Population and Activity of On-road
Vehicles in MOVES2014
&EPA
United States
Environmental Protection
Agency
-------
Population and Activity of On-road
Vehicles in MOVES2014
Assessment and Standards Division
Office of Transportation and Air Quality
U.S. Environmental Protection Agency
NOTICE
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.
&EPA
United States
Environmental Protection
Agency
EPA-420-R-16-003a
March 2016
-------
Table of Contents
1 Introduction 4
2 MOVES Vehicle and Activity Classifications 8
2.1 HPMS Class 8
2.2 Source Use Types 8
2.3 Regulatory Classes 9
2.4 Fuel Types 10
2.5 Road Types 11
2.6 Source Classification Codes (SCC) 12
2.7 Model Year Groups 13
2.8 Source Bins 13
2.9 Allowable Vehicle Modeling Combinations 15
2.10 Default Inputs and Fleet and Activity Generators 16
3 Data Sources 18
3.1 VIUS 18
3.2 Polk NVPP® and TIP® 18
3.3 EPA Sample Vehicle Counts 18
3.4 FHWA Highway Statistics 19
3.5 FTA National Transit Database 19
3.6 School Bus Fleet Fact Book 19
3.7 MOBILE6 20
3.8 Annual Energy Outlook & National Energy Modeling System 20
3.9 Transportation Energy Data Book 20
3.10 FHWA Weigh-in-Motion 20
3.11 Motorcycle Industry Council Statistical Annual 20
4 VMT by Calendar Year and Vehicle Type 21
4.1 Historic Vehicle Miles Traveled (1990 and 1999-2011) 21
4.2 Projected Vehicle Miles Traveled (2012-2050) 22
5 Vehicle Populations by Calendar Year 25
5.1 Historic Source Type Populations (1990 and 1999-2011) 25
5.2 Projected Vehicle Populations (2012-2050) 31
6 Fleet Characteristics 33
6.1 Source Type Definitions 33
6.2 Sample Vehicle Population 36
7 Vehicle Characteristics that Vary by Age 50
7.1 Age Distributions 50
7.2 Relative Mileage Accumulation Rate 58
8 VMT Distribution of Source Type by Road Type 65
9 Average Speed Distributions 67
9.1 Light-Duty Average Speed Distributions 67
9.2 Heavy-Duty Average Speed Distributions 71
10 Driving Schedules and Ramps 74
10.1 Driving Schedules 74
10.2 Ramp Activity 79
11 Hotelling Activity 81
-------
11.1 National Default Hotelling Rate 81
11.2 Hotelling Activity Distribution 83
12 Temporal Distributions 85
12.1 VMT Distribution by Month of the Year 86
12.2 VMT Distribution by Type of Day 87
12.3 VMT Distribution by Hour of the Day 88
12.4 Engine Starts and Parking 89
12.5 Hourly Hotelling Activity 92
12.6 Single and Multiday Diurnals 95
13 Geographical Allocation of Activity 96
13.1 Source Hours Operating Allocation to Zones 96
13.2 Engine Start Allocations to Zones 97
13.3 Parking Hours Allocation to Zones 98
14 Vehicle Mass and Road Load Coefficients 99
14.1. Source Mass and Fixed Mass Factor 100
14.2. Road Load Coefficients 100
15 Air Conditioning Activity Inputs 104
15.1 ACPenetrationFraction 104
15.2 FunctioningACFraction 105
15.3 ACActivityTerms 106
16 Conclusion and Areas for Future Research 108
17 Appendix A: Projected Source Type Populations by Year 110
18 Appendix B: Fuel Type and Regulatory Class Fractions for 1960-1981 112
19 Appendix C: 1990 Age Distributions 115
19.1 Motorcycles 115
19.2 Passenger Cars 115
19.3 Trucks 115
19.4 Intercity Buses 116
19.5 School Buses and Motor Homes 116
19.6 Transit Buses 116
20 Appendix D: Detailed Derivation of Age Distributions 118
20.1 2012-2050 Age Distribution Projections 118
20.2 1999-2010 Age Distributions 120
21 Appendix E: SCC Mappings 123
21.1 SCC Mappings between MOVES2014 and MOVES201 Ob 123
21.2 2011 SCC VMT Conversions 124
22 Appendix F: Calculation of Combination Truck Average Speed Distributions 127
23 Appendix G: Driving Schedules 130
24 AppendixH: MOVES2010b Source Masses 133
24.1 Motorcycles 134
24.2 Passenger Cars 135
24.3 General Trucks 135
24.4 Buses 137
24.5 Refuse Trucks 139
24.6 Motor Homes 140
25 Peer Review of Draft Report 143
-------
25.1 Kanok Boriboonsomsin, PhD, PE 143
25.2 Randall Guensler, PhD 150
26 References 160
-------
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 emissions produced
by onroad 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 to
estimate 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. An inventory
can be pictured as a stool; the three legs of the stool are the emission rates, activity, and
populations, while the seat is the inventory. The emission rates are inputs to the model specified
for various "processes" including running exhaust, start exhaust, and a number of evaporative
processes, among others. These processes also define the activity, populations, and technology
inputs required.
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 emissions modes are affected by
vehicle operation and the time that vehicles are parked. Emission rates are further categorized by
source bins with similar fuel type, regulatory classification, and other vehicle characteristics.
This report describes the sources and derivation for onroad vehicle population and activity
information and associated adjustments as stored in the MOVES2014 and MOVES2014aa
default databases. This data has been extensively updated from MOVES2010b and previous
versions of MOVES. Emission measurement and rates, correction factor values, and information
for nonroad equipment in the default database are described in other MOVES technical reports.l
There have not been any major updates between this final report and the earlier released public
draft in July 2015.2 However, this final report does have some notable revisions from the draft,
namely some added or improved explanatory tables and figures, a new introductory subsection
on vehicle model year groups, movement of content to different sections and appendices for
better readability, clarifications to ambiguous descriptions, and a new appendix documenting
peer review comments along with EPA's responses to those comments.
The MOVES2014 default database has a domain that encompasses all onroad (highway) vehicle
and nonroad equipment activity and emissions for the entire United States, Puerto Rico, and the
a In this report, "MOVES2014" refers to both MOVES2014 and MOVES2014a unless specified otherwise.
4
-------
Virgin Islands. Properly characterizing emissions from the onroad vehicle subset requires a
detailed understanding of the cars and trucks that make up the vehicle fleet and their patterns of
operation. The national default activity information in MOVES2014 provides a reasonable basis
for estimating national emissions. The most important of these inputs, such as VMT and
population estimates, come from long-term systematic national measurements.
Given the availability of these national measurements when MOVES2014 was being developed,
2011 was chosen as the base year for future year projections. Like in previous versions of
MOVES, users may analyze emission inventories in 1990 to correspond with the last Clean Air
Act amendments as well as every year from 1999 to 2050.
In addition to uncertainties associated with projections, the uncertainties and variability in the
default data contribute to the uncertainty in the resulting emission estimates. 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 that provided in the MOVES
defaults. MOVES has been specifically designed to accommodate the input of alternate, user-
supplied activity data for the most important parameters. EPA's Technical Guidance3 provides
more information on customizing MOVES2014 with local inputs.
This report documents the sources and calculations used to produce the default population and
activity data in the MOVES2014 database for computing national-level emissions. In particular,
this report will describe the data used to fill the tables listed below in Table 1-1.
Population and activity data are ever changing. As part of the MOVES development process, the
model undergoes major updates and review every few years. As MOVES progresses, the
development of fleet and activity inputs including projections will continue to be an important
area of focus and improvement.
-------
Table 1-1 MOVES database elements covered in this report
Database Table Name
AvgSpeedDistribution
DayVMTFraction
DriveSchedule
DriveScheduleAssoc
Drive Schedule Second
FuelType
HotellingActivityDistribution
HotellingCalendarYear
HourVMTFraction
HPMSVtypeYear
ModelYearGroup
MonthGroupHour
MonthVMTFraction
PollutantProcessModelYear
RegulatoryClass
RoadOpModeDistribution
RoadType
RoadTypeDistribution
Sample VehicleDay
Sample VehiclePopulation
Sample VehicleTrip
sec
SourceBinDistribution
Content Summary
Distribution of time among average speed bins
Distribution of VMT between weekdays and
weekend days
Average speed of each drive schedule
Mapping of which drive schedules are used for
each combination of source type and road type
Speed for each second of each drive schedule
Broad fuel categories that indicate the fuel
vehicles are capable of using
Distribution of hotelling activity to the various
operating modes
Rate of hotelling hours per rural restricted access
VMT
Distribution of VMT among hours of the day
Annual VMT by HPMS vehicle types
A list of years and groups of years corresponding
to vehicles with similar emissions performance
Coefficients to calculate air conditioning demand
as a function of heat index
Distribution of annual VMT among months
Assigns model years to appropriate groupings,
which vary by pollutant and process
Sorts vehicles into weight-rating based groups in
which emission regulations are applied
Operating mode distributions by source type, road
type, and speed bin
Distinguishes roadways by population density of
geographic area and by type of access, particularly
the use of ramps for entrance and exit
Distribution of VMT among road types
Identifies vehicles in the Sample VehicleTrip table
Fuel type and regulatory class distributions by
source type and model year.
Trip start and end times used to determine vehicle
start and soak times
Source Classification Codes that identify the
vehicle type, fuel type, road type and emission
process in MOVES output
Distribution of population among different vehicle
sub-types (source bins)
Report Sections
Section 9
Section 12
Section 10
Section 10
Section 10
Section 2
Section 1 1
Section 1 1
Section 12
Section 4
Section 2
Section 15
Section 12
Section 4
Section 2
Section 10
Section 2
Section 8
Section 12
Section 4
Section 12
Section 2
Section 4
-------
Table 1-1 MOVES database elements covered in this report
Database Table Name
SourceTypeAge
SourceTypeAgeDistribution
SourceTypeHour
SourceTypeModelYear
SourceTypePolProcess
SourceTypeYear
SourceUseType
SourceUseTypePhysics
Zone
ZoneRoadType
Content Summary
Rate of survival to subsequent age, relative
mileage accumulation rates, and fraction of
functional air conditioning equipment
Distribution of vehicle population among ages
The distribution of total daily hotelling among
hours of the day
Prevalence of air conditioning equipment
Indicates which source bin discriminators are
relevant for each source type and pollutant/process
Source type vehicle counts by year
Mapping from HPMS class to source type,
including source type names
Road load coefficients and vehicle masses for each
source type used to calculate vehicle specific
power (VSP) and scaled tractive power (STP)
Allocation of activity to zone (county)
Allocation of driving time to zone (county) and
road type
Report Sections
Section 7
Section 15
Section 7
Section 12
Section 15
Section 4
Section 5
Section 2
Section 14
Section 12
Section 13
7
-------
2 MOVES Vehicle and Activity Classifications
In developing MOVES, we needed to pull together information on vehicle activity and
emissions. 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
this to existing information on 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 the rates 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. 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).
Note that in MOVES2014, what we call the HPMS class for light-duty vehicles (25) denotes the
sum of the VM-1 values for long wheelbase and short wheelbase light-duty vehicles.
HPMSVTypelD 25 is new for MOVES2014 and replaces HPMSVTypelD 20 (passenger cars)
and 30 (other two-axle four-tire vehicles) in MOVES2010. As such, in MOVES2014 any VMT
input by HPMS class for passenger cars and light-duty trucks must be entered as a combined
value in the new HPMSVTypelD 25. This change in HPMS classes has come about as passenger
vehicles have evolved over time with the physical characteristics of "cars" and "trucks"
becoming less distinct. In response, US DOT changed the organization of HPMS classifications
and MOVES has evolved to reflect this change.
2.2 Source Use Types
The primary vehicle classification in MOVES is source use type, or, more simply, source type.
Source types are intended to be groups of vehicles with similar activity and usage patterns.
HPMS vehicle classes were differentiated into MOVES onroad source types.
Source types cannot be fully determined using field observations and must be paired with
additional information about the vehicle's activity to determine whether it typically travels short-
er long-haul routes (greater than 200 miles per day), whether it has specific travel routines such
as a refuse truck, and whether it is a commercial or personal vehicle. Estimates for short-
haul/long-haul and commercial/personal distributions relied on information from the federal
8
-------
Vehicle Inventory and Use Survey (VIUS). The MOVES2014 source types are listed in Table
2-1 along with the associated HPMS classes. More detailed source type definitions are provided
in Section 6.1.
Table 2-1 MOVES2014 onroad source types
sourceTypelD
11
21
31
32
41
42
43
51
52
53
54
61
62
Source Type Name
Motorcycles
Passenger Cars
Passenger Trucks (primarily personal use)
Light Commercial Trucks (primarily non-
personal use)
Intercity Buses (non-school, non-transit)
Transit Buses
School Buses
Refuse Trucks
Single Unit Short-Haul Trucks
Single Unit Long-Haul Trucks
Motor Homes
Combination Short-Haul Trucks
Combination Long-Haul Trucks
HPMSVTypelD
10
25
25
25
40
40
40
50
50
50
50
60
60
Description
Motorcycles
Light-Duty Vehicles
Light-Duty Vehicles
Light-Duty Vehicles
Buses
Buses
Buses
Single Unit Trucks
Single Unit Trucks
Single Unit Trucks
Single Unit Trucks
Combination Trucks
Combination Trucks
In MOVES, the distinction between light-duty (LD) and heavy-duty (HD) source types is
essential because light- and heavy-duty operating modes are assigned by source type and their
calculation differs for light- and heavy-duty vehicles. Light-duty vehicles (sourceTypelD 11,21,
31, and 32) use vehicle specific power (VSP) operating modes, which are dependent on the
measured mass of the test vehicle. Heavy-duty vehicles (sourceTypelD 41, 42, 43, 51, 52, 53, 54,
61, and 62) use scaled tractive power (STP) operating modes which are scaled by a fixed mass
factor since their emission rates correlates better with absolute vehicle power than vehicle
specific power. For more discussion on VSP and STP definitions, please refer to Section 14 of
this report and the MOVES2014 reports on light-duty and heavy-duty vehicle emission rate
development, respectively.4'5
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.
Thus, we must map the two schemas.
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 have an independent regulatory class.
Table 2-2 Regulatory classes in MOVES2014
regClassID
0
10
20
30
40
41
42
46
47
48
Regulatory Class Name
Doesn't Matter
MC
LDV
LOT
LHD<=10k
LHD<=14k
LHD45
MHD
HHD
Urban Bus
Description
Doesn't Matter
Motorcycles
Light-Duty Vehicles
Light-Duty Trucks
Class 2b Trucks with 2 Axles and 4 Tires (8,500 Ibs <
GVWR <= 10,000 Ibs)
Class 2b Trucks with 2 Axles and at least 6 Tires or Class 3
Trucks (8,500 Ibs < GVWR <= 14,000 Ibs)
Class 4 and 5 Trucks (14,00 Ibs < GVWR <= 19,500 Ibs)
Class 6 and 7 Trucks (19,500 Ibs < GVWR < =33,000 Ibs)
Class 8a and 8b Trucks (GVWR > 33,000 Ibs)
Urban Bus (see CFR Sec. 86.091 2)
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 Ibs. GVWR are Class 2a; in MOVES
they are considered light-duty trucks in regulatory class 30. Vehicles of 8,500-10,000 Ibs.
GVWR are Class 2b, and considered light heavy-duty vehicles (LHD) in regulatory classes 40 or
41.
In MOVES2014, we introduced a new regulatory class 40 for vehicles that are classified as light-
duty by FFIWA (because they have only two axles and four tires), and are thus mapped to source
type 30 (passenger trucks) or 31 (light-commercial trucks) in MOVES, but have a GVWR that
puts them in Class 2b, so are subject to heavy-duty emission standards. As described above,
these regulatory class 40 vehicles use light-duty (VSP-based) operating modes because they are
light-duty source types, but the new regulatory class maps them to emission rates that are more
consistent in how these vehicles are regulated. Meanwhile, Class 2b trucks with two axles and at
least six tires (colloquially known as "dualies") and Class 3 trucks are considered single unit
trucks by DOT, and therefore fall into regulatory class 41 and are modeled as the heavy-duty
source types using STP-based operating modes. In summary, the light-duty truck source types
(31 and 32) map only to regulatory classes 30 and 40 in MOVES2014, while the heavy-duty
vehicle source types (41 and above) map to regulatory classes 41 and above. Section 6.2
provides more information on the distribution of vehicles among regulatory classes.
2.4 Fuel Types
MOVES also classifies vehicles by the fuel they are designed to use. MOVES2014 models
vehicles and equipment powered by following fuel types: gasoline, diesel, E-85 (a nominal blend
of 85 percent ethanol and 15 percent gasoline), compressed natural gas (CNG), electricity, and
liquefied petroleum gas (LPG, only available for nonroad equipment). Note that in some cases, a
single vehicle can use more than one fuel; for example, flexible fuel vehicles are capable of
running on either gasoline or E-85. In MOVES, fuel type refers to the capability of the vehicle
10
-------
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
MOVES2014 technical report on the fuel supply.6 Table 2-3 below summarizes the fuel types
available in MOVES.
Table
2-3 A list of allowable fuel types to power vehicles/equipment in MOVES2014
fuelTypelD
1
2
3
4
5
9
defaultFormulationID
10
20
30
40
50
90
Description
Gasoline
Diesel Fuel
Compressed Natural Gas (CNG)
Liquefied Petroleum Gas (LPG)*
Ethanol (E-85) Capable
Electricity
* MOVES2014 models LPG use only in nonroad equipment.
It is important to note that not all fuel type/source type combinations can be modeled in
MOVES. That is, MOVES2014 will not model gasoline fueled long-haul combination trucks,
gasoline intercity buses, or diesel motorcycles. Though other natural gas vehicles such as CNG
refuse trucks can found in the US today, CNG transit buses are the most prevalent and well-
tested, and thus are currently the only onroad source type that may be modeled using CNG.
Similarly, flexible fuel (E85-compatible) and electric vehicles are only modeled for passenger
cars, passenger trucks, and light commercial trucks. None of the onroad (highway) source types
can be modeled as fueled by LPG. For more information on how MOVES models the impact of
fuels on emissions, please see the MOVES documentation on fuel effects.7
2.5 Road Types
MOVES calculates emissions separately for each of four road types and for "off-network"
activity when the vehicle is not moving. It also allows separate output for ramp and non-ramp, as
described in Section 10.2 below. The road type codes used in MOVES are listed in Table 2-4.
The four MOVES road types (2-5) are aggregations of FHWA functional facility types.
11
-------
Table 2-4 Road type codes in MOVES2014
roadTypelD
1
2
3
4
5
6
7
8
9
100
Description
Off Network
Rural Restricted Access
Rural Unrestricted Access
Urban Restricted Access
Urban Unrestricted Access
Rural Restricted without Ramps
Urban Restricted without Ramps
Rural Restricted only Ramps
Urban Restricted only Ramps
Nonroad
FHWA Functional Types
Off Network
Rural Interstate
Rural Principal Arterial, Minor Arterial, Major
Collector, Minor Collector & Local
Urban Interstate & Urban Freeway/Expressway
Urban Principal Arterial, Minor Arterial, Collector
& Local
The MOVES road types are based on two important distinctions in how FWHA classifies roads:
1) urban versus rural roadways are distinguished based on 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 choose to create their own additional road types, or 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 single numerical codes that identify the vehicle type, fuel type, road type,
and emission process. The SCCs were redesigned for MOVES2014 to directly relate to the
source use types and road types used by MOVES.
The new SCCs retain the previous 10-digit design, but use different numerical combinations to
avoid conflicts with existing codes. The new codes for onroad vehicles use MOVES numerical
identification (ID) codes 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.
12
-------
Building the SCC values in this way will allow 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 tables in the MOVES User
Guide3 or appropriate sections in this document for the descriptions of the sourceTypelD,
fuelTypelD and roadTypelD values currently used by MOVES. A description of mapping
between older SCCs in MOVES2010b and newer SCCs in MOVES2014 can be found in Section
21 (Appendix E: SCC Mappings). Emission processes are discussed in other MOVES reports on
emission rate development4'5 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 MOVES2014a, model year refers to the year in
which the vehicle was produced, built, and certified as compliant with emission standards.
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 2011 vehicle belongs to the "2011" model year group for estimating hydrocarbon running
exhaust emissions, but belongs to the "2011-2020" 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
MOVES2014 emission rate reports.4'5
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:
IFFEERRMMOOOOOOOOOO, where
• lisa placeholder,
• FF is a MOVES fuelTypelD,
-------
• EE is a MOVES engTechID,b
• RRisa. 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 and engine technologies and regulatory class are stored by model
year in the SampleVehiclePopulation table. The MOVES Source Bin Distribution Generator
combines information from these two tables (see Table 2-5) to create a detailed
SourceBinDistribution. These bins may vary by pollutant and process as indicated in the
SourceTypePolProcess table. In general, fuel type and model year group are relevant for all
emission calculations, but the relevance of regulatory class and model year group depend on the
pollutant and process being modeled. If desired, MOVES2014 can produce results by various
vehicle classifications—source type, SCC, or regulatory class—and by fuel type and model year.
Table 2-5 Data tables used to allocate source tyi
Generator Table Name
SourceTypePolProcess
PollutantProcessModelYear
SampleVehiclePopulation
Key Fields*
sourceTypelD
polProcessID
polProcessID
modelYearlD
sourceTypelD
modelYearlD
fuelTypelD
engTechID
regClassID
Additional Fields
isRegClassReqd
isMYGroupReqd
modelYearGroupID
stmyFuelEngFraction
stmyFraction
)e to source bin
Notes
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)
Assigns model years to appropriate
model year groups for each
polProcessID.
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. electric cars). This table
provides default fractions for the
Alternative Vehicle Fuel & Technology
(AFVT) importer.
* In these tables, the sourceTypelD and modelYearlD are combined into a single sourceTypeModelYearlD.
While details of the SourceTypePolProcess and PollutantProcessModelYear tables are discussed
in the reports on the development of the light- and heavy-duty emission rates4'5, the
SampleVehiclePopulation (SVP) table is a topic for this report and is discussed in Section 6.2.
b In MOVES2014, engTechID 1 is used for all fuel types except electric vehicles, where engTechID 30 is used
instead. Thus, in this version, engTechID is somewhat redundant with fuel type and adds no new information when
determining source bin distributions or calculating emissions.
14
-------
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 emissions testing or have negligible market
share cannot be directly modeled in MOVES2014.
Table 2-6 summarizes the allowable source type-fuel type combinations. Most of the gasoline
and diesel combinations exist with a few notable 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 in MOVES2014. Table 2-8 combines the information in the
two preceding tables. Each source type-fuel type combination contains all regulatory classes
listed, except for gasoline transit buses, which have been called out separately. Additional
discussion about decisions to include and exclude certain types of vehicles can be found in
Section 6.
Table 2-6 Matrix of the allowable source type-fuel type combinations in MOVES2014
(Allowable combinations are marked with an X)
Fuel Types
Gasoline
Diesel
CNG
ESS-Capable
Electricity
1
2
3
5
9
Source Use Types
Motorcycles
11
X
Passenger Cars
21
X
X
X
X
^rt
assenger Trucks
31
X
X
X
X
L~^
Lght Commercia
Trucks
32
X
X
X
X
Intercity Buses
41
X
Transit Buses
42
X
X
X
in
0
I
Cd
cn
43
X
X
Refuse Trucks
51
X
X
in
o
ffi
i-i £
" Oq
ST
B.
52
X
X
r
o
B
Oq
ffi
fiT
B.
53
X
X
£
o
&
X
o
rt>
VI
54
X
X
O
0
0* C^l
^'3"
§&
r-
o
en
61
X
X
o
0
^
R §
5' in
B HH
fa
r-
o
en
62
X
15
-------
Table 2-7 Matrix of the allowable source type-regulatory class combinations in MOVES2014
(Allowable combinations are marked with an X)
Regulatory Classes
MC
LDV
LOT
LHD<=10k
LHD<=14k
LHD45
MHD67
HHD8
Urban Bus
10
20
30
40
41
42
46
47
48
Source Use Types
Motorcycles
11
X
Passenger Cars
21
X
Passenger Trucks
31
X
X
Light Commercial
Trucks
32
X
X
Intercity Buses
41
X
X
X
X
Transit Buses
42
X
X
X
X
Gfl
0
g*
o_
Cd
1
cn
43
X
X
X
X
Refuse Trucks
51
X
X
X
X
Gfl
d ^
£•!
g"!
%%
<£,
oT
52
X
X
X
X
r
0
d ^
B °?
& m
g"!
B t/2
a 5'
<£,
oT
53
X
X
X
X
Motor Homes
54
X
X
X
X
Short-Haul
Combination Trucks
61
X
X
Long-Haul
Combination Trucks
62
X
X
Table 2-8 A summary of source type, fuel type, and regulatory class combinations in MOVES2014
sourceTypelD
11
21
31
32
41
/LO
43
51
52
53
54
61
62
fuelTypelD
1
1,2,5,9
1,2,5,9
1,2,5,9
2
1
2,3
,2
,2
,2
,2
,2
,2
2
regClassID
10
20
30,40
30,40
41, 42, 46, 47
42, 46, 47
48
41, 42, 46, 47
41, 42, 46, 47
41, 42, 46, 47
41, 42, 46, 47
41, 42, 46, 47
46,47
46,47
2.10 Default Inputs and Fleet and 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
16
-------
format. MOVES uses "generators" to create fine-scale information from user inputs and MOVES
defaults.
The MOVES Total Activity Generator (TAG) 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 MOVES2014 contains national
estimates for VMT, vehicle population, and vehicle age distributions for every possible analysis
year (1990 and 1999-2050). For national inventory runs, annual national activity is distributed
temporally and spatially using allocation factors.
The Source Bin Distribution Generator (SBDG) uses information on fuel type fractions,
regulatory class distributions, and similar information to estimate the number of vehicles
belonging to each source bin as a function of source type and model year. The SBDG maps the
activity data (by source types) to source bins which map directly to the MOVES emission rates.
There are a number of MOVES modules that generate operating mode distributions based on
vehicle activity inputs. The Rates Operating Mode Distribution Generator and the Link
Operating Mode Distribution Generator use information on speed distributions and driving
patterns (driving schedules) to develop operating mode fractions for each source type, road type,
and time of day. Similarly, the Evaporative Emissions Operating Mode Generator and the Start
Operating Mode Distribution Generator use MOVES inputs to develop operating mode
distributions for starts and vapor venting. The details of each these generators and other
MOVES2014 algorithms are described in the MOVES2014 Module Reference.8
-------
3 Data Sources
A number of organizations collect data relevant to this report. The most important sources used
to populate the national default vehicle population and activity portions of the MOVES database
are described here. These sources are referred to throughout this document by the abbreviated
name given in this description, but the reference citation is only given here.
The MOVES Technical Guidance3 provides detailed information on recommended data sources
for users developing their own inputs.
3.1 VIUS
Until 2002 the US Census Bureau conducted the Vehicle Inventory and Use Survey (VIUS)9 to
collect data on the physical characteristics and activity of US trucks every five years. The survey
is a sample of private and commercial trucks that were registered in the United States as of July
of the survey year. The survey excludes automobiles, motorcycles, government-owned vehicles,
ambulances, buses, motor homes, and nonroad equipment.
For MOVES, VIUS provides information to characterize trucks by source type and to estimate
age, fuel type, and regulatory class distributions as well as relative mileage accumulation rates.
MOVES2014 uses data from both the 1997 and 200210 surveys. While the survey includes a
large number of vehicles and was designed to be representative of the US fleet, information on
model year is not available for many of the older trucks. Thus, the distribution data for many
older model years is sparse and sometimes erratic. Note that the Census Bureau discontinued
VIUS in 2002, although there has been discussion recently about reinitiating the survey.
3.2 Polk NVPP® and TIP®
Acquired by UTS in July 2013, R.L. Polk & Co. was a private company providing automotive
information services. The company maintained two databases relevant for MOVES: the National
Vehicle Population Profile (NVPP®)11 and the Trucking Industry Profile (TIP®Net) Vehicles in
Operation12 database. The first focused on light-duty cars and trucks, the second focused on
medium and heavy-duty trucks. Both compiled data from state vehicle registration lists. For
MOVES2014, EPA used NVPP® and TIP® datasets purchased for 1999 and 201 l._Polk/fflS
data was used in determining vehicles populations by age, fuel type, and regulatory class. At the
time of these EPA data purchases Polk was independently operated, so we will continue to refer
to these datasets under the Polk name in this report.
3.3 EPA Sample Vehicle Counts
Neither VIUS nor the Polk dataset contained enough information separately to develop
distributions by regulatory class, fuel type, and age for each vehicle source type in MOVES, so
EPA combined these datasets, and incorporated additional data sources to cover vehicles types,
such as motorcycles, buses, and motor homes that were excluded from either the VIUS or Polk
datasets. The resulting sample vehicle counts dataset is the basis for the MOVES2014
SampleVehiclePopulation table and the 2011 age distributions. More details on how we
constructed the Sample Vehicle Counts dataset can be found in Section 6.2.
18
-------
3.4 FHWA Highway Statistics
Each year the US DOT Federal Highway Administration's (FHWA) Office of Highway Policy
Information publishes Highway Statistics. This volume summarizes a vast amount of roadway
and vehicle data from the Highway Performance Monitoring System, a national information
system that collects data from states and other sources on many facets of the US roadway
system.
In MOVES2014, vehicle miles traveled (VMT) and vehicle population data for the historic years
1990 and 1999-2011 come from four tables in Highway Statistics: MV-113, MV-1014, VM-115,
and VM-216, which we will reference by table name. For some years, the VMT values were
revised by FHWA in subsequent publications. Table 3-1 summarizes the data source and revision
date we used for each historical year.
able 3-1 Corresponding Highway Statistics data source for historical yeai
Year
1990
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
FHWA Publication Source (Publication/Revision Date)
Highway Statistics 1991
Highway Statistics 1999
Highway Statistics 2000
Highway Statistics 2001
Highway Statistics 2002
Highway Statistics 2003
Highway Statistics 2004
Highway Statistics 2005
Highway Statistics 2006
Highway Statistics 2007
Highway Statistics 2008
Highway Statistics 2010
Highway Statistics 2010
Highway Statistics 2011
(October 1992)
(October 2000)
(April 2011)
(April 2011)
(April 2011)
(April 2011)
(April 2011)
(April 2011)
(April 2011)
(April 2011)
(April 2011)
(December 20 12)
(December 20 12)
(March 2013)
3.5 FTA National Transit Database
The US DOT, Federal Transit Administration (FTA) summarizes financial and operating data
from mass transit agencies across the country in the National Transit Database (NTD).1? For
MOVES2014, we used 1999-2011 vehicle counts from the NTD Revenue Vehicle Inventory for
motor buses (MB) to determine fuel type distributions and populations.
3.6 School Bus Fleet Fact Book
The School Bus Fleet Fact Book includes estimates, by state, of the number of school buses and
total miles traveled.18 The Fact Book is published by Bobit Publications. School bus mileage
accumulation rates came from the 1997 Fact Book, originally used in MOBILE6. We have used
1999-2011 sales data from the 2009 and 2012 Fact Book to calculate age distributions.
-------
3.7 MOBILE6
MOBILE6 was a precursor to MOVES used to estimate highway vehicle emissions. In some
cases, we have used estimates from MOBILE6 model with only minor adaptation. In particular,
we used MOBILE6 data for some relative mileage accumulation rates, air conditioning usage
rates, and driving schedules. The MOBILE6 data is documented in technical reports, particularly
M6.FLT.002, Update of Fleet Characterization Data for Use in MOBILE6 - Final Report19
Additional MOBILE6 documentation is available online.20
3.8 Annual Energy Outlook & National Energy Modeling System
The Annual Energy Outlook (AEO)21 describes Department of Energy forecasts for future energy
consumption. The National Energy Modeling System (NEMS) is used to generate these
projections based on economic and demographic forecasts. Vehicle sales and miles traveled are
included in the projections because they strongly influence fuel consumption. Therefore, the
AEO is an important source of future projections in MOVES. For MOVES2014, we used
AEO2014 to forecast VMT and vehicle populations in years 2012-2050.
3.9 Transportation Energy Data Book
Each year Oak Ridge National Laboratory produces the annual Transportation Energy Data Book
(TEDB) for the Department of Energy. This book summarizes transportation and energy data
from a variety of sources, including EPA, FHWA, Polk, and Ward's Automotive, Inc. For
MOVES2014 we used information for estimating vehicle sales and survival fractions for historic
years 1990 and 1999-2011 from TEDB Edition 32, published in 2013.22
3.10 FHWA Weigh-in-Motion
FHWA compiles truck weight data by axle configuration and roadway type from individual
states' Weigh-in-Motion (WEVI) programs.23 The average weight for single unit trucks and
combination trucks was determined from FHWA's Vehicle Travel Information System (VTRIS)
W-3 Tables using data collected in 2011.
3.11 Motorcycle Industry Council Statistical Annual
The Motorcycle Industry Council (MIC) collects data on sales, ownership, and activity trends
each year. MIC's Statistical Annual summarizes this data,24 which we used in MOVES2014,
particularly the 1999-2011 sales of highway motorcycles.
20
-------
4 VMT by Calendar Year and Vehicle Type
For national level calculations, MOVES calculates source operating hours from national VMT
by vehicle type. The default database contains national VMT estimatesfor all analysis years,
which include 1990 and 1999-2050. Years 1991-1998 are excluded because there is no
regulatory requirement to analyze them and including them would increase model complexity.
Calendar year 1990 continues to be a base year because of the Clean Air Act Amendments of
1990.
The national VMT estimates are stored in the HPMSVTypeYear table0, 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 MOVES2014 national defaults, the
baseYearOffNetVMT is zero for all vehicle types. Additionally, the VMTGrowthFactor field is
not used in MOVES2014 and is set to zero for all vehicle types.
4.1 Historic Vehicle Miles Traveled (1990 and 1999-2011)
The HPMSBaseYearVMT field stores the total national VMT for each HPMS vehicle type for all
analysis years. For historical years 1990 and 1999-2011, the VMT is derived from the FHWA
VM-1 tables. In reporting years 2007 and later, the VM-1 data use an updated methodology25,
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
Generator8 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
0 In MOVES2014a, 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 MOVES2014, the default table, HPMSVTypeYear, continues to use annual miles by HPMS class in
MOVES2014a. Considering the default table has not changed in MOVES2014a, any discussion in this report on
annual VMT estimates will be in the context of annual miles traveled by HPMS class.
21
-------
reliable to retain the original 2000-2006 estimates because the information available for those
years does not fully meet the requirements of the new methodology."d 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 unknown, 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 4-1. The VMT
for 1990 and 1999 were EPA-adjusted from VM-1, 2000-2006 were FHWA-adjusted, and 2007-
2011 were unadjusted, other than the simple combination of the short and long wheelbase classes
into light-duty vehicles.
