in the Zone
table. In the default database for MOVES, the domain is the nation and the zones are the
counties. There is no national source for hours of parking by county, so we have used a VMT-
based allocation.
The allocation is determined using the VMT estimates for each county in each state as calculated
using Equation 14-2, where i represents each individual county and / is the set of all US
counties.
The county allocation values for parking hours sum to one (1.0) over all 50 states and
Washington D.C. The same SHPAllocFactor set is the default for all calendar years at the
National scale. County- and Project-level calculations do not use the default SHPAllocFactor
allocations at all. Instead, County and Project scales require that the user input all local activity.
Note that the same allocation values are used for the StartAllocFactor column, also saved in the
Zone table.
SHP Alloc Fact ori = CountyVMTjCountyVMT,
Equation
14-2
iel
140
-------
15. Vehicle Mass ami Road Load Coefficients
The MOVES model calculates emissions using a weighted average of emisson rates by operating
mode. For running exhaust emissions, the operating modes are defined by either vehicle specific
power (VSP) or scaled tractive power (STP). Both VSP and STP estimate the tractive power
exerted by a vehicle and are calculated based on a vehicle's speed and acceleration, but differ in
how they are scaled (or normalized). VSP is used for the motorcycle, light-duty vehicles and
light-duty truck regulatory classes 10, 20, and 30 and STP is used for heavy-duty regulatory
classes.
The SourceUseTypePhysics table describes the vehicle characteristics needed for the VSP and
STP calculations, including average vehicle mass, a fixed mass factor and three road load
coefficients for each combination of source type and regulatory class averaged over all ages. In
MOVES2014, the SourceUseTypePhysics table varied only by source type. However, regulatory
class and model year were added in MOVES3 as one of the key changes to model the Heavy-
Duty Greenhouse Gas Phase 2 rule83 which anticipates improvements to vehicle and trailer
design. MOVES uses values in the SourceUseTypePhysics table to calculate VSP and STP for
each source type/regulatory class combinations according to Equation 15-1 and Equation 15-2:
VSP =
STP =
Equation
Av + Bv2 + Cv3 + M ¦ (a + g ¦ sinO) ¦ v 15-1
M
Av + Bv2 + Cv3 + M ¦ (a + g ¦ siviQ) ¦ v Equation
fscale
where A, B and C are the road load coefficients in units of kW-s/m, kW-s2/m2 and kW-s3/m3
respectively. A is associated with tire rolling resistence, B with mechanical rotating friction as
well as higher order rolling resistance losses and C with aerodynamic drag. M is the source mass
for the source type in metric tons, g is the acceleration due to gravity (9.8 m/s2), v is the
instantaneous vehicle speed in m/s, a is the instantaneous vehicle acceleration in m/s , sin 0 is
the (fractional) road gradew and fscaie is a scaling factor. Note that the only difference between
the VSP and STP equations is the term in the denominator. For light-duty vehicles using VSP,
the power is normalized by the mass of the vehicle (fscaie = M). For heavy-duty vehicles, the
fscaie is similar, but not equal to the average source mass of the vehicle source type (fscaie ^
M).
When conducting light-duty emissions analysis, emissions data from individual vehicles are
assigned to VSP operating mode bins using Equation 15-1, with the individual vehicle's
w MOVES does not model grade at the national and county scale. Road grade may be entered at the project scale.
141
-------
measured weight as the source mass (hence the term "vehicle-specific"). When developing
emissions rates for MOVES, the emissions from individual vehicles are averaged across
operating mode bins to calculate average emission rates for each regulatory class. Because
individual vehicle weights within the same regulatory class vary, the absolute tractive power
produced by individual vehicle activity assigned to the same VSP-defined operating mode also
varies. In contrast, when MOVES calculates VSP from driving cycles and assigns operating
modes for an entire source type, the average source type mass is used instead.
For heavy-duty vehicles, STP is calculated with Equation 15-2, which is very similar to the VSP
equation except the tractive power is normalized by a fixed fscaie values for all vehicles within
the same regulatory class and model year group. The fscaie is used to bring the numerical range
of tractive power from heavy-duty vehicles into the same numerical range as the VSP values
when assigning operating modes. When developing emission rates for MOVES, operating modes
are assigned to individual vehicles using both the individual truck weight, and the common fscaie
value used for all heavy-duty vehicles from the same regulatory class, source type and model
year group. Because a common fscaievalue is used, individual vehicles assigned to the same
STP-defined operating mode bin are producing the same absolute tractive power, regardless of
differences in their individual source masses. When MOVES estimates STP and assigns
operating mode distributions for the heavy-duty fleet, it uses the average source type mass (M)
for each regulatory class, source type, and model year group in the numerator and uses the
common fscaie value which was used in the emission rate analysis.
Additional discussion regarding VSP and STP (including the selection of fscaie values) are
provided in the MOVES light-duty10 and heavy-duty11 exhaust emission rate reports,
respectively.
In both cases, MOVES derives operating mode distributions by combining second-by-second
speed and acceleration data from a specific drive schedule with the proper coefficients for a
specific source type. More information about drive schedules can be found in Section 9.1 The
following sections detail the derivation of values used in Equation 15-1 and Equation 15-2.
15.1. Source Mass ami Fixed Mass Factor
The two mass factors stored in the SourceUseTypePhysics table are the source mass and fixed
mass factor. The source mass represents the average weight of vehicles of a given regulatory
class within a source type, which includes the weight of the vehicle, occupants, fuel and payload
(M in Equation 15-1 and Equation 15-2) and the fixed mass factor represents the STP scaling
factor (fScaie m Equation 15-2). The mass factors in the SourceUseTypePhysics table are in units
of metric tons (1000 kilograms). The source masses are reported in this section both in units of
weight in lbs (used in the regulatory class defintions), and mass in kilograms (used in MOVES
calculations).
In MOVES4, the source masses of light-duty vehicles were unchanged from previous versions of
MOVES, as presented in Table 15-1 and documented in Appendix F.
142
-------
Table 15-1. Average Vehicle Weight and Mass for Motorcycles, Light-duty Vehicles, and Light-
duty Trucks Regulatory Classes
Source Type (sourceTypelD)
Regulatory Class (regClassID)
Average Vehicle
Weight (lbs)
Average Vehicle
Mass (kg)
Motorcycle (11)
Motorcycle (10)
628
285
Passenger Car (31)
Light-duty Vehicle, LDV (20)
3,260
1,479
Passenger Truck (31)
Light-duty Truck, LDT (30)
4,116
1,867
Light Commercial Truck (32)
Light-duty Truck, LDT (30)
4,541
2,060
The source masses for light heavy-duty trucks are based on a report from the National Research
Council.84 This report included data on empty vehicle weight ranges, typical payload capacity
and annual fleet VMT by truck class. For light heavy-duty trucks, the average source mass was
assumed to be the midpoint of the empty vehicle weight range plus 50 percent of the typical
payload capacity. The source mass for passenger trucks and light commercial trucks in
regulatory class 41 was calculated using the data presented for class 2b trucks only. A VMT-
weighted average mass was calculated for single-unit trucks in regulatory class 41 using data for
class 2b and 3 trucks, and single-unit trucks in regulatory class 42 were assigned a VMT-
weighted average for class 4 and 5 trucks.
Table 15-2. Average Vehicle Weight Mass for LHD2b3 and LHD45 Regulatory Classes by Source
Ilfie
Source Type (sourceTypelD)
Regulatory Class
(regClassID)
VMT-weighted
Average Vehicle
Weight (lbs)
VMT-weighted
Average Vehicle
Mass (kg)
Passenger Truck (31)
Light Commercial Truck (32)
LHD2b3 (41)
7,500
3,402
Refuse Truck (51)
Single-unit Short-haul Truck (52)
Single-unit Long-haul Truck (53)
Motor Home (54)
LHD2b3 (41)
7,879
3,574
LHD45 (42)
12,716
5,768
The source masses for medium and heavy heavy-duty single-unit trucks and combination trucks
were estimated based on weigh-in-motion data made available through FHWA's Vehicle Travel
Information System (VTRIS).85 This data source presents average gross vehicle weights by truck
type (single unit, single trailer and multi-trailer), axle count and state. An approximate mapping
between MOVES source types/regulatory classes and the VTRIS truck types/axle counts is
presented in Table 15-3. National average masses were calculated for regulatory classes 46, 47
and 49 in the single-unit truck and combination truck source types using 2013 VTRIS data,
weighted by VMT data for the same year by state, as presented in Highway Statistics Table VM-
2.
143
-------
Table 15-3 Average Vehicle Weight and Mass for MHD, HHD and Glider Regulatory Classes and
Source Type and VTRIS Vehicle Classes and Axle Count
Source Type
(sourceTypelD)
Regulatory
Class
(regClassID)
VTRIS Vehicle Class
and Axle Count
VMT-weighted
Average Vehicle
Weight (lbs)
VMT-weighted
Average Vehicle
Mass (kg)
Refuse Truck (51)
MHD (46)
Single-unit Trucks: 3-
axle
30,424
13,800
HHD (47)
-
45,645
20,704
Single-unit Short-haul
Truck (52)
Single-unit Long-haul
Truck (53)
Motor Home (54)
MHD (46)
Single-unit Trucks: 3-
axle
30,424
13,800
HHD (47)
Single-unit Trucks: 4-
axle
55,221
25,048
Combination Short-haul
Truck (61)
Combination Long-haul
Truck (62)
MHD (46)
Single Trailer Trucks: 4-
axles or less
30,891
14,012
HHD (47)
Single Trailer Trucks: 5-
axle, 6-axle, or more
54,741
24,830
All Multi-trailer Trucks
Glider (49)
Single Trailer Trucks: 5-
axle, 6-axle, or more
54,741
24,830
All Multi-trailer Trucks
The exception to the single-unit truck analysis described above is the average source mass for
class 8 (HHD) refuse trucks because these trucks are subject to a lower Federal weight limit due
to their typical vehicle length and axle configuration.86 These vehicles are assumed to have an
average source mass of 45,645 lbs, based on several studies of in-use refuse truck activity.87 88 89
90
The medium heavy- and heavy heavy-duty truck source masses were adjusted from the baseline
masses as calculated above to account for expected changes by model year due to both the
Heavy-Duty GHG Phase 1 and Phase 2 rules. With the Phase 1 rule, decreases were expected for
combination trucks. With the Phase 2 rule, weight reductions were also expected for single-unit
trucks. The changes in source masses from the baseline masses reflecting the Phase 1 and 2 rules
are shown in Table 15-4. The details of the analyses used to estimate the changes in source
masses can be found in the Phase 1 Regulatory Impact Analysis91 and in the docket for the Phase
2 rule.92 93
144
-------
Table 15-4 MHD and HHP Changes in Vehicle Weight by Model Year
Source Type*
Model Years
Change in Vehicle Weight
from Baseline (lbs)
Single-Unit Short-haul Truck
2021-2023
-4.4
2024-2026
-10.4
2027+
-16.5
Single-Unit Long-haul Truck
2021-2023
-7.9
2024-2026
-23.6
2027+
-39.4
Combination Short-haul Truck
2014+
-321
Combination Long-haul Truck
2014+
-400
* No change in vehicle weights is modeled for other sourcetypes.
The source masses for all medium heavy-duty and heavy heavy-duty buses are based on a report
from the American Public Transit Association (APTA).94 This report included data on the ranges
of seating capacity, curb weight and fully-loaded weight for different types and lengths of buses.
Lacking specific data on in-use bus masses, we assume that the average source mass is the mid-
point between the curb weight and fully-loaded weight. We also assume that seating capacity is
the driving variable for the curb weight and fully-loaded weight of a bus. Under this simplifying
assumption, linear functions of seating capacity for average and fully-loaded masses were
determined by bus type and length using the ranges presented in the APTA report.
To calculate national average source masses for medium heavy-duty and heavy heavy-duty
buses, the mass functions derived from the APTA report were weighted by bus activity data from
FTA's 2017 National Transit Database (NTD).19 The NTD contains estimates of transit bus
populations and VMT and also includes data on seating capacity, bus type and vehicle length.
For each entry in the NTD that described a type of bus, double decked bus or articulated bus, we
calculated an average mass and a fully-loaded mass based on seating capacity, bus type and
vehicle length. We assigned vehicles with fully-loaded masses between 19,500 and 33,000 lbs to
the medium heavy-duty regulatory class (regClassID 46) and vehicles above 33,000 lbs were
assigned to the heavy heavy-duty regulatory class (regClassID 48 for diesel and CNG transit
buses and regClassID 47 for all the remaining bus source type and fuel type combinations).
Using these regulatory class assignments, we calculated VMT-weighted average masses for
regulatory classes 46, 47, and 48 and applied them to all bus source types (i.e., sourceTypelDs
41, 42, and 43). In the future, we will consider updating the bus mass estimates to incoporate
data on mile-weighted average passenger load.
Because the APTA report did not include data for light heavy-duty buses, we calculated source
mass based on a number of assumptions regarding vehicle parameters from manufacturer
specifications for the most popular vehicle models in the NTD with a bus type of cutaway.
Specifically, we assumed the following:
• The fully-loaded vehicle mass is at the upper bounds of allowable GVWR for each class
(i.e., 14,000 lbs for regClassID 41 and 19,500 lbs for regClassID 42).
• Passenger capacity is 15 for regClassID 41 and 25 for regClassID 42 and the average
passenger weighs 175 lbs.
• Fuel tank capacity is 50 gallons (gasoline).
145
-------
If the average operating conditions for these vehicles are at 50 percent passenger and 50 percent
fuel capacity, then the average vehicle mass for LHD2b3 (regClassID 41) buses can be
calculated as 12,531 lbs and LHD45 (regClassID 42) buses as 17,156 lbs. The average weights
and source masses are shown in Table 15-5.
Table 15-5. Bus Weights and Mass by Regulatory Classes by Source Type
Source Type (sourceTypelD)
Regulatory Class
(regClassID)
Vehicle Weight
(lbs)
Vehicle Mass (kg)
Other Bus (41)
School Bus (43)
LHD2b3 (41)
12,531
5,684
Other Bus (41)
Transit Bus (42)
School Bus (43)
LHD45 (42)
17,156
7,782
Other Bus (41)
Transit Bus (42)
School Bus (43)
MHD (46)
25,060
11,367
Other Bus (41)
Transit Bus (42)
School Bus (43)
HHD (47)
34,399
15,603
Transit Bus (42)
Urban Bus (48)
34,399
15,603
The complete list of sourceMass and fixedMassFactor in MOVES4 is presented in Appendix J.
15.2. Road Load Coefficients
As indicated above, in MOVES, road load coefficients are used in the calculation of both VSP
and STP. A, B and C are the road load coefficients in units of kW-s/m, kW-s2/m2, and
kW-s3/m3, respectively. A is associated with rolling resistance, B with mechanical rotating
friction as well as higher order rolling resistance losses and C with aerodynamic drag. The
information available on road load coefficients varied by regulatory class.
15.2.1. Light-Duty and Motorcycles
Motorcycle road load coefficients, given in Equation 15-3 through Equation 15-5, were
empiricially derived in accordance with standard practice:95'96
A = 0.088 ¦ M Equation 15-3
5 = 0 Equation 15-4
C = 0.00026 + 0.000194 ¦ M Equation 15-5
146
-------
For light-duty vehicles, the road load coefficients were calculated according to Equation 15-6
through Equation 15-8:97
0-7457 ^
A = 50 ¦ 0 447 ' °"35 ' ™LHP@50mph Equation 15-6
0.7457
B = (50 ¦ 0 447)2 0 10 ' TRLHP@$omph Equation 15-7
0.7457
C = (50 ¦ 0 447)3 ' 0-55 ' TRLHP@50mPh Equation 15-8
In the three equations above, the first factor is the appropriate unit conversion to allow A, B and
C to be used in Equation 15-1 and Equation 15-2, the second factor is the power distribution into
each of the three load categories and the third is the tractive road load horsepower rating
(TRLHP). Average values for A, B and C for source types 21,31 and 32 were derived from
applying TRLHP values recorded in the Mobile Source Observation Database (MSOD)98 to
Equation 15-6 through Equation 15-8. While we expect light-duty road load coefficients to
improve over time due to the 2017 and Later Model Year Light-Duty Vehicle Greenhouse Gas
Emissions Rule, the impact of these changes have been directly incorporated into the emission
and energy rates.99 Therefore, these coefficients remain constant over time in the MOVES (if not
in the real-world) to avoid double counting the impacts of actual road load improvements in the
fleet.
15.2.2. Heavy-Duty Vehicles
For heavy-duty source types, no road load parameters were available in the MSOD. Therefore,
for the heavy-duty source types other than combination trucks, relationships of historical road
load coefficent to vehicle mass came from a study done by V.A. Petrushov,100 as shown in Table
15-6. These relationships are grouped by regulatory class; source type values were determined by
weighting the combination of weight categories that comprise the individual source typesx. As
noted in the table below, the B term is set to zero to reflect that the frictional forces that are
linearly related to vehicle speed in heavy-duty vehicles are very low when compared to the
rolling resistance and aerodynamic forces. The road load parameters for combination trucks have
been revised for model years 1960-2060 using the methods described in Section 15.2.2.2. The
x The A and C coefficients were derived in MOVES2010 based on the equations in Table 15-3 and the population
fraction of regulatory classes within the sourcetypes in MOVES2010. For MOVES2014 and MOVES3, we updated
the vehicle source masses, and we scaled the A coefficients from MOVES2010 according to the changes in vehicle
mass, because there is a direct relationship between rolling resistance and vehicle weight. In contrast, for all but the
combination trucks, the aerodynamic drag coefficients, C, are unchanged from MOVES2010 because we lacked new
data and there is not a direct relationship between arodynamic drag and vehicle weight.
147
-------
revised road load coefficients for heavy-duty source types other than combination trucks for
model years 2014-2060 are described in Section 15.2.2.3
Table 15-6 Road Load Coefficients for MY 1960-2013 Buses, Motor Homes and
Single-Unit Heavy-duty Trucks
Coefficient
8500 to 14000 lbs.
(3.855 to 6.350
metric ton)
14000 to 33000 lbs.
(6.350 to 14.968
metric ton)
>33000 lbs.
(>14.968 metric ton)
Buses and Motor
Homes
(kW-s\
A( m )
0.0996 -M
0.0875 -M
0.0661 -M
0.0643 -M
b( kwf)
\ mz J
0
0
0
0
c( kwf)
\ J
0.00147 +
5.22 X lO"5 ¦ M
0.00193 +
5.90 X 10"5 -M
0.00289 +
4.21 X 10"5 -M
0.0032 +
5.06 X 10"5 -M
15.2.2.1. Incorporation of Heavy-Duty Greenhouse Gas
Standards In MOVES
EPA set greenhouse gas (GHG) emission standards for heavy-duty vehicles in two separate
rulemakings, refered to in this report as the Phase 1101 and Phase 2102 HD GHG rules. The Phase
1 rulemaking became effective for the 2014 model year. The Phase 2 rulemaking became
effective in 2021 model year and is fully phased-in by the 2027 model year.
The road load coefficients in MOVES reflect the projected improvements to the vehicles in
different model year groups. The first model year group includes model years 1960-2013 to
reflect the time period prior to the first heavy-duty truck GHG emission standards. Due to
improvements in trailers over this time period, the first model year group is split into pre-2008
and 2008-2013 for combination tractor-trailers. The Phase 1 standards are applied to model years
2014-2017 (or through 2020 depending on category). The Phase 2 standards are phased-in using
model year groups 2021-2023, 2024-2026 and 2027-and-later. To account for the improvements
due to the HD GHG rules, road load forces were separated into individual road load coefficients
because significant improvements are expected in aerodynamic drag and rolling resistance,
particularly for tractor-trailers. The aerodynamic and rolling resistance components of the overall
road load are determined separately and updated in MOVES4 as a result of information on Phase
2 HD GHG rule implemention.y
The aerodynamic drag force, Faero as a function of speed is represented as:
y Due to a 2021 appeals court ruling vacating the portions of the Heavy-duty Greenhouse Gas Phase 2 standards
(HDGHG2) that apply to trailers, we revised MOVES inputs that describe weight, aerodynamics, rolling resistance
and "other efficiency" improvements for combination trucks of model year 2018 and later to better represent the
implemented program.
148
-------
Faero 2 P^d^fvair Equation 15-9
where p is the density of air, Cd is the aerodynamic drag coefficient, A is the frontal area of the
vehicle and Vair is the air speed relative to the vehicle as it is traveling. In zero wind conditions,
the relative air speed is equal to vehicle speed. Consequently, the aerodynamic drag component
of STP can be represented as:
STPaero = (7^-) "\pCdAfV3 Equation 15-10
\ J scale' ^
Thus, the C road load coefficient can be represented as:
_ 1
C - PQAr Equation 15-11
The quantity CdAf; shortened to CdA, is called the drag area and is used to characterize the overall
aerodynamic drag forces for a vehicle.
The tire rolling resistance force is represented using the A coefficient in the
SourceUseTypePhysics table. It is related to the coefficient of rolling resistance, Crr and source
mass M, using the following equation:
A = CRRMg Equation 15-12
where g is the gravitational acceleration.
Section 15.2.2.2 describes the analysis to update road load coefficients for combination long-
haul (sourceTypelD 62) and short-haul (sourceTypelD 61) trucks in MOVES. Section 15.2.2.3
describes the updates applied to heavy-duty source types other than combination trucks to
account for HD GHG Phase 1 and Phase 2 rulemakings. The details on the discussion of
incorporating Phase 1 and Phase 2 energy reductions from engine technology improvements into
MOVES can be found in the MOVES Heavy-Duty Emission Rates Report.11
While we expect road load coefficients for Heavy-Duty Pickups and Vans (regclassID 41) to
improve over time due to the Phase 1 and Phase 2 HD GHG rules, the impact of these changes
have been directly incorporated into the emission and energy rates.11 Since nearly all HD pickup
trucks and vans are certified on a chassis dynamometer, the improvements in road loads expected
from the greenhouse gas standards are modeled as total vehicle improvements without separating
out the engine and road load components. Therefore, these coefficients remain constant over
time in MOVES (if not in the real-world) to avoid double counting the impacts of actual road
load improvements in the fleet.
149
-------
15.2.2.2.
Combination Trucks for Model Years 1960-2060
MOVES3 included updates to both the aerodynamic and rolling resistance components of the
overall road load reflecting the greenhouse gas emissions standards for combination trucks. An
aerodynamic assessment of all model years of combination trucks was conducted to utilize a
consistent method in MOVES, and the aerodynamic values were updated for all model years to
reflect the aerodynamic technology analysis and projections in HD GHG Phase 2 rulemakings.
These values were further updated in MOVES4 to reflect the HD GHG Phase 2 rulemaking as
implemented. The average road load coefficients are updated by source type and regulatory class
through the beginModelYearlD and endModelYearlD fields in the SourceUseTypePhysics table.
Appendix J describes how the aerodynamic improvements were developed as part of the
rulemaking and how they were used to update MOVES.
15.2.2.3. Heavy-Duty Source Types other than Combination
Trucks for Model Years 2014-2060
For buses, refuse trucks, motor homes and long-haul and short-haul single-unit trucks
(sourceTypelDs 41 through 54), the A coefficient values determined through tire rolling
resistance reductions projected in the HD GHG Phase 1 and Phase 2 rulemakings were used
directly. The aerodynamic drag coefficient (C coefficient) was not updated for these heavy-duty
vehicles because no significant improvements in C coefficients is expected from the Phase 2
standards.103
The final road load coefficients for all regulatory classes and sourcetypes in MOVES4 are shown
in Appendix J.
16. Air Conditioning Activity Inputs
This section describes three inputs used in determining the impact of air conditioning on
emissions. The ACPenetrationFraction is the fraction of vehicles equipped with air conditioning.
FunctioningACFraction describes the fraction of these vehicles in which the air conditioning
system is working correctly. The ACActivityTerms relate air conditioning use to local heat and
humidity. These factors remain the same as in earlier versions of MOVES. More information on
air conditioning effects and how air conditioning affects Electric Vehicle energy consumption is
provided in the MOVES technical report on adjustment factors.104
16.1. ACPenetrationFraction
The ACPenetrationFraction is a field in the SourceTypeModelYear table that describes the
fraction of vehicles equipped with air conditioning. Default values, by source type and model
year, were taken from MOBILE6.105 Market penetration data by model year were gathered from
Ward's Automotive Handbook for light-duty vehicles and light-duty trucks for model years 1972
through 1995 for cars and 1975-1995 for light trucks. Rates in the first few years of available
150
-------
data were quite variable, so values for early model years were estimated by applying the 1972
and 1975 rates for cars and trucks, respectively. Projections beyond 1995 were developed by
calculating the average yearly rate of increase in the last five years of data and applying this rate
until a predetermined cap was reached. A cap of 98 percent was placed on cars and 95 percent on
trucks under the assumption that there will always be vehicles sold without air conditioning,
more likely trucks than cars. No data were available on heavy-duty trucks. While VIUS asks if
trucks are equipped with A/C, "no response" was coded the same as "no," making the data
unusable for this purpose. For MOVES, the light-duty vehicle rates were applied to passenger
cars and the light-duty truck rates were applied to all other source types (except motorcycles, for
which A/C penetration is assumed to be zero), as summarized in Table 16-1.