Table 4-1 Historic year VMT by HPMS vehicle class (millions of miles)
Year
1990
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
Motorcycles
11,404
13,619
12,175
11,120
11,171
11,384
14,975
13,773
19,157
21,396
20,811
20,822
18,513
18,500
Light-Duty
Vehicles
1,943,197
2,401,408
2,458,221
2,499,069
2,555,467
2,579,194
2,652,092
2,677,641
2,680,535
2,691,034
2,630,213
2,633,248
2,648,457
2,646,641
Buses
10,279
14,853
14,805
12,982
13,336
13,381
13,523
13,153
14,038
14,516
14,823
14,387
13,770
13,783
Single Unit
Trucks
70,848
100,534
100,486
103,470
107,317
112,723
111,238
109,735
123,318
119,979
126,855
120,207
110,738
103,515
Combination
Trucks
108,624
160,921
161,238
168,969
168,217
173,539
172,960
175,128
177,321
184,199
183,826
168,100
175,789
163,692
4.2 Projected Vehicle Miles Traveled (2012-2050)
The previous section describes historic fleet VMT. This section explains how EPA projected
those values into the future. The VMT growth in years beyond 2011 is based on the VMT
projections as described in AEO2014. Due to differences in methodology, the absolute VMT
values presented in AEO differ slightly from the HPMS values in VM-1 where the analysis years
overlap. Therefore, the projections in AEO were not used directly. Instead, percent changes from
d This text appears in a footnote to FHWA's Highway Statistics Table VM-1 for publication years 2000-2009.
22
-------
year to year in the projected values were calculated and applied to the HPMS data. Since
AEO2014 only projects out to 2040, VMT for years 2041-2050 were assumed to continue to
grow at the average growth rate over 2031-2040.
A mapping between the two data sources was necessary because the vehicle categories differed
between AEO and HPMS. AEO's light-duty percent growth 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 growth rates was mapped to the HPMS bus category. AEO's light medium and medium
heavy-duty growth rates were combined for mapping to the HPMS single unit truck category,
and AEO's heavy heavy-duty growth 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.
The percent growth over time was calculated for each of the groups described above and applied
by HPMS category to the 2011 base year VMT from the Table VM-1. The resulting values are
presented in Table 4-2 below.
-------
Table 4-2 VMT projections for 2012-2050 by HPMS class (millions of miles)
Year
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
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
2050
Motorcycles
18,776
19,030
19,073
19,162
19,375
19,590
19,756
19,931
20,107
20,284
20,454
20,627
20,807
20,997
21,205
21,426
21,662
21,897
22,133
22,378
22,625
22,867
23,086
23,293
23,493
23,687
23,880
24,060
24,217
24,436
24,657
24,880
25,105
25,332
25,561
25,792
26,025
26,261
26,498
Light-Duty
Vehicles
2,686,152
2,722,469
2,728,546
2,741,392
2,771,828
2,802,578
2,826,337
2,851,349
2,876,481
2,901,914
2,926,116
2,950,908
2,976,667
3,003,914
3,033,572
3,065,195
3,099,033
3,132,690
3,166,361
3,201,376
3,236,805
3,271,436
3,302,691
3,332,329
3,360,885
3,388,760
3,416,287
3,442,035
3,464,551
3,495,877
3,527,485
3,559,380
3,591,563
3,624,036
3,656,804
3,689,868
3,723,230
3,756,894
3,790,863
Buses
13,384
13,954
14,374
14,991
15,612
16,036
16,325
16,609
16,906
17,222
17,550
17,877
18,173
18,495
18,799
19,052
19,277
19,509
19,765
20,005
20,198
20,429
20,725
21,017
21,308
21,600
21,887
22,146
22,417
22,701
22,989
23,280
23,575
23,874
24,176
24,483
24,793
25,107
25,425
Single Unit
Trucks
103,284
108,811
113,054
118,343
123,348
126,693
128,737
130,692
132,833
135,237
137,759
140,171
142,243
144,418
146,389
147,999
149,382
150,824
152,391
153,916
155,034
156,435
158,246
159,910
161,452
162,945
164,353
165,603
166,905
168,431
169,970
171,524
173,091
174,673
176,270
177,881
179,507
181,147
182,803
Combination
Trucks
157,396
163,467
167,837
174,804
181,988
186,928
190,433
193,905
197,484
201,214
205,076
208,983
212,579
216,551
220,329
223,510
226,348
229,268
232,509
235,518
237,990
240,929
244,678
248,437
252,265
256,123
259,948
263,426
267,050
270,775
274,552
278,381
282,264
286,201
290,193
294,241
298,345
302,507
306,726
24
-------
5 Vehicle Populations by Calendar Year
MOVES uses vehicle populations to characterize emissions activity that is not directly dependent
on VMT. These data are also used to allocate VMT from HPMS class to source type and age.
(For more details, see Section 7) The default database stores historic estimates and future
projections of total US vehicle populations in 1990 and 1999-2050 by source type. All of these
values have been updated in MOVES2014 with improved data sources. 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 MOVES2014.
5.1 Historic Source Type Populations (1990 and 1999-2011)
MOVES populations for calendar years 1990 and 1999-2011 are derived top-down from
registration data in Table MV-1 of the Federal Highway Administration's annual Highway
Statistics report. In this table, vehicles are separated into four general vehicle categories:
motorcycles, passenger cars, trucks, and buses. These categories include government vehicles
and vehicles in Puerto Rico but do not account for vehicles in the Virgin Islands due to their
relatively small effects on national population estimates. Motorcycle and car data were used
without adjustment, but since MOVES populations are input by source type, allocations within
the general categories of trucks and buses were necessary, as shown in Figure 5-1.
Figure 5-1 Conceptual map of allocating FHWA MV-1 vehicle registration estimates to MOVES source types
FHWA (MV-1)
MOVES Source Types
Registrations
including
government
vehicles and
vehicles in
Puerto Rico
Source type
distributions
derived from
interpolation
between CY
1999 and 2011
-------
Trucks were separated into single unit and combination trucks using registration data in the
Highway Statistics Table VM-1. The remaining MV-1 truck registrations were allocated to the
light-duty trucks. Single unit and combination trucks were further allocated among their
respective source types using the EPA sample vehicle counts data (see Section 6.2.2). Since we
only had sample vehicle counts for calendar years 1999 and 2011, the 2000-2010 source type
allocations within the general truck categories were linearly interpolated between 1999 and 2011
rather than using the predictions for these years as in MOVES2010b. For example, we fit a linear
regression of the fraction of long-haul combination trucks out of total combination trucks
between 1999 and 2011 and then fit another regression for the short-haul combination truck
fraction. Regressions were fit in a similar fashion to allocate source type populations among
light-duty and single unit trucks. For reference, the interpolated fractions for MOVES2014 that
distribute the populations of light-duty, single unit, and combination trucks to the MOVES
source types by calendar year are shown below in Table 5-1.
Table 5-1 MOVES2014 linearly interpolated fractions to allocate truck populations to source types, such as
refuse trucks (sourceTypelD 51) among all single unit trucks (50s), by calendar year*
Year 31/30s
32/30s
51/SOs 52/50s
53/50s
54/50s 61/60s
62/60s
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
0.7496
0.7541
0.7586
0.7631
0.7676
0.7721
0.7767
0.7812
0.7857
0.7902
0.7947
0.7992
0.8037
0.2504
0.2459
0.2414
0.2369
0.2324
0.2279
0.2233
0.2188
0.2143
0.2098
0.2053
0.2008
0.1963
0.0155
0.0161
0.0166
0.0172
0.0178
0.0184
0.0190
0.0196
0.0201
0.0207
0.0213
0.0219
0.0225
0
0
0
0
0
0
0
0
0
0
0
0
0
7807
7786
7765
7745
7724
7703
7682
7662
7641
7620
7600
7579
7558
0
0
0
0
0
0
0
0
0
0
0
0
0
0462
0450
0438
0426
0414
0402
0390
0378
0366
0354
0341
0329
0317
0.1577
0.1604
0.1631
0.1657
0.1684
0.1711
0.1738
0.1765
0.1792
0.1819
0.1846
0.1873
0.1900
0.5744
0.5673
0
0
5601
5529
0.5457
0.5386
0
0
5314
5242
0.5171
0.5099
0
0
5027
4955
0.4884
0.4256
0.4327
0.4399
0.4471
0.4543
0.4614
0.4686
0.4758
0.4829
0.4901
0.4973
0.5045
0.5116
*Some fractions shown in this table may not sum exactly to one due to rounding. Fractions used to
calculate source type populations had more significant digits than shown and sum precisely to one.
These interpolated fractions were then multiplied by the FHWA populations of light-duty, single
unit, and combination trucks, respectively. This ensured that every source type population would
more or less track its Highway Statistics population, as shown for combination trucks in Figure
5-2, for single unit trucks in Figure 5-3, and light-duty trucks in Figure 5-4. Car and motorcycle
populations are reported directly in the Table MV-1 and thus were not subject to linear
interpolation adjustments. Note that 1990 source type fractions were not interpolated and were
instead retained from MOVES2010b.
26
-------
Figure 5-2 Combination truck source type populations interpolated for 1999-2011
3,000,000
2,500,000
a
o
i 2,000,000
I
T- 1,500,000
D Long-Haul
Combination
Trucks (62)
D Short-Haul
Combination
Trucks (61)
1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011
Year
Figure 5-3 Single unit truck source type populations interpolated for 1999-2011
9,000,000
8,000,000
a 7,000,000
b 6,000,000
5« ' '
'So
& 5,000,000
—
4,000,000
3,000,000
.0
| 2,000,000
1,000,000
D Motor Homes
(54)
• Long-Haul
Single Unit
Trucks (53)
D Short-Haul
Single Unit
Trucks (52)
D Refuse Trucks
(51)
1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011
Year
-------
Figure 5-4 Light-duty vehicle source type populations; light trucks interpolated for 1999—2011
300,000,000 n
250,000,000 -
a
o
i 200,000,000
I
150,000,000
I
13
.o
-*j
sS
Z
100,000,000
50,000,000
900,000
800,000
| 700,000
-^-i
| 600,000
'&
« 500,000
—
I 400,000
| 300,000
.o
| 200,000
100,000 -
D Light
Commercial
Trucks (32)
D Passenger
Trucks (31)
D Passenger
Cars (21)
D Motorcycles
(11)
1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011
Year
Figure 5-5 Bus source type populations in MOVES2014
D School Buses
(43)
D Transit Buses
(42)
D Intercity
Buses (41)
1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011
Year
Buses were allocated in a similar fashion as trucks, but using different data sources (see Figure
5-5). School bus estimates for all years 1999-2011 were taken from the Highway Statistics Table
MV-10 and transit bus estimates for these years were taken from the National Transit Database
(NTD) compiled by the Federal Transit Administration. The remainder of MV-1 bus
registrations were allocated to the intercity bus source type. Since school and transit bus
28
-------
registrations in Puerto Rico were not readily available, we estimated them by multiplying the US
transit or school bus registrations by the ratio of bus registrations in Puerto Rico to the total MV-
1 bus registrations. Note that the precipitous drop in bus populations from 2010 to 2011 is
reflected in the MV-1 bus registration data published by FHWA, which has been used in
MOVES2014 without adjustment.
29
-------
Table 5-2 Historic source type populations for calendar years 1990 and 1999-2011 (in thousands)
Year
1990
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
Motorcycle
4,281
4,174
4,368
4,925
5,026
5,392
5,813
6,259
6,770
7,254
7,869
8,046
8,125
8,553
Passenger
Car
145,112
134,480
135,670
139,709
137,996
137,745
138,642
138,779
137,742
138,354
139,501
138,743
133,313
128,078
Passenger
Truck
27,700
55,472
58,930
62,685
63,789
65,651
69,860
72,980
76,321
78,443
78,596
79,219
79,641
87,030
Light
Commercial
Truck
9,903
18,532
19,217
19,947
19,801
19,873
20,616
20,987
21,380
21,398
20,868
20,464
20,007
21,252
Intercity
Bus
60
81
81
81
79
81
83
85
88
91
96
94
89
18
Transit
Bus
59
56
60
61
65
65
65
65
66
67
65
67
68
66
School
Bus
511
595
609
611
620
634
650
660
672
680
687
684
694
587
Refuse
Truck
67
105
106
116
120
126
132
141
152
164
172
178
180
176
Single
Unit Short-
Haul
Truck
3,870
5,312
5,123
5,416
5,396
5,452
5,528
5,703
5,948
6,208
6,322
6,356
6,234
5,915
Single
Unit Long-
Haul
Truck
145
314
296
305
297
292
288
289
293
297
293
286
271
248
Motor
Home
927
,073
,055
,137
,155
,189
,228
,290
,370
,456
,509
,544
,540
,487
Combination
Short-Haul
Truck
1,177
,361
,368
,384
,335
,307
,293
,309
,353
,364
,319
,317
,266
,198
Combination
Long-Haul
Truck
705
,008
,043
,087
,080
,088
,108
,155
,228
,274
,268
,303
,289
,255
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 5-2. 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 Section 7.1.2.
30
-------
5.2 Projected Vehicle Populations (2012-2050)
The previous section described the historic fleet as it appeared in the data. This section presents
how EPA projected those vehicle populations into the future. This work is inherently dependent
on projections of both vehicle sales and scrappage. While future vehicle sales are commonly
included in economic forecasts, there are no reliable sources for projected national vehicle
scrappage. Therefore, we decided to use projected VMT growth as a surrogate for vehicle
population growth. In examining VMT per vehicle by HPMS class in historic years, this
surrogate appears reasonable. Figure 5-6 shows the VMT values of Table 4-1 divided by the
vehicle populations of Table 5-2 grouped by HPMS classification. At this level of aggregation,
VMT per vehicle is relatively constant with no clear trends over time.
Figure 5-6 MOVES2014 annual miles traveled per vehicle by HPMS class, 1999-2011
80,000
70,000
~ 60,000
I
50,000
£ 40,000
H
w
rt
^ 30,000
Jj 20,000
10,000
Motorcycles
•Light-Duty Vehicles
Buses
Single Unit Trucks
•Combination Trucks
1999
2001
2003
2005
Year
2007
2009
2011
Therefore, the AEO growth factors used to project future VMT as described in Section 4.2 were
used to project populations. Motorcycle growth was calculated using factors from light-duty
vehicles. Since these growth factors are by HPMS class, the 2011 source type populations were
aggregated by HPMS class before the growth factors were applied to the base populations. The
resulting HPMS class population projections are presented in Table 5-3. However, MOVES
cannot use populations in this format as it requires them to be disaggregated by source type. The
distribution projected HPMS class populations to source type was calculated with the same
algorithm used to produce age distributions. Please see Section 7.1.2.2 for a detailed discussion
on this topic. The resulting projected source type populations are tabulated in Section 17
(Appendix A).
-------
Table 5-3 Projected HPMS class populations for 2012-2050 (in thousands)
Year
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
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
2050
Motorcycles
8,571
8,687
8,706
8,747
8,844
8,943
9,018
9,098
9,178
9,260
9,337
9,416
9,498
9,585
9,680
9,781
9,888
9,996
10,103
10,215
10,328
10,439
10,538
10,633
10,724
10,813
10,901
10,983
11,055
11,155
11,256
11,357
11,460
11,564
11,668
11,774
11,880
11,988
12,096
Light-Duty Vehicles
236,285
239,479
240,028
241,178
243,868
246,584
248,692
250,904
253,126
255,371
257,508
259,695
261,966
264,368
266,983
269,767
272,745
275,707
278,670
281,752
284,871
287,918
290,669
293,277
295,790
298,244
300,667
302,932
304,914
307,671
310,453
313,260
316,092
318,951
321,835
324,745
327,681
330,642
333,632
Buses
704
734
757
789
822
844
860
875
890
906
923
941
956
974
990
,004
,015
,027
,041
,053
,063
,075
,091
,106
,122
,137
,152
,166
,180
,196
,210
,226
,241
,257
,273
,289
,304
,322
,338
Single Unit Trucks
8,198
8,637
8,973
9,393
9,790
10,056
10,218
10,373
10,543
10,733
10,934
11,126
11,290
11,463
11,620
11,747
11,858
11,978
12,107
12,234
12,335
12,454
12,606
12,745
12,877
13,007
13,129
13,238
13,346
13,472
13,599
13,731
13,864
13,998
14,135
14,273
14,411
14,550
14,691
Combination Trucks
2,471
2,566
2,635
2,745
2,857
2,935
2,990
3,045
3,100
3,159
3,220
3,281
3,338
3,400
3,459
3,510
3,554
3,600
3,650
3,698
3,737
3,783
3,842
3,901
3,961
4,021
4,081
4,136
4,193
4,251
4,311
4,371
4,432
4,494
4,556
4,620
4,684
4,750
4,816
32
-------
6 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.26'2? To
develop MOVES defaults, we have merged registration and survey data with activity
measurements in an effort identify key vehicle parameters such as weight, axle and tire
configuration, and typical trip range.
MOVES categorizes vehicles into thirteen source use 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, primarily through the SampleVehiclePopulation table, and how MOVES2014 estimates
and projects the number of vehicles in each category.
6.1 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. VIUS was our main source of information for
distinguishing these vehicles. Table 6-1 summarizes how the VIUS2002 parameters were used to
delineate the light-duty, single unit, and combination truck source types for MOVES2014.
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 (PREVIARY_TRIP) 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 Ibs. as light-duty and vehicles greater than or
equal to 10,000 Ibs. 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.
33
-------
Table 6-1 VIUS2002 parameters used to distinguish truck source types in MOVES2014
Source
Type
Passenger
Trucks
Light
Commercial
Trucks
Refuse
Trucks*
Single Unit
Short-Haul
Trucks*
Single Unit
Long-Haul
Trucks*
Combination
Short-Haul
Trucks
Combination
Long-Haul
Trucks
Axle
Arrangement
AXLE CONFIG
in(l,6/7,8)t
AXLE CONFIG
in(l,6,7,8)t
AXLE CONFIG
in (2,9,10,11)
AXLE CONFIG
<=21
AXLE CONFIG
in (2,9,10,11)
AXLE CONFIG
<=21
AXLE CONFIG
in (2,9,10,11)
AXLE CONFIG
<=21
AXLE CONFIG
>=21
AXLE CONFIG
>=21
Primary Distance
of Operation
Any
Any
TRIP PRIMARY
in (1,2,3,4)
TRIP PRIMARY
in (1,2,3,4)
TRIP PRIMARY
in (1,2,3,4)
TRIP PRIMARY
in (1,2,3,4)
TRIP PRIMARY
in (5,6)
TRIP PRIMARY
in (5,6)
TRIP PRIMARY
in (1,2,3,4)
TRIP PRIMARY
in (5,6)
Weight
ADM GVWin(l,2)&
VIUS GVW in (1,2,3)
ADM GVW in (1,2) &
VIUS_GVW in (1,2,3)
Any
ADM GVW>2&
VIUS GVW>3
Any
ADM GVW>2&
VIUS GVW>3
Any
ADM GVW>2&
VIUS GVW>3
Any
Any
Body Type
Any
Any
BODYTYPE
=21
BODYTYPE
=21
BODYTYPE
#1
BODYTYPE
#1
Any
Any
Any
Any
Operator
Classificat
ion
OPCLASS
=5
OPCLASS
^5
Any
Any
Any
Any
Any
Any
Any
Any
T In the MOVES2014 analysis, we did not constrain axle configuration of light-duty trucks, 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 trucks based primarily on their weight. Only 0.27 percent of light-duty trucks 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.
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 intercity, transit, and 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.
6.1.1 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."28 This category usually includes any registered motorcycles, motor scooters,
mopeds, and motor-powered bicycles. Neither the 2011 Polk dataset nor VIUS contain any
information on motorcycles. As noted in Section 5.1 information on motorcycle populations
comes from HPMS MV-1 registrations.
34
-------
6.1.2 Passenger Cars
Passenger cars are defined as any coupes, compacts, sedans, or station wagons with the primary
purpose of carrying passengers.28 All passenger cars (sourceTypelD 21) are categorized in the
light-duty vehicle regulatory class (regClassID 20). Cars were not surveyed in VIUS, but Polk
has a robust yet proprietary dataset of car registrations from all fifty states.
6.1.3 Light-Duty Trucks
Light-duty trucks include pickups, sport utility vehicles (SUVs), and vans.28 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). According to 2011 VM-1
vehicle classifications from FHWA, 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 (VIUS_GVW) to
differentiate between light-duty and single unit trucks. For the passenger trucks, there is a final
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.
6.1.4 Buses
MOVES has three bus source types: intercity (sourceTypelD 41), transit (sourceTypelD 42), and
school buses (sourceTypelD 43). Buses were not included in either VIUS or the Polk dataset, so
supplementary data sources were necessary. MOVES uses various US Department of
Transportation definitions for buses.
Transit buses are defined in the Federal Transit Administration's National Transit Database
(NTD), which states that they are buses owned by a public transit organization for the primary
purpose of transporting passengers on fixed routes and schedules.29 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.30 Intercity buses are, as defined by the Bureau of
Transportation Statistics, "interstate motor carrier of passengers with an average annual gross
revenue of at least one million dollars,"31 but MOVES also considers any bus that cannot be
categorized as either a transit or school bus to be an intercity bus, such as motor coaches and
airport shuttles.
6.1.5 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). With 2013 VM-1 updates to vehicle classifications, FHWA now
defines 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
35
-------
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 (TRIP_PRIMARY),
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 homes are not
included in VIUS.
6.1.6 Combination Trucks
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.32
6.2 Sample Vehicle Population
To match source types to emission rates, MOVES must associate each source type with specific
fuel types and regulatory classes. As vehicle markets shift, this mapping changes with model
year.
Much of default the information on fleet characteristics is stored in the SampleVehiclePopulation
table, which contains two fractions: 1) stmyFraction, and 2) stmyFuelEngFraction. The former
fraction defines the default fuel type distribution, which can be modified by the user through the
Alternative Fuel Vehicle and Technology (AVFT) table. The latter fraction forms the default
regulatory class distribution. Both fractions are computed using the EPA sample vehicle counts
dataset that joins 2011 national R.L. Polk vehicle registration data with 2002 Vehicle Inventory
and Use Survey (VIUS) classifications.
6.2.1 Fuel Type and Regulatory Class Distributions
The stmyFraction is the default national fuel type and regulatory class allocation for each source
type and model year. Written out mathematically in Equation 1, we define the stmyFraction as,
f(stmy)iijikil = — — ,
^" Equation 1
I,
jeJ,keK
-------
where the number of vehicles N in a given model year i, regulatory class j, fuel type /c, and
source type / is divided by the sum of vehicles across the set of all regulatory classes/ and all
fuel types K. That is, the denominator is the total for a given 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 and regulatory classes 30 and 40.
These values must sums to one for each source type and model year. A value of zero indicates
that the MOVES default population of vehicles of that source type, model year, fuel type, and
regulatory class is negligible or does not exist.
While stmyFraction indicates MOVES default values, the stmyFuelEngFraction allows the
modeling of non-default fuel type distributions. For each allowable combination of source type,
model year and fuel type, the StmyFuelEngFraction indicates the expected regulatory class
distribution, whether or not these vehicles exist in the default. Similar to the stmyFraction above,
we define StmyFuelEngFraction in Equation 2 as,
I,,
Equation 2
for number of vehicles N, model year i, regulatory class), fuel type /c, source type /, and the set
of all regulatory classes/. In this case, the denominator is the total for a given source type, model
year, and fuel type.. For example, for model year 2010 gasoline passenger trucks, the table will
list a StmyFuelEngFraction for regulatory class 30 and another for regulatory class 40. These
fractions sum to one for each combination of source type, model year and fuel type.
For example, while the stmyFraction indicates that the MOVES defaults assign zero fraction of
model year 2010 passenger trucks to the electricity fuel type, the StmyFuelEngFraction indicates
a default (hypothetical) regulatory class distribution if these vehicles existed. In this case,
MOVES would model any electric passenger trucks as belonging to regulatory class 30. The
stmyFraction is particularly important because users can edit fuel type distributions using the
Alternative Vehicle Fuel and Technology (AVFT) importer. For instance, a user can create a
future scenario in which there is a high penetration of electric passenger trucks. The
StmyFuelEngFraction allows MOVES to assign vehicles to regulatory class without requiring
this input from the user. This means an allowed StmyFuelEngFraction must never be zero.
As noted in Section 2.4, these fuel type fractions indicate the fuel capability of the vehicle and
not the fuel being used by the vehicle.6 In this report's nomenclature, ESS-capable and flexible
e 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 6-1 below shows the national
default fuel type fractions for all light-duty vehicles among the different MOVES fuel types.
37
-------
fuel vehicles are synomous—meaning they can accept either gasoline or E85 fuel. The amount of
E85 versus the amount of gasoline used out of all the fuel consumed by the vehicle is stored in
the fuel usage fraction. Discussion on fuel usage can be found in the MOVES2014 Fuel Supply
Report.6 MOVES2014 does not explicitly model hybrid electric cars but accounts for these
vehicles in calculating fleet-average energy consumption and CO2 rates.f
Figure 6-1 Default fuel type fractions for light-duty source types in MOVES2014, where being E85-capable
indicates flexible fuel vehicle populations and all default electric vehicle populations are zero
1.00-
0.75 -
0.50-
0.25 -
0.00-
1.00-
0.75-
0.50 -
1025-
o
a
£ o.oo -
&1.00-
£-
1J0.75-
t-
0.50-
0.25-
0.00-
1.00-
0.75-
0.50 -
0.25-
0.00-
^>^r^-
J^Ssi/s^ss/Y^
\r
A ,
J*
A
/V. ,T TT
<^J^^
'
I
5
I
1975 2000 2025 2050
Model Year
MOMLS Source Types
| j Passenger Cars
Passenger Trucks
Light Commercial Trucks
f 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.
38
-------
6.2.2 Sample Vehicle Counts
The SampleVehiclePopulation table fractions were developed by EPA using the sample vehicle
counts dataset referenced in Section 3, which primarily joins calendar year 2011 registration data
from R.L. Polk and the 2002 Vehicle Inventory and Use Survey (VIUS) results. The sample
vehicle counts dataset was generated by multiplying the 2011 Polk vehicle populations by the
source type allocations from VIUS.
While VIUS provide source type classifications, we relied primarily on the 2011 Polk vehicle
registration dataset to form the basis of the fuel type and regulatory class distributions in the
SampleVehiclePopulation table. We purchased the Polk dataset in April 2012, so it did not have
complete registration records for model year 2012 vehicles, and, therefore, model year 2012
vehicles were omitted from the SVP analysis. The Polk data was 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 Polk records by FHWA truck weight class were grouped into MOVES GVWR-based
regulatory classes, as shown in Table 6-2 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 Ibs.
Table 6-2 Initial mapping from FHWA truck classes to MOVES regulatory classes
Vehicle Category
Trucks
Trucks
Trucks
Trucks
Trucks
Trucks
Trucks
Trucks
Trucks
Trucks
Cars
FHWA Truck Weight Class
1
2a
2b
o
3
4
5
6
7
8a
8b
Weight Range (Ibs)
< 6,000
6,001-8,500
8,501 - 10,000
10,001 - 14,000
14,001 - 16,000
16,001 - 19,500
19,501-26,000
26,001 - 33,000
33,001-60,000
> 60,001
regClassID
30
30*
41*
41
42
42*
46
46
47
47
20
*After the Polk 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 Polk dataset did not distinguish between Class 2a (6,001-8,500 Ibs) and Class 2b
(8,501-10,000 Ibs) trucks, but MOVES regulatory classes 30, 40, and 41 all fall within Class 2,
we needed a secondary data source to allocate the Polk gasoline and diesel trucks between Class
39
-------
2a and 2b. We derived information from an Oak Ridge National Laboratory (ORNL) paper33
summarized in Table 6-3 to allocate the Polk Class 2 gasoline and diesel trucks into the
regulatory classes. Class 2a trucks fall in regulatory class 30 and Class 2b trucks fall in either
regulatory class 40 or 41.
Table 6-3 Fractions used to distribute Class 2a and 2b trucks
Fuel Type
Gasoline
Diesel
Truck Class
2a
0.975
0.025
2b
0.760
0.240
Additionally, the Polk dataset includes a variety of fuels, some that are included in MOVES and
others that are not. Only the Polk gasoline and diesel vehicles were included in our analysis; all
other alternative fuel vehicles were omitted. While MOVES2014 does model light-duty E-85 and
electric vehicles, and compressed natural gas (CNG) transit buses, these relative penetrations of
alternative fuel vehicles have been developed from secondary data sources rather than Polk
because Polk 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, and dedicated CNG bus populations from the
National Transit Database. The Table 6-4 illustrates how Polk fuels were mapped to MOVES
fuel types, and which Polk fuels were not used in MOVES.
The "N/A" mapping shown in Table 6-4 led us to discard 0.22 percent, roughly 530,000 vehicles
(mostly dedicated or aftermarket alternative fuel vehicles), of Folk'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 5.1, the total population
is not affected. We considered the Polk vehicle estimates to be a sufficient sample for the fuel
type and regulatory class distributions in the SampleVehiclePopulation table.
40
-------
Table 6-4 A list of fuels from the Polk dataset used to develop MOVES fuel type distributions
Polk Fuel Type
Unknown
Undefined
Both Gas and Electric
Gas
Gas/Elec
Gasoline
Diesel
Natural Gas
Compressed Natural Gas
Natr.Gas
Propane
Flexible (Gasoline/Ethanol)
Flexible
Electric
Cnvrtble
Conversion
Methanol
Ethanol
Convertible
MOVES fuelTypelD
N/A
N/A
1
1
1
1
2
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
MOVES Fuel Type
Gasoline
Gasoline
Gasoline
Gasoline
Diesel
Next we transformed the VIUS dataset into MOVES format. The VIUS vehicle data was first
assigned to MOVES source types using the constraints in Table 6-1 and then to MOVES
regulatory classes using the mapping described in Table 6-2, including the allocation between
Class 2a and 2b trucks from the ORNL study in Table 6-3. Similar to our fuel type mapping of
the Polk dataset, we chose to omit alternative fuel vehicles, as summarized below in Table 6-5.
-------
Table 6-5 Mapping of VIUS2002 fuel types to MOVES2014 fuel types
VIUS Fuel Type
Gasoline
Diesel
Natural gas
Propane
Alcohol fuels
Electricity
Gasoline and natural gas
Gasoline and propane
Gasoline and alcohol fuels
Gasoline and electricity
Diesel and natural gas
Diesel and propane
Diesel and alchol fuels
Diesel and electricity
Not reported
Not applicable
VIUS Fuel Code
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
MOVES fuelTypelD
1
2
N/A
N/A
N/A
N/A
1
1
1
1
2
2
2
2
N/A
N/A
MOVES Fuel Type
Gasoline
Diesel
Gasoline
Gasoline
Gasoline
Gasoline
Diesel
Diesel
Diesel
Diesel
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
k, the source type population fraction / for a specified source type / will be the number of VIUS
trucks N 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 3:
£—>lFl '
Equation 3
The VIUS data in our analysis spanned model year 1986 to 2002. The 2002 source type
distribution has been used for all distributions after MY 2002 and the 1986 distribution for all
prior to MY 1986.
From there the source type distributions from VIUS were multiplied by the Polk vehicle
populations to generate the sample vehicle counts by source type, as shown schematically in
Figure 6-2. Expressed in Equation 4, the sample vehicle counts are,
= P(Polk\jikil •
Equation 4
42
-------
where N is the number of vehicles used to generated the SampleVehiclePopulation table, P is the
2011 Polk vehicle populations, and / is the source type distributions from VIUS.
Figure 6-2 A schematic overview of how the 2011 Polk dataset and VIUS 2002 were joined to create EPA's
sample vehicle counts for MOVES2014. Note that data on buses, motorcycles, and motor homes was pulled
from other sources.
VIUS 2002
Polk 2011
Interim VIUS
Interim Polk
sourceTypelD
modelYearlD
modelYearlD
fuelTypelD
regClassID
totalCounts
fuelTypelD
regClassID
sourceTypeFractions
Household Units
INTERCITY BUSES
source jype
modelYearlD
fuelTypelD
regClassID
vehicleCounts
SAMPLEJD
AXLE_COKFIO
TRIP_PRIMARY
OPCLASS
FUEL
VTUS_GVW
ADM_MODELYEAR
ADM_GVW
TAB TRUCKS
MOTORCYCLES
J
SCHOOL BUSES
j
MOTOR HOMES
Sample Vehicle Counts
These sample vehicle counts by source type were then utilized to calculate the SVP fractions,
stmyFraction and stmyFuelEngFraction, as defined above. Due to a small sample size of vehicles
30 years old and older in both the Polk and VIUS datasets, MOVES2010b SVP fractions were
used for MY 1981 and earlier, which were generated following roughly the same procedure
outlined above but using a 1999 Polk vehicle registration dataset joined with VIUS. These
MOVES2010b SVP fractions for MY 1960-1981 are described in Section 18 (Appendix B).
MOVES2014 assumes no changes to fuel type distributions after model year 2011 except for
flexible-fuel (E85-capable) vehicles, which are assumed to displace gasoline vehicles based on
sales estimates as described below. MOVES2014 estimates any other population growth by
source type, as described earlier in Section 5.2 rather than growth for specific fuel types within a
source type.
All Class 2b and 3 trucks were initially assigned to regulatory class 41 until vehicles were sorted
into source types. Once the sample vehicle counts were available by source type, any light-duty
trucks (sourceTypelD 31 or 32) in the original LHD regulatory class less than 14,000 Ibs
(regClassID 41) were reclassified in the new LHD regulatory class less than 10,000 Ibs
(regClassID 40), whereas any heavy-duty vehicles (sourceTypelD 41 and above) remained in
regClassID 41. Similarly, any single unit trucks (sourceTypelD 52 and 53) in the LDT regulatory
class (regClassID 30) were reclassified in regClassID 41 as heavy-duty vehicles. We also moved
any regClassID 42 vehicles in combination truck source types to regClassID 46 because tractor-
trailers must be either Class 7 or 8 trucks. This ensures a clean break between light- and heavy-
43
-------
duty emission results and that the emission calculations use the appropriate fixedMassFactor
when calculating vehicle-specific power (VSP) for light-duty vehicles and scaled tractive power
(STP) for heavy-duty vehicles.
As noted above, the initial sample vehicle counts dataset did not contain motorcycles, buses, or
motor homes, so information on these source types was appended. Motor homes—even though
they are considered single unit vocational vehicles—cannot be identified in VIUS. In the
subsections below, we have provided more detailed descriptions by source type.
6.2.2.1 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. We acknowledge that some alternative fuel motorcycles have been prototyped and may
even be in small production, but they account for a negligible fraction of total US motorcycle
sales and cannot be modeled in MOVES2014.
6.2.2.2 Passenger Cars
Any passenger car is considered to be in the light-duty vehicle regulatory class (regClassID 20).
Cars were included in the Polk 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 (ESS-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
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
Polk dataset.
Similarly, for model years 2012 and later, we used Department of Energy car sales projections
from AEO2014's table labeled "Light-Duty Vehicle Sales by Technology Type" to derive
flexible fuel vehicle penetrations and applied them to the SVP fractions for regulatory class 20.34
All other alternative fueled cars were determined to have insignificant market shares now and
into the future.
While MOVES can model electric vehicles (fuelTypelD 9), the current market share of electric
cars is sufficiently small that we have set the default electric car population to zero. Users can
model an electric vehicle population by using the AVFT tool to redistribute market share.
Electric vehicles do not have any tailpipe emissions, but MOVES2014 has electric vehicle rates
for energy consumption, brakewear, and tirewear (electric vehicle brake and tirewear rates are
copied from gasoline vehicles). Please consult the MOVES2014 documentation on greenhouse
gases35 and brake and tirewear36, respectively, for more information on the development of the
energy and emission rates themselves.
6.2.2.3 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
44
-------
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 historic years (MY 2011 and earlier) and AEO light truck projections for
future years (MY 2012 and later) from the AEO2014 table on light-duty vehicle sales.34 The
flexible fuel vehicle penetration was applied to regClassID 30 for both E-85 (fuelTypelD 5)
passenger and light commercial trucks and then deducted from their gasoline counterparts in the
same regulatory class.