Table 16-1 AC penetration fractions in MOVES
Motorcycles
Passenger Cars
All Trucks and Buses
1972-and-earlier
0
0.592
0.287
1973
0
0.726
0.287
1974
0
0.616
0.287
1975
0
0.631
0.287
1976
0
0.671
0.311
1977
0
0.720
0.351
1978
0
0.719
0.385
1979
0
0.694
0.366
1980
0
0.624
0.348
1981
0
0.667
0.390
1982
0
0.699
0.449
1983
0
0.737
0.464
1984
0
0.776
0.521
1985
0
0.796
0.532
1986
0
0.800
0.544
1987
0
0.755
0.588
1988
0
0.793
0.640
1989
0
0.762
0.719
1990
0
0.862
0.764
1991
0
0.869
0.771
1992
0
0.882
0.811
1993
0
0.897
0.837
1994
0
0.922
0.848
1995
0
0.934
0.882
1996
0
0.948
0.906
1997
0
0.963
0.929
1998
0
0.977
0.950
1999+
0
0.980
0.950
16.2. FunctioningAC Fraction
The FunctioningACFraction field in the SourceTypeAge table (see Table 16-2) indicates the
fraction of the air-conditioning-equipped fleet with fully functional A/C systems, by source type
and vehicle age. A value of one means all systems are functional. This is used in the calculation
151
-------
of total energy to account for vehicles without functioning A/C systems. Default estimates were
developed for all source types using the "unrepaired malfunction" rates used for 1992-and-later
model years in MOBILE6. The MOBILE6 rates were based on the average rate of A/C system
failure by age reported in the 1997 Consumer Reports Magazine Automobile Purchase Issue and
assumptions about repair frequency during and after the warranty period. The MOBILE6 rates
were applied to all source types except motorcycles, which were assigned a value of zero for all
years.
Table 16-2 FunctioningACFraction by age (for all source types except motorcycles)
agelD
functioningACFraction
0
1
1
1
2
1
3
1
4
0.99
5
0.99
6
0.99
7
0.99
8
0.98
9
0.98
10
0.98
11
0.98
12
0.98
13
0.96
14
0.96
15
0.96
16
0.96
17
0.96
18
0.95
19
0.95
20
0.95
21
0.95
22
0.95
23
0.95
24
0.95
25
0.95
26
0.95
27
0.95
28
0.95
29
0.95
30
0.95
16.3. Air Conditioning Activity Demand
In the MonthGroupHour table, ACActivityTerms A, B and C are coefficients for a quadratic
equation that calculates air conditioning activity demand as a function of the heat index. These
terms are applied in the calculation of the A/C adjustment in the energy consumption calculator.
The methodology and the terms themselves were originally derived for MOBILE6 and are
documented in the report, Air Conditioning Activity Effects in MOBILE6,105 They are based on
analysis of air conditioning usage data collected in Phoenix, Arizona, in 1994.
152
-------
In MOVES, ACActivityTerms are allowed to vary by monthGroup and Hour, in order to provide
the possibility of different AJC activity demand functions at a given heat index by season and
time of day (this accounts for differences in solar loading observed in the original data).
However, the default data uses one set of coefficients for all MonthGroups and Hours. These
default coefficients represent an average AJC activity demand function over the course of a full
day. The coefficients are listed in Table 16-3.
able 16-3 Air conditioning activity coefficients
A
B
C
-3.63154
0.072465
-0.000276
The AJC activity demand function that results from these coefficients is shown in Figure 16-1. A
value of 1 means the AJC compressor is engaged 100 percent of the time; a value of 0 means no
AJC compressor engagement.
Heat Index (F)
Figure 16-1 Air conditioning activity demand as a function of heat index
153
-------
17.Conclusion ami Areas for Future Research
Properly characterizing emissions from vehicles requires a detailed understanding of the cars and
trucks that make up the vehicle fleet and their patterns of operation. The national default
information in MOVES4 provide a reliable basis for estimating national emissions. The most
important of these inputs are well-established: base year VMT and population estimates come
from long-term, systematic national measurements by US Department of Transportation. The
relevant characteristics for prevalent vehicle classes are well-known; base year age distributions
are well-measured and driving activity has been the subject of much study in recent years.
Still, the fleet and activity inputs do have significant limitations. In particular, local variations
from the national defaults can contribute to discrepancies in resulting emission estimates. Thus,
it is recommended to replace many of the MOVES fleet and activity defaults with local data
when available as explained in EPA's Technical Guidance.2
The fleet and activity defaults also are limited by the necessity of forecasting future emissions.
EPA utilizes annual US Department of Energy forecasts of vehicle sales and activity. The inputs
for MOVE3 were developed for a 2017 base year and much of the source data is from 2017 and
earlier. This information needs to be updated periodically to assure that the model defaults reflect
the latest available data and projections on the US fleet.
Moreover, for data that is specific to MOVES, we are also limited by available staff and funding.
Collecting data on vehicle fleet and activity is expensive, especially when the data is intended to
accurately represent the entire United States. Even when EPA does not generate data directly (for
example, compilations of state vehicle registration data), obtaining the information needed for
MOVES can be costly and, thus, dependent on budget choices.
Future updates to vehicle population and activity defaults will need to continue to focus on the
vehicles that contribute the most air pollution nationally, namely gasoline light-duty cars and
trucks and diesel heavy-duty trucks. Information collection on motorcycles, refuse trucks, motor
homes, diesel light-duty vehicles and gasoline heavy-duty vehicles will be a lower priority.
Similarly, in addition to updating the model defaults, we will need to consider whether the
current MOVES design continues to meet our modeling needs. Simplifications to the model to
remove categories, such as source types or road types, might simplify data collection and make
noticeable improvements in run time without affecting the validity of fleet-wide emission
estimates.
In addition to these general limitations, there are also specific MOVES data elements that could
be improved with additional research, including:
• Updates to the trip information used to generate evaporative activity to be consistent
with the new engine start and soak distributions based on the telematics data; this will
likely require modification to the MOVES code as well as updates to the default
database;
• Updated real-world highway driving cycles and operating mode distributions,
including incorporating ramp activity into the default highway driving cycles and
accounting for grade;
154
-------
• Additional instrumented vehicle data from a wider sample of heavy-duty vehicles to
better characterize off-network behavior including vehicle starts and soaks;
• Improved information on truck hotelling durations, locations and temporal
distributions, and variation in operating modes for different vehicle fuel types and
technologies;
• VSP/STP adjustments for road grade and vehicle load;
• Better data on activity changes with age, such as mileage accumulation rates, start
activity and soak distributions. Telematics will provide important insights here, but
gathering representative data for the oldest vehicles in the fleet will continue to be a
challenge;
• Updated estimates of vehicle scrappage rates used to project vehicle age distributions;
• Updated air conditioning system usage, penetration and failure rates;
• Finer vehicle type distinctions in temporal activity and road type distributions;
• Updated information on light-duty vehicle mass;
• Information on shifts in vehicle activity patterns with population shifts to electric,
shared, connected and automated vehicles.
At the same time, the fundamental MOVES assumption that vehicle activity varies by source
type and not by fuel type or other source bin characteristic may be challenged by the growing
market share of alternative vehicles such as autonomous, shared and electric vehicles which may
have distinct activity patterns. As we progress with MOVES, the development of vehicle
population and activity inputs will continue to be an essential area of research.
155
-------
Appendix A Fuel Type ami Regulatory Class Fractions from Previous
Versions of MOTES
Fuel type and regulatory class distributions for most source types are described in Section 5.2. In
the current version of MOVES, the fuel type and regulatory class distributions were unchanged
from previous versions of the model for the following source type and model year combinations:
• Passenger cars, school buses, refuse trucks, short-haul and long-haul single-unit trucks
and all combination trucks prior to model year 2000
• Passenger trucks and light commercial trucks prior to model year 1981
This appendix describes the derivation of these fuel type and regulatory class distributions.
• I, Distributions for Mod I . ,n I Mel I ' II
The fuel type distributions between 1960 and 1981 for each source type have been summarized
in Table A-l and Table A-2. Truck diesel fractions in Table A-l were derived using the 1999
IHS vehicle registrations and the 1997 VIUS,106 except for refuse trucks and motor homes. We
assumed 96 percent of refuse trucks were manufactured to run on diesel fuel in 1980 and earlier
according to the average diesel fraction from VIUS across all model years.
Table A-l Diesel fractions for truck source types
Source Type*
Model
Year
Passenger
Trucks
(31)
Light
Commercial
Trucks
(32)
Refuse
Trucks
(51)
Single-Unit
Trucks
(52 & 53)
Short-Haul
Combination
Trucks
(61)
Long-Haul
Combination
Trucks
(62)
1960-1979
0.0139
0.0419
0.96
0.2655
0.9146
1.0000
1980
0.0124
0.1069
0.96
0.2950
0.9146
1.0000
1981
0.0178
0.0706
0.96
0.3245
0.9146
1.0000
* All other trucks are assumed to be gasoline-powered. Motor homes values were estimated as
described in Section 5.2.
For the non-truck source types, school bus fuel type fractions were reused from MOBILE6,77
originally based on 1996 and 1997 IHS data, and passenger cars were split between gasoline and
diesel for 1960-1981 using the 1999 IHS vehicle registrations data. As in previous versions of
MOVES, motorcycles were assumed to be all gasoline.
156
-------
Tab
e A-2 Diesel fractions for non-truck source types
Source Type*
Model Year
Motorcycles (11)
Passenger Cars (21)
School Buses (43)
1960-1974
0
0.0069
0.0087
1975
0
0.0180
0.0087
1976
0
0.0165
0.0086
1977
0
0.0129
0.0240
1978
0
0.0151
0.0291
1979
0
0.0312
0.0460
1980
0
0.0467
0.0594
1981
0
0.0764
0.2639
*A11 other vehicles are assumed to be gasoline-powered. Values for Transit Buses and
Other Buses were estimated as described in Section 5.2.
The 1960-1981 regulatory class distributions were derived from the 1999 IHS data and VIUS.
Motorcycles (sourceTypelD 11 and regClassID 10) and passenger cars (sourceTypelD 21 and
regClassID 20) have one-to-one relationships between source types and regulatory classes for all
model years. Passenger trucks (sourceTypelD 31) and light commercial trucks (sourceTypelD
32) are split between fuel type and regulatory class (regClassID 30 and 40) as shown in Table A-
3.
Table A-3 Percentage by regulatory class and fuel type for passenger trucks (sourceTypelD 31) and
light commercial truck (sourceTypelD 32)
Passenger Trucks (31)
Light Commercial Trucks (32)
Gasoline
Diesel
Gasoline
Diesel
Model Year
LDT
(30)
LHD
(40)
LDT
(30)
LHD
(40)
LDT
(30)
LHD
(40)
LDT
(30)
LHD
(40)
1960-1966
81%
19%
38%
62%
24%
76%
7%
93%
1967
90%
10%
38%
62%
72%
28%
7%
93%
1968
88%
12%
38%
62%
67%
33%
7%
93%
1969
100%
0%
38%
62%
91%
9%
7%
93%
1970
99%
1%
38%
62%
80%
20%
7%
93%
1971
96%
3%
38%
62%
94%
6%
7%
93%
1972
96%
4%
38%
62%
75%
25%
7%
93%
1973
95%
5%
38%
62%
59%
41%
7%
93%
1974
95%
5%
38%
62%
65%
35%
7%
93%
1975
97%
3%
38%
62%
72%
28%
7%
93%
1976
95%
5%
38%
62%
88%
12%
7%
93%
1977
89%
11%
38%
62%
79%
21%
7%
93%
1978
85%
15%
38%
62%
81%
19%
7%
93%
1979
87%
13%
38%
62%
78%
22%
7%
93%
1980
90%
10%
38%
62%
74%
26%
40%
60%
1981
96%
4%
38%
62%
89%
11%
12%
88%
The school bus regulatory class fractions were reused from MOBILE6, originally based on 1996
and 1997 IHS data. The 1960-1981 regulatory class distributions for diesel-fueled single-unit and
combination trucks have been summarized in Table A-4 below. All 1960-1981 gasoline-fueled
single-unit and combination trucks fall into the medium heavy-duty (MHD) regulatory class
(regClassID 46).
157
-------
Table A-4 Percentange of MHD trucks (regClass 46) among diesel-fueled single-unit and
combination trucks*
Model Year
Refuse Trucks
(51)
Single-Unit
Trucks (52 & 53)
Short-haul Combination
Trucks (61)
Long-haul Combination
Trucks (62)
1960-1972
100%
0%
0%
0%
1973
100%
3%
8%
0%
1974
0%
6%
30%
0%
1975
0%
14%
3%
0%
1976
0%
44%
13%
0%
1977
0%
43%
31%
0%
1978
0%
36%
18%
0%
1979
0%
34%
16%
0%
1980
0%
58%
29%
5%
1981
0%
47%
31%
6%
* For these source types, all remaining trucks are in the HHD regulatory class (regClassID 47)
• 1 Distributions for Mod I '» ,n I M I ' >9
VIUS was our main source of information for determining fuel and regulatory class fractions for
these model years. Table A-5 summarizes how the VIUS2002 parameters were used to classify
the VIUS data to calculate fuel and regulatory class fractions for the light-duty, single-unit and
combination truck source types.
Axle arrangement (AXLE CONFIG) was used to define four categories: straight trucks with two
axles and four tires (codes 1, 6, 7, 8), straight trucks with two axles and six tires (codes 2, 9, 10,
11), all straight trucks (codes 1-21) and all tractor-trailer combinations (codes 21+). Primary
distance of operation (PRIMARYTRIP) was used to define short-haul (codes 1-4) for vehicles
with primary operation distances less than 200 miles and long-haul (codes 5-6) for 200 miles and
greater. The VIN-decoded gross vehicle weight (ADM_GVW) and survey weight (VIUS_GVW)
were used to distinguish vehicles less than 10,000 lbs. as light-duty and vehicles greater than or
equal to 10,000 lbs. as heavy-duty. Any vehicle with two axles and at least six tires was
considered a single-unit truck regardless of weight. We also note that refuse trucks have their
own VIUS vocational category (BODYTYPE 21) and that MOVES distinguishes between
personal (OPCLASS 5) and non-personal use.
158
-------
Table A-5 VIUS2002 parameters used to distinguish trucks in previous versions of MOVES
Source Type
Axle
Arrangement
Primary
Distance of
Operation
Weight
Body Type
Operator
Class
Passenger
Tracks
AXLE CONFIG
in(1,6,7,8)*
Any
ADM GVW in (1,2) &
VIUS GVW in (1,2,3)
Any
OPCLASS
=5
Light
Commercial
Tracks
AXLE CONFIG
in (1,6,7,8)*
Any
ADM GVW in (1,2) &
VIUS GVW in (1,2,3)
Any
OPCLASS
*5
Refuse
Tracks"
AXLE CONFIG
in (2,9,10,11)
TRIP PRIMARY
in (1,2,3,4)
Any
BODYTYPE
=21
Any
AXLE CONFIG
<=21
TRIP PRIMARY
in (1,2,3,4)
ADM GVW >2 &
VIUS GVW >3
BODYTYPE
=21
Any
Single-Unit
Short-Haul
Tracks"
AXLE CONFIG
in (2,9,10,11)
TRIP PRIMARY
in (1,2,3,4)
Any
BODYTYPE
#1
Any
AXLE CONFIG
<=21
TRIP PRIMARY
in (1,2,3,4)
ADM GVW >2 &
VIUS GVW >3
BODYTYPE
*21
Any
Single-Unit
Long-Haul
Tracks"
AXLE CONFIG
in (2,9,10,11)
TRIPPRIMARY
in (5,6)
Any
Any
Any
AXLE CONFIG
<=21
TRIPPRIMARY
in (5,6)
ADM GVW >2 &
VIUS GVW >3
Any
Any
Combination
Short-Haul
Tracks
AXLE CONFIG
>=21
TRIP PRIMARY
in (1,2,3,4)
Any
Any
Any
Combination
Long-Haul
Tracks
AXLE CONFIG
>=21
TRIPPRIMARY
in (5,6)
Any
Any
Any
In the MOVES2014 analysis, we did not constrain axle configuration of light-duty tracks, so there are some,
albeit very few, light-duty trucks that have three axles or more and/or six tires or more. These vehicles are
classified as light-duty tracks based primarily on their weight. Only 0.27 percent of light-duty tracks have such
tire and/or axle parameters and they have a negligible impact on vehicle populations and emissions.
For a source type with multiple rows, the source type is applied to any vehicle with either set of parameters.
Source Type Definitions
Motorcycles and passenger cars in MOVES borrow vehicle definitions from the FHWA
Highway Performance Monitoring System (HPMS) classifications from the Highway Statistics
Table MV-1. Source type definitions for school buses are taken from various US Department of
Transportation sources. While refuse trucks were identified and separated from other single-unit
trucks in VIUS, motor homes were not.
Light-Du' icks
Light-duty trucks include pickups, sport utility vehicles (SUVs) and vans.22 Depending on use
and GVWR, we categorize them into two different MOVES source types: 1) passenger trucks
(sourceTypelD 31) and 2) light commercial trucks (sourceTypelD 32). FHWA's vehicle
classification specifies that light-duty vehicles are those weighing less than 10,000 pounds,
specifically vehicles with a GVWR in Class 1 and 2, except Class 2b trucks with two axles or
more and at least six tires are assigned to the single-unit truck category.
VIUS contains many survey questions on weight; we chose to use both a VIN-decoded gross
vehicle weight rating (ADM GVW) and a respondent self-reported GVWR (VIUSGVW) to
differentiate between light-duty and single-unit trucks. For the passenger trucks, there is a final
159
-------
VIUS constraint that the most frequent operator classification (OPCLASS) must be personal
transportation. Inversely, light commercial trucks (sourceTypelD 32) have a VIUS constraint
that their most frequent operator classification must not be personal transportation.
Buses
Previous versions of MOVES had three bus source types: intercity (sourceTypelD 41), transit
(sourceTypelD 42) and school buses (sourceTypelD 43). Since the definition of sourceTypelDs
41 and 42 changed in MOVES3, only school bus distributions for model years prior to 2000
were retained in the current version of MOVES. According to FHWA, school buses are defined
as vehicles designed to carry more than ten passengers, used to transport K-12 students between
their home and school.
Siiigle-Uii neks
The single-unit HPMS class in MOVES consists of refuse trucks (sourceTypelD 51), short-haul
single-unit trucks (sourceTypelD 52), long-haul single-unit trucks (sourceTypelD 53) and motor
homes (sourceTypelD 54). FHWA's vehicle classification specifies that a single-unit truck as a
single-frame truck with a gross vehicle weight rating of greater than 10,000 pounds or with two
axles and at least six tires—colloquially known as a "dualie." As with light-duty truck source
types, single-unit trucks are sorted using VIUS parameters, in this case that includes axle
configuration (AXLE CONFIG) for straight trucks (codes 1-21), vehicle weight (both
ADM GVW and VIUS GVW), most common trip distance (TRIPPRIMARY) and body type
(BODYTYPE). All short-haul single-unit trucks must have a primary trip distance of 200 miles
or less and must not be refuse trucks and all long-haul trucks must have a primary trip distance of
greater than 200 miles. Refuse trucks are short-haul single-unit trucks with a body type (code 21)
for trash, garbage, or recyclable material hauling. Motor home distributions from previous
versions were not retained in the current version of MOVES, and therefore these vehicles are not
discussed further in this section.
Combinatioi :ks
A combination truck is any truck-tractor towing at least one trailer according to VIUS. MOVES
divides these tractor-trailers into two MOVES source types: short-haul (sourceTypelD 61) and
long-haul combination trucks (sourceTypelD 62). Like single-unit trucks, short-haul and long-
haul combination trucks are distinguished by their primary trip length (TRIP PRIMARY) in
VIUS. If the tractor-trailer's primary trip length is equal to or less than 200 miles, then it is
considered short-haul. If the tractor-trailer's primary trip length is greater than 200 miles, then it
is considered long-haul. Short-haul combination trucks are older than long-haul combination
trucks and these short-haul trucks often purchased in secondary markets, such as for drayage
applications, after being used primarily for long-haul trips.107
Fuel Type ai> Mm gulatory Class Dis111IIm>lions
The SampleVehiclePopulation table fractions were developed by EPA using the sample vehicle
counts data, which primarily joins calendar year 2011 registration data from IHS and the 2002
Vehicle Inventory and Use Survey (VIUS) results. The sample vehicle counts data were
generated by multiplying the 2011 IHS vehicle populations by the source type allocations from
VIUS.
160
-------
While VIUS provide source type classifications, we relied primarily on the 2011 IHS vehicle
registration data to form the basis of the fuel type and regulatory class distributions in the
SampleVehiclePopulation table. The IHS data were provided with the following fields: vehicle
type (cars or trucks), fuel type, gross vehicle weight rating (GVWR) for trucks, household
vehicle counts and work vehicle counts. We combined the household and work vehicle counts.
The MOVES distinction between personal and commercial travel for light-duty trucks comes
from VIUS.
The IHS records by FHWA truck weight class were grouped into MOVES GVWR-based
regulatory classes, as shown in Table A-6 below. As stated above, all passenger cars were
assigned to regClassID 20. The mapping of weight class to regulatory class is straightforward
with one notable exception: delineating trucks weighing more or less than 8,500 lbs.
Table A-6 Initial mapping from FHWA truck classes to MOVES regulatory classes
Vehicle Category
FHWA Truck Weight Class
Weight Range
(lbs.)
regClassID
Trucks
1
< 6,000
30
Trucks
2a
6,001-8,500
30*
Trucks
2b
8,501 - 10,000
41*
Trucks
3
10,001 - 14,000
41
Trucks
4
14,001 - 16,000
42
Trucks
5
16,001 - 19,500
42*
Trucks
6
19,501 -26,000
46
Trucks
7
26,001 -33,000
46
Trucks
8a
33,001 -60,000
47
Trucks
8b
>60,001
47
Cars
20
* After the IHS data had been sorted into source types (described later in this section), some regulatory classes
were merged or divided. Any regulatory class 41 vehicles in light-duty truck source types were reclassified into
the new regulatory class 40 (see explanation in Section 2.3), any regulatory class 30 vehicles in single-unit truck
source types were reclassified into regulatory class 41 and any regulatory class 42 vehicles in combination truck
source types were reclassified into regulatory class 46.
Since the IHS dataset did not distinguish between Class 2a (6,001-8,500 lbs.) and Class 2b
(8,501-10,000 lbs.) trucks, but MOVES regulatory classes 30, 40 and 41 all fall within Class 2,
we needed a secondary data source to allocate the IHS gasoline and diesel trucks between Class
2a and 2b. We derived information from an Oak Ridge National Laboratory (ORNL) paper,108
summarized in Table A-7, to allocate the IHS Class 2 gasoline and diesel trucks into the
regulatory classes. Class 2a trucks fall in regulatory class 30 and Class 2b trucks fall in
regulatory class 41.
161
-------
Table A-7 Fractions used to distribute Class 2a and 2b trucks
Truck Class
Fuel Type
Gasoline
Diesel
2a
0.808
0.255
2b
0.192
0.745
Additionally, the IHS dataset includes a variety of fuels, some that are included in MOVES and
others that are not. Only the IHS diesel, gasoline, or gasoline and another fuel were included in
our analysis; all other alternative fuel vehicles were omitted. While MOVES2014 did model
light-duty E-85 and electric vehicles, these relative penetrations of alternative fuel vehicles have
been developed from secondary data sources rather than IHS because IHS excludes some
government fleets and retrofit vehicles that could potentially be large contributors to these
alternative fuel vehicle populations. Instead, we used flexible fuel vehicle sales data reported for
EPA certification. Table A-8 illustrates how IHS fuels were mapped to MOVES fuel types and
which IHS fuels were not used in MOVES.
The "N/A" mapping shown in Table A-8 led us to discard 0.22 percent, roughly 530,000 vehicles
(mostly dedicated or aftermarket alternative fuel vehicles), of IHS's 2011 national fleet in
developing the default fuel type fractions. However, because the MOVES national population is
derived top-down from FHWA registration data, as outlined in Section 4.1, the total population
is not affected. We considered the IHS vehicle estimates to be a sufficient sample for the fuel
type and regulatory class distributions in the SampleVehiclePopulation table.
162
-------
Table A-8
ist of fuels from the IHS dataset used to develop MOVES fuel type distributions
IHS Fuel Type
MOVES fuelTypelD
MOVES Fuel Type
Unknown
N/A
Undefined
N/A
Both Gas and Electric
1
Gasoline
Gas
1
Gasoline
Gas/Elec
1
Gasoline
Gasoline
1
Gasoline
Diesel
2
Diesel
Natural Gas
N/A
Compressed Natural Gas
N/A
Natr.Gas
N/A
Propane
N/A
Flexible (Gasoline/Ethanol)
1
Gasoline
Flexible
1
Gasoline
Electric
N/A
Cnvrtble
N/A
Conversion
N/A
Methanol
N/A
Ethanol
1
Gasoline
Convertible
N/A
Next, we transformed the VIUS dataset into MOVES format. The VIUS vehicle data were first
assigned to MOVES source types using the constraints in Table A-5, and then to MOVES
regulatory classes using the mapping described in Table A-6, including the allocation between
Class 2a and 2b trucks from the ORNL study in Table A-7. Similar to our fuel type mapping of
the IHS dataset, we chose to omit alternative fuel vehicles, as summarized below in Table A-9.
163
-------
Table A-9 Map
ping of VIUS2002
fuel types to MOVES fuel types
VIUS Fuel Type
VIUS Fuel Code
MOVES fuelTypelD
MOVES Fuel Type
Gasoline
1
1
Gasoline
Diesel
2
2
Diesel
Natural gas
3
N/A
Propane
4
N/A
Alcohol fuels
5
N/A
Electricity
6
N/A
Gasoline and natural gas
7
1
Gasoline
Gasoline and propane
8
1
Gasoline
Gasoline and alcohol fuels
9
1
Gasoline
Gasoline and electricity
10
1
Gasoline
Diesel and natural gas
11
2
Diesel
Diesel and propane
12
2
Diesel
Diesel and alchol fuels
13
2
Diesel
Diesel and electricity
14
2
Diesel
Not reported
15
N/A
Not applicable
16
N/A
This process yielded VIUS data by MOVES source type, model year, regulatory class and fuel
type. The VIUS source type distributions were calculated in a similar fashion to the
SampleVehiclePopulation fractions discussed above for each regulatory class-fuel type-model
year combination. Stated formally, for any given model year i, regulatory class j, and fuel type
/c, the source type population fraction / for a specified source type I will be the number of VIUS
trucks JV in that source type divided by the sum of VIUS trucks across the set of all source types
L. The source type population fraction is summarized in Equation A-l:
f(VIUS)u
k.l
1 *
*—>leL
Equation A-l
The VIUS data in our analysis spanned model year 1986 to 2002. The 1986 distribution was used
for all prior to MY 1986.
From there the source type distributions from VIUS were multiplied by the IHS vehicle
populations to generate the sample vehicle counts by source type. Expressed in Equation A-2, the
sample vehicle counts are:
NCSVP);j,fc,j = P(Polk)i jXi ¦ f (VIUS)Equation A-2
164
-------
where JV is the number of vehicles used to generated the SampleVehiclePopulation table, P is the
2011 IHS vehicle populations and / is the source type distributions from VIUS.
Polk 2011
Vehicle Category
Model Year
Fuel Type
GVWR
Household Units
Work Units
INTERCITY BUSES
TRANSIT BUSES
SCHOOL BUSES
Interim Polk
Interim VIUS
r »
VIUS 2002
SAMPLE®
AXLECONFIG
TRIPPRIMARY
OPCLASS
FUEL
\TUS_GVW
ADMMODELYEAR
ADM_GVW
TAB TRUCKS
MOTORCYCLES
MOTOR HOMES
Sample Vehicle Counts
Figure A-l Flowchart of data sources of fuel and regulatory class distributions for model years
1982-1999
These sample vehicle counts by source type were then utilized to calculate the sample vehicle
population fractions, stmyFraction and stmyFuelEngFraction, as defined above. For simplicity,
we also moved the small number of LHD45 (regClassID 42) vehicles in combination truck
source types to MHD (regClassID 46). The source mass and road4oad coefficients for
combination trucks are only developed for MHD, HHD and Glider vehicles.
As noted above, the initial sample vehicle counts dataset did not contain buses, so information on
these source types was appended. In the subsections below, we have provided more detailed
descriptions by source type.
Appendix A.2.2.1 Motorcycles
The representation of motorcycles in the SampleVehiclePopulation table is straightforward. All
motorcycles fall into the motorcycle regulatory class (regClassID 10) and must be fueled by
gasoline.
Appendix A.2.2.2 Passenger Cars
Any passenger car is considered to be in the light-duty vehicle regulatory class (regClassID 20).
Cars were included in the IHS dataset purchased in 2012 and EPA's subsequent sample vehicle
counts dataset, which provided the split between gasoline and diesel cars in the
SampleVehiclePopulation table. Flexible fuel (E85-capable) cars were also included in the SVP
fuel type distributions but added after the sample vehicle counts analysis. We assume that a
flexible fuel vehicle would directly displace its gasoline counterpart. For model years 2011 and
165
-------
earlier, we used manufacturer reported sales to EPA in order to calculate the fraction of sales of
flexible fuel cars among sales of all gasoline and flexible fuel cars and added those penetrations
as the fraction of E85 (fuelTypelD 5) vehicles and deducted them from the gasoline cars in the
IHS dataset.
dix A.2.2.3 Light-Duty Trucks
Since passenger and light commercial trucks are defined as light-duty vehicles, they are
constrained to regulatory class 30 and 40. Within the sample vehicle counts, GVWR Class 1 and
2a trucks were classified as regulatory class 30 and Class 2b trucks with two axles and four tires
were classified as regulatory class 40. Both light-duty truck source types are divided between
gasoline and diesel using the underlying splits in the sample vehicle counts data. Passenger
trucks and light commercial trucks have similar but distinct distributions. Similar to cars, a
penetration of flexible fuel (E-85-capable) light-duty trucks was calculated using EPA
certification sales for MY 2011 and earlier.
dix A.2.2.4 Buses
Only school bus distributions from MOVES2014 for model years prior to 2000 were retained in
the current version of MOVES. The MOVES2014 school bus fuel type distributions were based
on MOBILE6 estimates, originally calculated from 1996 and 1997 IHS bus registration data, for
model years 1982-1996 and are summarized in Table A-10. The Union of Concerned Scientists
estimates that roughly one percent of school buses run on non-diesel fuels, so we have assumed
that one percent of school buses were gasoline fueled for MY 1997 and later.109 The school bus
regulatory class distributions were also derived from 2011 FHWA data110 as listed in Table A-
11, which were applied to model years prior to 2000 for both gasoline and diesel.
166
-------
Table A-10 Fuel type market shares by model year for school buses
Model Year
MOVES Fuel Type
Gasoline
Diesel
1982
67.40%
32.60%
1983
67.62%
32.38%
1984
61.55%
38.45%
1985
48.45%
51.55%
1986
32.67%
67.33%
1987
26.55%
73.45%
1988
24.98%
75.02%
1989
22.90%
77.10%
1990
12.40%
87.60%
1991
8.95%
91.05%
1992
1.00%
99.00%
1993
12.05%
87.95%
1994
14.75%
85.25%
1995
11.43%
88.57%
1996
4.15%
95.85%
1997-1999
1.00%
99.00%
Table A-ll Regulatory class fractions of school buses using 2011 FHWA data
Vehicle Type
MOVES regClassID
41
42
46
47
Total
School Buses
0.0106
0.0070
0.9371
0.0453
1
'I"I" A.2.2.5 Single-1 nit ar> I 1 urn Miration 'I i ucks
The fuel type and regulatory class distributions for the single-unit and combination trucks were
calculated directly from the EPA's sample vehicle counts datasets. The single-unit and short-haul
combination truck source types were split between gasoline and diesel only and long-haul
combination trucks only contained diesel vehicles. Single-unit vehicles were distributed among
all the heavy-duty regulatory classes (regClassIDs 41, 42, 46 and 47) and combination trucks
were distributed among the MHD and HHD regulatory classes (46 and 47) based on the
underlying sample vehicle data.
167
-------
Appendix B 1990 Age Distributions
In the current version of MOVES, the 1990 age distributions were unchanged from previous
versions of the model. This appendix describes their derivation; details on the derivations of the
other age distributions in MOVES may be found in Appendix C.
Motorcycles
The motorcycle age distributions are based on Motorcycle Industry Council estimates of the
number of motorcycles in use, by model year, in 1990. However, data for individual model years
starting from 1978 and earlier were not available. A logarithmic regression curve (R2 value =
0.82) was fitted to available data, which was then used to extrapolate age fractions for earlier
years beginning in 1978.
112. Passenger Cars
To determine the 1990 age fractions for passenger cars, we began with IHS NVPP® 1990 data
on car registration by model year. However, this data presents a snapshot of registrations on July
1, 1990 and we needed age fractions as of December 31, 1990. To adjust the values, we used
monthly data from the IHS new car database to estimate the number of new cars registered in the
months July through December 1990. Model Year 1989 cars were added to the previous estimate
of "age 1" cars and Model Year 1990 and 1991 cars were added to the "age 0" cars. Also the
1990 data did not detail model year for ages 15+. Hence, regression estimates were used to
extrapolate the age fractions for individual ages 15+ based on an exponential curve (R2 value
=0.67) fitted to available data.
113. :ks
For the 1990 age fractions for passenger trucks, light commercial trucks, refuse trucks, short-haul
and long-haul single-unit trucks and short-haul and long-haul combination trucks, we used data
from the TIUS92 (1992 Truck Inventory and Use Survey) database. Vehicles in the TIUS92
database were assigned to MOVES source types as summarized in Table B-l. TIUS92 does not
include a model year field and records ages as 0 through 10 and 11-and-greater. Because we
needed greater detail on the older vehicles, we determined the model year for some of the older
vehicles by using the responses to the questions "How was the vehicle obtained?" (TIUS field
"OBTAIN") and "When did you obtain this vehicle?" (TIUS field "ACQYR") and we adjusted
the age-11-and-older vehicle counts by dividing the original count by model year by the fraction
of the older vehicles that were coded as "obtained new."
168
-------
Table B-l VIUS1997 codes used for distinguishing truck source types
Source Type
Axle
Arrangement
Primary Area
of Operation
Body Type
Major Use
Passenger
Trucks
2 axle/4 tire
(AXLRE= 1,5,6,7)
Any
Any
personal
transportation
(MAJUSE=20)
Light
Commercial
Trucks
2 axle/4 tire
(AXLRE= 1,5,6,7)
Any
Any
any but personal
transportation
Refuse Trucks
Single-Unit
(AXLRE=2-4, 8-
16)
Off-road, local
or short-range
(AREAOP <=4)
Garbage hauler
(BODTYPE=30)
Any
Single-Unit
Short-Haul
Trucks
Single-Unit
(AXLRE=2-4, 8-
16)
Off-road, local
or short-range
(AREAOP<=4)
Any except
garbage hauler
Any
Single-Unit
Long-Haul
Trucks
Single-Unit
(AXLRE=2-4, 8-
16)
Long-range
(AREAOP>=5)
Any
Any
Combination
Short-Haul
Trucks
Combination
(AXLRE>=17)
Off-road, local
or medium
(AREAOP<=4)
Any
Any
Combination
Long-Haul
Trucks
Combination
(AXLRE>=17)
Long-range
(AREAOP>=5)
Any
Any
Other Buses
For 1990, we were not able to identify a data source for estimating age distributions of other
buses. Because the purchase and retirement of these buses is likely to be driven by general
economic forces rather than trends in government spending, we will use the 1990 age
distributions that were derived for short-haul combination trucks, as described above.
115. School IIii! :s
To determine the age fractions of school buses and motor homes, we used information from the
IHS TIP® 1999 database. School bus and motor home counts were available by model year.
Unlike the IHS data for passenger cars, these counts reflect registration at the end of the calendar
year and, thus, did not require adjustment. We converted model year to age and calculated age
fractions. Because we did not have access to 1990 data, these fractions were used for 1990.
116. isit Buses
For 1990 Transit Bus age distributions, we used the MOBILE6 age fractions since 1990 data on
transit buses was not available from the Federal Transit Administration database. MOBILE6 age
fractions were based on fitting curves through a snapshot of vehicle registration data as of July 1,
1996, which was purchased from IHS (then known as R.L. Polk Company). To develop a general
curve, the 1996 model year vehicle populations were removed from the sample because it did not
represent a full year and a best-fit analysis was performed on the remaining population data. The
best-fit analyses resulted in age distribution estimates for vehicles ages 1 through 25+. However,
since the vehicle sales year begins in October, the estimated age 1 population was multiplied by
0.75 to account for the fact that approximately 75 percent of the year's sales will have occurred
by July 1st of a given calendar year.
169
-------
Both Weibull curve fitting and exponential curve fitting were used to create the age distributions.
The nature of the Weibull curve fitting formula is to produce an "S" shaped curve, which is
relatively flat for the first third of the data, decreases rapidly for the next third and flattens again
for the final third. While using this formula resulted in a better overall fit for transit buses, the
flatness of the final third for each curve resulted in unrealistically low vehicle populations for the
older vehicle ages. For this reason, the original Weibull curve was used where it fit best and
exponential curves were fit through the data at the age where the Weibull curves began to flatten.
Table B-2 presents the equations used to create the age distribution and the years in which the
equations were used.
Table B-2 Curve fit equations for registration distribution data by age
Vehicle
Age
Equation
1-17
(( age *12.53214119\
y = 3462 * e U17.16909475J )
18-25+
24987.0776 * e"0-2000*^
170
-------
Appendix C Detailed Derivation of Age Distributions
Since purchasing registration data for all calendar years is prohibitively costly for historic years,
the base age distribution described in Section 6.1 and presented below is forecast and backcast
for all other calendar years in the model. While sales data for historic years are well known and
projections for future years are common in economic modeling, national trends in vehicle
survival for every MOVES source type at all ages are not well studied. For the analysis presented
in this appendix, a generic survival rate was scaled up or down for each calendar year based on
our assumptions of sales and changes in total populations. The following sections summarize the
derivation of the generic survival rate, the estimation of vehicle sales by source type and the
algorithms used to forecast and backcast age distributions for each year.
1 II, Vehicle Survival by I pe
The survival rate describes the fraction of vehicles of a given source type and age that remain on
the road from one year to the next. Although this rate changes from year to year, a single generic
rate was calculated from available data.
Survival rates for motorcycles were calculated based on a smoothed curve of retail sales and
2008 national registration data as described in a study conducted for the EPA.111 Survival rates
for passenger cars, passenger trucks and light commercial trucks came from NHTSA's
survivability Table 3 and Table 4.112 These survival rates are based on a detailed analysis of IHS
vehicle registration data from 1977 to 2002. We modified these rates to be consistent with the
MOVES format using the following guidelines:
• NHTSA rates for light trucks were used for both the MOVES passenger truck and light
commercial truck source types.
• MOVES calculates emissions for vehicles up to age 30 (with all older vehicles lumped
into the age 30 category), but NHSTA car survival rates were available only to age 25.
Therefore, we extrapolated car rates to age 30 using the estimated survival rate equation
in Section 3.1 of the NHTSA report. When converted to MOVES format, this caused a
striking discontinuity at age 26 which we removed by interpolating between ages 25 and
27.
• According to the NHTSA methodology, NHTSA age 1 corresponds to MOVES agelD 2,
so the survival fractions were shifted accordingly.
• Because MOVES requires survival rates for agelDs < 2, these values were linearly
interpolated with the assumption that the survival rate prior to agelD 0 is 1.
• NHTSA defines survival rate as the ratio of the number of vehicles remaining in the fleet
at a given year as compared to a base year. However, MOVES defines the survival rate as
the ratio of vehicles remaining from one year to the next, so we transformed the NHTSA
rates accordingly.
Quantitatively, the following piecewise formulas were used to derive the MOVES survival rates.
In them, sa represents the MOVES survival rate at age a and oa represents the NHTSA survival
rate at age a. When this generic survival rate is discussed below, the shorthand notation S0 will
represent a one-dimensional array of sa values at each permissible age a as described in
Equation C-l through Equation C-3 below:
171
-------
Age 0:
Equation C-l
Age 1:
s1 = l-
2(1-ff2)
Equation C-2
3
Ages 2-30:
°a-1
sa ~ s2...30 ~ ~
Equation C-3
With limited data available on heavy-duty vehicle scrappage, survivability for all other source
types came from the Transportation Energy Data Book.113 We used the heavy-duty vehicle
survival rates for model year 1980 (TEDB40, Table 3.17). The 1990 model year rates were not
used because they were significantly higher than rates for the other model years in the analysis
(i.e. 45 percent survival rate for 30 year-old trucks) and seemed unrealistically high. While
limited data exists to confirm this judgment, a snapshot of 5-year survival rates can be derived
from VIUS 1992 and 1997 results for comparison. According to VIUS, the average survival rate
for model years 1988-1991 between the 1992 and 1997 surveys was 88 percent. The comparable
survival rate for 1990 model year heavy-duty vehicles from TEDB was 96 percent, while the rate
for 1980 model year trucks was 91 percent. This comparison lends credence to the decision that
the 1980 model year survival rates are more in line with available data. TEDB does not have
separate survival rates for medium-duty vehicles; the heavy-duty rates were applied uniformly
across the bus, single-unit truck and combination truck categories. The TEDB survival rates were
transformed into MOVES format in the same way as the NHTSA rates.
The resulting survival rates are listed in the default database's SourceTypeAge table, shown
below in Table C-l. Please note that since MOVES does not calculate age distributions during a
run, these survival rates are not actively used by MOVES. However, they were used in the
development of the national age distributions stored in the SourceTypeAgeDistribution table and
remain in the default database for reference. In addition, the survival rates in the SourceTypeAge
table are listed by source type, but the values are identical for the grouping of vehicles listed in
the table header.
172
-------
Table C-l Vehicle survival rate bv age
Age
Motorcycles
Passenger
Cars
Light-duty Trucks
(Passenger and
Light Commercial)
Heavy-duty Vehicles
(Buses, Single-Unit Trucks
and Combination Trucks)
0
1.000
0.997
0.991
1.000
1
0.979
0.997
0.991
1.000
2
0.940
0.997
0.991
1.000
3
0.940
0.993
0.986
1.000
4
0.940
0.990
0.981
0.990
5
0.940
0.986
0.976
0.980
6
0.940
0.981
0.970
0.980
7
0.940
0.976
0.964
0.970
8
0.940
0.971
0.958
0.970
9
0.940
0.965
0.952
0.970
10
0.940
0.959
0.946
0.960
11
0.940
0.953
0.940
0.960
12
0.940
0.912
0.935
0.950
13
0.940
0.854
0.929
0.950
14
0.940
0.832
0.913
0.950
15
0.940
0.813
0.908
0.940
16
0.940
0.799
0.903
0.940
17
0.940
0.787
0.898
0.930
18
0.940
0.779
0.894
0.930
19
0.940
0.772
0.891
0.920
20
0.940
0.767
0.888
0.920
21
0.940
0.763
0.885
0.920
22
0.940
0.760
0.883
0.910
23
0.940
0.757
0.880
0.910
24
0.940
0.757
0.879
0.910
25
0.940
0.754
0.877
0.900
26
0.940
0.754
0.875
0.900
27
0.940
0.567
0.875
0.900
28
0.940
0.752
0.873
0.890
29
0.940
0.752
0.872
0.890
30
0.300
0.300
0.300
0.300
€2. Vehicle Sales by Soun »e
Knowing vehicle sales by source type for every calendar year is essential for estimating age
distributions in both historic and projected years. Since MOVES doesn't calculate age
distributions at run time, this information isn't stored in the default database.2 However, sales
data are used in the age distribution backcasting and projection algorithms, which are described
z Early versions of MOVES calculated age distributions at runtime and therefore required sales data to be stored in
the default database. Consequently, the SourceTypeYear table has a salesGrowthFactor column. Since MOVES no
longer needs this information, this column contains Os in the current MOVES default database.
173
-------
in subsequent sections. They are also used in calculating the age 0 fractions of vehicles in the
base age distribution, which is described in Section 6.1.1.
Historic motorcycles sales came from the Motorcycle Industry Council's 2015 Motorcycle
Statistical Annual,114 which contains estimates of annual on-highway motorcycle sales going
back to 1989. Sales for calendar years 2015-2020 were estimated as a constant proportion of the
total motorcycle stock, using the ratio of 2014 sales to population.
Historic passenger car sales came from the TEDB40 Table 4.6 estimate for total new retail car
sales.
Historic light truck sales came from the TEDB40 Table 4.7 estimate for total light truck sales.
These were then split into passenger truck and light commercial truck sales using the source type
distribution fractions described in Section 4.1.
Historic school bus sales came from the 2001, 2010, 2020, and 2023 publications of School Bus
Fleet Fact Book1* Each publication contains estimates for 10 years of historic annual national
sales. Sales for before 1990 were estimated as a constant proportion of the total school bus stock,
using the ratio of 1999 sales to population.
Historic transit bus sales were calculated from the Federal Transit Administration's National
Transit Database (NTD)19 data series on Revenue Vehicle Inventory and Rural Revenue Vehicle
Inventory. Since the annual publication does not necessarily contain all model year vehicles sold
in the year of publication, transit bus sales are instead estimated from 1-year-old buses. This
assumes 0 scrappage of new transit buses, which is consistent with the heavy-duty survival rate
presented in Table C-l. The 1-year-old transit bus populations were estimated from the NTD
active fleet vehicles using the definition of a transit bus as given in Section 5.1.4. Since the
Revenue Vehicle Inventory tables are not available for years before 2002, sales for 1990 and
1999-2001 were estimated as a constant proportion of the total transit bus stock, using the ratio
of 2002 sales to population.
Lacking a direct source of historic other bus sales, these were derived from the average sales rate
for school buses and transit buses. The ratio of total school and transit bus sales to school and
transit bus populations was applied to the other bus population, as shown in Equation C-4 below.
The historic populations for each of the bus source types were determined as described in
Section 4.1.
„ , school Sales transit n „ .
— ¦ — ¦ Popothey Equation C-4
PoVschool tOPtransit
Historic sales for heavy-duty trucks were derived from the TEDB40 Table 5.3 estimate for truck
sales by gross vehicle weight. These were translated to source type sales by calculating the
source type distribution for each weight class 3-8 from the 2020 IHS data. Since the 2020 IHS
data grouped short-haul (52) and long-haul (53) single-unit trucks, sales were further allocated to
the individual source types 52 and 53 using the source type distribution fractions described in
Section 4.1.
174
-------
Projected sales for all source types were derived from AEO2023. Because AEO vehicle
categories differ from MOVES source types, the AEO projected vehicle sales were not used
directly. Instead, ratios of vehicle sales to stock were calculated and applied to the projected
populations (see Section 4.2for the derivation of projected populations). Since AEO2023 only
projects out to 2050, sales for years 2051-2060 were assumed to continue to grow at the same
growth rate as between 2049 and 2050.
Table C-2 shows the mappings between AEO sales categories and MOVES source types. Where
multiple AEO categories are listed, their values were summed before calculating the sales to
stock ratios. These are the same groupings as presented for the stock categories in Table 4-3 and
more details on the selection of the groupings may be found in Section4.2. We acknowledge that
using sales projections from different vehicle types as surrogates for motorcycles and buses in
particular will introduce additional uncertainty into these projections.
The sales to stock ratios for each year and group were calculated and applied to the projected
source type populations using the mappings given in to derive projected sales for each source
type.