6.2.2.4 Buses
In line with the US Energy Information Administration (EIA) assumptions, all intercity buses in
MOVES are powered by diesel fuel.37 The following non-school bus regulatory class distribution
for intercity buses was applied to all model years based on 2011 FHWA data, as shown in Table
6-6.38
Table 6-6 Regulatory class fractions of school and non-school buses using 2011 FHWA data
Vehicle Type
Non-School Buses
School Buses
MOVES regClassID
41
0.1856
0.0106
42
0.0200
0.0070
46
0.1214
0.9371
47
0.6730
0.0453
Total
1
1
The National Transit Database (NTD) Revenue Vehicle Inventory (Form 408) closely tracks the
number of motor buses (MB) by fuel type each year and those statistics are used to develop the
MOVES fuel type distributions for transit buses. The mapping from NTD fuel types to MOVES
fuel types is summarized in Table 6-7.
45
-------
Table 6-7 Mapping National Transit Database fuel types to MOVES fuel types
NTD code
BD
BF
CN
DF
DU
EB
EP
ET
GA
GR
HD
HG
KE
LN
LP
MT
OR
NTD description
Bio-diesel
Bunker fuel
Compressed natural gas
Diesel fuel
Dual fuel
Electric battery
Electric propulsion
Ethanol
Gasoline
Grain additive
Hybrid diesel
Hybrid gasoline
Kerosene
Liquefied natural gas
Liquefied petroleum gas
Methanol
Other
fuelTypelD
2
N/A
o
J
2
1
N/A
N/A
N/A
1
N/A
2
1
N/A
3
N/A
N/A
N/A
MOVES Fuel
Description
diesel
CNG
diesel
diesel
gasoline
diesel
gasoline
CNG
While some other MOVES fuel types are included in the NTD, the transit bus fuel type
distributions were allocated between diesel, CNG, and gasoline only. Together these three fuel
types account for more than 99 percent of all transit buses in 2011, so no other alternative fuels
are allowed within the transit bus source type due to negligible market shares.
Biodiesel does not appear in the SampleVehiclePopulation table—in MOVES it is considered a
fuel subtype rather than a fuel type—so biodiesel buses were added to the diesel buses from the
NTD. Liquefied natural gas (LNG) comprises less than ten percent of all natural gas transit buses
and only about 1.5 percent of the whole transit bus fleet in 2011. Without any readily available
emission rate data on LNG buses, we grouped all natural gas fueled transit buses together. This
means we effectively model LNG buses as if they were powered by CNG. Due to limited data,
we assume that gasoline has a one-percent market share prior to model year 2000 and that diesel
has a 99 percent market share prior to MY 1990. All other market shares of transit bus fuel types
are derived using the NTD, as shown in Table 6-8. MOVES modelers can adjust these
distributions between the fuel types using the AVFT tool.
-------
Table 6-8 Fuel type market shares by model year for transit buses in MOVES2014
Model Year
1982-1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011+
MOVES Fuel Type
Gasoline
1.00%
1.00%
1.00%
1.00%
1.00%
1.00%
1.00%
1.00%
1.00%
1.00%
1.00%
0.85%
0.88%
0.91%
0.94%
0.89%
1.05%
1.18%
1.29%
1.61%
1.89%
2.14%
2.46%
Diesel
99.00%
98.30%
97.20%
94.40%
91.40%
90.50%
83.70%
89.20%
81.60%
84.10%
87.70%
91.57%
90.51%
89.09%
88.06%
86.85%
85.61%
84.73%
83.99%
82.91%
82.55%
81.96%
81.75%
CNG
0.00%
0.70%
1.80%
4.60%
7.60%
8.50%
15.30%
9.80%
17.40%
14.90%
11.30%
7.58%
8.60%
10.00%
10.99%
12.27%
13.34%
14.09%
14.72%
15.49%
15.56%
15.90%
15.79%
Urban transit buses are regulated separately from other heavy-duty vehicles, under 40 CFR
86.091-2.39 For this reason, CNG and diesel transit buses are each categorized in regulatory class
48. Lacking better data, we used a single regulatory class distribution from a study of diesel and
CNG transit buses, highlighted in the MOVES2014 heavy-duty emission rates report5, for
gasoline transit buses as shown in Table 6-9 below.
Table 6-9 Regulatory class fractions of gasoline transit buses in MOVES2014
MOVES Source Type & Fuel Type
Gasoline Transit Buses
MOVES regClassID
42
0.2683
46
0.0976
47
0.6341
Total
1
The MOVES2014 school bus fuel type distribution is based on MOBILE6 estimates, originally
calculated from 1996 and 1997 Polk bus registration data, for model years 1982-1996 are
summarized in Table 6-10. 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
47
-------
buses are gasoline fueled in MY 1997 and later.40 The school bus regulatory class distribution
was also derived from the 201 1 FHWA data in Table 6-6.
Table 6-10 Fuel type market shares by model year for school buses in MOVES2014
Model Year
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997+
MOVES Fuel Type
Gasoline
67.40%
67.62%
61.55%
48.45%
32.67%
26.55%
24.98%
22.90%
12.40%
8.95%
1.00%
12.05%
14.75%
11.43%
4.15%
1.00%
Diesel
32.60%
32.38%
38.45%
51.55%
67.33%
73.45%
75.02%
77.10%
87.60%
91.05%
99.00%
87.95%
85.25%
88.57%
95.85%
99.00%
6.2.2.5 Single Unit Trucks
The fuel type and regulatory class distributions for the single unit trucks are calculated directly
from the EPA's sample vehicle counts datasets, except motor homes. The single unit source
types are split between gasoline and diesel only. Single unit vehicle are distributed among the
heavy-duty regulatory classes (regClassIDs 41, 42, 46, and 47) based on the underlying sample
vehicle data. Motor home was not included as a VIUS body type response, so their fuel type and
regulatory class distributions have been developed through supplementary data sources. The fuel
type distribution for motor homes is unchanged from MOVES2010b (see Table 6-11), originally
based on interpolating information from the Recreation Vehicle Industry Association (RVIA) on
fuel type market shares.41
-------
Table
6-11 Fuel type market shares for motor homes in MOVES2014
Model Year
1982-1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010+
Percent of Diesel
15%
18%
21%
23%
26%
29%
32%
34%
37%
40%
41%
43%
44%
46%
47%
49%
50%
50%
Percent of Gasoline
85%
82%
79%
77%
74%
71%
68%
66%
63%
60%
59%
57%
56%
54%
53%
51%
50%
50%
The motor home regulatory class distribution, shown below in Table 6-12, is used across all
model years based on the same 2011 FHWA dataset38 referenced above for school and non-
school buses.
Table 6-12 Regulatory class fractions of motor homes using 2011 FHWA data
MOVES Source Type
Motor Homes
MOVES regClassID
41
0.2697
42
0.3940
46
0.2976
47
0.0387
Total
1
6.2.2.6 Combination Trucks
Combination trucks consist mostly of Class 8 trucks in the MOVES HHD regulatory class
(regClassID 47) but also contain some Class 7 trucks in the MHD regulatory class (regClassID
46), predominantly in short-haul. Similarly, almost all combination trucks are diesel fueled.
MOVES does not model gasoline long-haul combination trucks. Even for the short-haul source
type, gasoline combination trucks are being phased out rapidly. After model year 2005,
MOVES2014 assumes no gasoline combination trucks sales. These fuel type and regulatory class
trends come out of the sample vehicle counts dataset. There has been growing interest in natural
gas for freight transportation but currently this remains largely in the planning stages. There has
not been sufficient testing of these trucks to develop MOVES emission rates yet. We will
consider adding natural gas combination trucks as they become more prevalent and their
emissions are more thoroughly tested.
49
-------
7 Vehicle Characteristics that Vary by Age
Age is an important factor in calculating vehicle emission inventories, identifying high emitters,
and characterizing travel behavior. MOVES employs a number of different age dependent
factors, including deterioration of engine and emission after-treatment technology due to
tampering and malmaintenance, vehicle scrappage and fleet turnover, and mileage accumulation
over the lifetime of the vehicle. Deterioration effects are detailed in the MOVES2014 reports on
the development of light-duty and heavy-duty emission rates.4'5 In this section, there is
discussion of vehicle age distributions, survival rates, and relative mileage accumulation rates by
source type.
7.1 Age Distributions
A vehicle's age is simply the difference between its model year and the year of analysis. Age
distributions in MOVES vary by source type and range from zero to 30+ years, so that all
vehicles 30 years and older are modeled together. As such, 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, as summarized later in this section in Equation 9. Since
sales and scrappage rates are not constant, these distributions vary by calendar year. The age
distribution for each source type is stored in the SourceTypeAgeDistribution table, and fractions
from each source type's age distribution sum to one across a calendar year. MOVES age
distributions were compiled from a variety of data sources, which are discussed below. Age
distributions for the 2011 base year are summarized in Table 7-1; all other years are available in
the MOVES2014 default database SourceTypeAgeDistribution table.
7.1.1 Age Distributions from Registration Data
Ideally all historic age distributions could be derived from registration data sources for each
analysis year available in MOVES. However, acquiring such data is prohibitively costly, so
MOVES2014 only contains registration-based age distributions for two analysis years: 1990 and
2011. The following sections detail how these data were analyzed and used in MOVES2014.
7.1.1.1 1990 Age Distributions
MOVES2014 age distributions for calendar year 1990 have not been updated since the last
model release. Please refer to Section 19 (Appendix C) for more information on the 1990 age
distributions.
7.1.1.2 2011 Age Distributions
The 2011 age distributions for cars and trucks were derived from the sample vehicle counts
dataset, as discussed earlier in Section 3.3. This sample vehicle data includes eight of the thirteen
source types: passenger cars (21), passenger trucks (31), light commercial trucks (32), refuse
trucks (51), short-haul single unit trucks (52), long-haul single unit trucks (53), short-haul
combination trucks (61), and long-haul combination trucks (62). We were able to develop zero to
30+ year age distributions in 2011 for the eight source types mentioned.
For the source types that were not included in the sample vehicle data—specifically motorcycles,
motor homes, and buses—we calculated the 2011 age distributions from the MOVES2010b
50
-------
default 1999 distributions with the latest sales data available. This approach kept the
MOVES2010b base populations and scrappage rates but substituted in MY 1999-2011 sales. We
pulled sales for motorcycles (11) from the Motorcycle Industry Council's Statistical Annual
report24, for transit buses (42) from internal EPA estimates based on manufacturer reporting, and
for school buses (43) from the School Bus Fleet Fact Book1*. Since 2011 age distributions were
calculated independently, intercity bus (41) and motor home (54) sales data were based on
slightly different assumptions. Both of these source types used an average of Ward's Class 3-8
truck sales in Oak Ridge's Transportation Energy Data Boole22, transformed into MOVES source
types using the allocation of sample vehicle counts described in Section 6. For more information
on these data sources, please revisit Section 3.
Figure 7-1 and Table 7-1 show the fraction of vehicles by age (0-30+ years) and source type for
calendar year 2011. These 2011 age distributions became the basis for all the forecast age
distributions in Section 7.1.2.2 and all backcast age distributions in Section 7.1.2.3.
Figure 7-1 2011 age distributions by source type in MOVES2014
Source Type
Combination Long-haul Truck
~~ Combination Short-haul Truck
Intercity Bus
— Light Commercial Truck
— Motor Home
Motorcycle
Passenger Car
— Passenger Truck
Refuse Truck
~~ School Bus
Single Unit Long-haul Truck
Single Unit Short-haul Truck
Transit Bus
30
-------
Table 7-1 2011 age fractions by MOVES source type
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+
11
0.0585
0.0565
0.0614
0.1088
0.0968
0.0917
0.0803
0.0682
0.0583
0.0514
0.0436
0.0348
0.0263
0.0224
0.0215
0.0188
0.0142
0.0163
0.0133
0.0111
0.0088
0.0071
0.0053
0.0045
0.0044
0.0037
0.0031
0.0028
0.002
0.0016
0.0025
21
0.042
0.0472
0.043
0.0545
0.0597
0.0562
0.0562
0.0526
0.0551
0.055
0.0534
0.0575
0.05
0.0441
0.042
0.0354
0.0367
0.029
0.0249
0.0209
0.0178
0.015
0.0124
0.0097
0.008
0.0065
0.0053
0.0042
0.0025
0.0017
0.0016
31
0.0496
0.044
0.0335
0.0587
0.0626
0.0644
0.0677
0.0686
0.0638
0.0624
0.0562
0.0545
0.0504
0.0424
0.0372
0.0284
0.0274
0.025
0.0175
0.0142
0.012
0.0106
0.0108
0.0092
0.007
0.0071
0.0049
0.004
0.0024
0.0019
0.0016
32
0.0557
0.0482
0.0372
0.0668
0.0703
0.0743
0.077
0.0781
0.0724
0.0702
0.0647
0.055
0.0433
0.0273
0.0305
0.0203
0.0219
0.0137
0.0136
0.0073
0.007
0.0075
0.008
0.0073
0.0057
0.0053
0.0037
0.0031
0.0019
0.0015
0.0012
41
0.0477
0.0421
0.0353
0.0458
0.0601
0.0617
0.0638
0.062
0.0574
0.0538
0.0517
0.0492
0.0478
0.0362
0.0295
0.0244
0.0317
0.0244
0.0201
0.0148
0.0168
0.0188
0.0187
0.0174
0.018
0.0151
0.0132
0.0104
0.0041
0.0035
0.0047
42
0.0628
0.0385
0.0393
0.0555
0.0539
0.0389
0.0607
0.0498
0.0488
0.0495
0.057
0.0385
0.0374
0.0439
0.0401
0.0369
0.0303
0.0264
0.0219
0.019
0.0192
0.0281
0.0214
0.0168
0.0156
0.0131
0.0113
0.0088
0.0083
0.0045
0.0039
43
0.0368
0.0403
0.048
0.0529
0.0548
0.0644
0.0574
0.0565
0.0487
0.0511
0.0467
0.0508
0.047
0.0371
0.0345
0.0298
0.038
0.0184
0.0219
0.0177
0.0226
0.0255
0.0145
0.0173
0.0175
0.0153
0.0131
0.0101
0.0037
0.0027
0.0047
51
0.0334
0.0265
0.0351
0.0273
0.0956
0.0718
0.0677
0.0407
0.04
0.029
0.0357
0.0488
0.0702
0.0645
0.0312
0.0406
0.0521
0.0367
0.0167
0.0149
0.0233
0.0166
0.0256
0.0147
0.0132
0.0068
0.0068
0.0056
0.0025
0.0029
0.0035
52
0.035
0.0216
0.0231
0.0479
0.0629
0.0666
0.0577
0.0506
0.0438
0.0393
0.0427
0.0697
0.0591
0.0334
0.0459
0.0308
0.0423
0.0323
0.0225
0.0179
0.0162
0.022
0.0211
0.0188
0.0171
0.0154
0.0132
0.0113
0.0067
0.0067
0.0066
53
0.0237
0.015
0.0176
0.031
0.0544
0.0486
0.045
0.0333
0.0284
0.0238
0.059
0.1457
0.1267
0.0213
0.0175
0.0198
0.0338
0.0279
0.0777
0.0137
0.0213
0.0132
0.0535
0.017
0.0061
0.0064
0.0055
0.0048
0.0028
0.0028
0.0027
54
0.046
0.0406
0.034
0.0442
0.0579
0.0594
0.0615
0.0597
0.0553
0.0518
0.0498
0.0474
0.0461
0.0271
0.0417
0.0258
0.0305
0.0291
0.02
0.0175
0.013
0.0171
0.0221
0.0196
0.0191
0.0141
0.015
0.0152
0.0098
0.0057
0.0039
61
0.0219
0.0164
0.0213
0.0192
0.0629
0.0468
0.0455
0.0288
0.0256
0.0199
0.0391
0.0535
0.0482
0.049
0.0398
0.0556
0.0628
0.0524
0.038
0.0292
0.0272
0.0337
0.0343
0.0317
0.025
0.0174
0.0177
0.0145
0.0062
0.0073
0.0089
62
0.0478
0.0378
0.0501
0.0392
0.1371
0.1028
0.0971
0.0584
0.057
0.0415
0.0482
0.0766
0.0572
0.0381
0.0215
0.0234
0.0209
0.0127
0.0086
0.0052
0.004
0.0031
0.0031
0.0019
0.0032
0.0009
0.0009
0.0007
0.0003
0.0004
0.0004
52
-------
7.1.2 Forecasting and Backcasting Age Distributions
Since purchasing registration data for all calendar years is prohibitively costly for historic years,
an algorithm was developed to forecast and backcast age distributions from the 2011 age
distribution described above for all other calendar years in the model. In prior versions of
MOVES, these age distributions were calculated during the model run using sales estimates and
assuming a constant survival rate. In MOVES2014, age distributions for national level runs were
pre-calculated using updated sales estimates and assuming a dynamic survival rate. However,
while sales data for historic years are well known and projections for future years are common in
economic modeling, national trends in projected vehicle survival for every MOVES source type
at all ages are not well studied. For MOVES2014, 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 three sections detail the derivation of the generic survival rate and the algorithms used
to forecast and backcast age distributions using an adjusted survival rate in each year.
7.1.2.1 Generic Survival Rates
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. While the use of this generic rate is described in the next
couple of sections, its derivation is specified here.
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.42 Survival rates
for passenger cars, passenger trucks and light commercial trucks came from NHTSA's
survivability Table 3 and Table 4.43 These survival rates are based on a detailed analysis of Polk
vehicle registration data from 1977 to 2002. We modified these rates to 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. Effectively,
this results in a near constant survival rate until agelD 3 for light-duty vehicles and until
agelD 4 for heavy-duty vehicles.
• 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.
53
-------
Because MOVES agelD 30 is intended to represent all vehicles 30 years old and greater, this age
category can grow quite large as our age distribution algorithm eventually transfers all vehicles
to this age group. To assure that the population of very old vehicles does not grow excessively,
the generic survival rate for agelD 30 was set to 0.3. The actual survival rate of these age 30+
vehicles is unknown.
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 50 will
represent a one-dimensional array of sa values at each permissible age a as described in
Equation 5 through Equation 8 below:
1-0-7
Age 0: s0 = 1 Equation 5
A 1 2(1-oz) ^ . ,
Age 1: s-i = 1 — Equation 6
-
Age 2-29: sa = s2 29 = Equation 7
°a-2
Age 30: s30 = 0.3 Equation 8
With limited data available on heavy-duty vehicle scrappage, survivability for all other source
types came from the Transportation Energy Data Book. We used the heavy-duty vehicle survival
rates for model year 1980 (TEDB32, Table 3.14). 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, including setting
age 30+ survival rates to 0.3 for all source types.
The resulting survival rates are listed in the default database's SourceTypeAge table, shown
below in Table 7-2. Please note that since MOVES2014 does not calculate age distributions
during a run, these survival rates are not actively used by MOVES. However, they were used in
54
-------
the development of the national age distributions stored in the SourceTypeAgeDistribution table,
and remain in the default database for reference.
Table 7-2 MOVES survival rate by age and HPMS class
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
Motorcycles
1.000
0.979
0.940
0.940
0.940
0.940
0.940
0.940
0.940
0.940
0.940
0.940
0.940
0.940
0.940
0.940
0.940
0.940
0.940
0.940
0.940
0.940
0.940
0.940
0.940
0.940
0.940
0.940
0.940
0.940
0.300
Light-Duty Vehicles
Passenger
Cars
0.997
0.997
0.997
0.993
0.990
0.986
0.981
0.976
0.971
0.965
0.959
0.953
0.912
0.854
0.832
0.813
0.799
0.787
0.779
0.772
0.767
0.763
0.760
0.757
0.757
0.754
0.754
0.567
0.752
0.752
0.300
Passenger Trucks
Light Comm.
Trucks
0.991
0.991
0.991
0.986
0.981
0.976
0.970
0.964
0.958
0.952
0.946
0.940
0.935
0.929
0.913
0.908
0.903
0.898
0.894
0.891
0.888
0.885
0.883
0.880
0.879
0.877
0.875
0.875
0.873
0.872
0.300
Buses
1.000
1.000
1.000
1.000
0.990
0.980
0.980
0.970
0.970
0.970
0.960
0.960
0.950
0.950
0.950
0.940
0.940
0.930
0.930
0.920
0.920
0.920
0.910
0.910
0.910
0.900
0.900
0.900
0.890
0.890
0.300
Single Unit
Trucks
1.000
1.000
1.000
1.000
0.990
0.980
0.980
0.970
0.970
0.970
0.960
0.960
0.950
0.950
0.950
0.940
0.940
0.930
0.930
0.920
0.920
0.920
0.910
0.910
0.910
0.900
0.900
0.900
0.890
0.890
0.300
Combination
Trucks
1.000
1.000
1.000
1.000
0.990
0.980
0.980
0.970
0.970
0.970
0.960
0.960
0.950
0.950
0.950
0.940
0.940
0.930
0.930
0.920
0.920
0.920
0.910
0.910
0.910
0.900
0.900
0.900
0.890
0.890
0.300
7.1.2.2 2012-2050 Age Distributions
The 2012-2050 age distributions were derived from the 2011 age distribution described above
using population, survival, and sales projections. Age distributions are calculated from
population counts, if the populations are known by age:
55
-------
, _Pa
lay — 7T Equation 9
Here in Equation 9, fay is the age fraction to be calculated, 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 at the individual age level. For example, fy would be
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, projecting an age distribution forward one year involves removing the vehicles
scrapped in the base year and adding the new vehicles sold in the next year, as shown in
Equation 10:
Py+1' = P^ - lly + JVy+1' Equation 10
where Py+1 is the population (known at each age) of the next year, Py is the population in the
base year, Ry is the population of vehicles removed in the in the base year, and JVy+1 is new
vehicles sold in the next year. Please note that the final term only includes new vehicles at age 0;
if the equation is evaluated for any a > 0, the sales term is zero. Equation 10 can be used
algorithmically to forecast a known population distribution as follows:
1. Starting with the base population distribution (Py), remove the number of vehicles that
did not survive (Ry) at each age level.
2. Increase the population age index by one (for example, 3 year old vehicles are
reclassified as 4 year old vehicles).
3. Add new vehicle sales (Ny+1) as the age 0 cohort.
4. Combine the new age 30 and 31 vehicles into a single age 30 group.
5. This results in the next year population distribution (Py+i). If this algorithm is to be
repeated, Py+1 becomes Py for the next iteration.
Please see Section 20 (Appendix D: Detailed Derivation of Age Distributions) for more
information on how this algorithm was applied to derive the projected national default age
distributions in MOVES. The resulting age distributions are stored in the
SourceTypeAgeDistribution table.
In addition to producing the 2012-2050 default age distributions, a version of this algorithm was
implemented in the Age Distribution Projection Tool for MOVES2014.44 This tool can be used
to project future local age distributions from user-supplied baseline distributions, provided that
the baseline year is 2011 or later. This requirement ensures that the 2008-2009 recession is fully
-------
visible in the baseline. The differences between the default algorithm described above and the
algorithm used in the tool are as follows:
• In the tool, the generic survival rate for all vehicle types at age 30 is set to 1.0.
• Step 4 was modified so that in the tool, the new age 30 fraction is set equal to the new
age 31 fraction. The new age 31 fraction is then discarded.
• In the tool, the age distribution for ages 1-29 is then normalized such that the full
distribution (ages 0-30) sums to 1.0.
The first two bullets were implemented to retain the fraction of 30+ year old vehicles in the user-
inputted baseline distribution. This was done because local data frequently indicates a sizeable
fraction in this age bin. Since the default scrappage curve was designed to prevent this bin from
growing too large, the default algorithm would reduce this fraction in most cases. Therefore, the
age 30+ fraction is not modified and the resulting age distribution in each iteration of the
algorithm is normalized in the final step so that the full distribution sums to one. The sales rates
and scrappage assumptions are the same in the tool as they are in the national case. In general,
projections made with the tool tend to converge with the national age distributions the farther out
the projection year becomes. This is because local projections of sales and scrappage are
generally unavailable, and the national trends are the best available data.
7.1.2.3 1999-2010 Age Distributions
The method used to backcast the 1999-2010 age distributions from the 2011 distribution is very
similar to the forecasting method described above. For backcasting an age distribution one year,
Equation 10 of the previous section can be rewritten as Equation 11:
Equation 11
Essentially, this can be thought of as taking the base year's population distribution, removing the
vehicles sold (or added to the population) in that year, and then adding the vehicles that were
removed in the year before. This can be represented algorithmically as follows:
1. Starting with the base population distribution (Py), remove the age 0 vehicles (JVy).
2. Decrease the population age index by one (for example, 3 year old vehicles are
reclassified as 2 year old vehicles).
3. Add the vehicles that were removed in the previous year (Piy_i).
4. This results in the previous year population distribution (Py_i). If this algorithm is to be
repeated, Py-\ becomes Py for the next iteration.
Please see Section 20 Appendix D: Detailed Derivation of Age Distributions) for more
information on how this algorithm was applied to derive the historic national default age
distributions in MOVES. The resulting age distributions are stored in the
SourceTypeAgeDistribution table.
57
-------
7.2 Relative Mileage Accumulation Rate
MOVES uses a relative mileage accumulation rate (RMAR) in combination with source type
populations (see Section 5.1) and age distributions described earlier in this section to distribute
the total annual miles driven by each HPMS vehicle type (see Section 4) 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.
VMT is provided, either by default values or by user input, by the five Highway Performance
Monitoring System (HPMS) vehicle classifications. These classifications are further broken
down into the groupings of the MOVES source use types, as described in Section 2.1.
The RMAR is determined 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. By definition, new (age 0) passenger trucks and light commercial
trucks have a RMAR of one (1.0).g Based on the data, new passenger cars have a RMAR of
0.885. This means that when the VMT assigned to the HPMS class 25 is allocated to passenger
cars, passenger trucks and light commercial trucks, a passenger car of age 0 will be assigned only
88.5 percent of the annual VMT assigned to a passenger truck or light commercial truck of age 0.
The RMAR values for MOVES2014 for the source types 11 (motorcycles), 41 (intercity buses),
42 (transit buses), 43 (school buses) and 54 (motor homes) were not changed from the values
used in MOVES2010b. Passenger car and light-duty truck RMAR values were recalculated to
reflect the change in the HPMS vehicle classifications used for VMT input and the remaining
heavy-duty vehicle classifications were updated with data from the 2002 Vehicle Inventory and
Use Survey (VIUS) and recalculated.
7.2.1 Motorcycles
The RMAR values for motorcycles in MOVES2014 were not changed from MOVES2010b
estimates. The MOVES2010b RMAR values were calculated from MARs 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 Weibull
curve was fit to the data to develop the relative mileage accumulation rates used in MOVES.42
g 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.
-------
7.2.2 Passenger Cars, Passenger Trucks and Light Commercial Trucks
In MOVES2010b, passenger cars had their own HPMS vehicle classification. In MOVES2014,
they are grouped with passenger trucks and light commercial trucks. For MOVES2014, the
MOVES2010b passenger car RMAR values were adjusted to reflect the relative difference in
annual mile accumulation between passenger cars and the light trucks. Analysis of the data
determined that new passenger cars (age 0) accumulate only 88.5 percent of the annual miles
accumulated by new light trucks. Thus, all of the RMAR values for passenger cars were
adjusted to be 88.5 percent of their previous values.
The MOVES2010b RMAR values for passenger cars, passenger trucks and light commercial
trucks (sourceTypelD 21, 31 & 32) were taken from the NHTSA report on survivability and
mileage schedules.43 In the NHTSA analysis, annual mileage by age was determined for cars and
for trucks using data from the 2001 National Household Travel Survey. In this NHTSA analysis,
vehicles that were less than one year old at the time of the survey were classified as "age 1", etc.
NHTSA used a simple cubic regression to smooth the VMT by age estimates.
We used NHTSA's regression coefficients to extrapolate mileage to ages 26 through 30 not
covered by the report. Since passenger trucks had the highest MAR in what was then the light-
duty truck HPMS class, each source type's mileage by age was divided by passenger truck
mileage at age 1 to determine a relative MAR. For consistency with MOVES age categories, we
then shifted the relative MARs such that the NHTSA age 1 ratio was used for MOVES age 0,
etc. We used NHTSA's light truck VMT to determine relative MARs for both passenger trucks
and light commercial trucks.
Since a newer version of the National Household Travel Survey was available, we conducted a
preliminary analysis of the impact of updating the MARs based on the 2009 National Household
Travel Survey. This resulted in changes to the MOVES allocation of VMT by one percent or less
for each of the vehicle categories covered by the survey. As such, we feel that the MARs
developed from the 2001 survey are still reasonable for use in MOVES2014. However, the 2009
values may not fully represent current trends in vehicle usage due to the recent economic
downturn. A more complete analysis of all available mileage accumulation information in recent
years will be necessary to truly update these values.
59
-------
Table 7-3 NHTSA Vehicle Miles Traveled from 2001
Vehicle Age
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
31
32
Annual Vehicle Miles Traveled
Passenger Cars
14,417
13,803
13,692
13,415
13,183
12,301
12,253
11,709
11,893
11,855
10,620
9,986
10,248
9,515
9,168
8,636
8,941
7,267
8,890
8,759
6,878
7,242
6,350
5,745
4,130
Light Trucks
15,806
15,683
15,859
15,302
14,762
13,836
13,542
13,615
12,875
12,203
11,501
10,815
11,391
10,843
10,378
9,259
8,358
9,371
7,352
8,363
6,999
7,327
6,969
6,220
6,312
6,745
9,515
6,635
12,108
5,067
4,577
6,923
7.2.3
Buses
The RMAR values for all bus categories in MOVES2014 were not changed from MOVES2010b
estimates. The intercity bus (sourceTypelD 41) annual mileage accumulation rate is taken from
Motorcoach Census 2000.45 The data did not distinguish vehicle age, so the same MAR (59,873
miles per year) was used for each age. The school bus (sourceTypelD 43) annual mileage
accumulation rate (9,939 miles per year) is taken from the 1997 School Bus Fleet Fact Book.
The MOVES model assumes the same annual mileage accumulation rate for each age. The
60
-------
Transit Bus (category 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.46 The MOBILE6 results were extended
to calculate values for ages 26 through 30.
Table 7-4 Annual mileage accumulation of transit buses from 1994 Federal Transit Administration data
Age
1
2
3
4
5
6
7
8
9
10
Miles
*
*
46,791
41,262
42,206
39,160
38,266
36,358
34,935
33,021
Age
11
12
13
14
15
16
17
18
19
20
Miles
32,540
32,605
27,722
28,429
32,140
28,100
24,626
23,428
22,575
23,220
Age
21
22
23
24
25
26
27
28
29
30
Miles
19,588
22,939
26,413
23,366
11,259
23,228
21,515
25,939
20,117
17,515
* Insufficient data
7.2.4 Other Heavy-Duty Vehicles
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) were updated from MOVES2010b using the data from the 2002 Vehicle Inventory and
Use Survey (VIUS). The total reported annual miles traveled by truck in each source type, as
shown in Table 7-5, was divided by the vehicle population to determine the average annual miles
traveled per truck by source type.
61
-------
Table 7-5 VIUS2002 annual mileage by vehicle age
Age
0
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
0-3
Model
Year
2002
2001
2000
1999
1998
1997
1996
1995
1994
1993
1992
1991
1990
1989
1988
1987
1986
1999-2002
Average
Single Unit Trucks
Refuse
(51)
26,703
32,391
31,210
31,444
31,815
28,450
25,462
30,182
20,722
25,199
23,366
18,818
12,533
15,891
19,618
12,480
12,577
30,437
Short-Haul
(52)
21,926
22,755
24,446
23,874
21,074
21,444
16,901
15,453
13,930
13,303
11,749
13,675
11,332
9,795
9,309
9,379
4,830
23,250
Long-Haul
(53)
40,538
28,168
30,139
49,428
33,266
23,784
21,238
27,562
21,052
11,273
18,599
15,140
13,311
9,796
12,067
16,606
8,941
37,069
Combination Trucks
Short-Haul
(61)
119,867
114,983
110,099
105,215
100,331
95,447
90,563
85,679
80,795
75,911
71,026
66,142
61,258
56,374
51,490
46,606
41,722
61,240
Long-Haul
(62)
109,418
128,287
117,945
110,713
99,925
94,326
85,225
85,406
71,834
71,160
67,760
80,207
48,562
64,473
48,242
58,951
35,897
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 are at age 16 (the limit of the data). MOVES,
however, requires mileage accumulation rates for all ages to age 30. For MOVES2014, we
assumed that the relative mileage accumulation rate at age 30 would be the same as used for
MOVES2010b.
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 for the average of ages 0 to 3 as age 3 with ages 4 through 16 from
the data summarized in Table 7-6,
3) Ages 17 through 29 use values from interpolation between the values in age 16 and
age 30.
4) Age 30 uses the MOVES2010b relative mileage accumulation rate for age 30. These
rates were allocated to MOVES source types from MOBILE6 mileage accumulation
rates, which were derived from the 1992 TIUS as documented in the ARCADIS
report.46
62
-------
Table 7-6 Regression statistics for heavy-duty truck annual mileage accumulation rates (ages 4-16)
Measurement
Average 0-3*
Refuse
Truck (51)
30,437
Single Unit
Short-Haul (52)
23,250
Single Unit
Long-Haul (53)
37,069
Combination
Short-Haul (61)
61,240
Combination
Long-Haul
(62)
116,591
Intercept**
Slope**
36,315
-1,510
25,442
-1,209
36,305
-1,794
65,773
-3,447
119,867
-4,884
Age 30 RMAR
0.0320
0.0518
0.1025
0.0320
0.0571
* Average sample annual miles traveled for ages 0 through 3.
** Intercept and slope from ages 4 through 16.
The resulting relative mileage accumulation rates are shown in Table 7-7 below. Note that the
first four values are identical and then decline linearly to age 16 and then linearly to age 30 with
a different slope.
7.2.5
Motor Homes
Motor home relative mileage accumulation rates for MOVES2014 are unchanged from
MOVES2010b. For motor homes (sourceTypelD 54), the initial MARs were taken from an
independent research study47 conducted in October 2000 among members of the Good Sam
Club. The members are active recreation vehicle (RV) enthusiasts who own motor homes,
trailers and trucks. The average annual mileage was estimated to be 4,566 miles. The data did not
distinguish vehicle age, so the same MAR was used for each age.
63
-------
Table 7-7 Relative mileage accumulation rates for heavy-duty trucks in MOVES2014
agelD
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
Refuse (51)
1.0000
1.0000
1.0000
1.0000
0.9525
0.9050
0.8575
0.8099
0.7624
0.7149
0.6674
0.6199
0.5724
0.5249
0.4773
0.4298
0.3823
0.3573
0.3323
0.3073
0.2822
0.2572
0.2322
0.2072
0.1821
0.1571
0.1321
0.1071
0.0820
0.0570
0.0320
Short-Haul
Single Unit (52)
0.6864
0.6864
0.6864
0.6864
0.6484
0.6103
0.5723
0.5343
0.4962
0.4582
0.4202
0.3821
0.3441
0.3061
0.2680
0.2300
0.1920
0.1808
0.1696
0.1585
0.1473
0.1361
0.1249
0.1138
0.1026
0.0914
0.0802
0.0691
0.0579
0.0467
0.0355
Long-Haul
Single Unit
(53)
0.9729
0.9729
0.9729
0.9729
0.9165
0.8601
0.8036
0.7472
0.6908
0.6343
0.5779
0.5215
0.4650
0.4086
0.3522
0.2957
0.2393
0.2293
0.2194
0.2094
0.1994
0.1894
0.1795
0.1695
0.1595
0.1496
0.1396
0.1296
0.1197
0.1097
0.0997
Motor Home
(54)
0.0590
0.0590
0.0590
0.0590
0.0590
0.0590
0.0590
0.0590
0.0590
0.0590
0.0590
0.0590
0.0590
0.0590
0.0590
0.0590
0.0590
0.0590
0.0590
0.0590
0.0590
0.0590
0.0590
0.0590
0.0590
0.0590
0.0590
0.0590
0.0590
0.0590
0.0590
Short-Haul
Combination
(61)
0.5269
0.5269
0.5269
0.5269
0.4941
0.4613
0.4286
0.3958
0.3631
0.3303
0.2975
0.2648
0.2320
0.1993
0.1665
0.1338
0.1010
0.0950
0.0890
0.0830
0.0770
0.0710
0.0649
0.0589
0.0529
0.0469
0.0409
0.0349
0.0289
0.0229
0.0169
Long-Haul
Combination
(62)
1.0000
1.0000
1.0000
1.0000
0.9536
0.9072
0.8607
0.8143
0.7679
0.7215
0.6751
0.6286
0.5822
0.5358
0.4894
0.4430
0.3965
0.3723
0.3481
0.3238
0.2996
0.2753
0.2511
0.2268
0.2026
0.1783
0.1541
0.1298
0.1056
0.0814
0.0571
64
-------
8 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 vehicle class that is traveled on each of the MOVES five road
types. Users may supply the distribution VMT to vehicle classes for each road type for individual
counties when using County Scale, however, for National Scale, the default distribution is
applied to all locations.
The national default distribution of VMT to vehicle classes for each road type in MOVES2014
were derived to reflect the VMT data included in the 2011 National Emission Inventory (NEI)
Version I48 (July 31, 2013). This data is provided by states every three years as part of the NEI
project and is supplemented by EPA estimates, based on Federal Highway Administration
(FHWA) statistics49, when state supplied estimates are not available.
The 2011 NEI VI data50 is grouped by the source classification code (SCC) used at that time and
these older classifications do not cleanly map to the source types used by MOVES. As discussed
in Section 2.6, SCCs are now formed as a 10-digit concatenated string, including existing
identification codes for MOVES fuel type, source type, road type, and emission process. For
reference, we have included a comparison of the MOVES2010b SCCs and MOVES2014 fuel
types and regulatory classes in Section 21 (Appendix E: SCC Mappings).
The first seven digits of the 10-digit SCC (SCC7) indicate the vehicle classification. The SCC
road types map cleanly to the MOVES road types. The eighth and ninth digits of the 10-digit
SCC (SCC89) indicate the road type, as shown below in Table 8-1. The VMT was mapped to the
source types used by MOVES by calculating the fraction of VMT for each source type found in
each SCC classification result in a national MOVES2010b run for calendar year 2011. The
factors calculated from the MOVES2010b run are also shown in Section 21 (Appendix E: SCC
Mappings).
Table 8-1 Mapping of SCC road types to MOVES road types
SCC Road Type
Code (SCC89)
11
13
15
17
19
21
23
25
27
29
31
33
SCC Road Type
Rural Interstate
Rural Other Principal Arterial
Rural Minor Arterial
Rural Major Collector
Rural Minor Collector
Rural Local
Urban Interstate
Urban Other Freeways & Expressways
Urban Other Principal Arterial
Urban Minor Arterial
Urban Collector
Urban Local
MOVES
Road Type ID
2
o
J
3
3
o
5
o
6
4
4
5
5
5
5
MOVES Road Type
Rural Restricted Access
Rural Unrestricted Access
Rural Unrestricted Access
Rural Unrestricted Access
Rural Unrestricted Access
Rural Unrestricted Access
Urban Restricted Access
Urban Restricted Access
Urban Unrestricted Access
Urban Unrestricted Access
Urban Unrestricted Access
Urban Unrestricted Access
65
-------
Once the SCC VMT values have been mapped to MOVES source types and road types, the
national distribution of road type VMT by source type can be calculated from the NEI VMT
estimates, summarized in Table 8-2. The off network road type (roadTypelD 1) is not used and is
allocated none of the VMT.
Table 8-2 MOVES2014 road type distribution by source type
Source
Type
11
21
31
32
41
42
43
51
52
53
54
61
62
Description
Motorcycle
Passenger Car
Passenger Truck
Light Commercial Truck
Intercity Bus
Transit Bus
School Bus
Refuse Truck
Single Unit Short-Haul Truck
Single Unit Long-Haul Truck
Motor Home
Combination Short-Haul Truck
Combination Long-Haul Truck
Road Type*
Rural
Restricted
2
0.0805
0.0847
0.0859
0.0867
0.1409
0.1384
0.1384
0.2396
0.1635
0.1638
0.1234
0.2367
0.2476
Rural
Unrestricted
3
0.3019
0.2345
0.2754
0.2756
0.2812
0.2813
0.2813
0.2718
0.2869
0.2870
0.2876
0.2744
0.2705
Urban
Restricted
4
0.1913
0.2374
0.2178
0.2180
0.2196
0.2196
0.2196
0.2525
0.2346
0.2346
0.2255
0.2517
0.2543
Urban
Unrestricted
5
0.4263
0.4434
0.4209
0.4197
0.3583
0.3607
0.3607
0.2361
0.3150
0.3146
0.3635
0.2372
0.2276
All
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
* RoadTypelD = 1 (Off Network) is assigned no VMT.
66
-------
9 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, MOVES2014 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 9-1. The default average
speed distributions in MOVES201051 were based on much more limited data and travel demand
model output, and have been substantially updated in MOVES2014.52
Table 9-1 MOVES s
Bin
1
2
o
J
4
5
6
7
8
9
10
11
12
13
14
15
16
Average Speed (mph)
2.5
5
10
15
20
25
30
35
40
45
50
55
60
65
70
75
peed bin categories
Average Speed Range (mph)
speed < 2. 5 mph
2.5 mph <= speed < 7.5 mph
7.5 mph<= speed < 12.5 mph
12.5 mph <= speed < 17.5 mph
17.5 mph <= speed < 22.5 mph
22.5 mph <= speed < 27.5 mph
27.5 mph <= speed < 32.5 mph
32.5 mph <= speed < 37.5 mph
37.5 mph <= speed < 42.5 mph
42.5 mph <= speed < 47.5 mph
47.5 mph <= speed < 52.5 mph
52.5 mph <= speed < 57.5 mph
57.5 mph <= speed < 62.5 mph
62.5 mph <= speed < 67.5 mph
67.5 mph <= speed < 72.5 mph
72.5 mph<= speed
9.1 Light-Duty Average Speed Distributions
For MOVES2014, the light-duty average speed distributions are based on in-vehicle global
position system (GPS) data. The data was obtained through a contract with Eastern Research
Group (ERG), who subcontracted with TomTom to provide summarized vehicle GPS data.11
TomTom makes in-vehicle GPS navigation devices and supports cell-phone navigation
applications. ERG provided the US EPA with updated values for the AvgSpeedDistribution
calculated from the TomTom delivered data based on their consumers, where "virtually all" use
them in light-duty cars, trucks, and vans.
h Much of the following text and tables are excerpted from the ERG Work Plan (EPA-121019), submitted to US
EPA on January 11, 2012.
-------
Some of the characteristics of the TomTom GPS data are:
• Data is self-selective. Data is only recorded from users of TomTom GPS units and an
iPhone application. Additionally, TomTom data is only collected when the units are on.
This creates bias not only for users, but also for types of driving. Anecdotally, drivers
who own GPS units are less likely to use them when they drive in familiar areas in
comparison with unfamiliar areas. Compared to the default VMT by road type
information in MOVES, TomTom over-represents behavior on rural restricted access
roads, which suggests the higher usage of GPS on vacations and business trips.
• No information on vehicle type is available. TomTom suggests that "virtually all" the
vehicles are light-duty cars, trucks, and vans. MOVES allows for separate average speed
distributions for each source type. However, due to a lack of information on other source
types, the average speed distribution derived from the TomTom light-duty GPS data is
applied to all source types—although the combination long-haul trucks distribution was
adjusted as described at the end of this section. Other heavy-duty source types such as
single unit long-haul trucks were not adjusted. We recognize this as a potential
shortcoming, and look to incorporate source type specific average speed information in
the future.
• The average speed distributions are based on the average speed in each roadway segment,
not the average of all second-by-second speed measurements.
• Only data that is associated with the vehicle network is included in the average speed
delivery. As part of the quality control methods, TomTom excludes data that does not
"snap to the roadway grid" to remove points caused by loss of satellite signal and errors
while the TomTom unit is trying to acquire the satellite signal. TomTom uses data quality
control techniques to minimize data arising from non-light-duty-vehicle use, such as from
pedestrians, bicycles, and airplanes.
Some of the data characteristics present concerns regarding their representativeness of real-world
driving. Despite these concerns, the TomTom data presented a great improvement to the speed
distribution information used in previous versions of MOVES.
Under direction of EPA's contractor, ERG, TomTom queried its database of historic traffic
probes to produce a table of total distance and total time as a function of road type,
weekday/weekend, hour of the day, and average speed bin for the calendar year 2011 for the 50
states and the District of Columbia. TomTom delivered a table identifying the total distance and
total time of vehicles travelling at an average speed interval for all combinations of:
1. Identifier for Average Speed Bin (20 levels): average speeds were binned in 5 mph
increments, starting at 2.5mph: 0-2.5mph; 2.5mph-7.5mph; 7.5mph-12.5mph; ...
92.5mph-97.5mph.
2. Identifier for Month of the Year (12 levels).
3. Identifier for Day of the Week (2 levels): the period for weekday is Monday,
00:00:00 to Friday, 23:59:59, and the period for weekend is Saturday, 00:00:00 to
Sunday, 23:59:59.
68
-------
4. Identifier for Time of Day (24 levels): times are binned in one hour increments,
starting at midnight: 00:00:00 to 00:59:59; 01:00:00 to 01:59:59, ..., 23:00:00 to
23:59:59.
5. Identifier for Road Type (4 levels): TomTom used the information in Table 9-2 to
classify between the TomTom Functional Classes and the MOVES road type
description. TomTom also categorized the road types as rural or urban, according to
the Census definitions used in MOVES1.
Table 9-2 Correspondence between TomTom functional class, census information, and MOVES road types
MOVES Road Type
Description
Rural Restricted Access
Rural Unrestricted Access
Urban Restricted Access
Urban Unrestricted Access
Census Information for the
TomTom Roadway Segment
Rural
Rural
Urban
Urban
TomTom Functional
Road Class
Oandl
2 through 7
Oandl
2 through 7
TomTom first "snapped" their data points onto road segments. Off-network driving data was not
obtained from the TomTom data. Much of the TomTom data that does not "snap to the roadway
grid" is caused by loss of satellite signal and errors while the TomTom unit is trying to acquire
the satellite signal. Therefore, a difficult analysis would be required to separate real off-network
data from GPS error data, and even if the analysis could be done, the reliability of the results
would probably be unknown. As such, only data that was associated with the roadway grid was
used in the analysis.
Table 9-3 shows the method for using the internal TomTom data (Columns E through I) to
produce the desired output, which ERG used to produce the MOVES2014 tables. The example in
the table uses 16 observations that might have been recorded on two urban unrestricted roadway
segments (Column E) during TomTom personal navigation device use between 14:00:00 and
14:59:59 on a weekday in April 2011. Column F is an internal ID (1-5 occur on Segment A, and
11-21 occur on Segment B). Column G gives the length of the segment. Column H gives the
time that the device spent on the segment. Column I gives the average speed of the device on the
segment. The 16 observations are sorted by the average speed bin, which is given in Column J.
The total distance traveled and the total time spent in each combination of road type, month,
weekday/weekend, hour of the day, and average speed bin are given in Columns K and L.
TomTom provided Columns A, B, C, D, J, K, and L to ERG. The data in those columns was
purchased by ERG from TomTom and is provided under license terms that permit free
distribution to EPA and the public. The raw data in Columns E, F, G, H, and I were not provided
to ERG and the US EPA.
1 http://www.census.gov/geo/www/ua/2010urbanruralclass.html
-------
Table 9-3 Example of accumulating total distance and total time for the TomTom deliverable table
A
Road Type
(4 levels)
B
Month
(12
levels)
C
Weekday/
Weekend
(2 levels)
D
Hour
of the
Day
(24
levels)
E
Segment
F
Data
Point
G
Segment
Length
(feet)
H
Time
in
Segment
(s)
I
Average
Speed
in
Segment
(mph)
J
Average
Speed
Bin
(mph)
(20
levels)
K
Total of
Segment
Lengths
for this
Speed
Bin
(feet)
L
Total of
Segment
Times
for this
Speed
Bin (s)
Urban
Unrestricted
April
Weekday
14:00:00
to
14:59:59
A
B
A
B
B
B
B
B
B
A
A
A
B
B
B
B
5
16
1
11
12
15
18
20
21
2
3
4
13
14
19
17
300
250
300
250
250
250
250
250
250
300
300
300
250
250
250
250
15
12
10
8
9
8
8
9
8
9
8
9
7
7
7
6
13.64
14.20
20.45
21.31
18.94
21.31
21.31
18.94
21.31
22.73
25.57
22.73
24.35
24.35
24.35
28.41
15
20
25
30
550
1800
1650
250
27
60
47
6
Using the table delivered by TomTom, ERG calculated the time-based average speed distribution
for each road type, day, and hour of the day using the average speed bin (Column J) and the total
of segment times (Column L)j. ERG calculated the average speed distribution according to the
16 speed bins used in MOVES. Figure 9-1 plots the average speed distribution for one hour
(5pm) stored in the averageSpeedDistribution table in MOVES, which contains average speed
distributions for each hour of the day (24 hours). We are using the TomTom data to represent
national default average speed distribution in MOVES.
J MOVES uses time-based speed because the emission rates are time-based (e.g. gram/hour).
70
-------
Figure 9-1 Average speed distribution for 5pm (hourlD 17) for source types (11 through 54) stored in the
AvgSpeedDistribution table in MOVES2014
Dav
Weekday
Weekend
40 60 SO
Average Bin Speed
9.2 Heavy-Duty Average Speed Distributions
It has been shown that combination trucks travel at approximately 92 percent of the speed of
light-duty vehicles on restricted access roads53. Since the TomTom data was developed from
light-duty vehicles, the average speed distribution for both short-haul and long-haul combination
trucks was adjusted on rural and urban restricted road types to have an 8% lower average speed
than the respective TomTom average speed for light-duty vehicles. The equations and
assumptions used to adjust the combination truck average speed distributions are located in
-------
Section 22 (Appendix F: Calculation of Combination Truck Average Speed Distributions).
Figure 9-2 illustrates the results of this analysis.
Figure 9-2 Average weekday speed distribution for 5pm (hourlD 17) by source type stored in the
AvgSpeedDistribution table in MOVES
o
•43
'_>
PH
—
D
cx
^
I
Source
All other source types
Combination trucks
20 40 60
Average Bin Speed
72
-------
In the absence of additional data, all other heavy-duty vehicles (including single unit heavy-duty
vehicles) and all heavy-duty vehicles operating on unrestricted access roads, use the same
average speed distributions as light-duty vehicles. We recognize that these assumptions are less
than ideal, and we hope to update the heavy-duty average speed distributions using heavy-duty
data in the future. Nonetheless, MOVES energy consumption and emission estimates from
heavy-duty appear to be only moderately sensitive to changes in the average speed distribution.
The 8% speed decrease in average speed distribution on restricted access roadways for
combination trucks caused the total onroad predicted NOx emissions to decrease by only -0.5%
and the national onroad diesel fuel consumption to decrease by only -1.3%. Other researchers54
have found that other local inputs are significantly more important for emissions inventories than
average speed distributions, including population, age distribution, and the combination truck
fraction of heavy-duty VMT. Nonetheless, we strongly encourage MOVES users to use local
average speed distributions when using MOVES at the regional and county-level.
73
-------
10 Driving Schedules and Ramps
Drive schedule refers to a second-by-second vehicle speed trajectory. A drive schedule typically
includes all vehicle operation from the time the engine starts until the engine is keyed off, both
driving (travel) and idling time. Drive schedules are used in MOVES to determine the operating
mode distribution for most MOVES running processes for calculation of emissions and energy
consumption.
In brief, there is an emission rate (in grams per hour of vehicle operation) for each operating
mode of vehicle operation. 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), or scaled tractive
power (STP) for heavy-duty vehicles, is calculated from the driving schedules. This distinction
between VSP and STP is discussed in Section 14. 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.
10.1 Driving Schedules
A key feature of MOVES is the capability to accommodate a number of drive schedules to
represent driving patterns across source type, road type, and average speed. For the national
default case, MOVES2014 employs 49 drive schedules with various average speeds, mapped to
specific source types and road types.
MOVES stores all of the drive schedule information in four database tables. DriveSchedule
provides the drive schedule name, identification number, and the average speed of the drive
schedule. DriveScheduleSecond 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.
DriveScheduleAssoc defines the set of schedules which are available for each combination of
source use type and road type. Ramps use operating mode distributions directly and do not use
drive schedules to calculate operating modes. The RoadOpModeDistribution table lists
operating mode distributions used for ramps for each source use type, road type and speed bin,
discussed in further detail later in this section.
Table 10-1 through Table 10-6 MOVES driving cycles for combination trucks (61, 62) below list
the driving schedules used in MOVES2014. Some driving schedules are used for both restricted
access (freeway) and unrestricted access (non-freeway) driving. In most cases, these represent
atypical conditions, such as extreme congestion or unimpeded high speeds. In these conditions,
we assume that the road type itself has little impact on the expected driving behavior (driving
schedule). Normally, these conditions represent only a small portion of overall driving.
Similarly, some driving schedules are used for multiple source types where vehicle specific
information was not available.
74
-------
In the past, if there was no appropriate driving schedule to use for modeling an average speed
bin, MOVES would use the nearest schedule. MOVES2014 requires driving schedules that can
be used as the upper bound and the lower bound for all average speed bins. New default driving
schedules have been added to assure that all average speed bins have appropriate driving
schedules for all the MOVES average speed bins.
Table 10-1 MOVES driving cycles for motorcycles, passenger cars, passenger trucks, and light commercial
trucks (11,21,31,32)
ID
101
1033
1043
1041
1021
1030
153
1029
1026
1020
1011
1025
1019
1024
1018
1017
1009
158
Cycle Name
LD Low Speed 1
Final FC14LOSF
Final FC19LOS AC
Final FC17LOSD
Final FC11LOSF
Final FC14LOSC
LD LOS E Freeway
Final FC14LOSB
Final FC12LOSE
Final FC11LOSE
Final FC02LOSDF
Final FC12LOSD
Final FC11LOSD
Final FC12LOSC
Final FC11LOSC
Final FC11LOSB
Final FC01LOSAF
LD High Speed Freeway 3
Average
Speed
2.5
8.7
15.7
18.6
20.6
25.4
30.5
31.0
43.3
46.1
49.1
52.8
58.8
63.7
64.4
66.4
73.8
76.0
Unrestricted Access
Rural
X
X
X
X
X
X
X
X
Urban
X
X
X
X
X
X
X
X
X
Restricted access
Rural
X
X
X
X
X
X
X
X
X
X
X
Urban
X
X
X
X
X
X
X
X
X
X
X
75
-------
Table 10-2 MOVES driving cycles for intercity buses (41)
ID
398
404
201
405
202
203
204
205
206
251
252
253
254
255
397
Cycle Name
CRC E55 HHDDT Creep
New York City Bus
MD 5 mph Non-Freeway
WMATA Transit Bus
MD lOmph Non-Freeway
MD 15mph Non-Freeway
MD 20mph Non-Freeway
MD 25mph Non-Freeway
MD 3 Omph Non-Freeway
MD 3 Omph Freeway
MD 40mph Freeway
MD 50mph Freeway
MD 60mph Freeway
MD High Speed Freeway
MD High Speed Freeway Plus 5 mph
Average
Speed
1.8
3.7
4.6
8.3
10.7
15.6
20.8
24.5
31.5
34.4
44.5
55.4
60.4
72.8
77.8
Unrestricted access
Rural
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
Urban
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
Restricted access
Rural
X
X
X
X
X
X
X
X
X
X
X
X
X
Urban
X
X
X
X
X
X
X
X
X
X
X
X
X
Table 10-3 MOVES driving cycles for transit and school buses (42,43)
ID
398
201
404
202
405
401
203
204
205
402
206
251
252
403
253
254
255
397
Cycle Name
CRC E55 HHDDT Creep
MD 5 mph Non-Freeway
New York City Bus
MD 1 Omph Non-Freeway
WMATA Transit Bus
Bus Low Speed Urban*
MD 15mph Non-Freeway
MD 20mph Non-Freeway
MD 25mph Non-Freeway
Bus 30 mph Flow*
MD 3 Omph Non-Freeway
MD 3 Omph Freeway
MD 40mph Freeway
Bus 45 mph Flow*
MD 50mph Freeway
MD 60mph Freeway
MD High Speed Freeway
MD High Speed Freeway Plus 5 mph
Average
Speed
1.8
4.6
3.7
10.7
8.3
15
15.6
20.8
24.5
30
31.5
34.4
44.5
45
55.4
60.4
72.8
77.8
Unrestricted access
Rural
X
X
X
X
X
X
X
X
X
X
Urban
X
X
X
X
X
X
X
X
X
X
Restricted access
Rural
X
X
X
X
X
X
X
X
X
X
X
X
X
Urban
X
X
X
X
X
X
X
X
X
X
X
X
X
* To be consistent with the speed distributions described in Section 9, this speed represents the average
for the traffic the bus is traveling in, not the average speed of the bus, which is lower due to stops.
-------
Table 10-4 MOVES driving cycles for refuse trucks (51)
ID
398
501
301
302
303
304
305
306
351
352
353
354
355
396
Cycle Name
CRC E55 HHDDT Creep
Refuse Truck Urban
HD 5 mph Non-Free way
HD lOmph Non-Freeway
HD 15mph Non-Freeway
HD 20mph Non-Freeway
HD 25mph Non-Freeway
HD 3 Omph Non-Freeway
HD 3 Omph Freeway
HD 40mph Freeway
HD 50mph Freeway
HD 60mph Freeway
HD High Speed Freeway
HD High Speed Freeway Plus 5 mph
Average
Speed
1.8
2.2
5.8
11.2
15.6
19.4
25.6
32.5
34.3
47.1
54.2
59.4
71.7
77.8
Unrestricted access
Rural
X
X
X
X
X
X
X
X
X
X
X
X
Urban
X
X
X
X
X
X
X
X
X
X
X
X
Restricted access
Rural
X
X
X
X
X
X
X
X
X
X
X
X
X
Urban
X
X
X
X
X
X
X
X
X
X
X
X
X
Table 10-5 MOVES driving cycles for single unit trucks and motor homes (52,53,54)
ID
398
201
202
203
204
205
206
251
252
253
254
255
397
Cycle Name
CRC E55 HHDDT Creep
MD 5 mph Non-Freeway
MD 1 Omph Non-Freeway
MD 15mph Non-Freeway
MD 20mph Non-Freeway
MD 25mph Non-Freeway
MD 3 Omph Non-Freeway
MD 3 Omph Freeway
MD 40mph Freeway
MD 50mph Freeway
MD 60mph Freeway
MD High Speed Freeway
MD High Speed Freeway Plus 5 mph
Average
Speed
1.8
4.6
10.7
15.6
20.8
24.5
31.5
34.4
44.5
55.4
60.4
72.8
77.8
Unrestricted access
Rural
X
X
X
X
X
X
X
X
X
X
X
X
X
Urban
X
X
X
X
X
X
X
X
X
X
X
X
X
Restricted access
Rural
X
X
X
X
X
X
X
X
X
X
X
X
X
Urban
X
X
X
X
X
X
X
X
X
X
X
X
X
77
-------
Table 10-6 MOVES driving cycles for combination trucks (61,62)
ID
398
301
302
303
304
305
306
351
352
353
354
355
396
Cycle Name
CRC E55 HHDDT Creep
HD 5 mph Non-Free way
HD lOmph Non-Freeway
HD 15mph Non-Freeway
HD 20mph Non-Freeway
HD 25mph Non-Freeway
HD 3 Omph Non-Freeway
HD 3 Omph Freeway
HD 40mph Freeway
HD 50mph Freeway
HD 60mph Freeway
HD High Speed Freeway
HD High Speed Freeway Plus 5 mph
Average
Speed
1.8
5.8
11.2
15.6
19.4
25.6
32.5
34.3
47.1
54.2
59.4
71.7
77.8
Unrestricted access
Rural
X
X
X
X
X
X
X
X
X
X
X
X
X
Urban
X
X
X
X
X
X
X
X
X
X
X
X
X
Restricted access
Rural
X
X
X
X
X
X
X
X
X
X
X
X
X
Urban
X
X
X
X
X
X
X
X
X
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.001.55 "LD Low Speed 1" is a historic cycle
used in the development of speed corrections for MOBILES 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,56 with an average speed of 76 mph and a maximum speed of 90
mph. Fifteen new light-duty "final" cycles were developed by a contractor for MOVES based on
urban and rural data collected in California in 2000 and 2004.42 The new cycles were selected to
best cover the range of road types and average speeds modeled in MOVES.
Most of the driving schedules used for buses are borrowed directly from driving schedules used
for single unit trucks (described below). The "New York City Bus"57 and "WMATA Transit
Bus"58 drive schedules are included for urban driving that includes transit type bus driving
behavior. The "CRC E55 HHDDT Creep" 59 cycle was included to cover extremely low speeds
for heavy-duty trucks. The "Bus 30 mph Flow" and "Bus 45 mph Flow" cycles used for transit
and school buses were developed by EPA based on Ann Arbor Transit Authority buses
instrumented in Ann Arbor, Michigan.60 The bus "flow" cycles were developed using selected
non-contiguous snippets of driving from one stop to the next stop, including idle, to create cycles
with the desired average driving speeds. The bus "flow" cycles have a nominal speed used for
selecting the driving cycles that does not include the idle time and only considers the free-flow
speed between stops. The actual average speed of the cycle (including stops) are shown in
Section 23 (Appendix G: Driving Schedules). Note that the "Bus Low Speed Urban" bus cycle is
the last 450 seconds of the standard New York City Bus cycle.
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. The CRC E55 HHDDT Creep cycle was used
instead for restricted access driving of refuse trucks at extremely low speeds. All of the other
-------
driving cycles used for refuse trucks were borrowed from driving cycles developed for heavy-
duty combination trucks, described below.
Single unit and combination trucks use driving cycles developed specifically for MOVES, based
on work performed for EPA by Eastern Research Group (ERG), Inc. and documented in the
report "Roadway-Specific Driving Schedules for Heavy-Duty Vehicles."61 ERG analyzed data
from 150 medium and heavy-duty vehicles instrumented to gather instantaneous speed and GPS
measurements. ERG segregated the driving into restricted access and unrestricted access driving
for medium and heavy-duty vehicles, and then further stratified 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 schedules developed by ERG 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
available from the ERG analysis 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 new driving cycles (396 and 397) using simulations with the
EPA's Greenhouse Gas Emissions Model (GEM)62 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 were able to follow the
speed demands of the proposed driving cycles without exceeding maximum torque or power.
None of the driving schedules used to represent restricted access (freeway) driving contain
vehicle operation on entrance or exit ramps. The effect of ramp operation is added separately in
MOVES.
10.2 Ramp Activity
Ramp activity is the driving behavior of vehicles that occurs on entrance and exit ramps as
vehicles enter or leave restricted access roads. It includes all of the activity between operation
on the unrestricted road and operation on the restricted road.
The driving schedules used to represent restricted access (freeway) driving in MOVES2014 are
not intended to represent vehicle operation on entrance or exit ramps. Activity that occurs on the
79
-------
freeway in anticipation of ramps or occurring after entry (including activity on "weaving lanes")
is included in the non-ramp freeway driving schedules. The effect of ramp operation is calculated
separately. Instead of using driving schedules to generate operating mode distributions for
ramps, each average speed bin has an associated operating mode distribution that reflects the
power demand expected from ramp operation associated with each nominal average highway
speed for each of the source types. The operating mode distributions used for ramps in
MOVES2014 were estimated to represent the driving connecting to and from a freeway with the
given average speed. These operating mode distributions (i.e. the fractions of time spent in each
of the operating modes for each source type on each road type at each average speed) can be
found in the in the default MOVES2014 database (RoadOpModeDistribution table).
Each set of ramp operating modes is associated with a corresponding average highway speed that
does not include ramp operation. Since operating modes for ramp emissions are affected by the
distribution of the average speed bins on the surrounding roads, the determination of average
speeds for restricted access roads (both urban and rural) should not include the time or distance
of vehicles on ramps. However, the VMT on ramps should be included with restricted access
VMT.
The emission impact of ramp activity is combined with the other driving activity found in the
restricted access (freeway) driving cycles using a ramp fraction. This fraction defines the
fraction of all time spent on a road that occurs on entrance and exit ramps. The fraction used (8
percent) in MOVES2014 is derived from the ramp fraction value developed originally for the
MOBILE6 model.63
80
-------
11 Hotelling Activity
MOVES2014 defines "hotelling" as any long period of time that drivers spend in their vehicles
during mandated down times during long distance deliveries by tractor/trailer combination
heavy-duty trucks. During the mandatory down time, drivers can stay in motels or other
accommodations, but most of these trucks have sleeping spaces built into the cab of the truck and
drivers stay with their vehicles. Hotelling hours are included in MOVES2014 in order to account
for use of the truck engine (referred to as "extended idling") to power air conditioning, heat, and
other accessories and account for the use of auxiliary power units (APU), which are small on-
board power generators.
In MOVES2014, only the long-haul combination truck source use type (sourceTypelD 62) is
assumed to have any hotelling activity. All of the long-haul combination trucks are diesel
fueled. All source use types other than long-haul combination trucks have hotelling activity
fractions set to zero.
11.1 National Default Hotelling Rate
Federal law limits long-haul truck drivers to ten hours driving followed by a mandatory eight
hour rest period. These regulations are described in the Federal Register.64 In long-haul
operation, drivers will stop periodically along their routes. For MOVES, the total hours of
hotelling are estimated by using the national estimate of VMT by long-haul combination trucks
divided an estimated average speed to calculate total hours of driving. The total hours of driving
divided by ten gives the number of eight-hour rest periods needed and thus the national total
hotelling hours.
A method is needed to allocate these total hotelling hours to locations. For MOVES2014, we
decided to determine a "hotelling rate" (hours of hotelling per mile of travel) that could be used,
in combination with VMT information to allocate the hotelling hours, described in Equation 12
to Equation 15. We calculate a hotelling rate as the national total hours of hotelling divided by
the national total miles driven by long-haul trucks on rural restricted access (freeways) roads.
Driving time on all roads contributes to the total hotelling hours calculation. However, most
locations used for hotelling are located near the roadways (restricted access) most traveled by
long-haul trucks. In order to prevent large amounts of hotelling to be allocated to congested
urban areas, we decided to only use the VMT on rural restricted roads as the surrogate for
allocating the total hotelling hours.
The hotelling rate (hotelling hours per mile of rural restricted access travel by long-haul
combination trucks) is applied to the estimate of rural restricted access VMT by long-haul
combination trucks to estimate the default hotelling hours for any location, month or day. The
allocation of hotelling to specific hours of the day is described below in Section 12.5.
The MOVES2014 default hotelling rate was calculated using default national total VMT
estimates for calendar year 2011 shown in Table 11-1.
81
-------
Total Hours =
Total Vehicle Miles Traveled
Average Speed
Equation 12
Total Trips =
Total Hours
10 hours per trip
Equation 13
Hotelling Hours = Total Trips * 8 hours per trip
Equation 14
Hotelling Rate =
Hotelling Hours
Total Rural Restricted Miles Traveled
Equation 15
Where:
Total Hours is the calculated time long-haul combination trucks spend driving.
Total Vehicle Miles Traveled is the total miles traveled by diesel long-haul
combination trucks in the nation in calendar year 2011 on all road types taken
from MOVES defaults.
Average Speed is an estimate of the average speed (distance divided by time) for
diesel long-haul combination trucks on all road types while operating.
Total Trips is the calculated number of trips by long-haul combination trucks.
Hotelling Hours is the calculated amount of rest time for long-haul combination
trucks.
Rural Restricted Miles is the total miles traveled by diesel long-haul combination
trucks on only rural restricted access roads (freeways) in calendar year 2011 using
MOVES defaults.
Table 11-1 Calculation of hotelling hours from long-haul combination truck VMT
Description
Rural Restricted
Rural Unrestricted
Urban Restricted
Urban Unrestricted
Total annual VMT
Hours (58.3 mph)
Trips (10 hrs per trip)
Hotelling hours (8 hrs per trip)
Hotelling hours per mile on rural restricted roads
Annual Value
31,392,300,000
34,301,700,000
32,243,100,000
28,848,900,000
126,786,000,000
2,174,716,981
217,471,698
1,739,773,585
0.055414
units
miles
miles
miles
miles
miles
hours
trips
hours
hours/mile
82
-------
For the MOVES default, all hotelling activity is assumed to occur in counties with travel on rural
restricted access roads, and thus will occur primarily in rural areas of states.
The national rate of hotelling hours per mile of rural restricted access roadway VMT is stored in
the HotellingCalendarYear table for each calendar year. The same value calculated for 201 1 is
used as the default for all calendar years. The County Data Manager includes the
HotellingActivityDistribution table which provides the opportunity for states 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.
11.2 Hotelling Activity Distribution
Hotelling differs from simple parking. In MOVES, hotelling hours are divided into operating
modes which define the emissions associated with the type of hotelling activity. Long-haul
trucks are often equipped with sleeping berths and other amenities to make the drive rest periods
more comfortable. These amenities require power for operation. This power 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) to allow use of amenities without running either the truck engines or APUs.
Some of rest time may occur without use of amenities at all. Table 11-2 shows the hotelling
operating modes used in MOVES.
Table 11-2 Hotelling activity operating modes
OpModelD
200
201
203
204
Description
Extended Idling of Main Engine
Hotelling Diesel Auxiliary Power Unit (APU)
Hotelling Battery or AC (plug in)
Hotelling All Engines and Accessories Off
The HotellingActivityDistribution table (see Table 11-3 below) contains the MOVES default
values for the distribution of hotelling activity to the operating modes.
Table 11-3 Default hotelling activity distributions
beginModelYearlD
1960
1960
1960
1960
2010
2010
2010
2010
endModelYearlD
2009
2009
2009
2009
2050
2050
2050
2050
opModelD
200
201
203
204
200
201
203
204
opModeFraction
1
0
0
0
0.7
0.3
0
0
All of the hotelling hours for long-haul trucks of model years before 2010 are assumed to use
extended idle to power accessories. Starting with the 2010 model year, the trucks are assumed to
-------
use extended idle 70 percent of the time and use APUs 30 percent of the time based on EPA's
assessment of technologies used by tractor manufacturers to comply with the Heavy-Duty
Greenhouse Gas standards.
-------
12 Temporal Distributions
MOVES is designed to estimate emissions for every hour of every day type in every month of
the year. The vehicle miles traveled (VMT) are provided for MOVES2014 in terms of annual
miles. These miles are allocated to months, days and hours using allocation factors, either
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 (OHEVI).65 The report describes analysis of a
sample of 5,000 continuous traffic counters distributed throughout the United States. EPA
obtained the data used in the report and used it to generate the VMT temporal distribution inputs
in the form needed for MOVES2014.