Table C-2 Mapping AEO categories to source types for projecting vehicle populations
AEO Sales Category Groupings
MOVES Source Type
Total Car Sales1
11 - Motorcycle
21 - Passenger Car
Total Light Truck Sales1
+
Total Commercial Light Truck Sales11
31 - Passenger Truck
32 - Light Commercial Truck
Total Sales111
41 - Other Bus
42 - Transit Bus
43 - School Bus
Light Medium Subtotal Sales111
+
Medium Subtotal Sales111
51 - Refuse Truck
52 - Single-Unit Short-haul Truck
53 - Single-Unit Long-haul Truck
54 - Motor Home
Heavy Subtotal Sales111
61 - Combination Short-haul Truck
62 - Combination Long-haul Truck
I From AEO2023 Table 39: Light-Duty Vehicle Sales by Technology Type
II From AEO2023 Table 45: Transportation Fleet Car and Truck Sales by Type and Technology
III From AEO2023 Table 50: Freight Transportation Energy Use
175
-------
Age
0
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30+
Base Year A itributions
Tab e C-3 2020 age fractions by MOVES source type
11
21
31
32
41
42
43
51
52
53
54
61
0.047761
0.037546
0.057107
0.057107
0.045031
0.045031
0.045031
0.046470
0.046470
0.046470
0.046470
0.042869
0.052400
0.041352
0.063106
0.140304
0.058830
0.046336
0.058979
0.060318
0.059067
0.059067
0.043827
0.045291
0.046272
0.045642
0.064161
0.090038
0.054369
0.082068
0.058866
0.053975
0.045495
0.045495
0.047319
0.046055
0.043723
0.052911
0.063062
0.080902
0.062808
0.087860
0.059606
0.049131
0.053115
0.053115
0.036305
0.046192
0.043596
0.054667
0.057843
0.066182
0.066487
0.093850
0.052028
0.054158
0.051697
0.051697
0.041262
0.066533
0.041426
0.059893
0.056042
0.059659
0.058041
0.079109
0.049169
0.042716
0.050433
0.050433
0.036367
0.055583
0.039651
0.057165
0.047868
0.048493
0.049059
0.077841
0.043786
0.035514
0.032395
0.032395
0.031091
0.045340
0.035567
0.060114
0.042031
0.042520
0.050891
0.063694
0.038647
0.035715
0.029942
0.029942
0.026107
0.043162
0.033039
0.053182
0.036813
0.036637
0.037046
0.058783
0.041379
0.033047
0.040353
0.040353
0.019478
0.040796
0.023515
0.040409
0.037550
0.033283
0.037753
0.051184
0.045872
0.024895
0.031275
0.031275
0.017785
0.022894
0.018800
0.040849
0.029931
0.023765
0.037120
0.060321
0.043788
0.019612
0.014166
0.014166
0.017084
0.017332
0.039971
0.036269
0.022443
0.018943
0.041298
0.058819
0.055923
0.032297
0.017457
0.017457
0.016419
0.021724
0.048526
0.045040
0.038133
0.033914
0.039965
0.043701
0.057422
0.025863
0.039435
0.039435
0.024859
0.017525
0.061493
0.047569
0.039925
0.029767
0.042408
0.033745
0.050565
0.075982
0.039150
0.039150
0.039191
0.045234
0.060752
0.042595
0.039050
0.031243
0.040985
0.031658
0.029630
0.059385
0.046464
0.046464
0.041087
0.035303
0.054562
0.039156
0.039375
0.026351
0.026023
0.022260
0.026323
0.047669
0.038304
0.038304
0.042110
0.033956
0.042953
0.033351
0.039205
0.024091
0.020094
0.015748
0.028503
0.036373
0.029104
0.029104
0.046490
0.022496
0.044374
0.031378
0.033688
0.020613
0.029568
0.013383
0.019932
0.037159
0.025630
0.025630
0.041247
0.021725
0.033927
0.026976
0.030265
0.017589
0.028384
0.015676
0.023235
0.031347
0.024356
0.024356
0.036265
0.016992
0.027320
0.022473
0.025473
0.016038
0.028227
0.009650
0.024495
0.030762
0.028509
0.028509
0.031719
0.024432
0.021720
0.020538
0.022595
0.014391
0.029434
0.004137
0.021260
0.034838
0.028612
0.028612
0.035780
0.033098
0.016109
0.015239
0.019057
0.012273
0.025464
0.001592
0.019068
0.026211
0.027957
0.027957
0.041354
0.028281
0.011883
0.012193
0.014233
0.008508
0.020482
0.001241
0.015783
0.015993
0.014209
0.014209
0.032232
0.023972
0.009585
0.009815
0.013347
0.008593
0.012043
0.000953
0.014303
0.011716
0.016581
0.016581
0.027771
0.021106
0.008580
0.007195
0.009032
0.005820
0.010606
0.000666
0.011697
0.011972
0.012626
0.012626
0.024841
0.020950
0.007140
0.006689
0.008858
0.005957
0.007381
0.000234
0.012115
0.013910
0.014881
0.014881
0.022691
0.026790
0.005851
0.004924
0.007553
0.005018
0.005601
0.000225
0.006658
0.007677
0.010121
0.010121
0.021594
0.018307
0.005229
0.004031
0.005211
0.003611
0.004737
0.000072
0.007233
0.006233
0.008673
0.008673
0.018133
0.014518
0.003769
0.003459
0.003903
0.003003
0.002651
0.000027
0.006134
0.005228
0.007137
0.007137
0.015002
0.011214
0.003161
0.003253
0.003326
0.002686
0.002428
0.000018
0.007586
0.005831
0.007758
0.007758
0.012387
0.011864
0.067345
0.044129
0.029811
0.032701
0.024786
0.000117
0.024985
0.028002
0.108627
0.108627
0.065731
0.078464
176
-------
( 4. Historic A trifoutions
The base algorithm for backcasting age distributions is as follows:
1. Calculate the base population distribution (Py) by multiplying the base age distribution
(fy) and base population (Py).
2. Remove the age 0 vehicles (Ny).
3. Decrease the population age index by one (for example, 3-year-old vehicles are
reclassified as 2-year-old vehicles).
4. Add the vehicles that were removed in the previous year (Ry_1).
5. Convert the resulting population distribution into an age distribution using Equation 6-1.
6. Replace the new age 29 and 30+ fractions with the base year age 29 and 30+ fractions
and renormalize the new age distribution to sum to 1 while retaining the original age 29
and 30+ fractions.
7. This results in the previous year age distribution (/y_i). If this algorithm is to be
repeated, fy_1 becomes fy for the next iteration.
This is mathematically described with the following equation (reprinted from Section 6.1.2 for
reference):
Py-x = Py — Ny + Ry_i Equation 6-2
Unfortunately, as described in Section CI, the only survival information we have is a single
snapshot. Because vehicle populations and new sales change differentially (for example, the
historic populations shown in Section 4.1 leveled off during the recent recession; at the same
time, sales of most vehicle types plummeted), it is important to adjust the survival curve in
response to changes in population and sales. We did so by defining a scalar adjustment factor ky
that can be algebraically calculated from population and sales estimates. Its use in calculating the
scrapped vehicles with generic survival rate S0 is given by Equation C-5 Note that the open
circle operator (o) represents entrywise product; that is, each element in an array is multiplied by
the corresponding element in the other one and it results in an array with the same number of
elements. In this case, the scalar adjustment factor is applied to the scrappage rate (1 minus the
survival rate) at each age, which is then applied to the population of vehicles at each
corresponding age; this results in the number of removed vehicles by age.
Ry-i = ky_i ¦ (l — S0) o Py_1 Equation C-5
Substituting Equation C-5 into Equation 6-2 yields Equation C-6:
Py—i = Py — Ny + ky_1 ¦ (l — S0) o Py_1 Equation C-6
To solve for ky_x, Equation C-6 can be transformed into Equation C-7 using known total
populations and sales:
177
-------
Py-1 — Py — Ny + ky_1 • ^ ((l — 50) o Py_x j Equation C-7
However, this still leaves a Py-\ term, which is unavoidable because the total number of vehicles
removed is dependent on the age distribution of those vehicles. To solve Equation C-7, an
iterative approach was used. The first time the algorithm described above is run, Py_1 is
approximated by applying the base age distribution fy to the population of the previous year
Py-i- The scaling factor ky_1 is calculated using this approximation in Equation C-7 and then a
guess for Py-\ is calculated from Equation C-6. The guess for the resulting age distribution /y_x
is then calculated using the known Py_ 1. The algorithm is repeated for the same year using the
updated guess for the resulting age distribution. This is repeated until the resulting age
distribution matches the guessed age distribution at each age fraction within 1 x 10"6, which
occurred within 10 iterations for most source types and calendar years.
This algorithm was then repeated for each historic year from 2019 to 1999 and for each source
type using the following data sources:
• Total populations Py and Py-\ as described in Section 4.
• Generic survival rates S0 as described in Section CI.
• Vehicle sales Ny as described in Section C2.
• Base age distributions f2020 as described in Section 6.1.1. All other fy come from the
fy—\ of the previous iteration.
With all of this information, the age distributions were algorithmically determined for years
1999-2019 are stored in the SourceTypeAgeDistribution table of the default database.
( 5. I i ii I e Distributions
The base algorithm for forecasting age distributions is as follows:
1. Calculate the base population distribution (Py) by multiplying the base age distribution
(fy) and base population (Py).
2. Remove the vehicles that did not survive (Ry) at each age level.
3. Increase the population age index by one (for example, 3-year-old vehicles are
reclassified as 4-year-old vehicles).
4. Add new vehicle sales (Ny+1) as the age 0 cohort.
5. Convert the resulting population distribution into an age distribution using Equation 6-1.
6. Replace the new age 30+ fraction with the base year age 30+ fraction and renormalize the
new age distribution to sum to 1 while retaining the original age 0 and age 30+ fractions.
7. This results in the next year age distribution (fy+1). If this algorithm is to be repeated,
fy+1 becomes fy for the next iteration.
178
-------
This is mathematically described with the following equation (reprinted from Section 6.1.3 for
reference):
Equation 6-3
As with the backcasting algorithm, the scrapped vehicles need to be estimated by scaling the
generic survival rate. The equation governing vehicle removal discussed the previous section is
also applicable here. Taking careful note of the subscripts, Equation 6-3 and Equation C-5 can be
combined into Equation C-8:
To solve for ky, Equation C-8 can be transformed into Equation C-9 using the population and
sales totals:
This can be algebraically solved for ky and evaluated for each source type as all of the other
values are known. Please note that the iterative approach to solving this equation as described in
the back-casting section is not necessary here, as the number of scrapped vehicles depends on the
base age distribution, which is known. After ky is calculated, Equation C-8 is used to determine
Py+1- The resulting age distribution fy+1 is then calculated using the known Py+1-
This algorithm was then repeated for each projected year from 2021 to 2060 and for each source
type using the following data sources:
• Total populations Py and Py+l as described in Section 4.
• Generic survival rates S0 as described in Section CI.
• Vehicle sales Ny+1 as described in Section C2.
• Base age distributions f2020 as described in Section 6.1.1. All other fy come from the
fy+1 of the previous iteration.
With all of this information, the age distributions were algorithmically determined for years
2021-2060 and are stored in the SourceTypeAgeDistribution table of the default database. An
illustration of passenger car age distributions is presented in Figure C-l. For clarity, only four
years are shown: 2014, 2020, 2030 and 2040.
Py.|-1 — Py ky " (1 Sq) ° Py + Ny + ^
Equation C-8
Equation C-9
179
-------
8.0%-
7,0% -
6,0% -
g 5,0%-
u
2 4.0%-
tL<
u
m
< 3,0%-
2,(
0.1
0
Calendar Year
2014
2020
2030
2040
30
10 20
Passenger Car Age
Figure C-l Selected age distributions for passenger cars in MOVES4
180
-------
Appendix D
Driving Schedules
A key feature of MOVES is the capability to accommodate a number of drive schedules to
represent driving patterns across source type, roadway type and average speed. For the national
default case, MOVES4 employs 49 drive schedules with various average speeds, mapped to
specific source types and roadway types. These are unchanged from MOVES2014.
Table D-l below lists the driving schedules used in MOVES. Some driving schedules are used
for both restricted access (freeway) and unrestricted access (non-freeway) driving. Some driving
schedules are used for multiple source types or multiple road types where vehicle specific
information was not available.
181
-------
Table D-l MOVES default driving schedule statistics
Drive
Schedule
ID
Drive Schedule Name
Avg
Speed
Max
Speed
Idle
Time
(sec)
Percent of
Time Idling
Miles
Time (sec)
Minutes
Hours
101
LD Low Speed 1
2.5
10.00
280
46.5%
0.419
602.00
10.03
0.167
153
LD LOS E Freeway
30.5
63.00
5
1.1%
3.863
456.00
7.60
0.127
158
LD High Speed Freeway 3
76.0
90.00
0
0.0%
12.264
581.00
9.68
0.161
201
MD 5mph Non-Freeway
4.6
24.10
85
29.0%
0.373
293.00
4.88
0.081
202
MD lOmph Non-Freeway
10.7
34.10
61
19.6%
0.928
311.00
5.18
0.086
203
MD 15mph Non-Freeway
15.6
36.60
57
12.6%
1.973
454.00
7.57
0.126
204
MD 20mph Non-Freeway
20.8
44.50
95
9.1%
6.054
1046.00
17.43
0.291
205
MD 25mph Non-Freeway
24.5
47.50
63
11.1%
3.846
566.00
9.43
0.157
206
MD 30mph Non-Freeway
31.5
55.90
54
5.5%
8.644
988.00
16.47
0.274
251
MD 3 Omph Freeway
34.4
62.60
0
0.0%
15.633
1637.00
27.28
0.455
252
MD 40mph Freeway
44.5
70.40
0
0.0%
43.329
3504.00
58.40
0.973
253
MD 5 Omph Freeway
55.4
72.20
0
0.0%
41.848
2718.00
45.30
0.755
254
MD 60mph Freeway
60.1
68.40
0
0.0%
81.299
4866.00
81.10
1.352
255
MD High Speed Freeway
72.8
80.40
0
0.0%
96.721
4782.00
79.70
1.328
301
HD 5mph Non-Freeway
5.8
19.90
37
14.2%
0.419
260.00
4.33
0.072
302
HD lOmph Non-Freeway
11.2
29.20
70
11.5%
1.892
608.00
10.13
0.169
303
HD 15mph Non-Freeway
15.6
38.30
73
12.9%
2.463
567.00
9.45
0.158
304
HD 20mph Non-Freeway
19.4
44.20
84
15.1%
3.012
558.00
9.30
0.155
305
HD 25mph Non-Freeway
25.6
50.70
57
5.8%
6.996
983.00
16.38
0.273
306
HD 30mph Non-Freeway
32.5
58.00
43
5.3%
7.296
809.00
13.48
0.225
351
HD 30mph Freeway
34.3
62.70
0
0.0%
21.659
2276.00
37.93
0.632
352
HD 40mph Freeway
47.1
65.00
0
0.0%
41.845
3197.00
53.28
0.888
353
HD 50mph Freeway
54.2
68.00
0
0.0%
80.268
5333.00
88.88
1.481
354
HD 60mph Freeway
59.7
69.00
0
0.0%
29.708
1792.00
29.87
0.498
355
HD High Speed Freeway
71.7
81.00
0
0.0%
35.681
1792.00
29.87
0.498
396
HD High Speed Freeway Plus 5mph
76.7
86.00
0
0.0%
38.170
1792.00
29.87
0.498
397
MD High Speed Freeway Plus 5mph
77.8
85.40
0
0.0%
103.363
4782.00
79.70
1.328
182
-------
Drive
Schedule
ID
Drive Schedule Name
Avg
Speed
Max
Speed
Idle
Time
(sec)
Percent of
Time Idling
Miles
Time (sec)
Minutes
Hours
398
CRC E55 HHDDT Creep
1.8
8.24
107
42.3%
0.124
253.00
4.22
0.070
401
Bus Low Speed Urban
3.1
19.80
288
63.9%
0.393
451.00
7.52
0.125
402
Bus 12mph Non-Freeway
11.5
33.80
109
37.5%
0.932
291.00
4.85
0.081
403
Bus 30mph Non-Freeway
21.9
47.00
116
28.3%
2.492
410.00
6.83
0.114
404
New York City Bus
3.7
30.80
403
67.2%
0.615
600.00
10.00
0.167
405
WMATA Transit Bus
8.3
47.50
706
38.4%
4.261
1840.00
30.67
0.511
501
Refuse Truck Urban
2.2
20.00
416
66.9%
0.374
622.00
10.37
0.173
1009
Final FCOILOSAF Cycle (C10R04-
00854)
73.8
84.43
0
0.0%
11.664
569.00
9.48
0.158
1011
Final FC02LOSDF Cycle (C10R05-
00513)
49.1
73.06
34
5.0%
9.283
681.00
11.35
0.189
1017
Final FC11LOSB Cycle (C10R02-00546)
66.4
81.84
0
0.0%
9.567
519.00
8.65
0.144
1018
Final FC11LOSC Cycle (C15R09-00849)
64.4
78.19
0
0.0%
16.189
905.00
15.08
0.251
1019
Final FC11LOSD Cycle (C15R10-00068)
58.8
76.78
0
0.0%
11.922
730.00
12.17
0.203
1020
Final FC11LOSE Cycle (C15R11-00851)
46.1
71.50
1
0.1%
12.468
973.00
16.22
0.270
1021
Final FC11LOSF Cycle (C15R01-00876)
20.6
55.48
23
2.5%
5.179
905.00
15.08
0.251
1024
Final FC12LOSC Cycle (C15R04-00582)
63.7
79.39
0
0.0%
15.685
887.00
14.78
0.246
1025
Final FC12LOSD Cycle (C15R09-00037)
52.8
73.15
12
1.5%
11.754
801.00
13.35
0.223
1026
Final FC12LOSE Cycle (C15R10-00782)
43.3
70.87
0
0.0%
10.973
913.00
15.22
0.254
1029
Final FC14LOSB Cycle (C15R07-00177)
31.0
63.81
27
3.6%
6.498
754.00
12.57
0.209
1030
Final FC14LOSC Cycle (C10R04-00104)
25.4
53.09
41
8.0%
3.617
513.00
8.55
0.143
1033
Final FC14LOSF Cycle (C15R05-00424)
8.7
44.16
326
38.2%
2.066
853.00
14.22
0.237
1041
Final FC17LOSD Cycle (C15R05-00480)
18.6
50.33
114
16.1%
3.659
709.00
11.82
0.197
1043
Final FC19LOSAC Cycle (C15R08-
00267)
15.7
37.95
67
7.7%
3.802
870.00
14.50
0.242
183
-------
Appendix E
Total Idle Fraction Regression Coefficients
Table E-l displays the regression coefficients for the linear model used to estimate variation in
total idle fraction for light-duty vehicles presented in Equation 10-6 discussed in Section 10.2.3.
Table E-l Total idle fraction regression coefficients for light-duty vehicles trucks in urban counties
for weekdays
Variable
Coefficients
Comments
(Intercept)
0.20977
day ID 5
0.011262
Applicable when dayID=5
sourceTypeID31
0.001329
Applicable when sourceTypeID=31
county TypelDl
0.03058
Applicable when equation is used for an urban county
(county TypeID= 1)
idlcRcgionID 104
0.021342
Applicable when idlcRcgionID=104
idlcRcgionID 102
0.026097
Applicable when idlcRcgionID=102
idlcRcgionID 103
0.05461
Applicable when idlcRcgionID=103
idlcRcgionID 101
0.057216
Applicable when idlcRcgionID=101
monthID2
0.002789
Applicable when monthID=2
monthID3
-0.00429
Applicable when monthID=3
monthID4
-0.00609
Applicable when monthID=4
monthID5
-0.00412
Applicable when monthID=5
monthID6
-0.00264
Applicable when monthID=6
monthID7
0.002914
Applicable when monthID=7
monthID8
-0.00066
Applicable when monthID=8
monthID9
-0.00296
Applicable when monthID=9
monthlDlO
0.007288
Applicable when monthID=10
monthlDll
0.00585
Applicable when monthID=l 1
monthID12
0.007586
Applicable when monthID=12
idlcRcgionID 104:monthID2
-0.01478
Applicable when monthID=2 and idleRegionID=104
idlcRcgionID 102:monthID2
-0.00664
Applicable when monthID=2 and idleRegionID=102
idlcRcgionID 103 :monthID2
-0.0173
Applicable when monthID=2 and idleRegionID=103
idlcRcgionID 101 :monthID2
-0.01595
Applicable when monthID=2 and idleRegionID=101
idlcRcgionID 104:monthID3
-0.02666
Applicable when monthID=3 and idleRegionID=104
idlcRcgionID 102:monthID3
-0.01167
Applicable when monthID=3 and idleRegionID=102
idlcRcgionID 103 :monthID3
-0.04358
Applicable when monthID=3 and idleRegionID=103
idlcRcgionID 101 :monthID3
-0.0334
Applicable when monthID=3 and idleRegionID=101
idlcRcgionID 104:monthID4
-0.02855
Applicable when monthID=4 and idleRegionID=104
idlcRcgionID 102:monthID4
-0.01194
Applicable when monthID=4 and idleRegionID=102
idlcRcgionID 103 :monthID4
-0.04759
Applicable when monthID=4 and idleRegionID=103
idlcRcgionID 101 :monthID4
-0.03841
Applicable when monthID=4 and idleRegionID=101
idlcRcgionID 104:monthID5
-0.04011
Applicable when monthID=5 and idleRegionID=104
idlcRcgionID 102:monthID5
-0.01453
Applicable when monthID=5 and idleRegionID=102
idlcRcgionID 103 :monthID5
-0.05713
Applicable when monthID=5 and idleRegionID=103
idlcRcgionID 101 :monthID5
-0.0465
Applicable when monthID=5 and idleRegionID=101
idlcRcgionID 104:monthID6
-0.04388
Applicable when monthID=6 and idleRegionID=104
idlcRcgionID 102:monthID6
-0.01298
Applicable when monthID=6 and idleRegionID=102
idlcRcgionID 103 :monthID6
-0.05729
Applicable when monthID=6 and idleRegionID=103
idlcRcgionID 101 :monthID6
-0.05025
Applicable when monthID=6 and idleRegionID=101
idlcRcgionID 104:monthID7
-0.04935
Applicable when monthID=7 and idleRegionID=104
idlcRcgionID 102:monthID7
-0.0138
Applicable when monthID=7 and idleRegionID=102
idlcRcgionID 103 :monthID7
-0.06494
Applicable when monthID=7 and idleRegionID=103
idlcRcgionID 101 :monthID7
-0.05502
Applicable when monthID=7 and idleRegionID=101
184
-------
Variable
Coefficients
Comments
idlcRcgionID 104:monthID8
-0.04589
Applicable when monthID=8 and idleRegionID=104
idlcRcgionID 102:monthID8
-0.01495
Applicable when monthID=8 and idleRegionID=102
idlcRcgionID 103 :monthID8
-0.06051
Applicable when monthID=8 and idleRegionID=103
idlcRcgionID 101 :monthID8
-0.05
Applicable when monthID=8 and idleRegionID=101
idlcRcgionID 104:monthID9
-0.04807
Applicable when monthID=9 and idleRegionID=104
idlcRcgionID 102:monthID9
-0.02195
Applicable when monthID=9 and idleRegionID=102
idlcRcgionID 103 :monthID9
-0.06001
Applicable when monthID=9 and idleRegionID=103
idlcRcgionID 101 :monthID9
-0.04851
Applicable when monthID=9 and idleRegionID=101
idlcRcgionID 104: monthlD 10
-0.05049
Applicable when monthID=10 and idleRegionID=104
idlcRcgionID 102: monthlD 10
-0.03221
Applicable when monthID=10 and idleRegionID=102
idlcRcgionID 103: monthlD 10
-0.06831
Applicable when monthID=10 and idleRegionID=103
idlcRcgionID 101: monthlD 10
-0.05287
Applicable when monthID=10 and idleRegionID=101
idlcRcgionID 104: monthlD 11
-0.02092
Applicable when monthID=l 1 and idleRegionID=104
idlcRcgionID 102: monthlD 11
-0.0262
Applicable when monthID=l 1 and idleRegionID=102
idlcRcgionID 103: mo nth ID 11
-0.04514
Applicable when monthID=ll and idleRegionID=103
idlcRcgionID 101: mo nth ID 11
-0.04651
Applicable when monthID=l 1 and idleRegionID=101
idlcRcgionID 104: monthlD 12
-0.0075
Applicable when monthID=12 and idleRegionID=104
idlcRcgionID 102: monthlD 12
-0.02558
Applicable when monthID=12 and idleRegionID=102
idlcRcgionID 103: mo nth ID 12
-0.04263
Applicable when monthID=12 and idleRegionID=103
idlcRcgionID 101: mo nth ID 12
-0.04724
Applicable when monthID=12 and idleRegionID=101
Table E-2 shows a sample calculation of MOVES default total idle fractions using the
coefficients for passenger cars (sourceTypeID=21) in rural counties (countyTypeID=0) in
idleRegionID=101 (represented by New Jersey). The total idle fractions for all the sourceTypelD
21 and 32 derived from the TIF regression equation is available in the MOVES totalldleFraction
table.