The OHEVI 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 source hours operating (SHO) calculated by MOVES from the
VMT and speed distributions 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 12.5.
The temporal distribution for engine start and corresponding engine soak (parked) distributions
are calculated from vehicle activity data stored in the SampleVehicleDay and
SampleVehicleTrip tables of the MOVES database, shown below in Table 12-1. These tables
contain a set of vehicle trip activity information constructed to represent activity for each source
type. Evaporative emissions are also affected by the time of day and the duration of parking.
Some of the vehicles in the tables take no trips.
Table 12-1 SampleVehicleDay table
Source Type
sourceTypelD
11
21
31
32
41
42
43
51
52
53
54
61
62
Description
Motorcycle
Passenger Car
Passenger Truck
Light Commercial Truck
Intercity Bus
Transit Bus
School Bus
Refuse Truck
Single Unit Short-Haul Truck
Single Unit Long-Haul Truck
Motor Home
Combination Short-Haul Truck
Combination Long-Haul Truck
Number of Records
Weekday (daylD 5)
2214
821
834
773
190
110
136
205
112
123
5431
130
122
Weekend (daylD 2)
983
347
371
345
73
14
59
65
58
50
2170
52
49
85
-------
12.1 VMT Distribution by Month of the Year
In MOVES, VMT is entered as an annual value and allocated to month using the
MonthVMTFraction table. For MOVES, we use the data from the OHIM report, Figure 2.2.1
"Travel by Month, 1970-1995," but modified to fit MOVES specifications. The table shows
VMT/day taken from the OHIM report, normalized to one for January. For MOVES, we need the
fraction of total annual VMT in each month. The report values of VMT per day were used to
calculate the VMT in a month using the number of days in each month. The calculations in
Table 12-2 assume a non-leap year (365 days).
Table 12-2 Month VMTFraction
Month
January
February
March
April
May
June
July
August
September
October
November
December
Sum
Normalized
VMT/day
1.0000
1.0560
1.1183
1.1636
1.1973
1.2480
1.2632
1.2784
1.1973
1.1838
1.1343
1.0975
MOVES
Distribution
0.0731
0.0697
0.0817
0.0823
0.0875
0.0883
0.0923
0.0934
0.0847
0.0865
0.0802
0.0802
1.0000
FHWA does not report monthly VMT information by vehicle classification. But it is clear that
in many regions of the United States, motorcycles are driven much less frequently in the winter
months. For MOVES2014 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 2010.66 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 12-3 for motorcycles are only a national average and do not
reflect the strong regional differences that would be expected due to climate.
86
-------
Table 12-3 MonthVMTFraction for motorcycles
Month
January
February
March
April
May
June
July
August
September
October
November
December
Sum
Month ID
1
2
3
4
5
6
7
8
9
10
11
12
Distribution
0.0262
0.0237
0.0583
0.1007
0.1194
0.1269
0.1333
0.1349
0.1132
0.0950
0.0442
0.0242
1.0000
12.2 VMT Distribution by Type of Day
The DayVMTFraction distribution divides the weekly VMT into two 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 source type. MOVES uses the 1995 data displayed in Figure
2.3.2 of the OHIM report.
The DayVMTFraction needed for MOVES has only two categories; week days (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 12-4. The OHIM report explains that data for
"Sam" refers to data collected from Sam to 4am. Thus data labeled "midnight" belongs to and
was summed with the upcoming day.
Table 12-4 DayVMTFractions
Fraction
Weekday
Weekend
Sum
Rural
0.72118
0.27882
1.00000
Urban
0762365
0.237635
1.000000
We assigned the "rural" fractions to the rural road types and the "urban" fractions to the urban
road types. 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.
-------
12.3 VMT Distribution by Hour of the Day
HourVMTFraction uses the same data as for Day VMTFraction. 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 Day VMTFraction. 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 12-5 and Figure 12-1 summarize these default values.
Table 12-5 MOVES distribution of VMT by hour of the day
hourlD
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
Description
Hour beginning at 12:00 midnight
Hour beginning at 1 :00 AM
Hour beginning at 2:00 AM
Hour beginning at 3 :00 AM
Hour beginning at 4:00 AM
Hour beginning at 5:00 AM
Hour beginning at 6:00 AM
Hour beginning at 7:00 AM
Hour beginning at 8:00 AM
Hour beginning at 9:00 AM
Hour beginning at 10:00 AM
Hour beginning at 1 1 :00 AM
Hour beginning at 12:00 Noon
Hour beginning at 1 :00 PM
Hour beginning at 2:00 PM
Hour beginning at 3 :00 PM
Hour beginning at 4:00 PM
Hour beginning at 5:00 PM
Hour beginning at 6:00 PM
Hour beginning at 7:00 PM
Hour beginning at 8:00 PM
Hour beginning at 9:00 PM
Hour beginning at 10:00 PM
Hour beginning at 1 1 :00 PM
Sum of All Fractions
Urban
Weekday
0.00986
0.00627
0.00506
0.00467
0.00699
0.01849
0.04596
0.06964
0.06083
0.05029
0.04994
0.05437
0.05765
0.05803
0.06226
0.07100
0.07697
0.07743
0.05978
0.04439
0.03545
0.03182
0.02494
0.01791
1.00000
Weekend
0.02147
0.01444
0.01097
0.00749
0.00684
0.01036
0.01843
0.02681
0.03639
0.04754
0.05747
0.06508
0.07132
0.07149
0.07172
0.07201
0.07115
0.06789
0.06177
0.05169
0.04287
0.03803
0.03221
0.02457
1.00000
Rural
Weekday
0.01077
0.00764
0.00655
0.00663
0.00954
0.02006
0.04103
0.05797
0.05347
0.05255
0.05506
0.05767
0.05914
0.06080
0.06530
0.07261
0.07738
0.07548
0.05871
0.04399
0.03573
0.03074
0.02385
0.01732
1.00000
Weekend
0.01642
0.01119
0.00854
0.00679
0.00722
0.01076
0.01768
0.02688
0.03866
0.05224
0.06317
0.06994
0.07293
0.07312
0.07362
0.07446
0.07422
0.07001
0.06140
0.05050
0.04121
0.03364
0.02622
0.01917
1.00000
88
-------
Figure 12-1 Hourly VMT fractions by day type and road type
0.09
Rural Weekend
Urban Weekend
Rural Weekday
Urban Weekday
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 of the Day
12.4 Engine Starts and Parking
To properly estimate engine start emissions and 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 earlier. We have made only minor
changes for MOVE2014.
The first table, SampleVehicleDay, lists a sample population of vehicles, each with an identifier
(vehID), an indication of vehicle type (sourceTypelD), and an indication (day ID) 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
recent study of vehicle activity in Georgia.67 This change is described in greater detail in the
report describing evaporative emissions in MOVES2014.68
The second table, SampleVehicleTrip, lists the trips in a day made by each of the vehicles in the
SampleVehicleDay table. It records the vehID, day ID, 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 on 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. Light-duty vehicle trip and soak data was
copied to all the other source types (11, 41, 42, 43, 51, 52, 53, 54, 61, and 62) for both weekdays
(daylD 5) and weekends (daylD 2) for hours with no trips.
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 driving at all.
The data and processing algorithms used to populate these tables are detailed in two contractor
reports.69'70 The data comes from a variety of instrumented vehicle studies, summarized in Table
12-6. This data was cleaned, adjusted, sampled and weighted to develop a distribution intended
to represent average urban vehicle activity.
Table 12-6 Source data for sample vehicle trip information
Study
3 -City FTP
Study
Minneapolis
Knoxville
Las Vegas
Battelle
TxDOT
Study Area
Atlanta, GA; Baltimore, MD;
Spokane, WA
Minneapolis/St. Paul, MN
Knoxville, TN
Las Vegas, NV
California, statewide
Houston, TX
Study
Years
1992
2004-
2005
2000-
2001
2004-
2005
1997-
1998
2002
Vehicle Types
Passenger cars & trucks
Passenger cars & trucks
Passenger cars & trucks
Passenger cars & trucks
Heavy-duty trucks
Diesel dump trucks
Vehicle
Count
321
133
377
350
120
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 MOBILE6 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 12-7.
90
-------
Table 12-7 Synthesis of sample vehicles for source types lacking data
Source Type
Motorcycles
Passenger Cars
Passenger Trucks
Light Commercial Trucks
Intercity Buses
Transit Buses
School Buses
Refuse Trucks
Single Unit Short-Haul Trucks
Single Unit Long-Haul Trucks
Motor Homes
Combination Short-Haul trucks
Combination Long-Haul trucks
Based on
Direct Data?
No
Yes
Yes
No
No
No
No
No
Yes
No
No
Yes
Yes
Synthesized From
Passenger Cars
n/a
n/a
Passenger Trucks
Combination Long-Haul Trucks
Single Unit Short-Haul Trucks
Single Unit Short-Haul Trucks
Combination Short-Haul Trucks
n/a
Combination Long-Haul Trucks
Passenger Cars
n/a
n/a
The resulting trip-per-day estimates are summarized in Table 12-8. The same estimate for trips
per day is used for all ages of vehicles in any calendar year.
Table 12-8 Starts i
Source Type
Motorcycles
Passenger Cars
Passenger Trucks
Light Commercial Trucks
Intercity Buses
Transit Buses
School Buses
Refuse Trucks
Single Unit Short-Haul Trucks
Single Unit Long-Haul Trucks
Motor Homes
Combination Short-Haul trucks
Combination Long-Haul trucks
per day by source type
MOVES2014
Weekday
0.78
5.89
5.80
6.05
2.77
4.58
5.75
3.75
6.99
4.29
0.57
5.93
4.29
MOVES2014
Weekend
0.79
5.30
5.06
5.47
0.88
3.46
1.26
0.92
1.28
1.29
0.57
1.16
1.29
MOVES2014 now has inputs in the County Data Manager that 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 inputs override the default values provided by MOVES.
The same trip information that is used to determine the number of engine starts is also used to
determine the vehicle soak time. "Soak time" is the time between trips when the engine is off.
The soak times are used to estimate the activity in each of the operating modes for engine start
emissions. The base emission rate for engine starts is based on a 12-hour soak period. All
engine soaks greater than 12 hours assume the same engine start emission rate as for 12 hours.
However, for all engine soaks less than 12 hours, the base engine start emission rate is adjusted
based on soak time bins (operating modes).68 The distribution of operating modes in each hour of
the day is part of the calculation used to determine the engine start emissions for that hour of the
day.
-------
A more complete discussion of the relationship between engine soak time and emissions will be
found in the MOVES report covering engine start emission rates used in MOVES.4'5
12.5 Hourly Hotelling Activity
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 Vehicles61 combines data from several instrumented
truck studies and contains detailed information about truck driver behavior. While none of the
trucks 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 12-9
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.
Table 12-9 Hourly distribution of truck trips used to calculate hotelling hours
hourlD
1
2
o
J
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
Hour of the Day
Hour beginning at 12:00 midnight
Hour beginning at 1:00 AM
Hour beginning at 2:00 AM
Hour beginning at 3 : 00 AM
Hour beginning at 4:00 AM
Hour beginning at 5:00 AM
Hour beginning at 6:00 AM
Hour beginning at 7:00 AM
Hour beginning at 8:00 AM
Hour beginning at 9:00 AM
Hour beginning at 10:00 AM
Hour beginning at 1 1 :00 AM
Hour beginning at 12:00 Noon
Hour beginning at 1:00 PM
Hour beginning at 2:00 PM
Hour beginning at 3 : 00 PM
Hour beginning at 4:00 PM
Hour beginning at 5:00 PM
Hour beginning at 6:00 PM
Hour beginning at 7:00 PM
Hour beginning at 8:00 PM
Hour beginning at 9:00 PM
Hour beginning at 10:00 PM
Hour beginning at 1 1 :00 PM
Trip Starts
78
76
65
94
107
131
194
230
279
267
275
240
201
211
171
167
144
98
71
73
71
52
85
48
Trip Ends
171
167
144
98
71
73
71
52
85
48
78
76
65
94
107
131
194
230
279
267
275
240
201
211
92
-------
An estimate of the distribution of truck hotelling duration times is derived from a 2004 CRC
paper71 based on a survey of 365 truck drivers at six different locations. Table 12-10 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 12-10 Distribution of truck hotelling activity duration
Hotelling Duration
(hours)
2
4
6
8
10
12
14
16
Total
Fraction of Trucks
0.227
0.135
0.199
0.191
0.156
0.057
0.014
0.021
1.000
We assume that all hotelling activity begins at the trip ends shown in Table 12-9. But not all trip
ends have the same number of hotelling hours. The distribution of hotelling durations from
Table 12-10 is applied to the hotelling that occurs at each of these trip ends.
Table 12-11 illustrates the hotel 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 day 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 hour ID
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 12-10 is applied to the hour estimate in
each hotelling hours bin to calculate the weighted total idle hours for each hour of the day.
93
-------
Table 12-11 Calculation of hourly distributions of hotelling activity
hourlD
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
Totals
Weight
Trip
Starts
78
76
65
94
107
131
194
230
279
267
275
240
201
211
171
167
144
98
71
73
71
52
85
48
3428
Trip
Ends*
171
167
144
98
71
73
71
52
85
48
78
76
65
94
107
131
194
230
279
267
275
240
201
211
3428
2
hours
255.4
235.4
210.8
155.6
110.2
101.4
100.2
80.4
105.8
82
97.2
107.2
95.4
120
144.6
173.8
246.4
307.6
371
378.6
381.8
350
297
291.4
4799
0.227
4
hours
679
629.4
566.4
477.4
379.8
299.6
254.2
224.4
237.2
213.4
231.8
236
238.2
266.2
296.4
358
469.6
597.8
755.4
853.6
913
893.6
822.8
762
11655
0.135
6
hours
1204.8
1100
990
871.4
735.4
621.4
523.8
422.6
391.2
357.4
363.2
367.4
372.8
395
439.2
504.2
621.4
782
978.6
1143.8
1297.4
1368.6
1354
1305.6
18511
0.199
8
hours
1736
1643.6
1515.8
1342
1159
1015.4
879.4
744.4
660.8
555.6
517.2
511.4
504.2
526.4
573.8
633
764.2
928.2
1130.4
1328
1520.6
1658.8
1738.4
1780.6
25367
0.191
10
hours
2120.4
2118.6
2047
1885.6
1684.8
1486
1303
1138.4
1016.4
877.4
786.8
709.6
658.2
670.4
705.2
764.4
898.8
1057
1273.2
1474.2
1672.4
1843
1961.6
2070.8
32223
0.156
12
hours
2343.6
2408.8
2431 .4
2360.6
2216
2029.6
1828.8
1609
1440
1271.4
1142.4
1031.4
927.8
868.6
859.2
908.4
1030.2
1188.4
1407.8
1603
1815.2
1989.2
2113.4
2255
39079
0.057
14
hours
2495.4
2593
2654.6
2650.8
2600.4
2504.6
2360
2152.6
1965.8
1742
1566
1425.4
1283.4
1190.4
1128.8
1106.6
1184.2
1332.4
1539.2
1734.4
1949.8
2118
2256.2
2401.2
45935
0.014
16
hours
2638.2
2739.2
2806.4
2835
2823.6
2794.8
2744.4
2627.6
2497
2285.6
2091.8
1896
1707
1584.4
1484.4
1428.4
1453.8
1530.6
1693.2
1878.4
2081.2
2249.4
2390.8
2530
52791
0.021
Weighted Total
Idle Hours
1276
1234
1166
1056
930
823
728
630
581
507
479
457
434
447
476
526
635
767
933
1068
1194
1268
1289
1308
20213
Distribution
0.0628
0.0611
0.0577
0.0526
0.0458
0.0407
0.0357
0.0306
0.0289
0.0255
0.0238
0.0221
0.0221
0.0221
0.0238
0.0255
0.0323
0.0374
0.0458
0.0526
0.0594
0.0628
0.0645
0.0645
1.0000
* Assumes every trip ends 10 hours after it starts, such that all trips are 10 hours long. The first hour of hotelling in each hour bin
column sum is reduced by 60 percent to account for trip ends in a column that are not a full hour.
94
-------
The distribution calculated using this method is similar to the behavior observed in a
dissertation72 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 use results from a single study at a specific
location, MOVES2014 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 of the default
MOVES2014 database.
MOVES2014 uses this same default hourly distribution from
Table 12-11 for all days and locations, as shown below in Figure 12-2. 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.
Figure 12-2 Truck hotelling distribution by hour of the day in MOVES2014
0%
12:00 AM
12:00 PM
Hours of Day
12:00 AM
12.6 Single and Multiday Diurnals
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 soak time calculations are discussed earlier in Section 12.4. 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.70
95
-------
13 Geographical Allocation of Activity
MOVES is designed to model activity at a "domain" level and then to allocate that activity to
"zones." The MOVES2014 default database is populated for a domain of the entire United States
(including Puerto Rico and the Virgin Islands), and the default zones correspond to individual
counties. The MOVES design only allows for one set of geographic allocations to be stored in
the default database. While geographic allocations clearly change over time, the MOVES2014
defaults were developed using the data from calendar year 2011, and are used for all calendar
years. For this reason, the MOVES default allocation of activity is rarely used for any official
purpose by either EPA or local areas. 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. The MOVES geographic
allocation factors are stored in two tables, Zone and ZoneRoadType.
13.1 Source Hours Operating Allocation to Zones
Most of the emission rate calculations in MOVES2014 are based on emission rates by time units
(hour). Using time units for emissions is the most flexible approach, since the activity for some
processes (like leaks and idling) and some source types (like nonroad generators) are more
naturally in units of time. As a result, MOVES converts activity data to hours in many cases in
order to produce the hours needed for emissions calculations.
The national total source hours of operation (SHO) are calculated from the estimates of VMT
and speed as described in sections above. This total VMT for each road type is allocated to
county using the SHOAllocFactor field in the ZoneRoadType table. The allocation factors are
derived using 2011 VMT and MOVES default VMT.
In particular, the MOVES2014 default estimates for the VMT by county come from Version 1 of
the 2011 National Emission Inventory (NEI) analysis.48 These estimates are based on the
Highway Performance Monitoring System (HPMS) state level data collected by the Federal
Highway Administration73 annually for use in transportation planning. The HPMS state level
VMT is distributed to the individual counties in each state as part of the NEI analysis. This data
is reviewed and updated by the states as necessary prior to use in the NEI. The default inputs for
SHO AllocF actor in MOVES2014 were calculated using the VMT estimates obtained from
Version 1 of the 2011 NEI74 for each county by road type.
Vehicle miles traveled can be converted to hours of travel using average speeds. The average
speed estimates were taken directly from the AvgSpeedDistribution table of the MOVES default
database. The default average speed distributions do not vary by county or source type, but do
vary by road type, day type (weekday and weekend day) and hour of the day. The 2011 NEI
VMT was aggregated into the four MOVES road types in each county. The VMT by road type
in each county was then allocated to day type and hour of the day using the day type and hour
-------
distributions from the MOVES default database tables, DayMVTFraction and
HourVMTFracti on.
Using the nominal speeds for each average speed bin in the AvgSpeedDistribution table for each
hour of each day type and the corresponding VMT, the hours of vehicle operation (SHO) can be
calculated for each hour of the day on each road type for each day type in each county. The
average speed distribution is in units of time, so the distribution must be converted to units of
distance to be applied to the VMT values. For this step, we multiplied each value of each
distribution (in terms of time) by the corresponding nominal average speed value for that average
speed bin to calculate distance (hours * miles/hour). Then we divided each distance value in the
distribution by the sum of all distance values in that distribution to calculate the average speed
distribution in terms of distance.
Finally, we multiplied the total VMT corresponding to each average speed distance distribution
(by road type, by day type, by hour of the day) by each of the values in the distribution to
calculate the VMT corresponding to each average speed bin. We then calculated operating hours
by dividing the VMT in each average speed bin by the corresponding nominal average speed
value, shown in Equation 16.
SHO = VMT (miles) / Speed (miles per hour) Equation 16
Once the hours of operation have been calculated, the hours in each county were summed by
road type. The allocation factor for each county in Equation 17 was calculated by dividing the
county hours for each road type by the national total hours of operation for each road type.
SHOAllocFactor = County SHO / National SHO Equation 17
The county allocation values for each roadway type sum to one (1.0) for the nation. 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.
13.2 Engine Start Allocations to Zones
The allocation of the domain-wide count of engine starts to zones is stored in the
StartAllocFactor in the Zone table. In the default database for MOVES2014, the domain is the
nation and the zones are counties. There is no national source for data on the number of trip
starts by county, so for MOVES2014, we have used VMT to determine this allocation. VMT for
each county was taken from the most recent National Emission Inventory analysis for calendar
year 20II.74
VMT estimates for each county in each state and the allocation is calculated using Equation 18,
where i represents each individual county and / is the set of all US counties.
97
-------
CountyAllocatiorij = CountyVMTj J CountyVMTj Equation 18
iel
The county allocation values sum to one (1.0) for the nation. The same StartAllocFactor set is
the default for all calendar years at the National scale. County- and Project-level calculations do
not use the default StartAllocF actor allocations at all. Instead, County and Project scales require
that the user input all local activity.
13.3 Parking Hours Allocation to Zones
The allocation of the domain-wide hours of parking (engine off) to zones is stored in the
SHPAllocFactor in the Zone table. In the default database for MOVES2014, the domain is the
nation and the zones are the counties. There is no national source for hours of parking by county,
so for MOVES2014, we have used the same VMT-based allocation as used for the allocation of
starts in the StartAllocFactor (see above).
The county allocation values for parking hours sum to one (1.0) for the nation. 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.
In MOVES2014, hotelling hours (including extended idling and auxiliary power unit usage) are
calculated from long-haul combination truck VMT in each location and does have its own
allocation factors.
-------
14 Vehicle Mass and Road Load Coefficients
The MOVES model calculates emissions using a weighted average of emisson rates by operating
mode. This level of detail is required for microscale modeling, which in MOVES is called
project level analysis. For running exhaust emissions, the operating modes are defined by either
vehicle specific power (VSP) or scaled tractive power (STP). Both VSP and STP are calculated
based on a vehicle's speed and acceleration but differ in how they are scaled (or normalized).
VSP is used for light-duty vehicles (source types 11 through 32) and STP is used for heavy-duty
vehicles (source types 41 through 62).
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 source type averaged over all ages. MOVES uses these to calculate VSP
and STP for each source type according to Equation 19 and Equation 20:
VSP = (—} • v + (—} • v2 + (—} • v3 + (a + g • sin 0) • v Equation 19
\M/ \M/ \M/
Av + Bv2 + Cv3 + M • v • (at + g • sinO}
STP = —— Equation 20
Jscale
where A, 5, 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/s ), v
j
is the instantaneous vehicle speed in m/s, a is the instantaneous vehicle acceleration in m/s ,
sin 9 is the (fractional) road grade, and fscaie is a scaling factor.
When mapping actual emissions data to VSP bins with Equation 19, the vehicle's measured
weight is used as the source mass factor. In contrast, when calculating average VSP distributions
for an entire source type with MOVES, the average source type mass is used instead. STP is
calculated with Equation 20, which is very similar to the VSP equation except the denominators
are different. In the case of VSP, the power is normalized by the mass of the vehicle (fscaie =
M). For heavy-duty vehicles using STP, fscaie depends on their regulatory class and is used to
bring the numerical range of tractive power into the same numerical range as the VSP values
when assigning operating modes. Class 40 trucks use fscaie = 2.06, which is equal to the mass of
source type 32 in metric tons. This is because operating modes for passenger trucks and light-
commercial trucks are assigned operating modes using VSP, and using a fixed mass factor of
2.06 essentially calculates VSP-based emission rates. Running operating modes for all the heavy-
duty source types (buses, single unit, and combination trucks) are assigned using STP with
fscaie =17.1, which is roughly equivalent to the average running weight in metric tons of all
heavy-duty vehicles. Additional discussion regarding VSP and STP are provided in the MOVES
light-duty4 and heavy-duty5 emission rate reports, respectively.
99
-------
In both cases, operating mode distributions are derived 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 10.1. The
following sections detail the derivation of values used in Equation 19 and Equation 20.
14.1. Source Mass and 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 a given source type, which
includes the weight of the vehicle, occupants, fuel, and payload (M in the equations above), and
the fixed mass factor represents the STP scaling factor (fscaie m the equations above).
While the source masses for light-duty were unchanged from MOVES2010b, all of the heavy-
duty source masses were updated with newer data. Please see Section 24 (Appendix H:
MOVES2010b Source Masses) for a discussion of the MOVES2010b source masses. The heavy-
duty source masses for 2014+ model year vehicles heavy-duty vehicles were first updated to
account for the 2014 Medium and Heavy-Duty Greenhouse Gase Rule as discussed in Section
14.2. Then the heavy-duty source masses were updated with 2011 Weigh-in-Motion (WEVI) data
made available through FHWA's Vehicle Travel Information System (VTRIS). These data are
available from FHWA by state, road type, and HPMS truck type (single unit or combination).
The average national mass by truck type was calculated by weighting the masses with VMT by
state and road type using FHWA's Highway Statistics VM-2 Table. These average values then
needed to be allocated from the HPMS truck classification to source types. This allocation was
performed using the percent difference between the average WEVI HPMS mass and the average
MOVES2010b HPMS mass.k The MOVES2010b average masses were calculated by weighting
the source type masses with the updated 2011 VMT. The percentage difference between the
average single unit truck mass in MOVES2010b and the WEVI data was then applied to the
source masses of short-haul single unit trucks, long-haul single unit trucks, refuse trucks, and
motor homes. Likewise, the percentage difference between the average combination truck mass
in MOVES2010b and the WEVI data was applied to the source masses of short-haul and long-
haul combination trucks, including the 2014+ model year groups. These differences are shown in
Table 14-1, and the resulting source type masses are presented in Table 14-4.
Table 14-1 Weigh-in-Motion (WIM) masses weighted by VMT
HPMS Category
Single Unit Trucks
Combination Trucks
Average Weight (Ibs)
20,107
52,907
Percent Change from
MOVES2010b
11.7%
-21.7%
14.2. Road Load Coefficients
The information available on road load coefficients varied by regulatory class. Motorcycle road
load coefficients, given in Equation 21 through Equation 23, were empiricially derived in
accordance with standard practice75'76:
k For the WIM analysis, we only compared to the MOVES2010b masses because the 2014 Medium and Heavy-Duty
Rule impact is not assumed to begin phase-in until 2014.
1OO
-------
A = 0.088 • M Equation 21
5 = 0 Equation 22
C = 0.00026 + 0.000194 • M Equation 23
For light-duty vehicles, the road load coefficients were calculated according to Equation 24
through Equation 26:77
_ 0-7457 ,Q^,TRIHp Equation 24
50 • 0.447 1RLHr@5omPh
_ 0.7457 .QW.TRJHp Equation 25
(50 • 0.447)2 1RLHr@5omPh
_ 0.7457 Equation 26
(50 • 0.447)3 1KLHr@50mph
In each of the above equations, the first factor is the appropriate unit conversion to allow A, B,
and C to be used in Equation 19 and Equation 20, 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)78 to
Equation 24 through Equation 26. While we expect light-duty road load coefficients to improve
over time due to the Light-Duty Greenhouse Gas Rule, the impact of these changes have been
directly incorporated into the emission and energy rates. 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.
For the heavier vehicles, no road load parameters were available in the MSOD. For these source
types, relationships of road load coefficent to vehicle mass came from a study done by V.A.
Petrushov,79 as shown in Table 14-2. These relationships are grouped by regulatory class; source
type values were determined by weighting the combination of MOVES2010b weight categories
that comprise the individual source types. The final SourceMass, FixedMassFactor and road load
coefficients for all source types are listed in Table 14-4.
101
-------
Table 14-2 Road load coefficients for heavy-duty trucks, buses, and motor homes for MY 1960-2013 vehicles
Coefficient
(kW-s\
A 1 1
\ m J
(kW-s2\
p 1 I
( -2 )
fkW-s3\
r 1 1
( -3 )
8500 to 14000 Ibs
(3.855 to 6.350
metric ton)
0.0996 -M
0
0.00289 +
5.22 X id-5 • M
14000 to 33000 Ibs
(6.350 to 14.968
metric ton)
0.0875 -M
0
0.00193 +
5.90xlQ-5-M
>33000 Ibs
(>14.968 metric ton)
0.0661 -M
0
0.00289 +
4.21 XlO-5 -M
Buses and Motor
Homes
0.0643 -M
0
0.0032 +
5.06xlQ-5-M
In MOVES2014, the vehicle mass and road load coefficient were updated for 2014 and later
model year heavy-duty vehicles to account for the 2014 Medium and Heavy-Duty Greenhouse
Gase Rule.80 Table 14-3 contains the combination long-haul tractor and vocational vehicle tire
rolling resistance, coefficient of drag, and weight reductions expected from the technologies
which could be used to meet the standards. The value in the table reflects a 400 pound mass
reduction. As discussed in the regulatory impact analysis for the final rulemaking, EPA used a
sales mix of 10 percent Class 7 low roof, 10 percent Class 7 high roof, 45 percent Class 8 low
roof, and 35 percent Class 8 high roof based on feedback from the manufacturers.
The values in the table reflect a modeling assumption that 8 percent of all tractors (19.7 percent
of short-haul tractors) would be considered vocational tractors and therefore will only be
required to meet the vocational vehicle standards and not show any aerodynamic or weight
improvement. The weight reduction applied to short-haul tractors is 321 pounds, which is
calculated from the 400 pound weight reduction assumed for non-vocational tractors, reduced by
19.7 percent. The tire rolling resistance reduction is assumed to be 5 percent based on the data
derived in the tire testing program conducted by EPA. Comparatively tire rolling resistance is
reduced by 9.6 percent for long-haul tractors and 7 percent for short-haul tractors while
aerodynamic drag is reduced 12.1 percent for long-haul tractors and 5.9 percent for short-haul
tractors in model year 2014 and later. Further details on the rule's assumptions about reductions
to source mass and road load coefficients in the SourceUseTypePhysics table and discussion of
incorporating the rule's energy reductions from engine technology improvements into MOVES
can be found in the MOVES2014 Heavy-Duty Emission Rate Report.5
1O2
-------
Table 14-3 Estimated reductions in rolling resistance and aerodynamic drag coefficients from HD GHG
Phase 1 Rule for model years 2014 and later
Truck Type
Combination long-haul
Combination short-haul
Vocational vehicles (Single
unit tracks, refuse tracks,
motor homes, buses, and
light commercial tracks)
Reduction In Tire Rolling
Resistance Coefficient From
Baseline
9.6%
7.0%
5.0%
Reduction In
Aerodynamic Drag
Coefficient From Baseline
12.1%
5.9%
0%
Weight
Reduction
(Ibs)
400
321
0
These changes are represented in MOVES2014 through new aerodynamic coefficients
and weights, and they primarily affect short- and long-haul combination truck source types
beginning in MY 2014. The average vehicle mass and road load coefficients are updated by
source type through the beginModelYearlD and endModelYearlD fields in the
SourceUseTypePhysics table.
Table 14-4 MOVES2014 SourceUseTypePhysics table
sourceTypelD
11
21
31
32
41
41
42
42
43
43
51
51
52
52
53
53
54
54
61
61
62
62
Begin
Model
Year
1960
1960
1960
1960
1960
2014
1960
2014
1960
2014
1960
2014
1960
2014
1960
2014
1960
2014
1960
2014
1960
2014
End
Model
Year
2050
2050
2050
2050
2013
2050
2013
2050
2013
2050
2013
2050
2013
2050
2013
2050
2013
2050
2013
2050
2013
2050
Rolling
Term A
(kW-s/m)
0.0251
0.1565
0.2211
0.2350
1.2952
1.2304
1.0944
1.0397
0.7467
0.7094
1.5835
1.5043
0.6279
0.5965
0.5573
0.5294
0.6899
0.6554
1.5382
1.4305
1.6304
1.4739
Rotating
TermB
(kW-s2/m2)
0
0.0020
0.0028
0.0030
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
Drag
TermC
(kW-s3/m3)
0.0003
0.0005
0.0007
0.0007
0.0037
0.0037
0.0036
0.0036
0.0022
0.0022
0.0036
0.0036
0.0016
0.0016
0.0015
0.0015
0.0021
0.0021
0.0040
0.0038
0.0042
0.0037
Source Mass
(metric tons)
0.2850
1.4788
1.8669
2.0598
19.5937
19.5937
16.5560
16.5560
9.0699
9.0699
23.1135
23.1135
8.5390
8.5390
6.9845
6.9845
7.5257
7.5257
22.9745
22.8289
24.6010
24.4196
Fixed Mass
Factor (metric
tons)
0.2850
1.4788
1.8669
2.0598
17.1
17.1
17.1
17.1
17.1
17.1
17.1
17.1
17.1
17.1
17.1
17.1
17.1
17.1
17.1
17.1
17.1
17.1
103
-------
15 Air Conditioning Activity Inputs
This report 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. More information on air conditioning effects is provided in the MOVES technical
report on adjustment factors.81
15.1 ACPenetrationFraction
The ACPenetrationFraction is a field in the SourceTypeModelYear table. Default values, by
source type and model year were taken from MOBILE6.82 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 the 1995 for cars and 1975-1995 for light trucks. Rates in
the first few years of available data are 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 was 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), summarized
in Table 15-1.
104
-------
Table 15-1 AC penetration fractions in MOVES2014
1972-and-earlier
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999+
Motorcycles
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
Passenger Cars
0.592
0.726
0.616
0.631
0.671
0.720
0.719
0.694
0.624
0.667
0.699
0.737
0.776
0.796
0.800
0.755
0.793
0.762
0.862
0.869
0.882
0.897
0.922
0.934
0.948
0.963
0.977
0.980
All Trucks and Buses
0.287
0.287
0.287
0.287
0.311
0.351
0.385
0.366
0.348
0.390
0.449
0.464
0.521
0.532
0.544
0.588
0.640
0.719
0.764
0.771
0.811
0.837
0.848
0.882
0.906
0.929
0.950
0.950
15.2 FunctioningACFraction
The FunctioningACFraction field in the SourceTypeAge table (see Table 15-2) indicates the
fraction of the air-conditioning equipped fleet with fully functional A/C systems, by source type
and vehicle age. A value of 1 means all systems are functional. This is used in the calculation 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 a consumer study 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.
105
-------
Table 15-2 FunctioningACFraction by age (all source types except motorcycles)
agelD
0
1
2
o
J
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
functioningACFraction
1
1
1
1
0.99
0.99
0.99
0.99
0.98
0.98
0.98
0.98
0.98
0.96
0.96
0.96
0.96
0.96
0.95
0.95
0.95
0.95
0.95
0.95
0.95
0.95
0.95
0.95
0.95
0.95
0.95
15.3 ACActivityTerms
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 inMOBILE6%2 They are based on
analysis of air conditioning usage data collected in Phoenix, Arizona, in 1994.
In MOVES, ACActivityTerms are allowed to vary by monthGroup and Hour, in order to provide
the possibility of different A/C 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, for MOVES2014, the default data uses one set of coefficients for all MonthGroups and
Hours. These default coefficients represent an average A/C activity demand function over the
course of a full day. The coefficients are listed in Table 15-3.
106
-------
Table 15-3 Air conditioning activity coefficients
A
-3.63154
B
0.072465
C
-0.000276
The A/C activity demand function that results from these coefficients is shown in Figure 15-1. A
value of 1 means the A/C compressor is engaged 100 percent of the time; a value of 0 means no
A/C compressor engagement.
Figure 15-1 Air conditioning activity demand as a function of heat index
0.9
0.8
I 0.7
£ 0.6
g
I 0.5
I" °-4
jj 0.3
0.2
0.1
70 75 80 85 90 95
Heat Index (F)
100
105
110
1O7
-------
16 Conclusion and 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 MOVES2014 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
emission 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, and the uncertainties and
variability in this local data can contribute significant uncertainty in resulting emission estimates.
Thus it is often appropriate to replace many of the MOVES fleet and activity defaults with local
data as explained in EPA's Technical Guidance.3
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 MOVES2014 were developed for a 2011 base year and much of the source data is from 2011
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.
Updating the vehicle fleet data will be complicated by the fact that one of the primary data
sources for this document, the Census Bureau's Vehicle Inventory and Use Survey, has been
discontinued. EPA is currently working with DOT and other federal agencies to revive this
survey. Doing so becomes more important as the data gathered from the last survey (2002) ages.
A related complication is the cost of data. 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.
In addition to these general limitations, there are also specific MOVES data elements that could
be improved with additional research, including:
• real-world highway driving cycles and operating mode distributions;
• off-network behavior including vehicle starts and soaks;
• truck hotelling, particularly extended engine idling, and APU use;
• idling while loading/unloading, in traffic queues (i.e. tolls), or elsewhere;
• VSP/STP adjustments for speed, road grade, and loading;
• activity changes with age, such as mileage accumulation rates, start activity, and soak
distributions;
• updated estimates of vehicle scrappage rates used to project vehicle age distributions;
• further incorporation of data from instrumented vehicle studies;
• summaries from large-scale instrumented vehicle studies;
108
-------
• vehicle identification and sorting by size, sector, and vocation;
• activity weighting of source mass averages;
• air conditioning system usage, penetration, and failure rates;
• vehicle type distinctions in temporal activity;
• heavy truck and bus daily trip activity patterns; and
• ramp activity and operating mode distributions.
We expect many of these MOVES data limitations can be addressed through analysis of data
captured on instrumented vehicles. The recent emergence and availability of large streams of
activity data from GPS devices, data loggers, and other onboard diagnostic systems will likely
lead to a better understanding of travel behavior. These data streams often provide frequent
sampling of real-world driving for a large number of vehicles, so, while imperfect, they are
useful for improving the nationally representative default inputs in MOVES. EPA is actively
acquiring such data for future MOVES updates.
Future updates to vehicle population and activity defaults will need to continue to focus on the
most critical elements required for national fleet-wide estimates, 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. 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 make noticeable improvements in run time
without affecting the validity of fleet-wide emission estimates. EPA hopes to perform further
validation of MOVES activity data using fuel volumes reported from US Department of
Transportation in a separate technical report. This type of fuel volume validation and other
MOVES2014 validation work was initially presented to the MOVES Federal Advisory
Committee Act (FACA) Work Group.83
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 fuel vehicles, such as 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.
109
-------
17 Appendix A: Projected Source Type Populations by Year
Table 17-1; Source type populations (in thousands), as derived from HPMS populations in §5.2and the age distribution algorithm in §7.1.2.2
Year
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
2023
2024
2025
2026
2027
2028
2029
2030
2031
2032
2033
2034
Motorcycle
(11)
8571
8687
8706
8747
8844
8943
9018
9098
9178
9260
9337
9416
9498
9585
9680
9781
9888
9996
10103
10215
10328
10439
10538
Passenger
Car
(21)
128033
129764
130054
130666
132117
133583
134715
135907
137105
138317
139471
140653
141880
143179
144593
146100
147713
149317
150922
152591
154280
155930
157420
Passenger
Truck
(31)
86859
87924
88014
88345
89259
90198
90934
91718
92513
93324
94098
94892
95725
96598
97557
98575
99664
100741
101823
102952
104098
105216
106225
Light
Comm.
Truck
(32)
21393
21791
21960
22167
22492
22803
23043
23279
23508
23730
23939
24150
24361
24591
24833
25092
25368
25649
25925
26209
26493
26772
27024
Intercity
Bus
(41)
18
19
20
21
22
22
23
23
23
24
24
25
25
26
26
27
27
27
28
28
28
28
29
Transit
Bus
(42)
69
72
74
77
80
82
84
86
87
88
90
92
93
95
97
98
99
100
101
103
104
105
106
School
Bus
(43)
617
643
663
691
720
740
753
766
780
794
809
824
838
853
867
879
889
900
912
922
931
942
956
Refuse
Truck
(51)
185
195
203
213
223
230
235
239
243
247
252
256
260
264
267
269
272
274
277
280
283
286
290
Single
Unit
Short-
Haul
(52)
6194
6525
6777
7093
7392
7589
7709
7824
7953
8093
8242
8385
8510
8638
8752
8846
8927
9017
9114
9209
9286
9378
9493
Single
Unit
Long-
Haul
(53)
260
274
285
299
312
322
328
333
335
340
345
351
352
357
362
366
371
375
376
377
381
385
391
Motor
Home
(54)
1559
1643
1708
1788
1863
1915
1946
1977
2012
2053
2095
2134
2168
2204
2239
2266
2288
2312
2340
2368
2385
2405
2432
Combination
Short-Haul
(61)
1191
1234
1258
1306
1354
1380
1390
1400
1410
1422
1437
1453
1466
1482
1495
1505
1514
1527
1546
1567
1585
1609
1639
Combination
Long-Haul
(62)
1280
1332
1377
1439
1503
1555
1600
1645
1690
1737
1783
1828
1872
1918
1964
2005
2040
2073
2104
2131
2152
2174
2203
no
-------
Year
2035
2036
2037
2038
2039
2040
2041
2042
2043
2044
2045
2046
2047
2048
2049
2050
Motorcycle
(11)
10633
10724
10813
10901
10983
11055
11155
11256
11357
11460
11564
11668
11774
11880
11988
12096
Passenger
Car
(21)
158833
160194
161523
162835
164062
165135
166628
168135
169655
171189
172737
174299
175875
177465
179069
180688
Passenger
Truck
(31)
107181
108102
109001
109888
110717
111441
112449
113466
114490
115523
116567
117620
118683
119756
120838
121931
Light
Comm.
Truck
(32)
27263
27494
27720
27944
28153
28338
28594
28852
29115
29380
29647
29916
30187
30460
30735
31013
Intercity
Bus
(41)
29
30
30
30
31
31
32
32
32
33
33
34
34
34
35
35
Transit
Bus
(42)
108
109
111
113
114
115
117
118
120
121
123
124
126
127
129
131
School
Bus
(43)
969
983
996
1009
1021
1034
1047
1060
1074
1087
1101
1115
1129
1143
1158
1172
Refuse
Truck
(51)
293
296
299
301
304
306
309
312
315
318
321
324
328
331
334
337
Single
Unit
Short-
Haul
(52)
9599
9698
9795
9887
9968
10051
10147
10243
10342
10442
10543
10646
10749
10853
10958
11064
Single
Unit
Long-
Haul
(53)
396
401
405
409
413
416
420
424
428
432
436
440
445
449
453
458
Motor
Home
(54)
2457
2482
2508
2532
2553
2573
2596
2620
2646
2672
2698
2725
2751
2778
2805
2832
Combination
Short-Haul
(61)
1669
1701
1733
1766
1794
1822
1849
1876
1901
1925
1950
1975
2001
2028
2055
2083
Combination
Long-Haul
(62)
2232
2260
2288
2315
2342
2371
2402
2435
2470
2507
2544
2581
2619
2656
2695
2733
111
-------
18 Appendix B: Fuel Type and Regulatory Class Fractions
for 1960-1981
As noted in the text, the fuel type and regulatory class distributions in the
SampleVehiclePopulation table for model year 1981 and earlier have not changed from
MOVES2010b. Those fuel type distributions between 1960 and 1981 for each source type have
been summarized in Table 18-1 and Table 18-2. Many of the data sources for the fuel type
fractions are the same in MOVES2010b and MOVES2014. Truck diesel fractions in Table 18-1
are derived using a MOVES2010b sample vehicle counts dataset—similar to the MOVES2014
one—but with 1999 Polk vehicle registrations and the 1997 VIUS, 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. We
also assumed that 15 percent of these motor homes are diesel powered based on information
from the Recreation Vehicle Industry Association (RVIA), as noted above in Section 6.2.2.5.
Table 18-1 Diesel fractions for truck source types*
Model
Year
1960-1979
1980
1981
Source Type
Passenger
Trucks
(31)
0.0139
0.0124
0.0178
Light
Commercial
Trucks
(32)
0.0419
0.1069
0.0706
Refuse
Trucks
(51)
0.96
0.96
0.96
Single Unit
Trucks
(52 & 53)
0.2655
0.2950
0.3245
Motor Homes
(54)
0.15
0.15
0.15
Short-Haul
Combination
Trucks
(61)
0.9146
0.9146
0.9146
Long-Haul
Combination
Trucks
(62)
1.0000
1.0000
1.0000
* All other trucks are assumed to be gasoline powered
As in MOVES2010b, lacking both emission rate and population data, we assume in
MOVES2014 that all motorcycles will be gasoline powered, all intercity buses will be diesel
powered over all model years, and all transit buses will be run on diesel from 1960 to 1981.
School bus fuel type fractions are reused from MOBILE6, originally based on 1996 and 1997
Polk data. Passenger cars are split between gasoline and diesel for 1960-1981 using the
MOVES2010b sample vehicle counts dataset.
Table 18-2 Diesel fractions for non-truck source types*
Model
Year
1960-1974
1975
1976
1977
1978
1979
1980
1981
Source Type
Motorcycles
(11)
0
0
0
0
0
0
0
0
Passenger
Cars
(21)
0.0069
0.0180
0.0165
0.0129
0.0151
0.0312
0.0467
0.0764
Intercity Buses
(41)
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
Transit Buses
(42)
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
School Buses
(43)
0.0087
0.0087
0.0086
0.0240
0.0291
0.0460
0.0594
0.2639
All other vehicles are assumed to be gasoline powered
112
-------
The 1960-1981 regulatory class distributions have been derived from the MOVES2010b sample
vehicle counts dataset. 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 for both MOVES2010b and MOVES2014. 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 18-3.
Table 18-3 Percentage by regulatory class and fuel type for passenger trucks (sourceTypelD 31) and light
Model Year
1960-1966
1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
Passenger Trucks (31)
Gasoline
LDT
(30)
81%
90%
88%
100%
99%
96%
96%
95%
95%
97%
95%
89%
85%
87%
90%
96%
LHD
(40)
19%
10%
12%
0%
1%
3%
4%
5%
5%
3%
5%
11%
15%
13%
10%
4%
Diesel
LDT
(30)
38%
38%
38%
38%
38%
38%
38%
38%
38%
38%
38%
38%
38%
38%
38%
38%
LHD
(40)
62%
62%
62%
62%
62%
62%
62%
62%
62%
62%
62%
62%
62%
62%
62%
62%
Light Commercial Trucks (32)
Gasoline
LDT
(30)
24%
72%
67%
91%
80%
94%
75%
59%
65%
72%
88%
79%
81%
78%
74%
89%
LHD
(40)
76%
28%
33%
9%
20%
6%
25%
41%
35%
28%
12%
21%
19%
22%
26%
11%
Diesel
LDT
(30)
7%
7%
7%
7%
7%
7%
7%
7%
7%
7%
7%
7%
7%
7%
40%
12%
LHD
(40)
93%
93%
93%
93%
93%
93%
93%
93%
93%
93%
93%
93%
93%
93%
60%
88%
The bus and motor home source types each have a single regulatory class distribution for all
model years, as described in Section 6. The 1960-1981 regulatory class distributions for diesel-
fueled single unit and combination trucks have been summarized in Table 18-4 below. All 1960-
1981 gasoline-fueled single unit and combination trucks fall into the medium heavy-duty (MHD)
regulatory class (regClassID 46).
113
-------
Table 18-4 Percentage of MHD trucks (regClassID 46) among diesel-fueled single unit and combination
trucks*
Model Year
1960-1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
Source Type
Refuse Trucks
(51)
100%
100%
0%
0%
0%
0%
0%
0%
0%
0%
Single Unit Trucks
(52&S3)
0%
3%
6%
14%
44%
43%
36%
34%
58%
47%
Short-Haul Comb.
Trucks
(61)
0%
8%
30%
3%
13%
31%
18%
16%
29%
31%
Long-Haul Comb.
Trucks
(62)
0%
0%
0%
0%
0%
0%
0%
0%
5%
6%
*For these source types, all remaining trucks are in the HHD regulatory class (regClassID 47).
114
-------
19 Appendix C: 1990 Age Distributions
19.1 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.
19.2 Passenger Cars
To determine the 1990 age fractions for passenger cars, we began with Polk 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 Polk 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.
19.3 Trucks
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 19-1. Like VIUS97,
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 followed the practice used for the
1999 fractions and 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."
115
-------
Table 19-1 VIUS1997 codes used for distinguishing truck source types
Source Type
Passenger Trucks
Light Commercial
Trucks
Refuse Trucks
Single Unit Short-
Haul Trucks
Single Unit Long-
Haul Trucks
Combination Short-
Haul Trucks
Combination Long-
Haul Trucks
Axle Arrangement
2 axle/4 tire (AXLRE=
1,5,6,7)
2 axle/4 tire (AXLRE=
1,5,6,7)
Single Unit
(AXLRE=2-4, 8-16)
Single Unit
(AXLRE=2-4, 8-16)
Single Unit
(AXLRE=2-4, 8-16)
Combination
(AXLRE>=17)
Combination
(AXLRE>=17)
Primary Area of
Operation
Any
Any
Off-road, local or
short-range
(AREAOP <=4)
Off-road, local or
short-range
(AREAOP<=4)
Long-range
(AREAOP>=5)
Off-road, local or
medium
(AREAOP<=4)
Long-range
(AREAOP>=5)
Body Type
Any
Any
Garbage hauler
(BODTYPE=30)
Any except garbage
hauler
Any
Any
Any
Major Use
personal
transportation
(MAJUSE=20)
any but personal
transportation
Any
Any
Any
Any
Any
19.4 Intercity Buses
For 1990, we were not able to identify a data source for estimating age distributions of intercity
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.
19.5
School Buses and Motor Homes
To determine the age fractions of school buses and motor homes, we used information from the
Polk TIP® 1999 database. School bus and motor home counts were available by model year.
Unlike the Polk 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.
19.6
Transit 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 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.
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
116
-------
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 19-2 presents the equations used to create the age distribution and the years in which the
equations were used.
Table 19-2 Curve fit equations for registration distribution data by age
Vehicle
Age
1-17
18-25+
Equation
If age ,.12.53214119^
y = 3462 * e ^17.16909475^
24987.0776 * a'0-2000
)
*age
117
-------
20 Appendix D: Detailed Derivation of Age Distributions
20. 1 2012-2050 Age Distribution Proj ections
The base algorithm for forecasting age distributions is as follows:
1 . Starting with the base population distribution (Py), remove the number of vehicles that
did not survive (Ry) at each age level.
2. Increase the population age index by one (for example, 3 year old vehicles are
reclassified as 4 year old vehicles).
3. Add new vehicle sales (Ny+i) as the age 0 cohort.
4. Combine the new age 30 and 3 1 vehicles into a single age 30 group.
5. This results in the next year population distribution (Py+i). If this algorithm is to be
repeated, Py+1 becomes Py for the next iteration.
This is mathematically described with the following equation (reprinted from Section 7.1.2.2 for
reference):
Py+i = ~P^-~R^ + Ny+i Equation 10
Unfortunately, as described in Section 7.1.2.1, 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 5.1 level 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 determining the
population of vehicles removed and its relationship to the generic survival rate 50 is given by
Equation 27. 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.
flj = ky • (l - V) o ~Py Equation 27
Substituting Equation 27 into Equation 10 yields Equation 28:
~ (1 - sj) ° Ty + Ny+i Equation 28
Since both the value of the scalar adjustment factor and the actual distribution of the next year's
population are unknown, Equation 28 can't be used yet. However, by using an estimate of next
year's total population, it can be transformed into Equation 29:
Py+1 = Py - ky V ((1 - Sj) ° PJ) + JVy+1 Equation 29
^— 'a v /
118
-------
This was algebraically solved for ky and evaluated for each HPMS category1 using the following
information:
• Total populations Py and Py+1 by HPMS category. For analysis year 2011, this
information is described source type in Section 5.1 and simply needs to be summed by
HPMS category for use here. For years 2012+, this information is described in Section
5.2
• Survival 50 by HPMS category, which is described in Section 7.1.2.1.
• Population distribution Py by HPMS category. For analysis year 2011, this information
came from combining the total populations described in Section 5.1 with the age
distributions described in Section 7.1.1.2 and summing by HPMS category. For years
2012+, this comes from Py+1 of the previous year.
• New vehicle sales Ny+i by HPMS category, which are derived from AEO2014. The
projection of sales was calculated as a percentage of the total population using the vehicle
category mapping discussed in Section 4.2; this is converted to the number of new
vehicles by multiplying by the HPMS category population.
After determining ky by HPMS category, Equation 28 was used with the following information
to compute the next year's population and then age distribution by source type:
• Population distribution Py by source type. For analysis year 2011, this information
came from combining the total populations described in Section 5.1 with the age
distributions described in Section 7.1.1.2. For years 2012+, this comes from Py+1
of the previous year.
• The scalar adjustment factor ky and generic survival rate 50 applied by source
type using the HPMS to source type mapping described by Table 2-1. Please note
that limits were placed on the /cy(l — 50) term of Equation 28: the value of this
term for each age was restricted to being between 0 and 1.
• New vehicle sales JVy+1 determined as a percentage of the total population in
AEO2014 as discussed above; this is converted to the number of new vehicles by
multiplying by the total source type population.
With all of this information, the population distributions were algorithmically determined for
years 2012-2050. The resulting total source type populations (Py) are stored in the
SourceTypeYear table of the default database. The resulting age distributions are stored in the
SourceTypeAgeDistribution table. An illustration of passenger car age distributions is presented
in Figure 20-1. For clarity, only four years are shown: 2011, 2020, 2030, and 2040. The effects
1 Because vehicle survival rates use the categories of motorcycles, passenger cars, light-duty trucks, buses, single
unit trucks, and combination trucks, these were the categories used for determining the scalar adjustment factor.
Since these are essentially the HPMS categories used by MOVES with the additional subcategories of passenger car
and light-duty trucks, the term "HPMS category" is used here for simplicity.
119
-------
of the 2008-2009 recession are visible in the 0-3 year old cars in the 2011 age distribution. By
2020, the recession dip is moved to the 9-11 year old cars as expected.
Figure 20-1 Selected age distributions for passenger cars in MOVES2014
0.08
0.06
=
c
CU
0.04
0.02
0.00
Calendar Year
2020
2030
2040
10 20
Passeneer Car Aee
30
20.2 1999-2010 Age Distributions
The base algorithm for forecasting age distributions is as follows:
1. Starting with the base population distribution (Py), remove the age 0 vehicles (JVy).
2. Decrease the population age index by one (for example, 3 year old vehicles are
reclassified as 2 year old vehicles).
3. Add the vehicles that were removed in the previous year (Ry-i).
4. This results in the previous year population distribution (Py_i). If this algorithm is to be
repeated, Py-\ becomes Py for the next iteration.
This is mathematically described with the following equation (reprinted from Section 7.1.2.3 for
reference):
Py-! = Py - Ny + Ry_l
Equation 30
12O
-------
However, without detailed historical data for every year, we needed to estimate vehicle
removals. The equation governing vehicle removal discussed the previous section is also
applicable here. Taking careful note of the subscripts, Equation 27 and Equation 30 can be
combined into Equation 31:
P^ = ^-JVy + fry.! • (1 - Sj) o pj^ Equation 31
As in the forecasting situation, the value of the scalar adjustment factor and the actual
distribution of the previous year's population are unknown. With a similar strategy of using the
previous year's known total population, Equation 31 can be transformed into Equation 32:
Py-i = Py - Ny + ky^ Y ((l - SQ) ° Pylf) Equation 32
^—'a v /
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 properly solve Equation 32
for /ty-i and Py-\, a numerical method of approximation could be employed. However, due to
lack of resources, Py was used as a simple approximation of Py-\ on the left hand side of
Equation 32. The following sources were used to determine /ty-i by HPMS category:
• Total populations Py and Py-\ by HPMS category. For all historic analysis years, this
information is described source type in Section 5.1 and simply needs to be summed by
HPMS category across all ages for use here.
• Survival 50 by HPMS category, which is described in Section 7.1.2.1.
• Population distribution Py by HPMS category. For analysis year 2011, this information
came from combining the total populations described in Section 5.1 with the age
distributions described in Section 7.1.1.2 and summing by HPMS category. For other
years, this comes from Py-\ of the previous iteration.
• New vehicle sales Ny+i data, which was collected by source type from a variety of
sources. Each of these was summed by HPMS category. Motorcycles sales comes from
the Motorcycle Industry Council; sales data for passenger cars, passenger trucks, light
commercial trucks, refuse trucks, short-haul and long-haul single unit trucks, and short-
haul and long-haul combination trucks comes from TEDB and VIUS; transit buses
production estimates are based on EPA certification data; and school bus sales came from
the School Bus Fleet Fact Book. No sales data were available for intercity buses, so the
other bus categories were used as a surrogate. That is, the total transit bus production and
school bus sales as a percentage of the transit and school bus populations in each year
were applied to the intercity bus populations to estimate their sales. Similarly, no sales
data were available for motor homes, so a sales fraction was estimated by averaging the
sales of refuse, short-haul, and long-haul single unit trucks as a fraction of their total
population.
After determining /ty-i by HPMS category, Equation 31 was used with the following
information to compute the previous year's age distribution by source type:
121
-------
Population distribution Py by source type. For analysis year 201 1, this information
came from combining the total populations described in Section 5.1 with the age
distributions described in Section 7.1.1.2. For other years, this comes from Py-\
of the previous iteration.
The scalar adjustment factor /ty-i and generic survival rate 50 applied by source
type using the FIPMS to source type mapping described by Table 2-1. As with
before, limits were placed on the ky(l — 50) term, such that the value of this term
for each age was restricted to being between 0 and 1. Also, the Py-\ term used
when calculating the number of vehicles removed was approximated by Py.
New vehicle sales Wy+1, from the sources listed above and applied by source type.
With all of this information, the population distributions were algorithmically determined for
years 1999-2010. The resulting age distributions are stored in the SourceTypeAgeDistribution
table.
122
-------
21 Appendix E: SCC Mappings
21.1 SCC Mappings between MOVES2014 and MOVES2010b
The SCC values used in MOVES2010b and earlier versions of MOVES and MOBILE do not
have a one-to-one correspondence with the MOVES2014 SCC values. This makes it difficult to
compare results from MOVES2014 to those from earlier models. While MOVES2014 allows
output by fuel type and regulatory class (which were the primary identifiers for the earlier
SCCs), there are complications that prevent developing a simple mapping from the old to the
new. The most important complication is that the distribution of regulatory classes and fuel
types for each source type varies by model year, while typical inventories aggregate across
model years. This means any mapping would have to vary with calendar year and with user
vehicle population inputs. In addition, regulatory class groupings for light and light heavy-duty
trucks do not line up exactly with the GVWR groupings used in the earlier SCCs. Table 21-1
below compares MOVES2014 classification by fuel type and regulatory class to the older SCCs.
123
-------
Table 21-1 Comparison of MOVES2014 Fuel Types and Regulatory Classes to MOVES2010b Source
Classification Code (SCC) Vehicle Classes
Fuel Type
Gasoline
Diesel
MOVES2014 Regulatory Class
(RegClassID)
Motorcycles (10)
Passenger Cars (20)
Light-Duty Trucks (0-8500 Ibs
GVWR) (30)
Heavy-Duty Trucks and Buses (40,
41, 42, 46 & 47)
Passenger Cars (20)
Light-Duty Trucks (30)
Heavy-Duty Vehicles 2b (8501-10000
Ibs GVWR) with four tires (40)
Heavy-Duty Vehicles 2b with 6 tires
or more (8501-10000 Ibs GVWR)
Heavy-duty Vehicles (10001-14000
Ibs GVWR) (41)
Heavy-duty Vehicles (14001-19500
Ibs GVWR) (42)
Medium Heavy-Duty Vehicles
(19501-33000 Ibs GVWR) (46)
Heavy Heavy-Duty Vehicles (33001+
Ibs GVWR) (47)
Transit Buses (48)
MOVES2010b Source
Classification Code (SCC)
Motorcycles (01080)
Passenger Cars (01001)
Light-Duty Trucks (0-6000 Ibs
GVWR) (01020)
Light-Duty Trucks (6001-8500 Ibs
GVWR) (01040)
Heavy-Duty Trucks and Buses
(01070)
Passenger Cars (30001)
Light-Duty Trucks (30060)
Heavy Duty Vehicles 2b (8501-
10000 Ibs GVWR) (30071)
Light Heavy-duty Vehicles (10001-
19500 Ibs GVWR) (30072)
Medium Heavy-Duty Vehicles
(19501-33000 Ibs GVWR) (30073)
Heavy Heavy-Duty Vehicles
(33001+ Ibs GVWR) (30074)
Diesel Buses (30075)
21.2
2011 SCC VMT Conversions
The source classification code (SCC) used before MOVES2014 do not cleanly map to the source
types used by MOVES. In the 10-digit SCC, the first seven digits (SCC7) indicate the vehicle
classification. The SCC vehicle classifications were mapped to the source types used by MOVES
by calculating the fraction VMT for each source type found in each SCC classification result in a
national MOVES2010b run for calendar year 2011. The factors calculated from the
MOVES2010b run are shown in Table 21-2.
124
-------
Table 21-2 Mapping of previous SCC vehicle classifications to MOVES source types for calculation of road
type distributions
SCC
(7 digits)
2201001
2201020
2201020
2201040
2201040
2201070
2201070
2201070
2201070
2201070
2201070
2201070
2201070
2201070
2201080
2230001
2230060
2230060
2230071
2230071
2230072
2230072
2230073
2230073
2230073
2230073
2230073
Description
Gasoline Light-Duty Vehicles (Passenger Cars)
Gasoline Light-Duty Trucks (0-6,000 Ibs. GVWR)
Gasoline Light-Duty Trucks (0-6,000 Ibs. GVWR)
Gasoline Light-Duty Trucks (6,001-8,500 Ibs. GVWR)
Gasoline Light-Duty Trucks (6,001-8,500 Ibs. GVWR)
Gasoline Heavy-Duty Gasoline Vehicles (8501 Ibs.
and greater GVWR)
Gasoline Heavy-Duty Gasoline Vehicles (8501 Ibs.
and greater GVWR)
Gasoline Heavy-Duty Gasoline Vehicles (8501 Ibs.
and greater GVWR)
Gasoline Heavy-Duty Gasoline Vehicles (8501 Ibs.
and greater GVWR)
Gasoline Heavy-Duty Gasoline Vehicles (8501 Ibs.
and greater GVWR)
Gasoline Heavy-Duty Gasoline Vehicles (8501 Ibs.
and greater GVWR)
Gasoline Heavy-Duty Gasoline Vehicles (8501 Ibs.
and greater GVWR)
Gasoline Heavy-Duty Gasoline Vehicles (8501 Ibs.
and greater GVWR)
Gasoline Heavy-Duty Gasoline Vehicles (8501 Ibs.
and greater GVWR)
Gasoline Motorcycles
Diesel Light-Duty Vehicles (Passenger Cars)
Diesel Light-Duty Trucks (0-8,500 Ibs. GVWR)
Diesel Light-Duty Trucks (0-8,500 Ibs. GVWR)
Diesel Class 2b Heavy-Duty Vehicles (8501-10,000
Ibs. GVWR)
Diesel Class 2b Heavy-Duty Vehicles (8501-10,000
Ibs. GVWR)
Diesel Class 3, 4 & 5 Heavy-Duty Vehicles (10,001-
19,500 Ibs. GVWR)
Diesel Class 3, 4 & 5 Heavy-Duty Vehicles (10,001-
19,500 Ibs. GVWR)
Diesel Class 6 & 7 Heavy-Duty Vehicles (19,501-
33,000 Ibs. GVWR)
Diesel Class 6 & 7 Heavy-Duty Vehicles (19,501-
33,000 Ibs. GVWR)
Diesel Class 6 & 7 Heavy-Duty Vehicles (19,501-
33,000 Ibs. GVWR)
Diesel Class 6 & 7 Heavy-Duty Vehicles (19,501-
33,000 Ibs. GVWR)
Diesel Class 6 & 7 Heavy-Duty Vehicles (19,501-
33,000 Ibs. GVWR)
Source
Type
21
31
32
31
32
31
32
42
43
51
52
53
54
61
11
21
31
32
31
32
31
32
51
52
53
54
61
Description
Passenger Car
Passenger Truck
Light Commercial
Truck
Passenger Truck
Light Commercial
Truck
Passenger Truck
Light Commercial
Truck
Transit Bus
School Bus
Refuse Truck
Single Unit Short-
Haul Truck
Single Unit Long-
Haul Truck
Motor Home
Combination Short-
Haul Truck
Motorcycle
Passenger Car
Passenger Truck
Light Commercial
Truck
Passenger Truck
Light Commercial
Truck
Passenger Truck
Light Commercial
Truck
Refuse Truck
Single Unit Short-
Haul Truck
Single Unit Long-
Haul Truck
Motor Home
Combination Short-
Haul Truck
2011
Fractions
1.000000
0.779270
0.220730
0.779269
0.220731
0.450274
0.267803
0.000664
0.002476
0.000509
0.221958
0.030154
0.025802
0.000359
1.000000
1.000000
0.343599
0.656401
0.364691
0.635309
0.305092
0.694908
0.001726
0.623978
0.086570
0.025294
0.194650
125
-------
sec
(7 digits)
2230073
2230074
2230074
2230074
2230074
2230074
2230074
2230075
2230075
2230075
Description
Diesel Class 6 & 7 Heavy-Duty Vehicles (19,501-
33,000 Ibs. GVWR)
Diesel Class 8a & 8b Heavy-Duty Vehicles (33,001
Ibs. and greater GVWR)
Diesel Class 8a & 8b Heavy-Duty Vehicles (33,001
Ibs. and greater GVWR)
Diesel Class 8a & 8b Heavy-Duty Vehicles (33,001
Ibs. and greater GVWR)
Diesel Class 8a & 8b Heavy-Duty Vehicles (33,001
Ibs. and greater GVWR)
Diesel Class 8a & 8b Heavy-Duty Vehicles (33,001
Ibs. and greater GVWR)
Diesel Class 8a & 8b Heavy-Duty Vehicles (33,001
Ibs. and greater GVWR)
Diesel Buses
Diesel Buses
Diesel Buses
Source
Type
62
51
52
53
54
61
62
41
42
43
Description
Combination Long-
Haul Truck
Refuse Truck
Single Unit Short-
Haul Truck
Single Unit Long-
Haul Truck
Motor Home
Combination Short-
Haul Truck
Combination Long-
Haul Truck
Intercity Bus
Transit Bus
School Bus
2011
Fractions
0.067783
0.008531
0.100296
0.013800
0.000328
0.323425
0.553619
0.430859
0.122565
0.446576
126
-------
22 Appendix F: Calculation of Combination Truck Average
Speed Distributions
The average speed for each roadway type, day type, and hour can be calculated by multiplying
the average speed of each bin by the corresponding distribution of time as shown in Equation 33.
Here, v is the average speed of the distribution, vt is the average speed of bin i, and pt is the
proportion of time spent in bin i.
v = V IV Pi
Equation 33
= 2.5 • Pl + 5 • p2 + - + 70 • p15 + 75 • p16
To adjust the average speed for heavy-duty combination trucks, we redistributed the proportion
of time spent in each speed bin such that its contribution to the average speed was 92 percent of
the light-duty speed, as shown in Equation 18. This redistributed proportion of time in each
speed bin is given by p[.
^combination = (\J-s£) Alight-duty
, Equation 34
=r
To perform this redistribution, we defined two new variables, a and (3, where a^ is the fraction of
Pi that is shifted down one speed bin, and /?j is the fraction of pi shifted down two speed bins.
The new distribution at speed bin i (given by p[) starts with the original distribution (pt\ gains
the proportions moved down from the higher speed bins (ai+1 • pi+1 and /?j+2 • Pi+z), and loses
the proportion that is moved to a lower speed bin (at • pt and /?j • pi). This is shown in Equation
35:
Pi =Pi + Oi+i' Pi+i) + (A+2 ' Pi+z) ~ Oi' Pi) - (A ' Pi) Equation 35
For speed bins with an average speed of less than or equal to 60 mph, we only needed to shift
distributions using a fraction of one speed bin (or 5 mph). Thus we only calculated a^ and
set Pi = 0. Mathematically, reducing a bin's average speed by a certain fraction (77) can be
expressed with Equation 36:
(1 — 77) • PI = at • (Vi — 5) + (1 — «j) • PI Equation 36
Essentially, the fraction that is moved to the next slower bin (tfj) is multiplied by the slower
speed (note that each of the speed bins are 5 mph apart, so this is Vi — 5), and the fraction that
remains (1 — «j) is multiplied by the original speed vt. Since the average speed of the
combination trucks is 92 percent of cars, (1 — 77) = 92% and rj = 0.08.
By rearranging terms from Equation 20, and solving for a^ we obtain Equation 37:
127
-------
cci = —-— Equation 37
o
However, for speed bins > 65 mph, Equation 37 yields a^ greater than 1. Since that logically
can't happen, some of the distribution needed to be moved to the second next slower speed bin to
fully account for the 8 percent speed reduction. This is mathematically shown in Equation 38,
which is the logical extension of Equation 36:
(1 - 77) • vi = Pi • (vt - 10) + «i • Oj - 5) + (1 - «i - ft) • vt Equation 38
The difference between Equation 36 and Equation 38 is that an additional fraction (ft) is
removed from the original speed bin and is given the speed of two speed bins slower (or 10 mph
slower). With this additional factor, there is an infinite combination of solutions that could
satisfy Equation 38. We solved this problem with a linear equation solver by setting Equation 38
to a constraint (see Equation 39), adding the constraint that at + ft are less than or equal to 1
(see Equation 40), and choosing the solution that minimized ft.
«i • fa - 5) + ft • fa - 10) + vt • (n - at - ft) = 0 Equation 39
at + ft < 1 Equation 40
This linear program was used to solve for at and ft for each speed bin between 65 and 75 mph.
With at and ft known for each bin, the new distributions p[ were calculated.
An additional adjustment was made for the highest speed bins because we assumed that the
maximum speed bin had a triangular distribution with an average speed of 75 mph, see Figure
22-1. In the figure, the original speed distribution is shown in light gray. The darker gray is the
proportion of speed bin 55 that is moved out to the slower speed bin 50 mph, and the black areas
are the distributions from speed bin 60 and 65 that are moved in to speed bin 55 mph.
128
-------
Figure 22-1 An illustration of adjustments made to the average speed bin 55 mph for heavy-duty vehicles
0.20-
o.oo-
20 40 60
Average Speed Bin
SO
In the new distribution, all of the maximum speed bin fraction was redistributed to the 65 and 70
mph bins. Therefore, the new maximum speed bin (70 mph) was also assumed to have a
triangular distribution. Geometrically, 1/9* of a triangular distribution averaging 70 mph is faster
than 72.5 mph. Since the 75 mph speed bin is defined as any speed >72.5 mph, l/9th of the new
70 mph fraction (p{5) was reclassified as the new fraction for the 75 mph bin.