185
-------
Table E-2 Example total idle fractions for rural New Jersey passenger cars
sourceTypelD
monthID
daylD
idleRegionID
countyTypelD
TIF
21
1
2
101
0
0.2670
21
2
2
101
0
0.2538
21
3
2
101
0
0.2293
21
4
2
101
0
0.2225
21
5
2
101
0
0.2164
21
6
2
101
0
0.2141
21
7
2
101
0
0.2149
21
8
2
101
0
0.2163
21
9
2
101
0
0.2155
21
10
2
101
0
0.2214
21
11
2
101
0
0.2263
21
12
2
101
0
0.2273
21
1
5
101
0
0.2782
21
2
5
101
0
0.2651
21
3
5
101
0
0.2406
21
4
5
101
0
0.2337
21
5
5
101
0
0.2276
21
6
5
101
0
0.2254
21
7
5
101
0
0.2261
21
8
5
101
0
0.2276
21
9
5
101
0
0.2268
21
10
5
101
0
0.2327
21
11
5
101
0
0.2376
21
12
5
101
0
0.2386
186
-------
Appendix F
Source Masses for Light-Duty Vehicles
In MOVES4, the source masses of light-duty vehicles were unchanged from MOVES3,
MOVES2014 and MOVES2010b. This appendix describes the derivation of these source masses.
Information on the updated source masses for heavy-duty vehicles is provided in Section 15.
In MOVES2010b, weight data (among other kinds of information) were used to allocate source
types to source bins using a field called weightClassID. While that information is no longer used
in MOVES and has not been updated, it provides a reasonable basis for estimating source mass
for the MOVES source types. As described in Equation F-l, each source type's source mass was
calculated using an activity-weighted average of their associated source bins' midpoint weights:
where M is the source mass factor for the source type, fa is the age fraction at age a, ab is the
source bin activity fraction for source bin b and m is the vehicle midpoint mass. Table F-l lists
the vehicle midpoint mass for each weightClassID. The source bin activity fraction in
MOVES2010b is a calculated value of activity based on fuel type, engine technology, regulatory
class, model year, engine size and weight class.
M =
Equation
F-l
187
-------
Table F-l MQVES2010b weight classes
WeightClassID
Weight Class Name
Midpoint
Weight
0
Doesn't Matter
[NULL1
20
weight < 2000 pounds
1000
25
2000 pounds <= weight < 2500 pounds
2250
30
2500 pounds <= weight < 3000 pounds
2750
35
3000 pounds <= weight <3500 pounds
3250
40
3500 pounds <= weight < 4000 pounds
3750
45
4000 pounds <= weight < 4500 pounds
4250
50
4500 pounds <= weight < 5000 pounds
4750
60
5000 pounds <= weight < 6000 pounds
5500
70
6000 pounds <= weight < 7000 pounds
6500
80
7000 pounds <= weight < 8000 pounds
7500
90
8000 pounds <= weight < 9000 pounds
8500
100
9000 pounds <= weight < 10000 pounds
9500
140
10000 pounds <= weight < 14000 pounds
12000
160
14000 pounds <= weight < 16000 pounds
15000
195
16000 pounds <= weight < 19500 pounds
17750
260
19500 pounds <= weight < 26000 pounds
22750
330
26000 pounds <= weight < 33000 pounds
29500
400
33000 pounds <= weight < 40000 pounds
36500
500
40000 pounds <= weight < 50000 pounds
45000
600
50000 pounds <= weight < 60000 pounds
55000
800
60000 pounds <= weight < 80000 pounds
70000
1000
80000 pounds <= weight < 100000 pounds
90000
1300
100000 pounds <= weight < 130000 pounds
115000
9999
130000 pounds <= weight
130000
5
weight < 500 pounds (for MCs)
350
7
500 pounds <= weight < 700 pounds (for MCs)
600
9
700 pounds <= weight (for MCs)
700
The following sections detail how weight classes were assigned to light-duty vehicles in
MOVES.
Fl. Motorcycles
The Motorcycle Industry Council Motorcycle Statistical Annual provides information on
displacement distributions for highway motorcycles for model years 1990 and 1998. These were
mapped to MOVES engine displacement categories. Additional EPA certification data were
used to establish displacement distributions for model year 2000. We assumed that displacement
distributions were the same in 1969 as in 1990 and interpolated between the established values to
determine displacement distributions for all model years from 1990 to 1997 and for 1999. Values
for 2000-and-later model years are based on model year 2000 certification data.
188
-------
We then applied weight distributions for each displacement category as suggested by EPA
motorcycle experts. The average weight estimate includes fuel and rider. The weight
distributions depended on engine displacement but were otherwise independent of model year.
This information is summarized in Table F-2.
Table F-2 Motorcycle engine size and average weight distributions for selected model years
Displacement
Category
1969 MY
distribution
(assumed)
1990 MY
distribution
(MIC)
1998 MY
distribution
(MIC)
2000 MY
distribution
(certification
data)
Weight distribution
(EPA staff)
0-169 cc (1)
0.118
0.118
0.042
0.029
100%: <=500 lbs.
170-279 cc (2)
0.09
0.09
0.05
0.043
50%: <=500 lbs.
50%: 5001bs. -7001bs.
280+ cc (9)
0.792
0.792
0.908
0.928
30%: 500 lbs.-700 lbs.
70%: > 7001bs.
12. Passenger Cars
Passenger car weights come from the 1999 IHS dataset. The weightClassID was assigned by
adding 300 lbs. to the IHS curb weight and grouping into MOVES weight bins. For each fuel
type, model year, engine size and weight bin, the number of cars was summed and fractions were
computed. In general, entries for which data were missing were omitted from the calculations.
Also, analysis indicated a likely error in the IHS data (an entry for 1997 gasoline-powered
Bentleys with engine size 5099 and weight class 20). This fraction was removed and the 1997
values were renormalized. 1999 model year values were used for all 2000-and-later model years.
F3. Light-E :ks
Light truck weights came from VIUS1997 data, which combines information from two different
survey forms. The first form was administered for VIUS "Strata" 1 and 2 trucks: pickup trucks,
panel trucks, vans (including mini-vans), utility type vehicles (including jeeps) and station
wagons on truck chassis. The second form was administered for all other trucks. While both
surveys requested information on engine size, only the second form requested detailed
information on vehicle weight. Thus, for Strata 1 and 2 trucks, VIUS classifies the trucks only
by broad average weight category (AVGCK): 6,000 lbs. or less, 6,001-10,000 lbs., 10,001-
14,000 lbs., etc. To determine a more detailed average engine size and weight distribution for
these vehicles, we used an Oak Ridge National Laboratory (ORNL) light-duty vehicle database,
compiled from EPA test vehicle data and Ward's Automotive Inc.115 data, to correlate engine
size with vehicle weight distributions by model year.
For source types 31 and 32 (Passenger Trucks and Light Commercial Trucks) in regClassID 30
(Light-Duty Trucks):
• VIUS 1997 trucks of the source type in Strata 3, 4 and 5 were assigned to the appropriate
MOVES weight class based on VIUS detailed average weight information.
• VIUS 1997 trucks of the source type in Strata 1 and 2 were identified by engine size and
broad average weight category.
• Strata 1 and 2 trucks in the heavier (10,001-14,000 lbs., etc.) VIUS1997 broad categories
were matched one-to-one with the MOVES weight classes.
189
-------
• For trucks in the lower broad categories (6,000 lbs. or less and 6001-10,000 lbs.), we
used VIUS1997 to determine the fraction of trucks by model year and fuel type that fell
into each engine size/broad weight class combination (the "VIUS fraction").
• We assigned trucks in the ORNL light-duty vehicle database to a weightClassID by
adding 300 lbs. to the recorded curb weight and determining the appropriate MOVES
weight class.
• For the trucks with a VIUS1997 average weight of 6,000 lbs. or less, we multiplied the
VIUS 1997 fraction by the fraction of trucks with a given weightClassID among the
trucks in the ORNL database that had the given engine size and an average weight of
6,000 lbs. or less. Note, the ORNL database did not provide information on fuel type, so
the same distributions were used for all fuels.
Because the ORNL database included only vehicles with a GVW up to 8500 lbs., we did
not use it to distribute the trucks with a VIUS1997 average weight of 6,001-10,000 lbs.
Instead these were distributed equally among the MOVES weightClassID 70, 80, 90 and
100.
Note that the source mass for source types 31 and 32 in regClassID 41 (class 2b trucks) was
calculated as described in Section 15.
190
-------
Appendix G
NREL Fleet DNA Preprocessing Steps
This appendix discusses the preprocessing steps undertaken on the NREL's Fleet DNA database,
which is used to derive default activity for heavy-duty vehicles, including idle fractions (see
Section 10) and starts activity (see Section 12).
Prior to calculation and preprocessing, the data is collected from the database which involves
loading and combining all of the 1 Hz data from Fleet DNA into a single 1-dimensional data
array for each parameter. Each data file is arranged in the database by vehicle, day and parameter
as shown in Figure G-l. To create one contiguous array per parameter, the processing script
loads each parameter and appends it to the parameter from the previous day resulting in five 1-D
arrays of equal length which can be joined on index.
Fleet
DNA
| Raw
Data
j ~ Vehicle
T I
-> Day
Time.json
Latitude.json
~
Longitude.json
Engine Speed.json
Wheel Speed.json
T
Figure G-l Diagram of Fleet DNA database file structure
After collecting the data, a processing step is performed to ensure the data is an accurate
representation of a vehicle's activity. Two of the key activity analyses from this report are
vehicle soak lengths and starts which are defined by the engine speed parameter that indicates if
the vehicle is running or not. A start is calculated by identifying a transition of the engine speed
from 0 to greater than zero and a soak is the length of time the engine was off before it is started.
Both parameter calculations depend on the engine being off; however, in some instances the data
logger will shut off before recording a zero for engine speed raising the concern that starts and
soak times may be missed or not accurately categorized.
To account for these instances in data preprocessing an algorithm was developed to look at the
time stamp and identify large leaps or gaps from one data point to the next. If the algorithm finds
a gap, the engine speed is replaced with a zero at that point to indicate the vehicle's engine has
shut off.
One of the major questions with this time gap method is what time length would constitute an
engine-off event. If the selected time length is too short, then instances such as the logger
updating its timestamp from the GPS may be characterized as a start. Conversely, if the time
length is too long, starts and vehicle soaks may be missed. A possible scenario resulting in a
191
-------
mischaracterization of starts could be when the GPS updates the data logger's clock while
crossing a time zone or the logger pausing its recording for a few seconds when creating a new
log file. Depending on the type of data logger used, some will create a new file at a specified
time interval or when a file size limit is reached requiring the logger to shift computing power to
saving the file to memory. If the gap length is set to an hour or less, the algorithm may count
these normal logger operations as vehicle starts. Similarly, if the logger was taken off of a
vehicle on the west coast and placed on a vehicle on the east coast, the timestamp may jump 3
hours should the GPS update the internal clock to local time.
To avoid these types of timestamp jumps which may show for soak operation modes 101 through
106, the gap length was set to 6 hours for this analysis. Plots of vehicle soak distribution
weighted by start fraction for various gap lengths are provided in Figure G-2 and Figure G-3 to
demonstrate what effect changes in gap length might have. Finally, after running the gap filling
routine, the first and last days of data are eliminated to avoid counting incomplete or
unrepresentative operation when the data logger is being installed or removed.
Plots of vehicle soak distribution weighted by start fraction for gap lengths varying between 1
second and 30 hours are provided in Figure G-2 and Figure G-3 to demonstrate what effect
changes in gap length might have. Figure G-2 provides the distributions for source type 62 which
consists of combination long-haul trucks that have very few starts per day and Figure G-3
provides the distributions for source type 52 which consists of single-unit short-haul trucks that
have a large number of starts per day. Intuitively the gap length algorithm had the most
noticeable effect on source type 62 due to the high weighting placed on each start as a result of
having very few starts per day.
192
-------
ID 62 Day 5 Gap: 0.02min
ID 62 Day 5 Gap: 6.0min
ID 62 Day 5 Gap: 30.0min
0.14
0.12
0.10
0.08
0.06
0.04
ID 62 Day 5 Gap: 360.0min
ID 62 Day 5 Gap: 720.0min
3 6 9 12 15 18 21 24
Hour ID
ID 62 Day 5 Gap: 1800.Omin
Op Modes
M 101
" 102
103
104
105
106
M 107
IllUll
" 108
ilSl fc
0.10
t= 0.08
0.04
3 6 9 12 15 18 21 24
Hour ID
6 9 12 15 18 21 24
Hour ID
Figure G-2 Start fraction weights soak distribution weighted by gap length: source type 62
193
-------
ID 52 Day 5 Gap: 0.02min
ID 52 Day 5 Gap: 6.0min
ID 52 Day 5 Gap: 30.0min
cSdil
\ E
t -
Op Modes
101
102
103
104
105
106
107
108
Hi
9 12 15
Hour ID
0.08
0.06
0.04
18 21 24
9 12 15
Hour ID
x
£ 0.04
I'
J
ddiil
Bp
I c
9 12 15 18
Hour ID
ID 52 Day 5 Gap: 6Q.0min
ID 52 Day 5 Gap: 9Q.0min
ID 52 Day 5 Gap: 12Q.0min
0.00
.2 0.08
0.06
.2 0.04
6 9 12 15 18 21 24
Hour ID
ID 52 Day 5 Gap: 360.0min
0.02
0.00
9 12 15 18 21 24
Hour ID
ID 52 Day 5 Gap: 720.0min
t 0.06
(13
V)
X
o 0.04
0.02
0.00
aflilll
it. =
¦
li
Op Modes
101
102
103
104
105
Hi
106
107
108
3 6 9 12 15 18 21 24
Hour ID
ID 52 Day 5 Gap: 1800.Omin
0.08
0.04
S 0.02
6 9 12 15 18 21 24
Hour ID
6 9 12 15 18 21 24
Hour ID
6 9 12 15 18 21 24
Hour ID
Figure G-3 Start fraction weights soak distribution weighted by gap length: source type 52
194
-------
Appendix 11 Averaging Methods for Heavy-duty Telematics Activity Data
Telematics data provides great detail on vehicle activity, but to calculate MOVES inputs for
heavy duty starts, soaks and idle fractions, we need to compute averages across the available
data. Because different averaging methods lead to different results, we evaluated several
different approaches to calculating these averages. The following discussion uses the calculation
of the idle fraction by sourcetypelD and daylD (weekend and weekday) to illustrate the strengths
and weaknesses of each of the methods.
Evaluated Methods
Initally, we used Method 1 (Equation H-l) to average the idle fractions across all vehicles
within the same sourcetypelD and day ID (weekday vs weekend). Method 1 could also be
referred to as an average of ratios. We initially chose to use Method 1 because it is simple to
implement and it equally weights each vehicle in the sample.
Method 1 -
"Average
of Ratios"
Idle fractions d =
£ Idle fractiorii
n
i = individual vehicle ID
n = vehicles sampled within each sourcetype
s = source type ID
d = day type ID
Equation
H-l
However, to estimate a representative idle fraction, the vehicle should be weighted by its
contribution to real-world activity. By weighting each individual vehicle idle fraction equally,
Method 1 over-represents the vehicles with little real-world activity (with possibly unrealistic
idle fractions) and under-respresents the idle fractions from the vehicles with the most activity.
We then considered Method 2, shown in Equation H-2, which is referred to as the "Sum over
Sum Method." In Method 2, the average is weighted according to vehicle activity. Vehicles that
are operated for long work days will have more operating hours and idle hours than vehicles that
are only operated intermittently. Multiplying the Idle fraction estimated from Method 2 by the
the total operating hours, £ operating hoursh will yield the total idle hours, £ idle hoursh
measured in our sample. This property assures that the relationship between idle hours and
operating time is consistent between our model estimates and the source data.
£ idle hoursi
Idle fractions a =
Method 2 - ' Z operating hourst
Equation
Sum over n
gum„ i = individual vehicle ID H-2
5 = source type ID
d = day type ID
One disadvantage of Method 2 is that the Idle fraction is dependent on the instrumentation time.
For example, a vehicle that is instrumented for two months will be weighted twice as much in the
195
-------
idle fraction calculation as a vehicle with the same duty cycle that is only measured for one
month. In some cases, using more information from vehicles that are instrumented longer would
be a desirable property; however, if instrumentation times are not random, this can skew the
average to overrepresent certain groups of vehicles. For example, we hope in the future to
develop idle, start and soak inputs from multiple datasets with different instrumentation times,
such as the Fleet DNA dataset with an average instrumentation time of 35 days/vehicle and the
HD SCR (CE-CERT) data with an 86-day average instrumentation time. Using Method 2 to
combine data from these two datasets, the HD SCR vehicle data would be weighted twice as
much as the vehicles in the Fleet DNA sample. This is undesirable, because we have no reason to
assume that the HD SCR sample vehicles are more representative of the national fleet than the
Fleet DNA sample vehicles.
In Method 3. we propose using a "Normalized Sum over Sum" approach as shown in Equation
H-3. Method 3 is similar to Method 2, except that the sum of idle hours and the operating hours
from each vehicle is divided (or normalized) by the number of days each sample vehicle was
instrumented. Method 3 controls for the different lengths of time each vehicle is instrumented
and the Idle fraction is weighted most heavily by the vehicles with the most daily average
activity, rather than the most measured activity. Method 3 (Equation H-3) is the current approach
we are using for developing MOVES inputs and is equivalent to
Equation 10-7 presented in Section 10.
Method 3 -
"Normalized
Sum over
Sum"
Idle fractions a =
Y /idle hourSi /
^ V 'daysj
Y /operating hourst / \
^ V 'daySJ
i = individual vehicle ID
day Si = # of days vehicle, is instrumented
s = source type ID
d = day type ID
Equation
H-3
The methods we explained above are also applicable to estimating start fractions. The start
fraction determines at fraction of total daily starts occur at each hour of the day. The following
table contains the equations for the start fractions for each of the three methods.
Start fractionhsd
Method 1 - h= hour of the day
"Average of / = individual vehicle ID
Ratios" n = # of sampled vehicles
s = source type ID
d = day type ID
Y, Start fractionh t
n
Equation
H-4
196
-------
Method 2 -
"Sum over
Sum"
Start fractionhs d =
Method 3 -
"Normalized
Sum over
Sum"
h= hour of the day
i = individual vehicle ID
s = source type ID
d = day type ID
Start fractionhsd =
£ startshi
£ startsi
rstartshi , \
^ V 'daysj
Y /starts^, \
^ V ' dayst J
h= hour of the day
i = individual vehicle ID
day st = days vehicle, is instrumented
s = source type ID
d = day type ID
Equation
H-5
Equation
H-6
The three different averaging method was also applied for calculating the soak fractions in the
table below. The soak fraction determines the distribution of starts occuring for the 8 different
start operating modes in MOVES (or soak lengths) as defined in Table 12-3.
Method 1 -
"Average of
Ratios"
Method 2 -
"Sum over
Sum"
Soak fractionho sd =
h= hour of the day
i = Vehicle ID
n = # of sampled vehicles
s = source type ID
d = day type ID
Soak fractionho s d =
h= hour of the day
i = Vehicle ID
o = operating mode (soak length)
s = source type ID
d = day type ID
Y, Soak fractionhio
n
2
^ startshi
Soak fractionhosc
Method 3 -
"Normalized
Sum over
Sum"
^ V ' day Si)
„ /startshji . \~
^ V ' daysj
h= hour of the day
i = Vehicle ID
o = operating mode (soak length)
day st = days vehicle, is instrumented
s = source type ID
d = day type ID
Equation
H-7
Equation
H-8
Equation
H-9
197
-------
H2. Comparison of Evaluated Methods
Figure H-l graphically compares the idle fractions calculated using Method 1 and Method 3,
using the data for single-unit short-haul trucks. In the graph, the darker colors represent idling
that is over 1 hour in duration, classified as extended idle. The height of the bars represents the
total idle fraction. For refuse trucks and single-unit short haul trucks there is a significantly
higher idle fraction on the weekends, when using method 3. This implies that the refuse and
single unit truck vehicles that operate most on the weekend, also have higher idle fractions.
Method 3 appropriately weights the idle fractions from each vehicle according to its average
daily activity.
Weekend | Weekday
M Extended
Workday
J- J> J1
-------
0.10
Single-Unit Short-Haul | Weekday
C
o
•fi 0.08
u
TO
3 0.06
c
"B 0.04
u
(C
g 0.02
i/i
0.00
6 9 12 15 18 21 24
Hour of the Day
Single-Unit Short-Haul | Weekday
6 9 12 15 18 21 24
Hour of the day
Method 1 "Average of Ratios''
Method 3 "Normalized Sum over Sum'
Figure H-2 Start fraction and Soak fraction calculated using Method 1 and Method 3.
H3. Future Work
The previous methods are all based on the assumption that the sample of vehicles are
representative of the entire vehicle population. However, as presented in Section 12.12.2.2, there
is significant variation in idle fractions and starts per day by truck vocation within the MOVES
sourcetypes. For example, parcel delivery trucks and concrete mixers are are both single unit
short haul trucks, but parcel delivery trucks have many more starts per day in the Fleet DNA
database.
The truck samples we are currently using (FleetDNA) and which we intend to use in the future
(CE-CERT), made efforts to collect data from a variety of important vocational classes.
Ftowever, the truck samples in these programs were not systematically chosen to be
representative of U.S. truck vocations. To address this deficiency, we would like to use a method
that weights each vehicle according to its average activity as well as the population of each
vocation. The proposed Method 4 "Vocation and Activity Weighted fraction" would use a
weighting factor to weight the vehicles within each vocation according to how many vehicles
were sampled, compared to how many exist in the national population.
199
-------
Method 4 -
"Vocation
and
Activity
Weighted
fraction"
Idle fraction^ d =
2(
idle hourSi
/
dayst
x w.
'v^j
v, foperatinq hours; / w \
£ ( /days, x
i = individual vehicle ID
dayst = # of days vehicle, is instrumented
v = vehicle vocation
wv= (population/sample size of each vocation, v)
s = source type ID
d = day type ID
Equation
H-10
In practice, we are not yet able to implement Method 4, because we are unable to accurately map
instrumented truck vocations to national truck populations because we lack information on both
parts of the equation.
1) We lack information on the total number of vehicles in each vocation. The IHS vehicle
registration data provides sufficient information to classify trucks by the MOVES
sourcetype, but not by vocation or specific firm. Some are characterized by the industry
sector of the firm that owns the truck, but, with large populations of trucks classified in
sectors such as: individual, general freight, government/miscellaneous, lease/rental,
wholesale/retail, manufacturing and services, these sector distinctions are insufficient to
determine the vocation of the truck. For example, should a "service truck" be classified as
a utility truck or a single unit box delivery truck?
2) The trucks in the Fleet DNA database only represent a subset of truck vocations classified
by these industrial sectors. For example, we do not have instrumented truck data from
many of the industry sectors in the registration data including: agriculture/farm,
petroleum, landscaping, mining, logging and emergency vehicles,
Additional work is needed to have confidence that the additional data needs and complexity of
Method 4 would yield meaningful improvements in emissions accuracy.
200
-------
Appendix I Road Load ( oeffiecient for Combination Trucks In ill) GHG
Rule
In the HD GHG rules, certification test procedures were developed to evaluate the aerodynamic
performance of tractors and trailers. The test procedures varied between Phase 1 and Phase 2 of
the standards. Trailers were not included in the Phase 1 program and tractor aerodynamic
performance was measured at no wind conditions. While trailers were ultimately not regulated in
Phase 2, new test procedures were developed for trailers that approximate a wind-averaged drag
performance. Wind-averaged drag reflects a vehicle's average performance for a range of yaw
angles (the angle of attack of the air during travel) at a given vehicle speed and wind speed and is
more representative of real-world performance. The wind-averaged drag result modeled in the
Phase 2 rule is determined by an average of drag values two yaw positions which represents a
vehicle speed of 65 mph and a wind speed of 7 mph. In the tractor analysis, the drag value is
represented by the aerodynamic drag area, CdA. In the trailer analysis, the drag value is
represented as a reduction in drag area, ACdA, relative to a commonly available baseline trailer
that is not equipped with aerodynamic devices.
The GHG rules also create bins for aerodynamic certification, so that a precise drag value is not
needed to certify every tractor. A representative aerodynamic value from each bin is used, along
with other aspects of the powertrain and vehicle, as an input into the Greenhouse Gas Emissions
Model (GEM) to determine a vehicle configuration's CO2 emissions result. Tractors are
categorized in the rule by their roof height and cab type - sleeper cabs and day cabs - and
different aerodynamic bins exist for each category and a mid-point from each bin is used as the
GEM input. The trailer analysis used the bottom boundaries of the bins for GEM input values,
which represent a conservative estimate of aerodynamic improvements. For this analysis,
midpoints of the bins were used to reflect average performance within the trailer bins. Bin I
represents no improvement, so a ACdA value of 0 m2 was used in this analysis. Non-box trailers
including flatbed and tank trailers, have standards based on tire technologies in the HD Phase 2
GHG program and aerodynamic improvements for those trailer types are neither expected nor
included in this analysis. The CdA bin structures for tractors and trailers are shown respectively
below in Table 1-1 and Table I-2116 117 The trailer bin structure is common to all box van trailer
types.