This process was repeated for both short- and long-haul combination trucks on restricted access
road types for every hour and day type combination.
129
-------
23 Appendix G: 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, MOVES2014 employs 49 drive schedules with various average speeds, mapped to
specific source types and roadway types.
Table 23-1 below lists the driving schedules used in MOVES2014. 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.
130
-------
Table 23-1 MOVES2014 default driving schedule statistics
drive
schedule id
101
153
158
201
202
203
204
205
206
251
252
253
254
255
301
302
303
304
305
306
351
352
353
354
355
396
397
drive schedule name
LD Low Speed 1
LD LOS E Freeway
LD High Speed Freeway 3
MD 5mph Non-Freeway
MD lOmph Non-Freeway
MD 15mph Non-Freeway
MD 20mph Non-Freeway
MD 25mph Non-Freeway
MD 30mph Non-Freeway
MD 30mph Freeway
MD 40mph Freeway
MD 50mph Freeway
MD 60mph Freeway
MD High Speed Freeway
HD 5mph Non-Freeway
HD lOmph Non-Free way
HD 15mph Non-Freeway
HD 20mph Non-Freeway
HD 25mph Non-Freeway
HD 3 Omph Non-Free way
HD 3 Omph Freeway
HD 40mph Freeway
HD 50mph Freeway
HD 60mph Freeway
HD High Speed Freeway
HD High Speed Freeway Plus 5 mph
MD High Speed Freeway Plus 5 mph
avg
speed
2.5
30.5
76.0
4.6
10.7
15.6
20.8
24.5
31.5
34.4
44.5
55.4
60.1
72.8
5.8
11.2
15.6
19.4
25.6
32.5
34.3
47.1
54.2
59.7
71.7
76.7
77.8
max
speed
10.00
63.00
90.00
24.10
34.10
36.60
44.50
47.50
55.90
62.60
70.40
72.20
68.40
80.40
19.90
29.20
38.30
44.20
50.70
58.00
62.70
65.00
68.00
69.00
81.00
86.00
85.40
idle
time
(sec)
280
5
0
85
61
57
95
63
54
0
0
0
0
0
37
70
73
84
57
43
0
0
0
0
0
0
0
percent of
time idling
46.5%
1.1%
0.0%
29.0%
19.6%
12.6%
9.1%
11.1%
5.5%
0.0%
0.0%
0.0%
0.0%
0.0%
14.2%
11.5%
12.9%
15.1%
5.8%
5.3%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
miles
0.419
3.863
12.264
0.373
0.928
1.973
6.054
3.846
8.644
15.633
43.329
41.848
81.299
96.721
0.419
1.892
2.463
3.012
6.996
7.296
21.659
41.845
80.268
29.708
35.681
38.170
103.363
time (sec)
602.00
456.00
581.00
293.00
311.00
454.00
1046.00
566.00
988.00
1637.00
3504.00
2718.00
4866.00
4782.00
260.00
608.00
567.00
558.00
983.00
809.00
2276.00
3197.00
5333.00
1792.00
1792.00
1792.00
4782.00
minutes
10.03
7.60
9.68
4.88
5.18
7.57
17.43
9.43
16.47
27.28
58.40
45.30
81.10
79.70
4.33
10.13
9.45
9.30
16.38
13.48
37.93
53.28
88.88
29.87
29.87
29.87
79.70
hours
0.167
0.127
0.161
0.081
0.086
0.126
0.291
0.157
0.274
0.455
0.973
0.755
1.352
1.328
0.072
0.169
0.158
0.155
0.273
0.225
0.632
0.888
1.481
0.498
0.498
0.498
1.328
131
-------
Table 23-1 MOVES2014 default driving schedule statistics
drive
schedule id
398
401
402
403
404
405
501
1009
1011
1017
1018
1019
1020
1021
1024
1025
1026
1029
1030
1033
1041
1043
drive schedule name
CRC E55 HHDDT Creep
Bus Low Speed Urban (nominal 15 mph)
Bus 30 mph Flow (nominal 30 mph)
Bus 45 mph Flow (nominal 45 mph)
New York City Bus
WMATA Transit Bus
Refuse Truck Urban
Final FC01LOSAF Cycle (C10R04-
00854)
Final FC02LOSDF Cycle (C10R05-
00513)
Final FC11LOSB Cycle (C10R02-00546)
Final FC11LOSC Cycle (C15R09-00849)
Final FC11LOSD Cycle (C15R10-00068)
Final FC11LOSE Cycle (C15R11-00851)
Final FC11LOSF Cycle (C15RO 1-00876)
Final FC12LOSC Cycle (C15R04-00582)
Final FC12LOSD Cycle (C15R09-00037)
Final FC12LOSE Cycle (C15R10-00782)
Final FC14LOSB Cycle (C15R07-00177)
Final FC14LOSC Cycle (C10R04-00104)
Final FC14LOSF Cycle (C15R05-00424)
Final FC17LOSD Cycle (C15R05-00480)
Final FC19LOSAC Cycle (C15R08-
00267)
avg
speed
1.8
3.1
11.5
21.9
3.7
8.3
2.2
73.8
49.1
66.4
64.4
58.8
46.1
20.6
63.7
52.8
43.3
31.0
25.4
8.7
18.6
15.7
max
speed
8.24
19.80
33.80
47.00
30.80
47.50
20.00
84.43
73.06
81.84
78.19
76.78
71.50
55.48
79.39
73.15
70.87
63.81
53.09
44.16
50.33
37.95
idle
time
(sec)
107
288
109
116
403
706
416
0
34
0
0
0
1
23
0
12
0
27
41
326
114
67
percent of
time idling
42.3%
63.9%
37.5%
28.3%
67.2%
38.4%
66.9%
0.0%
5.0%
0.0%
0.0%
0.0%
0.1%
2.5%
0.0%
1.5%
0.0%
3.6%
8.0%
38.2%
16.1%
7.7%
miles
0.124
0.393
0.932
2.492
0.615
4.261
0.374
11.664
9.283
9.567
16.189
11.922
12.468
5.179
15.685
11.754
10.973
6.498
3.617
2.066
3.659
3.802
time (sec)
253.00
451.00
291.00
410.00
600.00
1840.00
622.00
569.00
681.00
519.00
905.00
730.00
973.00
905.00
887.00
801.00
913.00
754.00
513.00
853.00
709.00
870.00
minutes
4.22
7.52
4.85
6.83
10.00
30.67
10.37
9.48
11.35
8.65
15.08
12.17
16.22
15.08
14.78
13.35
15.22
12.57
8.55
14.22
11.82
14.50
hours
0.070
0.125
0.081
0.114
0.167
0.511
0.173
0.158
0.189
0.144
0.251
0.203
0.270
0.251
0.246
0.223
0.254
0.209
0.143
0.237
0.197
0.242
132
-------
24 Appendix H: MOVES2010b Source Masses
Light-duty source masses were unchanged from MOVES2010b. In addition, the heavy-duty
source masses originally come from MOVES2010b, although they have been updated as
described in Section 14.1.
In MOVES2010b, weight data (among other kinds of information) were used to allocate source
types to source bins using a field called weightClassID. As described in Equation 41, each source
type's source mass was calculated using an activity-weighted average of their associated source
bins' midpoint weights:
M =
i ab • 77T
ja]ja \ V. „. /i Equation 41
ia) a
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 24-1 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. This calculation is outside the scope of this
document, but more information can be found in the MOVES2010b SDRM.
133
-------
Table 24-1 MOVES weight classes
WeightClassID
0
20
25
30
35
40
45
50
60
70
80
90
100
140
160
195
260
330
400
500
600
800
1000
1300
9999
5
7
9
Weight Class Name
Doesn't Matter
weight < 2000 pounds
2000 pounds <= weight < 2500 pounds
2500 pounds <= weight < 3000 pounds
3000 pounds <= weight < 3500 pounds
3500 pounds <= weight < 4000 pounds
4000 pounds <= weight < 4500 pounds
4500 pounds <= weight < 5000 pounds
5000 pounds <= weight < 6000 pounds
6000 pounds <= weight < 7000 pounds
7000 pounds <= weight < 8000 pounds
8000 pounds <= weight < 9000 pounds
9000 pounds <= weight < 10000 pounds
10000 pounds <= weight < 14000 pounds
14000 pounds <= weight < 16000 pounds
16000 pounds <= weight < 19500 pounds
19500 pounds <= weight < 26000 pounds
26000 pounds <= weight < 33000 pounds
33000 pounds <= weight < 40000 pounds
40000 pounds <= weight < 50000 pounds
50000 pounds <= weight < 60000 pounds
60000 pounds <= weight < 80000 pounds
80000 pounds <= weight < 100000 pounds
100000 pounds <= weight < 130000 pounds
130000 pounds <= weight
weight < 500 pounds (for MCs)
500 pounds <= weight < 700 pounds (for MCs)
700 pounds <= weight (for MCs)
Midpoint Weight
[NULL]
1000
2250
2750
3250
3750
4250
4750
5500
6500
7500
8500
9500
12000
15000
17750
22750
29500
36500
45000
55000
70000
90000
115000
130000
350
600
700
The following sections detail how weight classes were assigned to the various source types in
MOVES.
24.1 Motorcycles
The Motorcycle Industry Council "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 was 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.
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 24-2.
134
-------
Table 24-2 Motorcycle engine size and average weight distributions for selected model years
Displacement
Category
0-169 cc (1)
170-279 cc (2)
280+ cc (9)
1969 MY
distribution
(assumed)
0.118
0.09
0.792
1990 MY
distribution
(MIC)
0.118
0.09
0.792
1998 MY
distribution
(MIC)
0.042
0.05
0.908
2000 MY
distribution
(certification
data)
0.029
0.043
0.928
Weight distribution (EPA
staff)
100%: <=5001bs
50%: <=5001bs
50%: 5001bs -7001bs
30%: 500 lbs-700 Ibs
70%: >7001bs
24.2 Passenger Cars
Passenger car weights come from Polk. The weightClassID was assigned by adding 300 Ibs to
the Polk 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 was missing were omitted from the calculations. Also, analysis
indicated a likely error in the Polk 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.
24.3 General Trucks
24.3.1 Light-Duty Trucks
Determining weight categories for light trucks was fairly complicated. The VIUS1997 data
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 Ibs or
less, 6,001-10,000 Ibs, 10,001-14,0001bs, 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.84
data, to correlate engine size with vehicle weight distributions by model year.
In particular, for source types 31 and 32 (Passenger Trucks and Light Commercial Trucks):
• VIUS1997 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 Ibs, etc) VIUS1997 broad categories
were matched one-to-one with the MOVES weight classes.
• For trucks in the lower broad categories (6,000 Ibs or less and 6001-10,000 Ibs), we used
VIUS 1997 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").
135
-------
• We assigned trucks in the ORNL light-duty vehicle database to a weightClassID by
adding SOOlbs to the recorded curb weight and determining the appropriate MOVES
weight class.
• For the trucks with a VIUS1997 average weight of 6,000 Ibs or less, we multiplied the
VIUS1997 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 Ibs 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 Ibs, we did
not use it to distribute the trucks with a VIUS1997 average weight of 6,001-10,000 Ibs.
Instead these were distributed equally among the MOVES weightClassID 70, 80, 90 and
100.
24.3.2 Single Unit Trucks
Source types 52 and 53 (long- and short-haul single unit trucks) also included some trucks in
VIUS1997 Strata 1 and 2, thus a similar algorithm was applied.
• VIUS1997 trucks of the source type in Strata 3, 4, and 5 were assigned to the appropriate
MOVES weight class based on VIUS1997 detailed average weight information.
• VIUS1997 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 Ibs, etc) VIUS1997 broad categories
were matched one-to-one with the MOVES weight classes.
• For trucks in the lower broad categories (6,000 Ibs-or-less and 6001-10,000 Ibs), 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 did not believe the ORNL light-duty vehicle database adequately represented single
unit trucks. Thus, the trucks with a VIUS 1997 average weight of 6,000 Ibs or less and an
engine size less than 5 liters were distributed equally among the MOVES weight classes
20, 25, 30, 35, 40, 45, 50, and 60. Because no evidence existed of very light trucks
among the vehicles with larger engines (5 liter or larger), these were equally distributed
among MOVES weight classes 40, 45, 50 and 60.
• The trucks with a VIUS 1997 average weight of 6,001-10,000 Ibs were distributed equally
among the MOVES weight classes 70, 80, 90 and 100.
24.3.3 Combination Trucks
Long- and short-haul combination trucks (source types 61 and 62) did not include any vehicles
of VIUS 1997 Strata 1 or 2. Thus we used the detailed VIUS 1997 average weight information
and engine size information to assign engine size and weight classes for all of these trucks.
When VIUS2002 became available, we updated values that had been based on VIUS 1997. The
VIUS2002 contains an estimate of the average weight (vehicle weight plus cargo weight) of
136
-------
1998-2002 model year vehicle or vehicle/trailer combination as it was most often operated when
carrying a typical payload during 2002. These estimates were used to determine the MOVES
weightClassID categories for these trucks. Any vehicles with a zero or missing value for the
average weight and without a weight classification in the WeightAvgCK field were excluded
from the analysis for determining the average weight distributions.
Since there is a smaller number of gasoline trucks among the single unit and refuse trucks, all
model years (1998-2002) were combined to determine a single weight distribution to use for
these model years. The VIUS1997 based estimates were retained for light-duty trucks (source
types 31 and 32) and for all model years prior to 1998.
In cases where distributions were missing (no survey information), distributions from a nearby
model year with the same source type was used. Weight distributions for all 2003 and newer
model years were set to be the same as for the 2002 model year for each source type.
24.4 Buses
For intercity buses, we used information from Table II-7 of the FT A 2003 Report to Congress46
that specified the number of buses in various weight categories. This information is summarized
in below in Table 24-3. Note the FTA uses the term "over-the-road bus" to refer to the class of
buses roughly equivalent to the MOVES intercity bus category. The FTA weight categories
were mapped to the equivalent MOVES weight classes.
Table 24-3 FTA estimates of bus weights
Weight (Ibs)
0-20,000
20,000-30,000
30,000-40,000
40,000-50,000
total
MOVES Weight
ClassID
400
500
MOVES Weight
Range (Ibs)
33,000-40,000
40,000-50,000
Number of
buses (2000)
173,536
392,345
120,721
67,905
754,509
Bus type
school & transit
school & transit
school & transit & intercity
intercity
Table 24-4 1999 bus population comparisons by data source
Data Source
FHWA MV-1
FHWA MV-10
(excludes PR)
FHWA adjusted for PR
FTA NTD
APT A85 ***
Polk TIP®
School Bus Fleet Fact
Book
Motorcoach Census45**
Total Buses
732,189
728,777
Intercity Buses
44,200
Transit Buses
55,706
75,087
School Buses
592,029*
594,800
460,178
429,086
* Includes some church & industrial buses.
** Includes Canada.
*** Includes trolleybuses.
137
-------
Using the 1999 bus population estimates in Table 24-4, we were able to estimate the fraction of
all buses that were intercity buses and then to estimate the fraction of intercity buses in each
weight bin. In particular:
Estimated number of intercity buses in 2000:
754,509 * (84,4547(84,454+55,706+592,029)) = 87,028
Estimated number of intercity buses 30,000-40,000 Ibs:
87,028-67,905 = 19,123
Estimated intercity bus weight distribution:
Class 400 = 19,123/87,028 = 22%
Class 500 = 67,905/87,028 = 78%
This distribution was used for all model years.
For transit buses, we took average curb weights from Figure II-6 of the FT A Report to
Congress46 and added additional weight to account for passengers and alternative fuels. The
resulting in-use weights were all in the range from 33,850 to 40,850. Thus all transit buses were
assigned to the weight class "400" (33,000 - 40,000 Ibs) for all model years. This estimate could
be improved if more detailed weight information for transit buses becomes available.
For school buses, we used information from a survey of California school buses. While this data
is older and may not be representative of the national average distribution, it was the best data
source available. The California data86 provided information on number of vehicles by gross
vehicle weight class and fuel as detailed in Table 24-5.
Table 24-5 California school bus study weight classes and fuel types
LHDV
MHDV
HHDV
Total
Gas
2740
467
892
4099
Diesel
4567
2065
11639
18271
Other
8
2
147
157
Total
7315
2534
12678
To estimate the distribution of average weights among the MOVES weight classes, we assumed
that the Light Heavy-Duty (LHDV) school buses were evenly distributed among weightClassIDs
70, 80, 90, 100, and 140. Similarly, we assumed the Medium Heavy-Duty (MHDV) school buses
were evenly distributed among weightClassIDs 140, 160, 195, 260, and 330 and the Heavy
Heavy-Duty (HHDV) school buses were evenly distributed among weightClassIDs 195, 260,
330, and 440.
The final default weight distributions for buses are summarized in Table 24-6.
138
-------
Table 24-6 Weight distributions for buses by fuel type
Weight Class
70
80
90
100
140
160
195
260
330
400
500
Intercity Buses (41)
Diesel
0.2197
0.7800
Transit Buses (42)
Diesel & Gas
1.0000
School Buses (43)
Diesel
0.0500
0.0500
0.0500
0.0500
0.0726
0.0226
0.1819
0.1819
0.1819
0.1593
Gas
0.1337
0.1337
0.1337
0.1337
0.1565
0.0228
0.0772
0.0772
0.0772
0.0544
24.5
Refuse Trucks
Because the sample of Refuse Trucks in VIUS was small, the weight distributions were
calculated for model year groups rather than individual model years, shown below in Table 24-7.
As for other trucks, the WeightClass was determined from the VIUS reported average weight.
139
-------
Table 24-7 Refuse truck SizeWeight fractions by fuel type
Gasoline
Engine Size
3-3. 5L
>5L
>5L
>5L
>5L
>5L
>5L
>5L
>5L
>5L
Sum
Diesel
Engine Size
3.5-4L
4-5L
4-5L
4-5L
>5L
>5L
>5L
>5L
>5L
>5L
>5L
>5L
>5L
>5L
>5L
>5L
Sum
Weight (Ibs.)
5000-6000
7000-8000
9000-10000
10000-14000
14000-16000
16000-19500
19500-26000
26000-33000
33000-40000
50000-60000
Weight (Ibs.)
10000-14000
10000-14000
14000-16000
16000-19500
9000-10000
10000-14000
14000-16000
16000-19500
19500-26000
26000-33000
33000-40000
40000-50000
50000-60000
60000-80000
80000-100000
100000-130000
Pre-1997
0.009074
0.148826
0.070720
0.135759
0.199961
0.055085
0.205341
0.022105
0.153129
0
1.000000
Pre-1998
0.007758
0
0
0
0.006867
0.011727
0.022960
0.063128
0.099782
0.102077
0.237485
0
0.336484
0.111730
0
0
1.000000
1997 and
Newer
0
0
0
0.324438
0.593328
0
0
0
0
0.082234
1.000000
1998
0
0
0
0
0.009593
0
0
0
0.035378
0.019625
0.103922
0.283642
0.338511
0.196424
0
0.012904
1.000000
1999
0
0
0
0
0
0
0
0.011367
0.026212
0.067419
0.088975
0.275467
0.326902
0.193238
0.010420
0
1.000000
2000
0
0
0.015505
0
0
0
0
0.047200
0.052132
0.072106
0.085991
0.165624
0.384612
0.176831
0
0
1.000000
2001
0
0
0
0.011670
0
0.019438
0
0
0.018329
0.043877
0.042678
0.266357
0.315133
0.282517
0
0
1.000000
2002 and
Newer
0
0.006614
0
0
0
0
0
0
0.026079
0
0.046966
0.194716
0.474469
0.224995
0.013081
0.013081
1.000000
24.6
Motor Homes
No detailed information was available on average engine size and weight distributions for motor
homes. We assumed all motor home engines were 5 L or larger. As a surrogate for average
weight, we used information on gross vehicle weight provided in the Polk TIP® 1999 database
by model year and mapped the Polk GVW Class to the MOVES weight bins. These values are
likely to overestimate average weight. The Polk TIP® information did not specify fuel type, so
we assumed that the heaviest vehicles in the Polk database were diesel-powered and the
remainder were powered by gasoline. This led to the weight distributions in Table 24-8 and
Table 24-9.
140
-------
Table 24-8 Weight fractions for diesel motor homes by model year
Polk GVW bin
MOVES weight
class
Model Year
1975-and-earlier
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999+
3
140
4
160
5
195
6
260
7
330
8
400
Diesel
0.171431
0.637989
0.68944
0.423524
0.096922
0.462916
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0.792112
0.340639
0.292308
0.574539
0.899344
0.537084
0.941973
0.868333
0.912762
0.932659
0.881042
0.855457
0.791731
0.72799
0.73298
0.173248
0
0
0
0
0
0
0
0
0
0.029828
0.018755
0.012168
0
0
0
0
0
0.000203
0.000835
0.001474
0.013381
0.085493
0.148917
0.128665
0.614798
0.619344
0.551548
0.345775
0.45546
0.635861
0.553807
0.666905
0.267
0
0
0.000436
0.000277
0.000387
0.001067
0
0.030174
0.049
0.014845
0.009183
0.010761
0.022962
0.022498
0.015469
0.043052
0.043628
0.063712
0.01901
0.471873
0.354386
0.163195
0.229529
0.193167
0.335069
0.736656
0.006629
0.002181
0.005531
0.00155
0.002667
0
0
0.03
0.030096
0.036732
0.083285
0.089534
0.087164
0.093335
0.082792
0.149939
0.296399
0.385085
0.144844
0.159622
0.17468
0.184208
0.111299
0.357508
0.233886
0
0
0.000277
0
0
0
0.027853
0.052667
0.042094
0.020592
0.023438
0.018667
0.013113
0.014289
0.012511
0.018387
0.020545
0.044356
0.037509
0.030531
0.026264
0.032456
0.028628
0.040423
0.029458
141
-------
Table 24-9 Weight fractions for gasoline motor homes by model year
Polk GVW bin
MOVES weight class
Model Year
1975-and-earlier
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999+
3
140
4
160
5
195
6
260
7
330
8
400
Gasoline
1
1
1
1
1
1
0.747723
0.732235
0.714552
0.641577
0.692314
0.720248
0.606635
0.459429
0.551601
0.543354
0.612025
0.54464
0.583788
0.481099
0.52997
0.435959
0.221675
0.288222
0.170133
0
0
0
0
0
0
0.252277
0.267765
0.285448
0.358423
0.307686
0.279752
0.393365
0.540571
0.448399
0.456646
0.322022
0.373999
0.361277
0.361146
0.198479
0.289453
0.433334
0.581599
0.392451
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0.065952
0.081361
0.054935
0.157755
0.271551
0.274588
0.344991
0.13018
0.288411
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0.149004
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
142
-------
25 Peer Review of Draft Report
This section contains comments on the draft report ofPopulation and Activity ofOn-road
Vehicles inMOVES2014 from two peer reviewers and EPA's responses to those comments. The
reviewers were selected by a third-party contractor, ICF International, facilitating a peer review
of the MOVES2014 technical reports. The submitted peer review comments are publicly
available on the EPA Science Inventory database.8?
25.1 Kanok Boriboonsomsin, PhD, PE
Dr. Boriboonsomsin has been a researcher at the University of California at Riverside since
2005. He currently holds the position of Assistant Research Engineer at the College of
Engineering's Center for Environmental Research and Technology (CE-CERT) and received a
PhD in Transportation Engineering from University of Mississippi in 2004. Dr. Boriboonsomsin
previously reviewed the MOVES2010b Population and Activity Report.
25.1.1 General Comments
This is a review of the Draft Report on Population and Activity ofOn-road Vehicles in
MOVES2014, referred to as the "Fleets Report", prepared by the EPA Office of Transportation
and Air Quality. I was also a peer reviewer of the Draft MOVES2009 Highway Vehicle
Population and Activity Data., which helped me identify and understand changes made to the
national default values for vehicle population and activity inputs in MOVES2014 during the time
of this review.
Overall, the Fleets Report is well written and organized, with sensible use of examples, tables,
and figures. I appreciate the addition of Section 2 (MOVES Vehicle and Activity
Classifications), which will help readers understand early on the various ways in which vehicles
and their activities are classified in the context of MOVES. I find the description of analytical
methods and procedures to be sufficiently clear with appropriate use of mathematical equations
to help explain complex calculations such as in Section 9.2 (Heavy-Duty Average Speed
Distributions) [Now Appendix F: Calculation of Combination Truck Average Speed
Distributions]. I also appreciate the list of areas for future research in Section 16 (Conclusion and
Areas for Future Research), which informs research directions for improving the vehicle
population and activity data inputs in future updates of MOVES.
In terms of the vehicle population and activity inputs, I find that the national default values in
MOVES2014 have been appropriately updated by using more recent data from Polk (2011),
AEO (2014), and TEDB (2013). Perhaps, the most important development in this vehicle
population and activity update is the use of nationwide GPS dataset to develop average speed
distributions for light-duty vehicles. This is an exciting time for vehicle activity research due to
the increasing availability of large-scale, high-resolution instrumented vehicle data from a
variety of sources. As indicated in the Fleets Report, many of the limitations in the current
MOVES vehicle activity inputs can be addressed through analysis of such instrumented vehicle
data.
143
-------
25.1.2 Detailed Comments
Detailed comments and suggestions are provided below. These are made with the understanding
of the challenges of developing nationally representative default values for MOVES vehicle
population and activity inputs under the limited resources that the EPA has.
25.1.2.1 Section 1 - Introduction
An early explanation of the analysis years considered in this Fleets Report (e.g., 2011 being the
base year) would be helpful to readers.
Response: A short paragraph has been added to the Introduction to address the possible analysis
years and 2011 base year.
25.1.2.2 Section 2.3 - Regulatory Classes
The mapping between multiple vehicle classification schemes has always been a challenging
topic. The introduction of a new regulatory class 40 is well thought out, and the rationale for it is
well explained.
Response: Further explanation on regulatory class 40 can be found in the MOVES2014 report
on emission rates of heavy-duty vehicles.5
25.1.2.3 Section 2.4 - Fuel Types
The population of CNG-fueled refuse trucks is growing and emissions test data of these trucks
are increasingly available. This source type-fuel type combination may be considered for
modeling in future versions of MOVES.
Response: Yes, CNG refuse trucks have a rapidly growing market share. More than half of the
refuse trucks sold today are manufactured to run on natural gas88; however, despite this rapid
growth, there are about 8,800 natural gas-fueled refuse trucks according to 2014 industry
estimates88, which constitutes less than 10 percent of the US refuse truck fleet. While developing
MOVES2014, EPA decided not to include CNG refuse trucks but will consider adding them to
future versions of MOVES. Report text has been edited to mention CNG refuse trucks.
25.1.2.4 Section 2.8 - Allowable Vehicle Modeling Combinations
Tables Table 2-6 and Table 2-7 provide a very good summary of allowable vehicle modeling
combinations in MOVES2014.
Where would shuttle buses (e.g., those used to pick up and drop off passengers at airports) fit in
Table 2-7?
Response: As discussed in Section 6.1.4, any buses in MOVES not utilized for urban public
transit or service to and from a school are considered intercity buses. Since airport shuttles
cannot be classified as either transit or school buses, they fall into the MOVES intercity bus
source type.
144
-------
25.1.2.5 Section 4.1 - Historic Vehicle Miles Traveled (1990 and 1999-2011)
Does FHWA publish the methodology used to adjust VMT data for 2000-2006? If not, the
average ratio method used appears reasonable.
Response: FHWA did not publish their methodology of adjusting the VMT data. The language in
this section has been edited for clarity.
25.1.2.6 Section 4.2 - Projected Vehicle Miles Traveled (2012-2050)
The methods used to project VMT for future years are appropriate.
25.1.2.7 Section 5.1 - Historic Source Type Populations (1990 and 1999-2011)
It is described that "the 2000-2010 distributions among source types within the general truck
categories were linearly interpolated between 1999 and 2011". However, the 2000-2010 truck
population distributions in Figures Figure 5-2, Figure 5-3, and Figure 5-4 do not show linear
trends. Please clarify the linear interpolation that was performed.
Response: We are not interpolating the populations but the fractions of each source type out of
its general populations, such as long-haul combination trucks out of total combinations trucks,
between 1999 and 2011. Those source type fractions were then multiplied by the general MV-1
populations. Without interpolation of these source type allocations, source type populations
would not have necessarily followed the MV-1 populations trends from 1999 to 2011, which
would have implied that we knew more about the source type populations than we actually did.
We have added clarifying text and Table 5-1 with these 1999-2011 source type population
fractions to this section.
25.1.2.8 Section 5.2 - Projected Vehicle Populations (2012-2050)
The use of VMT growth as a surrogate for vehicle population growth is reasonable per the
analysis of VMT per vehicle trends shown in [Figure 5-6, previously represented in a table].
25.1.2.9 Section 6.2.1 - Fuel Type and Regulatory Class Distributions
Data on actual fuel type used by ESS-capable vehicles are available for 100 vehicles in
California, which may be used in future updates,
(http://www.dot.ca.gov/research/researchreports/reports/2015/fmal_report_task_l 919.pdf).
Response: The use of E85 fuels varies geographically and is an area of great uncertainty and
change. MOVES default values can be considered as our initial national estimate for E85 use.
Users should consider using more locally-specific and up-to-date information when running
MOVES. A more detailed discussion ofE85 use inMOVES2014 can be found in the report on the
default fuel supply.6
According to AEO2014, hybrid electric and plug-in hybrid electric vehicles are projected to
grow from 2.2% of total cars and light truck sales in 2011 to 6.1% in 2040. Would they warrant
their own category with respect to fuel type in future versions of MOVES?
145
-------
Response: Hybrid electric and plug-in hybrid electric vehicles certainly present some modeling
challenges in MOVES. We have added text to this section of the report discussing these
challenges.
25.1.2.10 Section 8 - VMT Distribution of Source Type by Road Type
In Table 8-2, it is my personal opinion that some numbers are not intuitive. For example, I would
think that refuse trucks are operated mostly in urban areas, but they are reported to have about
the same VMT fraction in rural and urban areas. In another example, combination long-haul
trucks have roughly the same VMT distribution as combination short-haul trucks although I
would expect them to have higher VMT fraction on rural restricted access roads. The numbers in
Table 8-2 are derived from the 2011 NEIVI, which is probably the most appropriate source of
this type of data at this time. These numbers may be compared with numbers derived from large-
scale GPS datasets for each source type in the future.
Response: It is difficult to find data on the activity of refuse trucks on a national basis and many
states have trouble finding appropriate data for this category. Similarly, it is difficult to
distinguish long-haul and short-haul trucks in most data sources. We are considering
simplifying our source type categories in future versions of MOVES.
25.1.2.11 Section 9.1 - Light-Duty Average Speed Distributions
It may be of interest to compare some of the average speed distributions derived from TomTom
dataset with those derived from traffic monitoring systems. For example, California has the
Freeway Performance Measurement System or PeMS (http://pems.dot.ca.gov/). Average speed
distributions can be derived using a subset of TomTom data on California freeways and compare
to those derived from PeMS. This would help understand potential biases, if any, in TomTom
data. It is understood this will incur additional analyses (and costs) by TomTom as the raw data
are not provided to ERG and EPA.
Response: Acquiring a California subset of the TomTom data and comparing it to PeMS data is
an interesting idea to under stand potential inherent bias—over sampling of long-distance trips
and trips with routes unfamiliar to the driver. In the hope of making national comparisons, EPA
is actively identifying other datasets like PeMS that represent typical driving in the US to
compare against TomTom data.
In Figure 9-1, it is observed that the highest average speed fraction for urban unrestricted access
road is not in the lowest average speed bin (< 2.5 mph) although one would expect a significant
amount of idle time at signalized intersections. This may be due to the length of intersection
segments being much longer than a typical length of traffic queue, which causes the zero speed
while idling in the queue to be canceled out by relatively higher speeds before joining the queue.
I am not sure how much the shift in this average speed distribution would impact emission
inventories at the national scale. If the impact would be significant enough, these intersection
segments may be divided into shorter segments in future analyses.
Response: Without the raw TomTom data, one can only speculate about the amount of time these
GPS-equipped vehicles spent at urban intersections. We will consider how to improve this
146
-------
information in future data collection efforts. You are correct that the road segments used by
MOVES are much larger than what might be used specifically for intersection modeling. All of
the low speed driving cycles used in MOVES contain a significant amount of idling time that
would reflect the time at idle at signalized intersections.
25.1.2.12 Section 9.2 - Heavy-Duty Average Speed Distributions
The adjustment made in this section is well done.
Response: We have moved the details of the heavy-duty average speed adjustment to the
appendix, and left a summary in the main text.
25.1.2.13 Section 10.2 - Ramp Activity
What data were used to estimate operating mode distributions for ramp activity?
Response: Ramp operating modes in MOVES2014 were not based on data and were carried over
from previous MOVES versions. The distributions were derived from existing operating mode
distributions using engineering judgement. EPA has begun to analyze data from second-by-
second measurements of real world vehicle activity to develop new ramp operating mode
distributions for the next major release of MOVES.
The ramp fraction may be determined using either PeMS or TomTom data. It is understood that
the latter will incur additional analyses (and costs) by TomTom as the raw data are not provided
to ERG and EPA.
Response: TomTom data was not provided with any geospatial coordinates, which would be
necessary for identifying ramps, and acquiring this new level of detail would take additional
analysis and cost beyond the original data pur chase.
25.1.2.14 Section 11.1- National Default Hotelling Rate
The assumptions made in this section can be validated using large-scale GPS datasets of
commercial trucks, for example, the truck GPS dataset maintained by the American
Transportation Research Institute (ATRI) (http://atri-online.org/2014/10/28/truck-gps-data-for-
tracking-freight-flows/).
Response: Many states are also looking into truck hotelling and may be able to provide local
estimates for truck hotelling hours.
25.1.2.15 Section 11.2 - Hoteling Activity Distribution
There are studies that provide data on APU and truck electrification usage that may be
considered in future updates. For example:
• Frey, H. C., P.-Y. Kuo, and C. Villa. (2008). Methodology for characterization of long-haul
truck idling activity under real-world conditions. Transportation Research Part D, 13, 516-
523.
147
-------
• National Renewable Energy Laboratory (NREL)'s Truck Stop Electrification Testing
(http://www.nrel.gov/transportation/fleettest_truck_stop_electrificati on.html).
Response: The current MOVES design does not currently allow for the application of county -
by-county control programs that affect the amount ofhotelling hours for national scale
modeling. However, users may include specific hotelling hour estimates in the HotellingHours
table in their county databases (CDBs). County specific CDBs are normally used for most
official inventory estimates (such as the National Emission Inventory (NEI) and EPA rules,
such as the recent light-duty vehicle Tier 3 rule), so providing appropriate hotelling hour
estimates is done on a case-by-case basis using the available information at the time of the
MOVES runs. EPA has used data provided by states during the NEI process from sources, such
as you cite, to keep the CDBs we use and the inventory estimates generated up-to-date and
accurate.
25.1.2.16 Section 12 - Temporal Distributions
Temporal distributions of VMT rely heavily on the 1996 OHTM report. Traffic monitoring
systems, such as PeMS, may be considered for use as a source of more recent data, especially
for restricted access roads. Note that in the case of PeMS, VMT are estimated separately for
cars and trucks, which can be used to represent light-duty source types and heavy-duty source
types, respectively.
Response: EPA has begun analyzing data from instrumented vehicles as a source for much of
the activity information used by MOVES, including temporal distributions, and would like to
use this new data in future versions of MOVES. We also plan to investigate data from traffic
monitoring systems, such as PeMS.