Table 1-1 Phase 2 GHG Aerodynamic Drag Area Bin Structure for Tractors [m2]
High-roof Sleeper Cab
High-roof Day Cab
Low-roof Sleeper &
Day Cabs
Mid-roof Sleeper &
Day Cabs
Tractor
CdA Bin
CdA range
CdA input
CdA range
CdA input
CdA range
CdA input
CdA range
CdA input
I
>6.9
7.15
>7.2
7.45
>5.4
6.00
>5.9
7.00
II
6.3-6.8
6.55
6.6.7.1
6.85
4.9-5.3
5.60
5.5-5.8
6.65
III
5.7-6.2
5.95
6.0-6.5
6.25
4.5-4.8
5.15
5.1-5.4
6.25
IV
5.2-5.6
5.40
5.5-5.9
5.70
4.1-4.4
4.75
4.7-5.0
5.85
V
4.7-5.1
4.90
5.0-5.4
5.20
3.8-4.0
4.40
4.4-4.6
5.50
VI
4.2-4.6
4.40
4.5-4.9
4.70
3.5-3.7
4.10
4.1-4.3
5.20
VII
<4.1
3.90
<4.4
4.20
<3.4
3.80
<4.0
4.90
201
-------
Table 1-2 Phase 2 GHG Aerodynamic Drag Area Bin Structure for Box Van Trailers [m
Trailer ACdA Bin
ACdA range
ACdA input for GEM
Midpoint of ACdA range
I
<0.09
0.0
0
II
0.10-0.39
0.1
0.25
III
0.40-0.69
0.4
0.55
IV
0.70-0.99
0.7
0.85
V
1.00-1.39
1.0
1.2
VI
1.40-1.79
1.4
1.6
VII
>1.80
1.8
1.9
The tractor and trailer bin structures were used to estimate adoption rates of improved
aerodynamic technologies. For tractors, EPA conducted such analyses for Phase 1 GHG and
Phase 2 GHG rulemakings, for both their respective baselines and the rulemaking scenarios. For
tractor certification in the GHG rules, different tractor types are assumed to be matched with
specific trailer types. High-roof tractors are matched with 53-foot box van trailers. Mid-roof
tractors are matched with tank trailers and low-roof tractors are matched with flatbed trailers.
The Phase 1 GHG baseline analysis was used for model years prior to implementation of the
Phase 1 GHG rule (pre-2014 model years). The Phase 2 GHG baseline analysis was used for
model years 2014 through 2020, which are predominantly the Phase 1 GHG implementation
years. The Phase 2 GHG technology penetration analysis was the basis for the adoption rates for
model years 2021 and later, with different rates for different types of cabs and each of the major
steps established in the rulemaking - model years 2021-2023, 2024-2026 and 2027 and beyond.
The bin-weighted average CdA (i.e., the "CdA input" from Table J-l) was then calculated by
model year group. For the high-roof sleeper cab and high-roof day cab subcategories, the effect
of the trailer skirt was removed to calculate the CdA of a tractor-trailer combination with a
baseline trailer. Through extensive testing in the Phase 2 GHG rulemaking development, the
trailer skirt was estimated to have Trailer Bin III performance of 0.55 m2, as seen in Table 1-3.
Table 1-3 Tractor aerodynamic technology adoption rates by model year groups
Tractor
Tractor Bin
1960-2013
Phase 1 GHG
Phase 2 GHG
Phase 2 GHG
Phase 2 GHG
Bin
CdA input [m2]
2014-2020
2021-2023
2024-2026
2027+
I
7.15
25%
0%
0%
0%
0%
II
6.55
70%
10%
0%
0%
0%
III
5.95
5%
70%
60%
40%
20%
IV
5.40
0%
20%
30%
40%
30%
O
V
4.90
0%
0%
10%
20%
50%
VI
4.40
0%
0%
0%
0%
0%
O
£
VII
3.90
0%
0%
0%
0%
0%
"So
Mean C,iA (w/ skirl) |nr|
6.67
5.9
5.68
5.52
5.26
0
Skirt effect [m2]
0.55
0.55
0.55
0.55
0.55
Mean CdA (w/o skirt) |nr|
7.22
6.45
6.23
6.07
5.81
202
-------
Tractor
Bin
Tractor Bin
CdA input [m2]
1960-2013
Phase 1 GHG
2014-2020
Phase 2 GHG
2021-2023
Phase 2 GHG
2024-2026
Phase 2 GHG
2027+
I
7.45
25%
0%
0%
0%
0%
II
6.85
70%
30%
0%
0%
0%
5«
-a
III
6.25
5%
60%
60%
40%
30%
(J
IV
5.70
0%
10%
35%
40%
30%
es
¦o
V
5.20
0%
0%
5%
20%
40%
o
o
~-
VI
4.70
0%
0%
0%
0%
0%
1
-fl
©X
VII
4.20
0%
0%
0%
0%
0%
n
Mean CdA (w/ skirt) [m2]
6.97
6.375
6.005
5.82
5.665
Skirt effect [m2]
0.55
0.55
0.55
0.55
0.55
Mean CdA (w/o skirt) [m2]
7.52
6.925
6.555
6.37
6.215
I
7.00
100%
15%
10%
0%
0%
£
II
6.65
0%
15%
10%
20%
20%
~-
O
III
6.25
0%
70%
70%
60%
50%
V
4>
IV
5.85
0%
0%
10%
20%
30%
Xfl
V
5.50
0%
0%
0%
0%
0%
©
£
VI
5.20
0%
0%
0%
0%
0%
-i
VII
4.90
0%
0%
0%
0%
0%
s
Mean CdA |m2]
7.00
6.4225
6.325
6.25
6.21
I
7.00
100%
20%
10%
0%
0%
%
£
II
6.65
0%
20%
10%
20%
20%
es
(j
III
6.25
0%
60%
70%
60%
50%
s*
•o
IV
5.85
0%
0%
10%
20%
30%
o
V
5.50
0%
0%
0%
0%
0%
~-
1
¦a
§
VI
5.20
0%
0%
0%
0%
0%
VII
4.90
0%
0%
0%
0%
0%
Mean CdA [m2]
7.00
6.48
6.325
6.25
6.21
&>
I
6.00
100%
15%
10%
0%
0%
-s
«s
II
5.60
0%
15%
10%
20%
20%
~-
III
5.15
0%
70%
70%
60%
50%
IV
4.75
0%
0%
10%
20%
30%
&>
«*¦
V
4.40
0%
0%
0%
0%
0%
©
2
VI
4.10
0%
0%
0%
0%
0%
i
if
VII
3.80
0%
0%
0%
0%
0%
o
-J
Mean CdA |m2|
6.00
5.345
5.24
5.16
5.12
I
6.00
100%
20%
10%
0%
0%
%
II
5.60
0%
20%
10%
20%
20%
es
(j
III
5.15
0%
60%
70%
60%
50%
es
•o
IV
4.75
0%
0%
10%
20%
30%
o
V
4.40
0%
0%
0%
0%
0%
s
1
VI
4.10
0%
0%
0%
0%
0%
£
o
-J
VII
3.80
0%
0%
0%
0%
0%
Mean CdA [m2]
6.00
5.41
5.24
5.16
5.12
203
-------
A survey conducted by the North American Council for Freight Efficiency (NACFE) was used to
estimate that trailer aerodynamic technologies were not in significant use prior to 2008.118
Therefore, the model years between 1960-2007 reflect the time period prior to the use of trailer
aerodynamic improvements. The model year groups of 2008-2014 and 2014-2017 reflect
voluntary improvements to trailer aerodynamics. Since trailers were not regulated in the Phase 2
HDGHG rulemaking and lacking data on further voluntary improvements, trailer aerodynamics
for model years 2018 and beyond are modeled as MY2017.
Table 1-4 shows the trailer technology adoption rates were used to determine the average ACdA
by model year group for for several trailer categories. Long box vans represent 53-ft box van
trailers. Short box vans are 50 feet and shorter, and the shortest ones are often pulled in tandem.
However, for simplicity and consistency with the compliance framework of the HD GHG Phase
2 rule, a single-trailer configuration is the basis for this analysis for both long and short trailers.
204
-------
Table 1-4 Trailer aerodynamic techno
ogy adoption rates by model year groups
Trailer Bin
1960-
2008-
2014-
2018-2020
Phase 2
Phase 2
Phase 2
2007
2013
2017
GHG
2021-2023
GHG
2024-2026
GHG
2027+
I
100%
65%
55%
55%
55%
55%
55%
II
0%
0%
0%
0%
0%
0%
0%
Q£
fi
III
0%
35%
40%
40%
40%
40%
40%
>
*
IV
0%
0%
5%
5%
5%
5%
5%
O
pD
V
0%
0%
0%
0%
0%
0%
0%
fi
o
VI
0%
0%
0%
0%
0%
0%
0%
N-1
VII
0%
0%
0%
0%
0%
0%
0%
Average
ACdA [m2]
0
0.1925
0.2625
0.2625
0.2625
0.2625
0.2625
I
100%
100%
100%
100%
100%
100%
100%
II
0%
0%
0%
0%
0%
0%
0%
fi
III
0%
0%
0%
0%
0%
0%
0%
M
o
pfi
t:
o
-fi
IV
0%
0%
0%
0%
0%
0%
0%
V
0%
0%
0%
0%
0%
0%
0%
VI
0%
0%
0%
0%
0%
0%
0%
t/5
VII
0%
0%
0%
0%
0%
0%
0%
Average
ACdA [m2]
0
0
0
0
0
0
0
Q£
I
100%
100%
100%
100%
100%
100%
100%
s*
II
0%
0%
0%
0%
0%
0%
0%
*
o
-Q
III
0%
0%
0%
0%
0%
0%
0%
W)
fi
IV
0%
0%
0%
0%
0%
0%
0%
o
o
V
0%
0%
0%
0%
0%
0%
0%
VI
0%
0%
0%
0%
0%
0%
0%
3
VII
0%
0%
0%
0%
0%
0%
0%
"S
Average
ACdA [m2]
0
0
0
0
0
0
0
OA
fi
I
100%
100%
100%
100%
100%
100%
100%
etf
M
O
pD
II
0%
0%
0%
0%
0%
0%
0%
III
0%
0%
0%
0%
0%
0%
0%
t:
o
IV
0%
0%
0%
0%
0%
0%
0%
pfi
QC
V
0%
0%
0%
0%
0%
0%
0%
5-
VI
0%
0%
0%
0%
0%
0%
0%
1
VII
0%
0%
0%
0%
0%
0%
0%
t
a
a.
Average
ACdA [m2]
0
0
0
0
0
0
0
205
-------
The average ACdA values by model year group for tractor-trailer combinations were determined
by estimating the distribution of each trailer category within each tractor subcategory. Following
the analysis performed for the HD GHG Phase 2 rulemaking, the distribution in Table 1-5 was
used. Trailers in the non-aero category are incompatible with aerodynamic improvements and are
assumed to be matched entirely within the low-roof and mid-roof tractor types and no
aerodynamic improvements are applied to these trailers. Trailers with work-performing
equipment that impedes the use of some aerodynamic devices are considered partial-aero trailers.
These trailers are assumed to be used in short haul operations and assigned to high roof day cab
tractors. The remaining trailers are full-aero box vans capable of adopting a range of
aerodynamic devices and we assume these trailer types are used in long haul with sleeper cab
tractors. Using a combination of data from the 2002 VIUS database and trailer production
results from ACT Research, over 70 percent of the full-aero capable trailers are assumed to be
long box vans (longer than 50-feet). Partial-aero box vans used in short-haul applications,
however, are more than 60 percent short trailer (50 feet and shorter).
Table 1-5 Trailer category distribution by tractor category
Trailer Category
Sleeper Cabs
Day Cabs
Low-roof
Mid-roof
High-roof
Low-roof
High-roof
Full-aero long
0%
0%
73%
0%
0%
Full-aero short
0%
0%
27%
0%
0%
Partial-aero long
0%
0%
0%
0%
36%
Partial-aero short
0%
0%
0%
0%
64%
Non-aero
100%
100%
0%
100%
0%
We assume no aerodynamic improvements for trailers pulled by low- and mid-roof tractors, so
all aerodynamic improvements for these vehicles come from the tractors only. Aerodynamic
improvements for the high-roof tractors pulling box trailers are calculated by combining the
aerodynamic drag estimates from the tractor and trailer. The average trailer ACdA values by
model year group and tractor category are listed in Table 1-6. Trailer aerodynamic improvements
are calculated using the trailer distribution shown in Table 1-5 and the adoption rates of Table
1-4. The average CdA for a tractor-trailer combination by model year can be calculated by
subtracting the average trailer ACdA values from the average tractor CdA values in Table 1-3.
Table 1-6 Average trailer ACdA values by tractor category and model year group [m2]
——^^tylodel years
Category ———-
Pre-2008
2008-2013
2014-2017
2018-2020
2021-2023
2024-2026
2027+
High-roof sleeper cab
0
0.140
0.192
0.192
0.192
0.192
0.192
High-roof day cab
0
0
0
0
0
0
0
The resulting drag values that include aerodynamic improvements from tractors and trailers are
shown below.
206
-------
Table 1-7 Drag area, CdA [m2], by tractor-trailer subcategory and model year group
^^ Model years
Category ~~~
Pre-2008
2008-2013
2014-2017
2018-2020
2021-2023
2024-2026
2027+
High-roof sleeper cab
7.2200
7.0798
6.2584
6.2584
6.0384
5.8784
5.6184
High-roof day cab
7.5200
7.5200
6.9250
6.9250
6.5550
6.3700
6.2150
Mid-roof sleeper cab
7.0000
7.0000
6.4225
6.4225
6.3250
6.2500
6.2100
Mid-roof day cab
7.0000
7.0000
6.4800
6.4800
6.3250
6.2500
6.2100
Low-roof sleeper cab
6.0000
6.0000
5.3450
5.3450
5.2400
5.1600
5.1200
Low-roof day cab
6.0000
6.0000
5.4100
5.4100
5.2400
5.1600
5.1200
Vocational tractor
6.0000
6.0000
6.0000
6.0000
6.0000
6.0000
6.0000
In MOVES, the values for sleeper cab tractors (with trailers) used for long-haul combination
trucks (sourceTypelD 62) and the values for day cab tractors (with trailers) are used for short-
haul combination trucks (sourceTypelD 61). Both the sleeper cab and day cab categories contain
a mix of high-roof, mid-roof and low-roof types. Day cab tractors also contain a vocational
tractor subcategory, for which the aerodynamic requirements of the Phase 2 rule do not apply.
They are of a low-roof height configuration and assumed to have the aerodynamic characteristics
of pre-2008 MY low-roof tractors for all model years. The combined average CdA for the
MOVES combination trucks shown in Table 1-9 was calculated using the distribution from Table
1-8 and the drag areas from Table 1-7.
Table 1-8 Roof height distribution within cab types
Roof height
Sleeper Cab
Day Cab
Low-roof
5%
47%
Mid-roof
15%
0%
High-roof
80%
45%
Vocational
0%
8%
Table 1-9 Average CdA for each source type by model year group weighted by roof height
Pre-2008
2008-2013
2014-2017
2018-2020
2021-2023
2024-2026
2027+
Sleeper cab (sourceType 62)
7.1260
7.0136
6.2373
6.2373
6.0415
5.8982
5.6822
Day cab (sourceType 61)
6.6840
6.6840
6.1390
6.1390
5.8926
5.7717
5.6832
To convert from CdA to the C coefficient, Equation 15-11 was used with an estimate for air
density. A national annual MOVES run produced an average temperature of 61°F. At standard
atmospheric air pressure, the air density is 1.22 kg/m3. The resulting C coefficient values are
shown in Table I-10.
Table 1-10 C coefficients [kW-s3/m3] of source types 61 and 62 by model year group
2014-2017
2018-2020
2021-2023
2024-2026
2027+
Sleeper cab (sourceType 62)
0.00380
0.00380
0.00369
0.00360
0.00347
Day cab (sourceType 61)
0.00374
0.00374
0.00359
0.00352
0.00347
The Phase 1 and Phase 2 GHG emission standards also project improvements to the tire rolling
resistance. MOVES3 reflected these improvements through revisions to the A coefficient in the
SourceUseTypePhysics table. It is related to the coefficient of rolling resistance, Cm and source
massM, using the following equation where g is the gravitational acceleration:
207
-------
A — CRRMg
Equation 1-1
For combination tractor-trailers, the tires typically differ by axle position (steer, drive and
trailer). The HD GHG Phase 1 and Phase 2 rulemakings developed adoption rates of lower
rolling resistance tires for the steer and drive tires for all model years.119 120 The overall rolling
resistance of the vehicle is a weighted average of rolling resistance over axle based on axle
loading.
„ „ Msteer „ ^drive „ Mtrailer Equation 1-2
^RR ~ ^RR,steer ^ ¦" ^RR,drive ^ ¦" ^RR,trailer ^
Tire rolling resistance for tractor-trailers was updated using the same tractor type distributions
described in Table 1-8. Rolling resistance distributions, based on tire rolling resistance levels
from the GHG rules are shown in Table 1-11.
Table 1-11 Crr by axle and tractor type
Tire Crr
Tire Crr value
Pre-2014
Phase 1
2018-2020
Phase 2
Phase 2 GHG
Phase 2
level
[kg/metric
ton]
GHG
2014-2017
GHG
2021-2023
2024-2026
GHG
2027+
Base
7.8
100%
10%
10%
5%
5%
5%
a
1
6.6
0%
70%
70%
35%
15%
10%
S-H
2
5.7
0%
20%
20%
50%
60%
50%
{*5
Q
in
3
4.9
0%
0%
0%
10%
20%
35%
£
Avg Crr |kg/metric t«n|
7.8
6.54
6.54
6.04
5.78
5.615
~-
.g
Base
8.1
100%
10%
10%
5%
5%
5%
1
6.9
0%
70%
70%
35%
15%
10%
>
2
6.0
0%
20%
20%
50%
60%
50%
O
©
'C
Q
3
5.0
0%
0%
0%
10%
20%
35%
i
Avg Crr
Ikg/metric ton|
8.1
6.84
6.84
6.32
6.04
5.845
gx
1
6.5
0%
0%
0%
0%
0%
0%
3
2
6.0
100%
100%
100%
100%
100%
100%
i-l
3
5.1
0%
0%
0%
0%
0%
0%
y
H
4
4.7
0%
0%
0%
0%
0%
0%
Avg Crr | kg/metric ton|
6.0
6.0
6.0
6.0
6.0
6.0
Base
7.8
100%
30%
30%
5%
5%
5%
a
1
6.6
0%
60%
60%
35%
25%
20%
.a
S-H
2
5.7
0%
10%
10%
50%
55%
50%
~_
CO
3
4.9
0%
0%
0%
10%
15%
25%
O.
Avg Crr [kg/metric ton]
7.8
6.87
6.87
6.04
5.91
5.785
—
a
Base
8.1
100%
30%
30%
15%
10%
5%
o
1
6.9
0%
60%
60%
35%
25%
10%
s
-------
Table 1-11 (Continued) Crr
Tire Crr
Tire Crr value
Pre-2014
Phase 1
2018-2020
Phase 2
Phase 2 GHG
Phase 2
level
[kg/metric
ton]
GHG
2014-2017
GHG
2021-2023
2024-2026
GHG
2027+
Base
7.8
100%
30%
30%
5%
5%
5%
.g
1
6.6
0%
60%
60%
35%
15%
10%
vh
CO
3
4.9
0%
0%
0%
10%
20%
35%
pD
Avg Crr
[kg/metric ton]
7.8
6.87
6.87
6.04
5.78
5.615
u
Si
Base
8.1
100%
30%
30%
5%
5%
5%
a
1
6.9
0%
60%
60%
35%
15%
10%
2
6.0
0%
10%
10%
50%
60%
50%
o
-
•c
Q
3
5.0
0%
0%
0%
10%
20%
35%
-C
W)
Avg Crr
[kg/metric ton]
8.1
7.17
7.17
6.32
6.04
5.845
n
1
6.5
0%
0%
0%
0%
0%
0%
2
6.0
100%
100%
100%
100%
100%
100%
3
5.1
0%
0%
0%
0%
0%
0%
H
4
4.7
0%
0%
0%
0%
0%
0%
Avg Crr [kg/metric ton]
6.0
6.0
6.0
6.0
6.0
6.0
Base
7.8
100%
40%
40%
5%
5%
5%
.s
1
6.6
0%
50%
50%
35%
25%
20%
vh
2
6.0
0%
10%
10%
50%
65%
85%
s
•fi
Q
3
5.0
0%
0%
0%
0%
0%
0%
!£
Avg Crr
[kg/metric ton]
8.1
7.29
7.29
6.63
6.435
6.195
-J
1
6.5
100%
100%
100%
100%
100%
100%
2
6.0
0%
0%
0%
0%
0%
0%
V-l
2
6.0
0%
10%
10%
50%
50%
55%
•c
Q
3
5.0
0%
0%
0%
0%
20%
30%
9*
CJ
Avg Crr
[kg/metric ton]
8.1
7.29
7.29
6.63
6.19
5.895
o
>
1
6.5
100%
100%
100%
100%
100%
100%
2
6.0
0%
0%
0%
0%
0%
0%
?-(
-------
The average Crr values of each tire type were weighted based on a typical loading of a heavy-
duty vehicle: 42.5 percent over the trailer axle, 42.5 percent over the drive axle, and 15 percent
over the steer axle.aa The result is shown in Table 1-12.
Table 1-12 Crr [kg/metric ton] by tractor category
Pre-2014
2014-2017
2018-2020
2021-2023
2024-2026
2027+
High-roof sleeper cab
7.163
6.438
6.438
6.142
5.984
5.876
High-roof day cab
7.163
6.628
6.628
6.142
5.984
5.876
Low and Mid-roof sleeper cab
7.375
6.840
6.840
6.486
6.384
6.263
Low-roof day cab
7.375
6.909
6.909
6.486
6.384
6.263
Vocational tractor
7.375
6.909
6.909
6.530
6.283
6.110
Using the roof height distributions in Table 1-8, the resulting Crr values are:
Pre-
2014
2014-2017
2018-2020
2021-2023
2024-2026
2027+
Sleeper cab (sourceType
62)
7.2050
6.5185
6.5185
6.2109
6.0640
5.9537
Day cab (sourceType 61)
7.2794
6.2298
6.2298
5.8124
5.6932
5.5880
To calculate the A coefficient, Equation 1-1 was used in combination with the source mass values
and Crr values from Table 1-13. Resulting A coefficients by model year group are shown in
Table 1-14.
Table 1-14 A coefficient values [kW-s/m] by model year group
2014-2017
2018-2020
2021-2023
2024-2026
2027+
Sleeper cab (sourceType 62)
1.576
1.576
1.502
1.466
1.440
Day cab (sourceType 61)
1.509
1.509
1.407
1.379
1.353
aa This distribution is equivalent to the federal over-axle weight limits for an 80,000 GVWR 5-axle tractor-trailer:
12,000 pounds over the steer axle, 34,000 pounds over the tandem drive axles (17,000 pounds per axle) and 34,000
pounds over the tandem trailer axles (17,000 pounds per axle).
210
-------
Appendix J MOVES4 Sou reel 'seTy pePhysics Table
Table J-l MOVES4 SourceUseTypePhysics Table
Source
Type
ID
Reg
Class
ID
Begin
Model
Year
End
Model
Year
Rolling
Term A
(kW-s/m)
Rotating
Term B
(kW-s2/m2)
Drag
Term C
(kW-s3/m3)
Source Mass
(metric tons)
Fixed
Mass
Factor
(metric
tons)
11
10
1960
2060
0.0251
0
0.000315
0.285
0.285
21
20
1960
2060
0.156461
0.002002
0.000493
1.4788
1.4788
31
30
1960
2060
0.22112
0.002838
0.000698
1.86686
1.86686
31
41
1960
2009
0.22112
0.002838
0.000698
3.40194
2.05979
32
30
1960
2060
0.235008
0.003039
0.000748
2.05979
2.05979
32
41
1960
2009
0.235008
0.003039
0.000748
3.40194
2.05979
41
41
1960
2009
1.29515
0
0.003715
5.68398
2.05979
41
41
2014
2060
1.23039
0
0.003715
5.68398
5
41
42
1960
2009
1.29515
0
0.003715
7.78184
2.05979
41
42
2014
2020
1.23039
0
0.003715
7.78184
5
41
42
2021
2023
1.00646
0
0.003715
7.78184
5
41
42
2024
2026
0.974469
0
0.003715
7.78184
5
41
42
2027
2060
0.926484
0
0.003715
7.78184
5
41
46
1960
2009
1.29515
0
0.003715
11.3666
17.1
41
46
2014
2020
1.23039
0
0.003715
11.3666
7
41
46
2021
2023
1.00646
0
0.003715
11.3666
7
41
46
2024
2026
0.974469
0
0.003715
11.3666
7
41
46
2027
2060
0.926484
0
0.003715
11.3666
7
41
47
1960
2009
1.29515
0
0.003715
15.6028
17.1
41
47
2014
2020
1.23039
0
0.003715
15.6028
10
41
47
2021
2023
1.00646
0
0.003715
15.6028
10
41
47
2024
2026
0.974469
0
0.003715
15.6028
10
41
47
2027
2060
0.926484
0
0.003715
15.6028
10
42
42
1960
2009
1.0944
0
0.003587
7.78184
2.05979
42
42
2014
2020
1.03968
0
0.003587
7.78184
5
42
42
2021
2023
1.03968
0
0.003587
7.78184
5
42
42
2024
2026
1.03968
0
0.003587
7.78184
5
42
42
2027
2060
0.913879
0
0.003587
7.78184
5
42
46
1960
2009
1.0944
0
0.003587
11.3666
17.1
42
46
2014
2020
1.03968
0
0.003587
11.3666
7
42
46
2021
2023
1.03968
0
0.003587
11.3666
7
42
46
2024
2026
1.03968
0
0.003587
11.3666
7
42
46
2027
2060
0.913879
0
0.003587
11.3666
7
42
47
1960
2009
1.0944
0
0.003587
15.6028
17.1
42
47
2014
2020
1.03968
0
0.003587
15.6028
10
42
47
2021
2023
1.03968
0
0.003587
15.6028
10
211
-------
Source
Type
ID
Reg
Class
ID
Begin
Model
Year
End
Model
Year
Rolling
Term A
(kW-s/m)
Rotating
Term B
(kW-s2/m2)
Drag
Term C
(kW-s3/m3)
Source Mass
(metric tons)
Fixed
Mass
Factor
(metric
tons)
42
47
2024
2026
1.03968
0
0.003587
15.6028
10
42
47
2027
2060
0.913879
0
0.003587
15.6028
10
42
48
1960
2009
1.0944
0
0.003587
15.6028
17.1
42
48
2014
2020
1.03968
0
0.003587
15.6028
10
42
48
2021
2023
1.03968
0
0.003587
15.6028
10
42
48
2024
2026
1.03968
0
0.003587
15.6028
10
42
48
2027
2060
0.913879
0
0.003587
15.6028
10
43
41
1960
2009
0.746718
0
0.002176
5.68398
2.05979
43
41
2014
2060
0.709382
0
0.002176
5.68398
5
43
42
1960
2009
0.746718
0
0.002176
7.78184
2.05979
43
42
2014
2020
0.709382
0
0.002176
7.78184
5
43
42
2021
2023
0.637734
0
0.002176
7.78184
5
43
42
2024
2026
0.603684
0
0.002176
7.78184
5
43
42
2027
2060
0.569634
0
0.002176
7.78184
5
43
46
1960
2009
0.746718
0
0.002176
11.3666
17.1
43
46
2014
2020
0.709382
0
0.002176
11.3666
7
43
46
2021
2023
0.637734
0
0.002176
11.3666
7
43
46
2024
2026
0.603684
0
0.002176
11.3666
7
43
46
2027
2060
0.569634
0
0.002176
11.3666
7
43
47
1960
2009
0.746718
0
0.002176
15.6028
17.1
43
47
2014
2020
0.709382
0
0.002176
15.6028
10
43
47
2021
2023
0.637734
0
0.002176
15.6028
10
43
47
2024
2026
0.603684
0
0.002176
15.6028
10
43
47
2027
2060
0.569634
0
0.002176
15.6028
10
51
41
1960
2009
1.58346
0
0.003572
3.57431
2.05979
51
41
2014
2060
1.50429
0
0.003572
3.57431
5
51
42
1960
2009
1.58346
0
0.003572
5.76818
2.05979
51
42
2014
2020
1.50429
0
0.003572
5.76818
5
51
42
2021
2023
1.50429
0
0.003572
5.76818
5
51
42
2024
2026
1.50429
0
0.003572
5.76818
5
51
42
2027
2060
1.32227
0
0.003572
5.76818
5
51
46
1960
2009
1.58346
0
0.003572
13.8001
17.1
51
46
2014
2020
1.50429
0
0.003572
13.8001
7
51
46
2021
2023
1.50429
0
0.003572
13.8001
7
51
46
2024
2026
1.50429
0
0.003572
13.8001
7
51
46
2027
2060
1.32227
0
0.003572
13.8001
7
51
47
1960
2009
1.58346
0
0.003572
20.7044
17.1
51
47
2014
2020
1.50429
0
0.003572
20.7044
10
51
47
2021
2023
1.50429
0
0.003572
20.7044
10
212
-------
Source
Type
ID
Reg
Class
ID
Begin
Model
Year
End
Model
Year
Rolling
Term A
(kW-s/m)
Rotating
Term B
(kW-s2/m2)
Drag
Term C
(kW-s3/m3)
Source Mass
(metric tons)
Fixed
Mass
Factor
(metric
tons)
51
47
2024
2026
1.50429
0
0.003572
20.7044
10
51
47
2027
2060
1.32227
0
0.003572
20.7044
10
52
41
1960
2009
0.627922
0
0.001603
3.57431
2.05979
52
41
2014
2060
0.596526
0
0.001603
3.57431
5
52
42
1960
2009
0.627922
0
0.001603
5.76818
2.05979
52
42
2014
2020
0.596526
0
0.001603
5.76818
5
52
42
2021
2023
0.558348
0
0.001603
5.76619
5
52
42
2024
2026
0.558348
0
0.001603
5.76344
5
52
42
2027
2060
0.53568
0
0.001603
5.76069
5
52
46
1960
2009
0.627922
0
0.001603
13.8001
17.1
52
46
2014
2020
0.596526
0
0.001603
13.8001
7
52
46
2021
2023
0.558348
0
0.001603
13.7981
7
52
46
2024
2026
0.558348
0
0.001603
13.7953
7
52
46
2027
2060
0.53568
0
0.001603
13.7926
7
52
47
1960
2009
0.627922
0
0.001603
25.0484
17.1
52
47
2014
2020
0.596526
0
0.001603
25.0484
10
52
47
2021
2023
0.558348
0
0.001603
25.0464
10
52
47
2024
2026
0.558348
0
0.001603
25.0437
10
52
47
2027
2060
0.53568
0
0.001603
25.0409
10
53
41
1960
2009
0.557262
0
0.001474
3.57431
2.05979
53
41
2014
2060
0.529399
0
0.001474
3.57431
5
53
42
1960
2009
0.557262
0
0.001474
5.76818
2.05979
53
42
2014
2020
0.529399
0
0.001474
5.76818
5
53
42
2021
2023
0.484929
0
0.001474
5.76461
5
53
42
2024
2026
0.458989
0
0.001474
5.75747
5
53
42
2027
2060
0.458989
0
0.001474
5.75033
5
53
46
1960
2009
0.557262
0
0.001474
13.8001
17.1
53
46
2014
2020
0.529399
0
0.001474
13.8001
7
53
46
2021
2023
0.484929
0
0.001474
13.7965
7
53
46
2024
2026
0.458989
0
0.001474
13.7894
7
53
46
2027
2060
0.458989
0
0.001474
13.7822
7
53
47
1960
2009
0.557262
0
0.001474
25.0484
17.1
53
47
2014
2020
0.529399
0
0.001474
25.0484
10
53
47
2021
2023
0.484929
0
0.001474
25.0449
10
53
47
2024
2026
0.458989
0
0.001474
25.0377
10
53
47
2027
2060
0.458989
0
0.001474
25.0306
10
54
41
1960
2009
0.68987
0
0.002105
3.57431
2.05979
54
41
2014
2060
0.655376
0
0.002105
3.57431
5
54
42
1960
2009
0.68987
0
0.002105
5.76818
2.05979
213
-------
Source
Type
ID
Reg
Class
ID
Begin
Model
Year
End
Model
Year
Rolling
Term A
(kW-s/m)
Rotating
Term B
(kW-s2/m2)
Drag
Term C
(kW-s3/m3)
Source Mass
(metric tons)
Fixed
Mass
Factor
(metric
tons)
54
42
2014
2020
0.655376
0
0.002105
5.76818
5
54
42
2021
2023
0.519058
0
0.002105
5.76818
5
54
42
2024
2026
0.519058
0
0.002105
5.76818
5
54
42
2027
2060
0.493498
0
0.002105
5.76818
5
54
46
1960
2009
0.68987
0
0.002105
13.8001
17.1
54
46
2014
2020
0.655376
0
0.002105
13.8001
7
54
46
2021
2023
0.519058
0
0.002105
13.8001
7
54
46
2024
2026
0.519058
0
0.002105
13.8001
7
54
46
2027
2060
0.493498
0
0.002105
13.8001
7
54
47
1960
2009
0.68987
0
0.002105
25.0484
17.1
54
47
2014
2020
0.655376
0
0.002105
25.0484
10
54
47
2021
2023
0.519058
0
0.002105
25.0484
10
54
47
2024
2026
0.519058
0
0.002105
25.0484
10
54
47
2027
2060
0.493498
0
0.002105
25.0484
10
61
46
1960
2007
1.64062
0
0.004077
14.0122
17.1
61
46
2008
2009
1.64062
0
0.004077
14.0122
17.1
61
46
2014
2017
1.509
0
0.003745
13.8666
7
61
46
2018
2020
1.509
0
0.003745
13.8666
7
61
46
2021
2023
1.407
0
0.003594
13.8666
7
61
46
2024
2026
1.379
0
0.003521
13.8666
7
61
46
2027
2060
1.353
0
0.003467
13.8666
7
61
47
1960
2007
1.64062
0
0.004077
24.8298
17.1
61
47
2008
2009
1.64062
0
0.004077
24.8298
17.1
61
47
2014
2017
1.509
0
0.003745
24.6842
10
61
47
2018
2020
1.509
0
0.003745
24.6842
10
61
47
2021
2023
1.407
0
0.003594
24.6842
10
61
47
2024
2026
1.379
0
0.003521
24.6842
10
61
47
2027
2060
1.353
0
0.003467
24.6842
10
62
46
1960
2007
1.73882
0
0.004347
14.0122
17.1
62
46
2008
2009
1.73882
0
0.004278
14.0122
17.1
62
46
2014
2017
1.576
0
0.003805
13.8308
7
62
46
2018
2020
1.576
0
0.003805
13.8308
7
62
46
2021
2023
1.502
0
0.003685
13.8308
7
62
46
2024
2026
1.466
0
0.003598
13.8308
7
62
46
2027
2060
1.44
0
0.003466
13.8308
7
62
47
1960
2007
1.73882
0
0.004347
24.8298
17.1
62
47
2008
2009
1.73882
0
0.004278
24.8298
17.1
62
47
2014
2017
1.576
0
0.003805
24.6484
10
62
47
2018
2020
1.576
0
0.003805
24.6484
10
214
-------
Source
Type
ID
Reg
Class
ID
Begin
Model
Year
End
Model
Year
Rolling
Term A
(kW-s/m)
Rotating
Term B
(kW-s2/m2)
Drag
Term C
(kW-s3/m3)
Source Mass
(metric tons)
Fixed
Mass
Factor
(metric
tons)
62
47
2021
2023
1.502
0
0.003685
24.6484
10
62
47
2024
2026
1.466
0
0.003598
24.6484
10
62
47
2027
2060
1.44
0
0.003466
24.6484
10
31
41
2010
2060
0.22112
0.002838
0.000698
3.40194
5
32
41
2010
2060
0.235008
0.003039
0.000748
3.40194
5
41
41
2010
2013
1.29515
0
0.003715
5.68398
5
41
42
2010
2013
1.29515
0
0.003715
7.78184
5
41
46
2010
2013
1.29515
0
0.003715
11.3666
7
41
47
2010
2013
1.29515
0
0.003715
15.6028
10
42
42
2010
2013
1.0944
0
0.003587
7.78184
5
42
46
2010
2013
1.0944
0
0.003587
11.3666
7
42
47
2010
2013
1.0944
0
0.003587
15.6028
10
42
48
2010
2013
1.0944
0
0.003587
15.6028
10
43
41
2010
2013
0.746718
0
0.002176
5.68398
5
43
42
2010
2013
0.746718
0
0.002176
7.78184
5
43
46
2010
2013
0.746718
0
0.002176
11.3666
7
43
47
2010
2013
0.746718
0
0.002176
15.6028
10
51
41
2010
2013
1.58346
0
0.003572
3.57431
5
51
42
2010
2013
1.58346
0
0.003572
5.76818
5
51
46
2010
2013
1.58346
0
0.003572
13.8001
7
51
47
2010
2013
1.58346
0
0.003572
20.7044
10
52
41
2010
2013
0.627922
0
0.001603
3.57431
5
52
42
2010
2013
0.627922
0
0.001603
5.76818
5
52
46
2010
2013
0.627922
0
0.001603
13.8001
7
52
47
2010
2013
0.627922
0
0.001603
25.0484
10
53
41
2010
2013
0.557262
0
0.001474
3.57431
5
53
42
2010
2013
0.557262
0
0.001474
5.76818
5
53
46
2010
2013
0.557262
0
0.001474
13.8001
7
53
47
2010
2013
0.557262
0
0.001474
25.0484
10
54
41
2010
2013
0.68987
0
0.002105
3.57431
5
54
42
2010
2013
0.68987
0
0.002105
5.76818
5
54
46
2010
2013
0.68987
0
0.002105
13.8001
7
54
47
2010
2013
0.68987
0
0.002105
25.0484
10
61
46
2010
2013
1.64062
0
0.004077
14.0122
7
61
47
2010
2013
1.64062
0
0.004077
24.8298
10
62
46
2010
2013
1.73882
0
0.004278
14.0122
7
62
47
2010
2013
1.73882
0
0.004278
24.8298
10
61
49
1960
2007
1.64062
0
0.004077
24.8298
17.1
61
49
2008
2009
1.64062
0
0.004077
24.8298
17.1
215
-------
Source
Type
ID
Reg
Class
ID
Begin
Model
Year
End
Model
Year
Rolling
Term A
(kW-s/m)
Rotating
Term B
(kW-s2/m2)
Drag
Term C
(kW-s3/m3)
Source Mass
(metric tons)
Fixed
Mass
Factor
(metric
tons)
61
49
2010
2013
1.64062
0
0.004077
24.8298
17.1
61
49
2014
2017
1.509
0
0.003745
24.6842
17.1
61
49
2018
2020
1.509
0
0.003745
24.6842
17.1
61
49
2021
2023
1.407
0
0.003594
24.6842
17.1
61
49
2024
2026
1.379
0
0.003521
24.6842
17.1
61
49
2027
2060
1.353
0
0.003467
24.6842
17.1
62
49
1960
2007
1.73882
0
0.004347
24.8298
17.1
62
49
2008
2009
1.73882
0
0.004278
24.8298
17.1
62
49
2010
2013
1.73882
0
0.004278
24.8298
17.1
62
49
2014
2017
1.576
0
0.003805
24.6484
17.1
62
49
2018
2020
1.576
0
0.003805
24.6484
17.1
62
49
2021
2023
1.502
0
0.003685
24.6484
17.1
62
49
2024
2026
1.466
0
0.003598
24.6484
17.1
62
49
2027
2060
1.44
0
0.003466
24.6484
17.1
216
-------
Appendix K AVFT Tool
The AVFT Tool is a user tool available in the MOVES GUI that can be used to develop the
AVFT (Alternate Vehicle Fuel and Technology) table, which allows users to modify the fraction
of vehicles capable of using different fuels and technologies. The purpose of this tool is to
project future fuel type distributions based on the combination of local historic data and
projected national trends. The projections are applied to model years beyond the user-specified
last complete model year in the input data to the user-specified analysis year. The last complete
model year is needed, as partial model years are common in vehicle registration data. For
simplicity, the last complete model year will be referred to as the "base model year" throughout
this appendix.
The tool contains the following methods to project future fuel type distributions: proportional,
national average, known fractions, and constant.
The proportional, national average, and known fraction projection methods are dependent on
knowing the national default fuel type distributions. These are calculated from the default
SampleVehiclePopulation table by summing the stmyFraction values associated with each source
type, model year, fuel type, and engine technology combination (see Section 5.2 for information
on how the default SampleVehiclePopulation table was calculated). In effect, this calculation is
aggregating over the regClassID column in the SampleVehiclePopulation table to calculate a
nationally representative AVFT table. In keeping with the naming convention of the AVFT
columns, the resulting sum of the stmyFraction column will be referred to as the fuelEngFraction
throughout this appendix.
The following subsections describe each projection method. Note that these projection methods
are selected by source type, so different source types can use different methods. As such, the
algorithm descriptions below assume the algorithm is operating on a single source type for
clarity.
Kl. Proportional
This method projects future fuel type distributions based on proportional differences between the
local and the national distributions in the base model year. The intention with this method is to
preserve differences between local conditions and the national average, while still accounting for
expected changes in national fuel type distribution trends.
To implement this method, the ratio between the input AVFT fuelEngFraction and the national
default fuelEngFraction is calculated for the base model year for each fuel type and engine
technology. This ratio is subject to the boundary limits of 0.5 and 2.0, which effectively limits
the proportional projection to stay between 50% lower and 200% higher than the national default
fraction for any given fuel type and engine technology combination. These limits were chosen
because there is considerable uncertainty as to the future geographic distribution of EVs, and we
did not want extreme differences between the national averages and local conditions in the base
model year to inappropriately bias the projected data. For example, if an area has no EVs in the
base model year, a proportional projection without a lower boundary limit would result in no
EVs in the future. Given the projected future EV sales fractions incorporated into MOVES (see
217
-------
Section 5.2), this does not seem reasonable. Conversely, if an area is an early adopter of EVs in
the base model year, a proportional projection without an upper boundary limit could result in an
unrealistic projection of 100% EVs. We chose boundary limits of 50% and 200% (instead of
looser limits, like 33% and 300% or 25% and 400%) because the ratio is based on a single model
year and we wanted to prevent the single model year from introducing too much bias in the
results.
After calculating the ratio and applying the boundary limits, the projected fuelEngFraction is
calculated by multiplying the base model year ratio with the national default fuelEngFraction for
all model years between the base model year and the analysis year. Since the base model year
ratio is calculated independently for each fuel type, the resulting projected fuelEngFraction
values are normalized so that the sum of the fuelEngFraction values by model year sum to 1.0.
K2. Nations rage
This method applies the national default fuel type distributions for all model years beyond the
base model year in the input data. The intention of this method is to allow users to provide
AVFT inputs for source types where local data do not reflect local vehicle activity (for example,
long haul source types), or where local data are not available for a particular source type.
To implement this method, the national default fuelEngFraction for each model year, fuel type,
and engine technology is used as-is for the projected fuelEngFraction.
K3. Known Fractions
This method allows the user to provide known fuel fractions for specific fuel types. The intention
with this method is to allow users to input values mandated by local programs (such as a ZEV
program), and the tool will handle fuel types not explicitly covered by the local program.
Essentially, this method applies the proportional method for all model years, fuel types, and
engine technologies not included in the user provided fractions.
To implement this method, the known fractions are used directly for the projected
fuelEngFraction values. Then, like the proportional method, the ratio between the input AVFT's
fuelEngFraction and the national default fuelEngFraction is calculated for the base model year
for each fuel type and engine technology. This ratio is subject to the same boundary limits of
50% and 200%.
For all model years, fuel types, and engine technologies not included in the input known
fractions, the projected fuelEngFraction is calculated by multiplying the base model year ratio
with the national default fuelEngFraction. Finally, the projected fuelEngFraction values are
normalized so that the sum of the fuelEngFraction by model year sums to 1.0, without changing
the input known fractions.
Constant
This method uses the fuelEngFraction for the base model year for each fuel type and engine
technology as-is for the projected fuelEngFraction values for all projected model years.
218
-------
18. References
1 Motor Vehicle Emission Simulator (MOVES) technical reports are available on the US
Environmental Protection Agency website:
https://www.epa.gov/moves/moves-technical-reports
2 US EPA, MOVES4 Technical Guidance: Using MOVES to Prepare Emission Inventories for
State Implementation Plans and Transportation Conformity, . EPA-420-B-23-001. Ann Arbor,
MI
https://www.epa.gov/state-and-local-transportation/policv-and-technical-guidance-state-and-
local-transportation#emission
3 USEPA (2015). U.S. Environmental Protection Agency Peer Review Handbook. EPA/100/B-
15/001. Prepared for the U.S. Environmental Protection Agency under the direction of the EPA
Peer Review Advisory Group. Washington, D.C. 20460. October 2015.
epa.gov/sites/production/files/2020-Q8/documents/epa peer review handbook 4th edition.pdf
4 USEPA (2017). Population and Activity of Onroad Vehicles in MO VES20IX - Draft Report.
Draft report and peer-review documents. Record ID 328870. EPA Science Inventory. September
2017.
https://cfpub.epa.gov/si/si public record report.cfm?dirEntryId=328870
5 USEPA (2019). Exhaust Emission Rates of Heavy-Duty Onroad Vehicles in
MOVES CTI NPRM - Draft Report. Draft report and peer-review documents. Record ID
347135. EPA Science Inventory. December 2019.
https://cfpub.epa.gov/si/si public record report.cfm?dirEntryId=347135
6 US FHWA, Table VM-1, "Annual Vehicle Distance Traveled in Miles and Related Data,"
Highway Statistics, 1990-2021, Washington, DC: May 2023.
https://www.fhwa.dot.gov/policvinformation/statistics/2021/vml.cfm
7 USEPA, Frequently Asked Questions about Heavy-Duty "Glider Vehicles" and "Glider Kits",
EPA-420-F-15-904, Ann Arbor, MI: July 2015.
https://nepis.epa.gov/Exe/ZvPDF.cgi?Dockev=P100MUVI.PDF
8USEPA (2023). Fuel Supply Defaults: Regional Fuels and the Fuel Wizard inMOVES4. EPA-
420-R-23-025 Office of Transportation and Air Quality. US Environmental Protection Agency.
Ann Arbor, MI. August 2023
https://www.epa.gov/moves/moves-onroad-technical-reports
219
-------
9 US EPA, Development of Gasoline Fuel Effects in the Motor Vehicle Emissions Simulator
(MOVES2009), EPA-420-P-09-004, Ann Arbor, MI: August 2009.
http://www.epa.gov/otaq/models/moves/techdocs/420pQ9004.pdf
10 USEPA (2023). Exhaust Emission Rates for Light-Duty Onroad Vehicles in MOVES4. EPA-
420-R-23-028. Office of Transportation and Air Quality. US Environmental Protection Agency.
Ann Arbor, MI. August 2023.
https://www.epa.gov/moves/moves-technical-reports
11 USEPA (2023). Exhaust Emission Rates of Heavy-Duty Onroad Vehicles inMOVES4. EPA-
420-R-23-027. Office of Transportation and Air Quality. US Environmental Protection Agency.
Ann Arbor, MI. August 2023.
https://www.epa.gov/moves/moves-technical-reports
12 US FHWA, Annual Vehicle Miles Travelled and Related Data: Procedures Used to Derive
Data Elements Contained in Highway Statistics Table VM-1 for Years 2009 and after and 2007
and 2008 Historical Data, FHWA-PL-11-031, Washington, DC: August 2011.
http://www.fhwa.dot.gov/ohim/vml methodology 2007.pdf
13 US Energy Information Administration (EIA), Annual Energy Outlook 2023, Washington, DC:
February 2023.
14 US Federal Highway Administration (FHWA), Office of Highway Policy Information (OHPI),
Table MV-1, "State Motor-Vehicle Registrations," Highway Statistics, 1990-2021, Washington,
DC.
https://www.fhwa.dot.gov/policyinformation/statistics/2021/mvl.cfm
15IHS, Inc. (formerly R.L. Polk & Co, now part of S&P Global), National Vehicle Population
Profile®, Southfield, MI; 1999, 2011, and 2014.
https ://www. spglobal. com/en/enterpri se/btp/P oik, html
16 Vehicle in Operation Registration Data, IHS Markit-Polk. July 2020.
https://www.spglobal.com/mobilitv/en/products/automotive-market-data-analvsis.html
17 US EPA, MOVES2010 Highway Vehicle Population and Activity Data, EPA-420-R-10-026,
Washington, DC: November 2010.
https://nepis.epa.gov/Exe/ZvPDF.cgi?Dockev=P100ABRQ.pdf
220
-------
18 Bobit Publications, School Bus Fleet Fact Book, Torrance, CA: 1992, 2002, 2004-2023.
http ://www. schoolbusfleet. com
19 US Federal Transit Administration (FTA), National Transit Database, 2002-2021.
https://www.transit.dot.gov/ntd
20 Koupal, J., T. DeFries, C. Palacios, S. Fincher and D. Preusse (2014). Motor Vehicle
Emissions Simulator Input Data. Transportation Research Record: Journal of the Transportation
Research Board, 2427, 63-72. DOI: doi: 10.3141/2427-07.
21 Yoon, S., Georgia Institute of Technology, A New Heavy-Duty Vehicle Visual Classification
and Activity Estimation Method for Regional Mobile Source Emissions Modeling (student
thesis), Atlanta, GA: August 2005.
https://smartech.gatech.edu/bitstream/handle/1853/7245/seungiu yoon 200508 phd.pdf
22 US FHWA, "Vehicle Type Codes and Descriptions," Highway Performance Monitoring
System Field Manual, Washington, DC: last updated 4 April 2011.
http ://www.fhwa. dot, gov/ohim/hpmsmanl/chapt3. cfm
23 Weatherrby, M., Jackson, D., Koupal, J., and DenBleyker, A., Eastern Research Group, Inc.
(ERG), Analysis of IHS Registration Data and Preparation of WA 5-08 Task 1 Deliverables,
Memorandum to David Brzezinski, EPA. EPA EP-C-12-017, Work Assignment 5-08, Austin,
TX: July, 2017.
24 US FTA, National Transit Database, "NTD Glossary," last updated 19 February 2015.
https://www.transit.dot.gov/ntd
25 US FHWA, Office of Planning, Environment, and Realty, "Planning Glossary," updated 21
March 2012.
https://www.fhwa.dot.gov/planning/glossary/
26 Stanard, A., Fincher, S., Kishan, S., and Sabisch, M., Eastern Research Group, Inc. (ERG),
Data Analyses on Drayage Heavy-Duty Vehicles. EPA EP-C-12-017, Work Assignment 0-2,
Austin, TX: 7 December 2012.
27 US Environmental Protection Agency. 2022 EPA Automotive Trends Report. Data available at
www.epa.gov/automotive-trends/explore-automotive-trends-data. Accessed July, 2023
221
-------
28 US EPA, Optimization Model for reducing Emissions of Greenhouse Gases from
Automobiles (OMEGA). 2023.
https://www.epa.gov/regulations-emissions-vehicles-and-engines/optimization-model-reducing-
emissions-greenhouse-gases#overview
29 US Environmental Protection Agency. 2022 EPA Automotive Trends Report. Data available at
www.epa.gov/automotive-trends/explore-automotive-trends-data. Accessed July, 2023
30 US EPA, Optimization Model for reducing Emissions of Greenhouse Gases from
Automobiles (OMEGA). 2023.
https://www.epa.gov/regulations-emissions-vehicles-and-engines/optimization-model-reducing-
emissions-greenhouse-gases#overview
31 California Air Resources Board, 2021 Advanced Clean Trucks Regulation.
https://ww2.arb.ca.gov/rulemaking/2019/advancedcleantrucks
32 US Government Publishing Office (GPO), Code of Federal Regulations, Title 40 - Protection
of Environment, Vol. 19, CFR 86.091-2, "Definitions," 1 July 2012.
http://www.gpo.gov/fdsvs
33 US EPA, 2019 Annual Production Volume Reports into Engine and Vehicle Compliance
Information System.
34 Brakora, J. 2018. Estimated Glider Vehicle Production for Use in EPA MOVES Models.
Jessica Brakora. Cleaner Trucks Initiative - Docket EPA-HQ-OAR-2019-0055. Draft
memorandum to the docket.
35 The Age Distribution Projection Tool for MOVES is a macro-enabled spreadsheet and is
available for download at:
https://www.epa.gov/moves/tools-develop-or-convert-moves-inputs
36 US EPA, Use of Data from "Development of Emission Rates for the MOVES Model, " Sierra
Research, March 3, 2010, EPA-420-R-12-022, Ann Arbor, MI: August 2012.
37 National Highway Traffic Safety Administration, Final Rulemaking for Model Years 2024-206
Light-Duty Vehicle Corporate Average Fuel Economy Standards, Technical Support Document,
March 2022.
222
-------
38 US EPA, Population and Activity of Onroad Vehicles inM0VES3, EPA-420-R-21-012, April
2021.
https://www.epa.gov/moves/moves-technical-reports
39 Browning, L., Chan, M., Coleman, D., and Pera, C., ARCADIS Geraghty & Miller, Inc.,
Update of Fleet Characterization Data for Use inM0BILE6: Final Report, EPA420-P-98-016,
Mountain View, CA: 11 May 1998.
http://nepis.epa.gov/Exe/ZyPDF.cgi?Dockev=P 1001ZUK.PDF
40 US Census Bureau, 2002 Vehicle Inventory and Use Survey, EC02TV-US, Washington, DC:
December 2004.
https://www.census.gov/librarv/publications/2002/econ/census/vehicle-inventory-and-use-
survev.html
41 Browning, L., Chan, M., Coleman, D., and Pera, C., ARCADIS Geraghty & Miller, Inc.,
Update of Fleet Characterization Data for Use inM0BILE6: Final Report, EPA420-P-98-016,
Mountain View, CA: 11 May 1998.
http://nepis.epa.gov/Exe/ZyPDF.cgi?Dockev=P 1001ZUK.PDF
42 National Household Travel Survey, 2017 NHTS Average Annual Vehicle Miles of Travel Per
Vehicle (Best Estimate).
https://nhts.ornl.gov
43 US EPA, The National Emission Inventory (NEI).
https://www.epa.gov/air-emissions-inventories/national-emissions-inventory-nei
44 U.S. Department of Transportation Federal Highway Administration Office of Highway Policy
Information Highway Statistics Series.
https://www.fhwa.dot.gov/policvinformation/statistics.cfm
45 CRC Project A-100, Improvement of Default Inputs for MOVES and SMOKE-MOVES.
46 U.S. EPA, 2015. Technical Support Document, Preparation of Emissions Inventories for the
Version 6.2, 2011 Emissions Modeling Platform. Office of Air Quality Planning and Standards.
47 National Emissions Inventory Collaborative. 2016beta Emissions Modeling Platform. 2019.
http://views.cira.colostate.edu/wiki/wiki/10197
223
-------
48 Sierra Research, Inc., Development of Speed Correction Cycles, M6.SPD.001, EPA 68-C4-
0056, Sacramento, CA: 26 June 1997.
49 Hart, C., Koupal, J., and Giannelli, R., US EPA, EPA's Onboard Analysis Shootout: Overview
and Results, EPA420-R-02-026, Ann Arbor, MI: October 2002.
https://nepis.epa.gov/Exe/ZyPDF.cgi?Dockev=P10005PG.txt
50 DieselNet, Emission Test Cycles: New York Bus. Last accessed 8 June 2015.
https://www.dieselnet.com/standards/cvcles/nybus.php
51 Melendez, M. T., J; Zuboy, J (2005). Emission Testing of Washington Metropolitan Area
Transit Authority (WMATA) Natural Gas and Diesel Transit Buses. NREL/TP-540-36355.
December 2005.
http://www.afdc. energy. gov/pdfs/3 63 5 5. pdf
52 Clark, N. and Gautam, M., West Virginia Research Corporation, Heavy-Duty Vehicle Chassis
Dynamometer Testing for Emissions Inventory, Air Quality Modeling, Source Apportionment
and Air Toxics Emissions Inventory, CRC Project E55/59, Morgantown, VW: 12 July 2005.
https://crcao.org/wp-content/uploads/2019/05/E-55 59 Final Report 23AUG2007.pdf
53 Sensors, Inc., On-Road Emissions Testing of 18 Tier 1 Passenger Cars and 17 Diesel Powered
Public Transport Buses, EPA-420-R-02-030. Saline, MI: 22 October 2002.
https://nepis.epa. gov/Exe/ZyPDF.cgi?Dockev=P10005S8.txt
54 Eastern Research Group, Inc., Roadway-Specific Driving Schedules for Heavy-Duty Vehicles,
EPA420-R-03-018, Austin, TX: August 2003.
https://nepis.epa.gov/Exe/ZyPDF.cgi?Dockev=P100LWCT.txt
55 US EPA, Greenhouse Gas Emissions Model (GEM) for Medium- and Heavy-Duty Vehicle
Compliance, Simulation Model v2.0.1, May 2013.
https://www.epa.gov/regulations-emissions-vehicles-and-engines/greenhouse-gas-emissions-
model-gem-medium-and-heavy-duty
56 Liu, H., D. Sonntag, D. Brzezinski, C. R. Fulper, D. Hawkins and J. E. Warila (2016).
Operations and Emissions Characteristics of Light-Duty Vehicles on Ramps. Transportation
Research Record: Journal of the Transportation Research Board, 2570, 1-11. DOI:
doi: 10.3141/2570-01.
224
-------
57 US EPA. Amendments to Vehicle Inspection Maintenace Program Requirements
Incorporating the Onboard Diagnositc Check. 66 FR 18156, April 5, 2001
58 Walkowicz, K., K. Kelly, A. Duran and E. Burton (2014). Fleet DNA Project Data. National
Renewable Energy Laboratory.
https://www.nrel.gov/transportation/fleettest-fleet-dna.html
59 Controller Area Network (CAN) Overview. National Instruments. Accessed August 1, 2014.
http://www.ni.com/white-paper/2732/en
60 ATRI. Compendium of Idling Regulations. Updated August 2015.
https://truckingresearch.org/2021/10/idling-regulations-compendium/
61 Kotz, A. and K. Kelly. MOVES Activity Updates Using Fleet DNA Data: Interim Report.
NREL/TP-5400-70671. National Renewable Energy Laboratory. Golden, CO. January 2019.
https://www.nrel.gov/docs/fyl9osti/70671.pdf
62 Greenhouse Gas Emissions and Fuel Efficiency Standards for Medium- and Heavy-Duty
Engines and Vehicles— Phase 2. 81 FR 206 (October 25, 2016) Section III.D.l.c.v "Idle
Reduction", page 73593.
63 National Academies Press. "Commercial Motor Vehicle Driver Fatigue, Long-Term Health,
and Highway Safety: Research Needs." Aug 12, 2016.
https://www.ncbi.nlm.nih.gov/books/NBK384967
64 Greenhouse Gas Emissions and Fuel Efficiency Standards for Medium- and Heavy-Duty
Engines and Vehicles— Phase 2. 81 FR 206 (October 25, 2016) Section III.D.l.a "Tractor
Baselines for Costs and Effectiveness", page 73587.
65 Greenhouse Gas Emissions and Fuel Efficiency Standards for Medium- and Heavy-Duty
Engines and Vehicles— Phase 2. 81 FR206 (October 25, 2016) Section III.D.l.c.v "Idle
Reduction", page 73593.
66 NACFE (2018). 2018 Annual Fleet Fuel Study. North American Council for Freight
Efficiency.
http s ://nacfe. or g/annual -fl eet-fuel -studi es/
225
-------
67 Schoettle, B., M. Sivak and M. Tunnell. A Survey of Fuel Economy and Fuel Usage by Heavy-
Duty Truck Fleets. SWT-2016-12. The University of Michigan. Sustainable Worldwide
Transportation. October 2016.
http://umich.edu/~umtriswt/PDF/SWT-2016-12.pdf
68USEPA. 2014 National Emissions Inventory, version 2: Technical Support Document. Section
6.8.4. Office of Air Quality Planning and Standards. Research Triangle Park, North Carolina.
July 2018.
https://www.epa.gov/sites/production/files/2018-07/documents/nei2014v2 tsd 05iul2018.pdf
69 Lutsey, N., C.-J. Brodrick, D. Sperling and C. Oglesby (2004). Heavy-Duty Truck Idling
Characteristics: Results from a Nationwide Truck Survey. Transportation Research Record,
1880 (1), 29-38. DOI: 10.3141/1880-04.
70 TRB (2019). Guide to Truck Activity Data for Emissions Modeling. Transportation Research
Board. National Academies of Sciences, Engineering, Medicine. The National Academies Press.
Washington, DC.
https://www.nap.edu/catalog/25484/guide-to-truck-activitv-data-for-emissions-modeling
71 Festin, S., US FHWA, Summary of National and Regional Travel Trends: 1970-1995,
Washington, DC: May 1996.
http://www.fhwa.dot.gov/ohim/bluebook.pdf
72 Motorcycle crash data from NHTSA, Fatality Analysis Reporting System (FARS).
http://www.nhtsa.gov/FARS Raw 2010 data last updated 11 December 2012.
73 Guensler, R., Yoon, S., Li, H., and Elango, V., Georgia Institute of Technology, Atlanta
Commute Vehicle Soak and Start Distributions and Engine Starts per Day: Impact on Mobile
Source Emission Rates, EPA/600/R-07/075, Atlanta, GA: April 2007.
http://nepis.epa. gov/Adobe/PDF/P 100AE2E.pdf
74 USEPA. Evaporative Emissions from Onroad Vehicles in MOVES4. EPA-420-R-23-023 .
Office of Transportation and Air Quality. US Environmental Protection Agency. Ann Arbor, MI.
August 2023.
https://www.epa.gov/moves/moves-technical-reports
226
-------
75 Sierra Research, Inc., Development of Trip and Soak Activity Defaults for Passenger Cars and
Trucks inMOVES2006, SR2006-03-04, EPA Contract EP-C-05-037, Work Assignment No. 0-
01, Sacramento, CA: 27 March 27, 2006.
76 USEPA, Development of Evaporative Emissions Calculations for the Motor Vehicle Emissions
Simulator MOVES2010, EPA-420-R-12-027, Ann Arbor, MI: September 2012.
https://nepis.epa.gov/Exe/ZyPDF.cgi?Dockev=P100F3ZY.txt
77 Browning, L., Chan, M., Coleman, D., and Pera, C., ARCADIS Geraghty & Miller Inc.,
Update of Fleet Characterization Data for Use in MOBILE6 - Final Report, EPA420-P-98-016,
Mountain View, CA: 11 May 1998.
https://nepis.epa.gov/Exe/ZvPDF.cgi?Dockev=P1001ZUK.txt
78 Lutsey, N., Brodrick, C., Sperling, D., and Oglesby, C., "Heavy-Duty Truck Idling
Characteristics: Results from a Nationwide Truck Survey," Transportation Research Record
(1880), pg. 29-38, Washington, DC: January 2004.
79 Indale, G., University of Tennessee, Effects of Heavy-Duty Diesel Vehicle Idling Emissions on
Ambient Air Quality at a Truck Travel Center and Air Quality Benefits Associated with
Advanced Truck Stop Electrification Technology (PhD dissertation), Knoxville, TN: May 2005.
http://trace.tennessee.edu/utk graddiss/2085
80 FHWA. Jason's Law Truck Parking Survey Results and Comparative Analysis.
ops.fhwa.dot.gov/freight/infrastructure/truck parking/iasons law/truckparkingsurvev/ch3.htm
81 USEPA. 2020 National Emissions Inventory. Office of Air Quality Planning and Standars.
Research Triangle Park, North Carolina. 2023.
https://www.epa.gov/air-emissions-inventories/2020-national-emissions-inventory-nei-data
82 US Census Bureau. 2020 Census.
https://www.census.gov/2020results
83 Greenhouse Gas Emissions and Fuel Efficiency Standards for Medium- and Heavy-Duty
Engines and Vehicles—Phase 2, 81 FR 73478, October 25, 2016.
84 National Research Council, Technologies and Approaches to Reducing the
Fuel Consumption of Medium- and Heavy-Duty Vehicles. Table 2-1. Washington, DC: The
National Academies Press. 2010. https://doi.org/10.17226/12845.
227
-------
85 US FHWA, Vehicle Travel Information System. Table W-3: 2013,
http://www.fhwa.dot.gov/ohim/ohimvtis.cfm
86 US FHWA, Bridge Formula Weights, FHWA-HOP-06-105, August 2006.
87 Sandhu, G. S., H. C. Frey, S. Bartelt-Hunt, and E. Jones. In-Use Measurement of the Activity,
Fuel Use, and Emissions of Front-Loader Refuse Trucks. Atmospheric Environment, Vol. 92,
2014, pp. 557-565
https://doi.Org/10.1016/i.atmosenv.2014.04.036
88 Sandhu, G. S., H. C. Frey, S. B artel t-Hunt, and E. Jones. In-Use Activity, Fuel Use, and
Emissions of Heavy Duty Diesel Roll-off Refuse Trucks. Journal of the Air & Waste
Management Association, Vol. 65, No. 3, 2015, pp. 306-323.
https://doi.org/10.1080/10962247.2014.99Q587
89 Sandhu, G. S., H. C. Frey, S. Bartelt-Hunt, and E. Jones. Real-World Activity, Fuel Use, and
Emissions of Diesel Side-Loader Refuse Trucks. Atmospheric Environment, Vol. 129, 2016, pp.
98-104.
https://doi.Org/10.1016/i.atmosenv.2016.01.014
90 Sandhu, G. S. Evaluation of Activity, Fuel Use, and Emissions of Heavy-Duty Diesel and
Compressed Natural Gas Vehicles. Ph.D. North Carolina State University, Raleigh, NC, USA,
2015.
http ://www.lib .ncsu.edu/resolver/1840.16/10222
91 US EPA, Final Rulemaking to Establish Greenhouse Gas Emissions Standards and Fuel
Efficiency Standards for Medium- and Heavy-Duty Engines and Vehicles: Regulatory Impact
Analysis, EPA-420-R-11-901, Ann Arbor, MI: August 2011.
92 Memorandum to the Docket "FRM - Tractor-Trailer Inputs to MOVES" Docket No. EPA-HQ-
OAR-2014-0827-2221.
93 Memorandum to the Docket "FRM - Vocational Inputs to MOVES" Docket No. EPA-HQ-
OAR-2014-0827-2222.
94 American Public Transportation Association, An Analysis of Transit Bus Axle Weight Issues,
Table 7: November 2014.
228
-------
95 US Government Publishing Office (GPO), Code of Federal Regulations, Title 40 - Protection
of Environment, Vol. 17, CFR 86.529-78, "Road load force and inertia weight determination," 1
July 2004.
http://www.gpo.gov/fdsvs
96 Steven, H., United Nations Economic Commisssion for Europe, Worldwide Harmonised
Motorcycle Emissions Certification Procedure, UN/ECE-WP 29 - GRPE (informal document
no. 9, 46th GRPE, 13-17 January 2003, agenda item 3), Geneva, Switzerland: 28 December
2002.
http://www.unece.org
97 Warila, J., "Derivation of Mean Energy Consumption Rates within the MOVES Modal
Framework," 14th Coordinating Research Council On-Road Vehicle Emissions Workshop
(poster), San Diego, CA: 29-31 March 2004.
98 US EPA, IM240 andEvap Technical Guidance, EPA420-R-00-007, Ann Arbor, MI: April
2000.
https://nepis.epa.gov/Exe/ZyPDF.cgi?Dockev=P1008F0I.txt
99 Final Rule for Model Year 2017 and Later Light-Duty Vehicle Greenhouse Gas Emissions and
Corporate Average Fuel Economy Standards, 77 FR 62624, October 15, 2012.
100 Petrushov, V., "Coast Down Method in Time-Distance Variables," SAE International, SAE
970408, Detroit, MI: 24 February 1997.
http://www.sae.org
101 US EPA, Final Rulemaking to Establish Greenhouse Gas Emissions Standards and Fuel
Efficiency Standards for Medium- and Heavy-Duty Engines and Vehicles: Regulatory Impact
Analysis, EPA-420-R-11-901, Ann Arbor, MI: August 2011.
https://nepis.epa.gov/Exe/ZyPDF.cgi?Dockev=P100EG9C.txt
102 Greenhouse Gas Emissions and Fuel Efficiency Standards for Medium- and Heavy-Duty
Engines and Vehicles— Phase 2. Federal Register Volume 81, No. 206 (October 25, 2016)
Pages 73587, 73608-73611.
229
-------
103 US EPA, Final Rulemaking to Establish Greenhouse Gas Emissions Standards and Fuel
Efficiency Standards for Medium- and Heavy-Duty Engines and Vehicles: Regulatory Impact
Analysis, EPA-420-R-11-901, Ann Arbor, MI: August 2011
104 US EPA, Emission Adjustments for Temperature, Humidity, Air Conditioning, and Inspection
and Maintenance for On-road Vehicles in MOVES4.
https://www.epa.gov/moves/moves-technical-reports
105 Koupal, J., Air Conditioning Activity Effects in MOBILE6, M6. ACE.001, EPA420-R-01-054,
Ann Arbor, MI: November 2001.
https://nepis.epa.gov/Exe/ZvPDF.cgi?Dockev=P100226H.txt
106 US Census Bureau, 1997 Vehicle Inventory and Use Survey, EC97TV-US, Washington, DC:
October 1999.
www.census.gov/librarv/publications/1997/econ/census/vehicle-inventory-and-use-survev.html
107 Stanard, A., Fincher, S., Kishan, S., and Sabisch, M., Eastern Research Group, Inc. (ERG),
Data Analyses on Drayage Heavy-Duty Vehicles. EPA EP-C-12-017, Work Assignment 0-2,
Austin, TX: 7 December 2012.
108 Davis, S. and Truitt, L., ORNL, Investigation of Class 2b Trucks (Vehicles of8,500 to 10,000
lbs GVWR), ORNL/TM-2002.49, Oak Ridge, TN: March 2002.
https://citeseerx.ist.psu.edU/viewdoc/download:i sessionid=042E66B7C5DCC456376A489C50F
F419F?doi=10.1.1.5 66.65 0&rep=rep 1 &tvpe=pdf
109 Union of Concerned Scientists (personal communication).
http://www.ucsusa.org
110 Data from 2011 was provided by US FHWA to replace Table II.2 and II.3 in the 1997 Federal
Highway Cost Allocation Study, specifying vehicles by weight (email correspondence), January
2013.
111 US EPA, Use of Data from "Development of Emission Rates for the MOVES Model, " Sierra
Research, March 3, 2010, EPA-420-R-12-022, Ann Arbor, MI: August 2012.
https://nepis.epa.gov/Exe/ZyPDF.cgi?Dockev=P100F lA5.txt
230
-------
112 National Highway Traffic and Safety Administration (NHTSA), "Vehicle Survivability and
Travel Mileage Schedules," DOT HS 809 952, Springfield, VA: January 2006.
http://www-nrd.nhtsa.dot.gov/Pubs/809952.pdf
113 Davis, S., Diegel, S., and Boundy, R., Oak Ridge National Laboratory (ORNL), Center for
Transportation Analysis, Transportation Energy Data Book (TEDB) Edition 40.
https ://tedb. ornl. gov
114 Motorcycle Industry Council, Motorcycle Statistical Annual, Irvine, CA: 2015.
https://www.mic.org
115 Ward's Automotive Inc.
http://www.wardsauto.com
116 US Government Publishing Office (GPO), Code of Federal Regulations, Title 40 - Protection
of Environment, Vol. 36, CFR 1037.520, "Modeling C02 emissions to show compliance for
vocational vehicles and tractors," 25 October 2016.
http://www.gpo.gov/fdsvs
117 US Government Publishing Office (GPO), Code of Federal Regulations, Title 40 - Protection
of Environment, Vol. 36, CFR 1037.515, "Determining CO2 emissions to show compliance for
trailers," 25 October 2016.
http://www.gpo.gov/fdsvs
118 North American Council for Freight Efficiency, 2015 Annual Fleet Fuel Study. 6 May 2015.
https://nacfe.org/downloads/nacfe-2015-annual-fleet-fuel-study/
119 Greenhouse Gas Emissions and Fuel Efficiency Standards for Medium- and Heavy-Duty
Engines and Vehicles. Federal Register Volume 76, No. 179 (September 15, 2011) Page 57211.
120 Greenhouse Gas Emissions and Fuel Efficiency Standards for Medium- and Heavy-Duty
Engines and Vehicles— Phase 2. Federal Register Volume 81, No. 206 (October 25, 2016)
Pages 73587, 73608-73611.
231
------- |