25.1.2.17 Section 12.1 - VMT Distribution by Month of the Year
Container volumes at ports around the US may be considered for use as a surrogate of VMT
distribution by month of year for short-haul and long-haul combination trucks. For example,
https://www.portoflosangeles.org/mari time/stats.asp.
Response: This is a helpful suggestion, but we think instrumented vehicle data and traffic
monitoring systems will likely be better sources for national average VMT temporal
distributions. We intend to investigate these for a future version of MOVES.
25.1.2.18 Section 12.2 - VMT Distribution by Type of Day
Data from traffic monitoring systems may be used to estimate DayVMTFraction for each
month.
Response: States may and often provide their own data for modeling their areas, so updating
the national average default distributions has not been a high priority. However, as traffic
monitoring and instrumented vehicle data have become more available, we hope to make
updates in the future.
148
-------
25.1.2.19 Section 12.4 - Engine Starts and Parking
More recent instrumented vehicle data are available on NREL's Transportation Secure Data
Center website (http://www.nrel.gov/transportation/secure_transportation_data.html) for
passenger vehicles and on NREL's Fleet DNA website
(http://www.nrel.gov/transportation/fleettest_fleet_dna.html) for commercial vehicles.
Response: EPA has gained access to the NREL FleetDNA dataset but has not yet determined
its applicability for MOVES. EPA is also currently analyzing start and soak information for
light-duty vehicles using a large telematics dataset.
25.1.2.20 Section 12.5 - Hourly Hotelling Activity
In future updates, the hourly hotelling activity may be estimated from large-scale GPS datasets
of long-haul trucks such as ATRI's.
Response: We agree that data from instrumented vehicles will likely provide better estimates of
truck hotelling activity than survey or self-reported data. EPA is exploring a variety of heavy-
duty vehicle activity datasets, including streams collected from GPS and/or OBD devices, to
update MOVES default hotelling information such as hourly rates.
25.1.2.21 Section 14.2 - Road Load Coefficients
The road load coefficients for light-duty vehicles were set to remain constant over time despite
the Light-Duty Greenhouse Gas Rule (because the improvements in these coefficients have
already been incorporated into the energy and emission rates). However, the road load
coefficients for 2014 and later model year heavy-duty vehicles were updated in light of the
2014 Medium and Heavy-Duty Greenhouse Gas Rule. Shouldn't the impact of the 2014 Rule
be expected to reflect in [future energy and emission rates]?
Response: Energy rates have been updated to reflect engine efficiency improvements set out in
the 2014 Heavy-Duty Greenhouse Gas Rule, as described in the MOVES2014 technical report
documenting the heavy-duty vehicle emission rates.5 Please refer to Chapter 5 of the 2014 HD
GHG Rule's Regulatory Impact Analysis (RIA) for further information on modeling emissions
from heavy-duty vehicles in future years.80
25.1.2.22 Section 16 - Conclusion and Areas for Future Research
The national default values for vehicle population and activity inputs in MOVES2014 were
developed for the base year of 2011. It may be of interest to validate the 2012-2014 projections
for some of these inputs with actual data that are available for those years. This will allow the
assumptions made in the projections to be adjusted if necessary.
Response: In the past EPA has validated MOVES default activity estimates according to fuel
volumes reported through tax receipt data from FHWA.83 We hope to do a similar analysis for
MOVES2014.
149
-------
25.2 Randall Guensler, PhD
Dr. Guensler is a Professor at Georgia Institute of Technology's School of Civil and
Environmental Engineering. He has served as chairmen of the Transportation Research Board
Committee on Transportation and Air Quality (1997-2002) and as an advisor on EPA's Mobile
Sources Technical Review Subcommittee of the Clean Air Act Advisory Committee (1995-
2001). Dr. Guensler holds a PhD in Civil Engineering from the University of California at Davis.
25.2.1 General Comments
Thank you for the opportunity to participate in the peer review of the USEP A's Population and
Activity ofOn-road Vehicles in MOVES2014 Documentation. I have provided suggested edits
using revision marks and comments in the margins of the Word document. At various points in
the paper, I have suggested edits to move text explaining tables so that the text appears before the
table is presented. There are a number of sections in the document that I suggest be summarized
in a single paragraph, shipping the detailed text off to an Appendix, to improve readability.
25.2.2 Detailed Comments
The most important issues that I believe could be addressed in the document are summarized
below.
25.2.2.1 Section 1 - Introduction
Somewhere up front in this paper a very brief overview of emissions sources and modeling goals
should be added. How MOVES works, in a nutshell, and what data are needed to run MOVES.
This can also differentiate between baseline emissions by source type and correction factors. VSP
can be addressed here as well as internal driving cycles. Then, the document can refer back to the
general discussion when needed.
Response: We have updated the introduction to provide better context, to make the function of this
report more clear, and to point users to other documents that address many of these questions.
A big picture issue throughout the entire document is to set the stage for the reader as to why they
should be using local-specific or regional-specific data. This is a common theme throughout my
comments.
Response: This report documents how EPA developed national-scale default data related to fleet
characteristics and travel behavior. As explained in the introduction, information on customizing
MOVES with local inputs is detailed in the MOVES2014 Technical Guidance.3
25.2.2.2 Section 2 - MOVES Vehicle and Activity Classifications
The MOVES Vehicle and Activity Classification section really needs an overview designed to
introduce the reader to the content of the Chapter. This overview can help the reader understand
that the emission rates need to be properly linked to the concepts of vehicle classes, vehicle
source types, regulatory classes, etc.
Response: We have added a paragraph to the beginning of this section describing the purpose of
the section and the need to link emission rates to vehicle activity.
150
-------
The paper could probably use a paragraph or two associated with the difficulty in mapping
FHWA vehicle classes and EPA vehicle classes. Papers by Yoon (2006) and Liu (2015) offer
some insight into these issues. Yoon discusses these in the context of visual classes for
observational data, although that paper would need to be updated. Providing this in an Appendix
might prove helpful to users. This applies in [Section] 3 as well.
Response: We have added information in Section 2.2 and Section 6 that discussing the
difficulties of vehicle classification for MOVES modeling and have added citations to the papers
referenced.
There is a problem with MOVES implementation at a higher level that, if resolved, would
significantly improve modeling efforts. As outlined [in Section 2.2] and elsewhere, it is important
to structure MOVES for users to enter mutually exclusive technology groups that can be derived
from license plate observational data. Anything that can be added to the documentation to help
users better classify their vehicle input based upon field observations will be appreciated by users.
Comment 8 also suggests the development of a table to instruct users.
Response: Source types cannot be fully resolved using field observations. Text has been added
to Section 2.2 describing our use of a combination of field measurements along with survey and
registration data. Guidance for users developing local data is provided in the Technical
Guidance.3
Comment 8: Another table here showing the mutually exclusive lines for source type, regulatory
class, and fuel type might be helpful.
Response: Section 2.8 describes the mutually exclusive combinations of source types, fuel types,
and regulatory classes that can be modeled in MOVES2014.
I suggest adding a new section to introduce the use of model year distributions.
Response: An introduction to model year groups has been added, see Section 2.7
There are a number of detailed explanations that probably belong in Appendices rather than in the
text to improve readability (and initial clarity).
Response: Both before and after this peer review, we have moved certain detailed explanations
where appropriate to appendices for improved readability and clarity. In this final report, we
have created appendices on SCC mappings between MOVES versions, forecast and backcast
algorithms for age distributions, average speed distributions of combination trucks, and
responses to peer review comments regarding this report.
The SCC classes are another big picture issue with MOVES, in that these contribute to the
mutually exclusive technology groups. The concept is complex and needs to be explained better
in the text. I suggest the addition of a table for clarity.
Response: The new SCCs do not add another grouping, but just clearly label the specific
combination of technologies. We have changed Section 2.6 to improve the description of the
SCC classes. A table has been added to Section 21 (Appendix E: SCC Mappings) to compare the
MOVES regulatory classes to the old SCC categories to help map between them.
151
-------
The audience needs a connection between SCC and regulatory class in Section 2.6. At the same
time, Table 2-5 loses the audience due to complexity. An overview paragraph would help here.
This is one of the most complicated sections and general improvements would help the audience.
Specific comments are provided in the document markup.
Response: The new SCCs are not based on regulatory class. The complexity comes when trying
to use regulatory class to map back to the old SCCs. We have revised this section and Table 2-5
to make the discussion clearer.
Table 2-8 appears to be the key table for the entire chapter. If the text is rewritten, I would
suggest pointing all of the explanations and discussions so that they result in the reader reaching
the table with full understanding of the content of that table. A paragraph is needed after Table
2-8 to let the reader know that everything they do from here on out is to generate the data that will
be used by the 80 groups represented in this table.
Response: We do not consider Table 2-8 to be of primary importance. We have revised the entire
section to make it clearer that the aim of this section is to define MOVES terminology and that
later sections talk about how we actually estimate populations and activity for the different
categories. Table 2-8 simply identifies the vehicle technology combinations for which we have
emissions data.
Table 2-9 [now Table 1-1] is excellent and can be used to organize the presentation of materials
before and after. Listing in order of use in the document, rather than alpha order, will help the
structure.
Response: We have moved this entire section, including the table to the introduction to help
provide context for all of the paper. Listing the MOVES table names in Table 1-1 alphabetically
makes it more useful for reference purposes. The table of contents will assist the reader in
understanding the order of how topics will appear in the report.
25.2.2.3 Section 3 - Data Sources
Data sources introduction should be expanded significantly to inform the reader about what they
need for modeling. Given the sensitivity and capabilities of MOVES, A goal here should be to
shift users to locally-sourced data rather than national defaults.
Response: The data sources described in Section 3 were used to develop national default values
of activity-related inputs found in the MOVES2014 database. Users are encouraged to
incorporate locally-source data for project- or county-scale modeling where possible. As
indicated in the report's introduction, there is separate technical guidance that provides details
on developing locally-sourced inputs for project- and county-scale modeling.3 We have now
repeated that information at the beginning of Section 3.
25.2.2.4 Section 4 - VMT by Calendar Year and Vehicle Type
As indicated in Comment 34, buses and HD Trucks experience different growth rates. A
separate data source should be found for the next set of updates. At the very least, local data
should be recommended for buses of all types (these data can be obtained from transit agencies).
152
-------
Comment 34: This seems a bit shaky. Bus activity growth will not parallel HDV truck growth.
There are completely different causal factors in play. Single unit and large trucks are also likely
to grow at different rates. Not sure I have a reasonable alternative to propose though. At least the
uncertainty should be acknowledged here.
Response: Lacking better information, it is unclear how VMTgrowth in buses will relate to
growth in other heavy-duty vehicle types. We agree that the uncertainty of assuming bus growth
follows heavy-duty freight growth should be acknowledged and have added text to that effect.
25.2.2.5 Section 5 - Vehicle Populations by Calendar Year
Changes in vehicle ownership and mileage accrual rates are generally different. These sources
can be obtained from registration databases coupled with I/M programs. This would be a
worthwhile small study to sponsor.
Response: MOVES does not currently use vehicle registration databases from I/M areas for
relative MARs or survival rates of light-duty vehicles. Only using data from a small sample of
selected areas may generate issues of representativeness, but we are interested in reassessing the
viability of registration data from I/M areas in the future.
25.2.2.6 Section 6 - Fleet Characteristics
The materials presented [in reference to modeling flexible fuel vehicles running on either
gasoline or E85 in Section 6.2.1] (Comment 41) are very confusing for the reader and serve to
reinforce the need for users to obtain their own regional/local input data. The discussion can be
simplified for clarity or expanded with detail for clarity.
Comment 41: This whole paragraph [discussing flexible fuel vehicles] is confusing (and
reinforces the need to allow the user to provide direct inputs rather than relying on the internal
algorithms for assignment), as noted earlier. If this paragraph remains, it should either be
expanded to provide the exact details of the internal method, or reduced to avoid confusion.
Response: The text has been revised to add a footnote that more clearly explain that information
on MOVES vehicle fuel consumption comes from two separate fractions. The fuel type fraction
delineates whether the vehicle is capable of running on single or multiple fuels and the fuel
usage fraction describes how much of the total fuel consumed is a specific fuel, particularly for
light-duty flexible fuel vehicles that are using E8 5 or gasoline. As noted in the text, discussion on
fuel usage is in a separate technical report on the MOVES2014fuel supply.6 Given the spatial
variation ofE85 use, user-supplied data would be preferable over national defaults for localized
flexible fuel vehicle modeling.
Comment 48 identifies an internal problem in MOVES that causes problems for users in matching
local fleet composition.
Comment 48: This is the problem [with the SampleVehiclePopulation table] identified earlier
associated with single fixed assignments inside MOVES. The user cannot control these
allocations later in the process. See the discussion in Liu et al., 2015.
153
-------
Response: While the AVFT importer is designed to make it easy for users to change fuel type
fractions by model year, we thought it unlikely that many users would have information on
regulatory class distributions, so we did not create an importer for this data.
25.2.2.7 Section 7 - Vehicle Characteristics that Vary by Age
The discussion on survival modeling could be significantly improved (see comments) and
caveats should be added. A number of comments are also provided on model year distribution
values, especially for the oldest vehicle groups. Plus, the detailed text in this section would fit
better as an appendix. A focused peer review of this section is probably warranted (see
comments).
Comment 55: This is a fairly weak justification.... If a method is applied later, stipulate the
method and basis here.
Response: To thoroughly model scrappage in MOVES, we need projections of vehicle retirement
by source type and age for every calendar year 2012+. This is the justification for using base
survival rates from TEDB and scaling them as necessary based on sales and VMT projections.
The text here has been amended to include a more detailed justification.
Comment 56: As noted below, a reader needs to see plots over time here to assess impact of
assumptions. This can be done by overlaying future fleets by calendar year. The 30+ group
should be growing slightly, or remaining stable, rather than shrinking over time, as folks hold
onto vehicles longer. These are 1985 and older vehicles today. Vehicles in today's fleet are more
durable. Need to reassure users that the failure rate assumption is reasonable with an independent
confirmation.
Response: MOVES2014 has generic survival curves by vehicle classifications independent of
model year. With this limited data, we chose to scale these generic survival curves to ensure that
changes in population would be evident over time. We do not have enough data to adequately
assess how 30+ year old vehicle populations are changing over time, but this is a topic of
interest for EPA. We have added a figure towards the end of Section 20.1 (in Appendix^):
Detailed Derivation of Age Distributions^ with overlaying future fleets. Due to our assumptions
in the national case, most source types do not see an uptick in 30+ vehicles (combination trucks
being the exception). However, as noted in the paragraph regarding the user tool, projected
local age distributions retain their 30+ population fractions.
Comment 58: Need to be careful here. Differential retirements in model year groups could
represent technology durability/acceptability issues. Need to double-check prior to discounting
sources. Caveat whole paragraph by reassuring the audience that you did the best you could and
that users can specify their own future fleets and ignore the retirement rates.
Response: We agree that double-checking is necessary before discounting sources, which is why
5-year survival rates derived from V1US 1992 and 1997 were used to justify our choice. The
paragraph is also caveatedwith the acknowledgement of how limited the data are. We provide
user guidance in separate documents.
154
-------
Comment 60: Showing a survival curve is a good idea. Survival curves can be developed
separately for various technology groups if desired.
Response: Vehicle survival rates by age andHPMS class are shown in Table 7-2.
Comment 61:1 suggest moving all of the text from this point down to the end of the section into
an Appendix. It is not needed here and adds to reader confusion for something that is rarely
needed by a user. The bottom line is that you have made adjustments to the rates to try to help
the predictions match the data without adjusting rate parameters. You can say that here and refer
the user to an appendix.
Response: We agree and have moved the detailed description of implementing the algorithm to
Appendix D (Section 20).
Comment 62: Based on this text, it looks like the 0.3 value has been discarded, which is probably
a good thing. This needs to be clarified and applicable text in this section corrected as needed.
Response: In the Age Distribution Projection Tool, which users are encouraged to use to project
local age distributions into the future, this is correct: the 0.3 value has been discarded. However,
this is only true for the tool; the national analysis retains this assumption.
Comment 63: It is not clear why [for] 1999-2010 this needs to be modeled with a survivor model
at all, rather than simply interpolated between your 1990 and 2011 data sets, using sales figures
for control given that there are no better data in between). Given that survival rates are so
different across the country (e.g. New England vs. Arizona), and by technology as it entered the
fleet, I'm not convinced that the detailed approach is warranted.
Response: Our chosen approach provides a consistent method for generating age distributions
across all analysis years in MOVES. Again, our user guidance encourages the use of local age
distributions for regional or state-level analysis.
Comment 64: Given the 140 pages to review, there is not enough time to perform a full technical
analysis of Section 7.1.1 would suggest that the equations be sent out for a separate and focused
peer review... Finally, the users need to be reassured that they can specify the composition of
the future fleet off model using their own sales and survival functions.
Response: Thank you for the suggestion. Work on vehicle sales growth and retirement rates was
presented as a poster at the 25th Coordinating Research Council (CRC) Real World Emissions
Workshop in March 2015. EPA has discussed the possibility of publishing this research in a
peer-reviewed journal sometime in the future. Our guidance documents allow users to employ
different methodologies to generate age distributions.
Mileage accrual for the older vehicles is also a potential issue (see comments).
Comment 70: You have confounding effects from the recession in Section 7.2. The traffic
volumes on freeways declined significantly during that period, but have been on the rise. MARs
warrant a double check with post-2009 data. Given the vehicle purchase delays, the
155
-------
accumulation rates will likely vary even more by model year cluster. I don't have a better
answer, but I question the stability assumption.
Response: We agree with the assertion that the 2001 and 2009 data are similar does not address
the issue of whether the 2009 data is anomalous in regards to a trend in mileage accumulation.
However, without a full analysis of the trend, we are left with the choice of using the 2001 data
or the 2009 data. Text has been added to the explanation of our choice to continue to use the
2001 data—since it was unaffected by the 2008-2009 recession—in the report.
Comment 72: These are fairly significant assumptions that cannot be verified from the
information provided. Older vehicles are relegated to different service activities, so these are
important assumptions to verify, especially given the age of the 1992 TIUS data.
Response: Information about the oldest model years in the fleet are scarce, and due to the
smaller number of vehicles, any data will have significant variability. It will be difficult to reduce
the uncertainty in the mileage accumulation rates for the oldest vehicles without significant
effort. Without clear information that the activity of these older vehicles has changed
dramatically in recent years, updating this information will have a low priority.
The Single unit long-haul truck distribution in Figure 7-1 is so different than the other curves that
it warrants a detailed explanation.
Response: All the age distributions for single unit trucks were constructed primarily from
national registration data, but there was a particular small sample of long-haul trucks in
VIUS2002, especially in more current years. Due to this small sample size, the long-haul single
unit age distribution is probably affected greatly by minor changes in population.
The Cubic Regression approach [in Section 7.2.2] is not clearly defined.
Response: The text of Section 7.2.2 was updated to state clearly that the regression used was a
simple cubic fit and not, for example, a spline fit. Given the very good fit reported by NHTSA
and the use of the regression in their model, we did not see a need to reconsider this choice.
Table 7-3 is good. Similar tables should be provided for other classes.
Response: We have added a table to the report with the raw mileage accumulation rates for
passenger cars and light-duty trucks copied from the NHTSA reference document.
I could not replicate the data in Table 7-6. Please see comments.
Response: We have updated the text explaining how the data was generated to include more
details. The updated method will produce the reported statistics.
25.2.2.8 Section 8 - Average Speed Distributions
I have some expertise in the availability and resolution of TomTom data. The use of these data as
outlined in the document appears problematic. Comments are provided throughout Section 9.1
and 9.21 cannot recommend the use of these data in this fashion. I recommend that additional
research in this area be undertaken.
156
-------
Comment 79: You have nailed most of the potential biases in the bullets [in Section 9.1]. This
was enough to keep us from using these data for our research efforts. There are significant
vehicle class, lane choice, operating condition, and geographic biases that likely result. Given the
tremendous sensitivity of MOVES to the selected duty cycle, I am not inclined to recommend the
use of the derived average speed for hours or selection of driving cycle weightings without much
more information to evaluate this effort and comparative studies with other data sources.
Response: We agree that there are limitations to this dataset, which is why we listed the biases in
the report. However, using the default nationally representative average speed distribution from
TomTom GPS data is a substantial improvement over the previous default average speed
distributions inMOVES2010. We added a sentence in the introduction paragraph of Section 9
that discusses the state of the default average speed distribution inMOVES2010, and a citation
to a MOVES FAC A work group presentation that included comparisons of the updated the
TomTom average speed distributions with theMOVES2010 average speed distributions. The
MOVES2010 average speed distributions for urban areas were based on average speed
estimates from a survey of urban travel demand models, and the rural speeds were based on
chase-car studies conducted in California. Changing the average speed distribution does not
always have a large impact on the vehicle emissions. We have added text in Section 9.2 that
discusses cases in which MOVES is not strongly sensitive to changes in the average speed
distributions.
Comment 80: This [assumption of MOVES average speed distributions based on the average
speed in each roadway segment, not second-by-second speed measurements] is even more
troubling, because it indicates that the most disaggregate TomTom data were not employed.
Response: The average speed distribution is intended to be the average distribution of the
average link speeds within a modeling domain. It is not intended to be the average speed
distribution of a collection of second-by-second speeds.
Comment 81:1 cannot recommend this [TomTom average speed analysis] or use of the results.
Another independent data source is needed to verify these results. Naturalistic driving data
[SHRP 2 Naturalistic Driving Study] or ATRI [truck GPS data from the American
Transportation Research Institute] data.
Response: We did not have another comparable telematics or data set available to compare the
results to the TomTom dataset. We agree this would be a valuable comparison going forward. As
we mention in the Introduction and the introductory paragraph of Section 3, local users have the
option to (and should wherever possible) replace the default activity inputs, including average
speed distributions, with their own local data. Please also see the comments above regarding the
sensitivity of average speed distributions on results.
Comment 82: A consistent method at the national level can have a significant bias and still be
useful, as long as the bias is consistent over time. That is, you can look at percentage changes
over time and even if the magnitude of the predicted value is consistently off by 20%, the results
are useful. The problem here [with using TomTom data to represent national default average
speed distribution in MOVES] is that regional agencies will likely use the same distributions in
157
-------
county or regional El development. I would suggest that the guidance here inform regional and
project level users that they need to develop their own speed distributions.
Response: The intention of this report is to document the default population and activity in
MOVES. Recommended modeling approaches for regional and project level users is contained
in the technical guidance documents.3 For example, in the development of SIP and regional
transportation conformity analysis, a local speed distribution is a necessary input.
Comment 83: Trucks operate in the two right-hand lanes. Field studies clearly show that the
speed distributions in these lanes are very different than inside lanes, and trucks speed
distributions can also differ in these lanes. A more appropriate data source is the ATA data set
collected from trucks. I do not recommend "adjusting" the TomTom data for use here. You also
do not need to show all of the equations below to tell the audience that you manually adjusted the
values by [tell the audience in one paragraph of text]. You would be better off just showing the
initial and shifted results in a comparative table. Providing all of these equations is an oversell of
the quality of the data and the assumptions made. These equations can be moved to an Appendix
if you decide move forward with this method.
Response: We have added text to Section 9.2 that points out that it would be preferable to have
heavy-duty specific data on heavy-duty trucks, including combination trucks. We also state that
we will prioritize updating the heavy-duty specific average speed data in the future. We have
also added information regarding the energy and emissions sensitivity of the average speed
adjustment on combination truck speeds. In addition, we have moved the equations and detailed
discussion to an appendix.
25.2.2.9 Section 10 - Driving Schedules and Ramps
The use of the driving cycle weighting is an issue in MOVES (see comment 92 and 93). Use of
local driving cycles is preferable when such data are available.
Comment 92: All of this assumes that the driving cycles are representative of these average
speed cutpoints. I agree that the approach is probably better than the previous approach of using
a "closed" cycle, but no compelling argument has been made that the weighting of the cycles
employed in the latest algorithms matches real world composite driving for a facility. Some of
the cycles were generated to make sure that we have adequate emission rate data for the model
bins, not necessarily to be representative of onroad operations. [Tis] is not as big a deal at the
national level (provided that all analyses backcast emissions for previous years and do not mix
these outputs with the results of previous analyses that employed MOVES2010). However, there
is no compelling reason to advocate that this default approach be used in regional or local
analyses without corroboration.
Response: The driving schedules used by MOVES are derived from real world driving behavior
rather than emission certification cycles, and while it is difficult to assure they are representative
of all driving in the associated bin, we feel they are reasonable estimates for national defaults.
In addition, MOVES is specifically designed to accept the use of local driving schedules and
operating modes when this type of data is available, and EPA guidance explains that this type of
data is preferred, especially for project-level analysis.89
158
-------
Comment 93: The creep cycle was designed to assess emission rates for high inertial load lug
operations required to get freight loads moving at low speeds (in freight yards as I recall).
Matching this by average speed bin, based upon TomTom data, and weighting that bin may be a
huge stretch and may even overstate emissions. Unfortunately, the only way to assess whether
the method is viable is to do verification data collection, probably by extensive video analysis.
Response: Trucks driving at extremely low speeds (similar to the average speed of the creep
cycle) will experience inertial load lug operations similar to those occurring in the creep cycle,
due to stop-and-go operation. At the national and regional level, most truck operation will occur
at speeds much higher than the creep cycle, so that any differences between the creep cycle and
actual low-speed truck behavior will have little effect on overall emission estimates for trucks. In
cases where the focus of the analysis is specifically on low speed operation, EPA recommends
using project-level analysis with user supplied driving schedules.
It is not clear to users how they should handle activity on weaving and exit lanes. Comments [97-
99] address this issue.
Comment 97: It is unclear whether the schedule includes any activity on weaving lanes (lanes that
run between an entry ramp and the next exit ramp when ramps are close together). My assumption
has always been (based upon Sierra Research presentation years ago) that weaving areas upstream
of ramps were part of the freeway activity (and freeway driving cycles) and that ramps began at
the gore area. Is there any way to confirm this and state it in the text?
Response: The text in Section 10.2 has been updated to make is clear that the total operation of
vehicles on restricted roads (freeways) was divided so that ramp activity could be separated.
Activity that occurs on the freeway in anticipation of ramps or occurring after entry is included in
the non-ramp freeway.
Comment 98: It would be helpful to establish how these distributions were developed. A clear
definition of start and end of ramp is warranted for user application. Perhaps some diagrams
would support this. As I recall, the ramp cycles used car following data collected from gore area
to the arterial and vice-versa, including any off-freeway weaving areas. It may be important to let
the reader know that the HCM "area of influence" (about 450m upstream and downstream of the
ramp) is not included in ramp activity but in freeway activity.
Response: The ramp operating mode distributions used in MOVES2014 are not based on data
collected from ramp activity. The existing set of operating modes for ramps were selected to
represent the different average speed bins. Updating the handling of ramps related to freeway
driving is a high priority for future versions of MOVES.
25.2.2.10 Section 12 - Temporal Distributions
Section 12.3 provides defaults for temporal distributions. Again, local data are preferred given
the variability noted across urban areas.
Response: EPA guidance generally advocates using local information in preference to MOVES
defaults in almost all cases. MOVES specifically includes an importer for user inputs for
temporal allocations in the County Data Manager and Project Data Manager to make it easier
for users to provide their own data.
159
-------
26 References
1 Motor Vehicle Emission Simulator (MOVES) technical reports are available on the US Environmental
Protection Agency website (http://www.epa.gov/oms/models/moves/moves-reports.htm). Please refer to
the MOVES2009 or 2010 report for any MOVES2014 report that has not been finalized and published.
2 US Environmental Protection Agency (EPA), Office of Transportation and Air Quality (OTAQ),
Population and Activity ofOn-road Vehicles inMOVES2014: Draft Report, EPA Science Inventory,
Record ID 309336, Ann Arbor, MI: July 2015,
http ://cfpub .epa.gov/si/si_public_record_report.cfm?dirEntryId=3 093 3 6.
3 US EPA, MOVES2014 Technical Guidance: Using MOVES to Prepare Emission Inventories for State
Implementation Plans and Transportation Conformity, EPA-420-B-15-007, Ann Arbor, MI: January
2015, http://www.epa.gov/otaq/models/moves/documents/420bl5007.pdf
4 US EPA, Exhaust Emission Rates for Light-Duty On-road Vehicles inMOVES20l4, EPA-420-R-15-
005, Ann Arbor, MI: October 2015, http://www3.epa.gov/otaq/models/moves/documents/420rl5005.pdf
5 US EPA, Exhaust Emission Rates for Heavy-Duty On-road Vehicles inMOVES20l4, EPA-420-R-15-
004, Ann Arbor, MI: November 2015,
http://www3.epa.gov/otaq/models/moves/documents/420rl5015a.pdf.
6 US EPA, Fuel Supply Defaults: Regional Fuels and the Fuel Wizard in MOVES2014, EPA-420-R-15-
002, Ann Arbor, MI: 2015, http://www.epa.gov/otaq/models/moves/moves-reports.htm.
7 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/420p09004.pdf.
8 US EPA, MOVES2014 Module Reference, last updated 7 October 2014,
http://www.epa.gov/otaq/models/moves/moves-algorithms.htm.
9 US Census Bureau, 7997 Vehicle Inventory and Use Survey, EC97TV-US, Washington, DC: October
1999, https://www.census.gov/svsd/www/vius/1997.html.
10 US Census Bureau, 2002 Vehicle Inventory and Use Survey, EC02TV-US, Washington, DC: December
2004, http://www.census.gov/svsd/www/vius/2002.html.
11 IHS, Inc. (formerly RL. Polk & Co.), National Vehicle Population Profile®, Southfield, MI; 1999,
https://www.ihs.com/btp/polk.html.
12 IHS, Inc. (formerly RL. Polk & Co.), Trucking Industry Profile TIP® Vehicles in Operation,
Southfield, MI: 1999, https://www.ihs.com/btp/polk.html.
l6o
-------
13 US Federal Highway Administration (FHWA), Office of Highway Policy Information (OHPI), Table
MV-1, "State Motor-Vehicle Registrations," Highway Statistics 2011, Washington, DC: March 2013,
https://www.fhwa.dot.gov/policyinformation/statistics/2011/mvl .cfm.
14 US FHWA, Table MV-10, "Bus Registrations," Highway Statistics 2011, Washington, DC: December
2012, https://www.fhwa.dot.gov/policyinformation/statistics/2011/mvlO.cfm.
15 US FHWA, Table VM-1, "Annual Vehicle Distance Traveled in Miles and Related Data - 2011 by
Highway Category and Vehicle Type," Highway Statistics 2011, Washington, DC: March 2013,
https://www.fhwa.dot.gov/policyinformation/statistics/2011/vml.cfm.
16 US FHWA, Table VM-2, "Functional System Travel," Highway Statistics 2011, Washington, DC:
March 2014, https://www.fhwa.dot.gov/policyinformation/statistics/201 l/vm2.cfm.
17 US Federal Transit Administration (FTA), "Revenue Vehicle Inventory: Details by Transit Agency
(Form A-30)," National Transit Database, 1999-2011, http://www.ntdprogram.gov.
18 Bobit Publications, School Bus Fleet Fact Book, Torrance, CA: 1997, 2009, and 2012,
http://www.schoolbusfleet.com.
19 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, http://www.epa.gov/otaq/models/mobile6/m6flt002.pdf.
20 US EPA, MOBILE6, http://www.epa.gov/otaq/m6.htm,
21 US Energy Information Administration (EIA), Annual Energy Outlook 2014, DOE/EIA-0383(2014),
Washington, DC: April 2014, http://www.eia.gov/forecasts/aeo/pdf/0383(2014).pdf.
22 Davis, S., Diegel, S., and Boundy, R., Oak Ridge National Laboratory (ORNL), Center for
Transportation Analysis, Transportation Energy Data Book (TEDB) Edition 32, Oak Ridge, TN: 2013,
http://cta.ornl .gov/data/index. shtml.
23 US FHWA, Vehicle Travel Information System (VTRIS), last updated 21 March 2012,
http://www.fhwa.dot.gov/ohim/ohimvtis.cfm.
24 Motorcycle Industry Council, Statistical Annual, Irvine, CA: 1997, 2009, and 2012,
http://mic.org/StatAnnual.aspx.
25 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.
26 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.
27 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/seungju_yoon_200508_phd.pdf.
28 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.
29 US FTA, National Transit Database, "NTD Glossary," last updated 19 February 2015,
http://www.ntdprogram.gov/ntdprogram/Glossary.htm.
30 US FFiWA, Office of Planning, Environment, and Realty, "Planning Glossary," updated 21 March
2012, http://www.fhwa.dot.gov/planning/glossary/glossary_listing.cfm?sort=defmition.
31 US Department of Transportation (DOT) Bureau of Transportation Statistics, "Appendix B -
Glossary," National Transportation Statistics, Washington, DC: April 2015,
http://www.bts.gov/publications/national_transportation_statistics.
32 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.
33 Davis, S. and Truitt, L., ORNL, Investigation of Class 2b Trucks (Vehicles of 8,500 to 10,000 Ibs
GVWR), ORNL/TM-2002.49, Oak Ridge, TN: March 2002, http://www-
cta.ornl.gov/cta/Publications/Reports/ORNL_TM_2002_49.pdf.
34 US EIA, "Light-Duty Vehicle Sales by Technology Type," reference case, Annual Energy Outlook
2014, released April 2014, http://www.eia.gov/oiaf/aeo/tablebrowser.
35 US EPA, Greenhouse Gas and Energy Consumption Rates for On-Road Vehicles: Updates for
MOVES2014, EPA-420-R-15-003, Ann Arbor, MI: October 2015,
http://www3.epa.gov/otaq/models/moves/documents/420rl5003.pdf.
36 US EPA, Brake and Tire Wear Emissions from On-road Vehicles inMOVES2014, EPA-420-R-14-018,
Ann Arbor, MI: November 2015, http://www3.epa.gov/otaq/models/moves/documents/420rl5018.pdf.
37 US EIA, "Transportation Sector Energy Use by Fuel Type Within a Mode," reference case, Annual
Energy Outlook 2014, released April 2014, http://www.eia.gov/oiaf/aeo/tablebrowser.
38 Data from 2011 was provided by US FHWA to replace Table II.2 and II.3 in the 1997 Federal Highway
Cost Allocation Study (http://www.fhwa.dot.gov/policy/hcas/final/index.htm) specifying vehicles by
weight (email correspondence), January 2013.
39 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/fdsys.
40
Union of Concerned Scientists (personal communication), http://www.ucsusa.org.
41 Bryan, M., Recreational Vehicle Industry Association (personal communication), October 19, 2009,
http: //www. rvia. org.
162
-------
42 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,
http://www.epa.gov/oms/models/moves/documents/420rl2022.pdf.
43 National Highway Traffic and Safety Administration (NHTSA), "Vehicle Survivability and Travel
Mileage Schedules," DOTHS 809 952, Springfield, VA: January 2006, http://www-
nrd.nhtsa.dot.gov/Pubs/809952.pdf
44 The Age Distribution Projection Tool for MOVES2014 is a macro-enabled spreadsheet and is available
for download at http://www.epa.gov/oms/models/moves/tools.htnrffleet-2014.
45 American Bus Association, Motorcoach Census 2000, conducted by R. L. Banks and Associates, Inc.,
Washington, DC: July 2000.
46 Browning, L., Chan, M., Coleman, D., and Pera, C., ARCADIS Geraghty & Miller, Inc., Update of
Fleet Characterization Data for Use inMOBILE6: Final Report, EPA420-P-98-016, Mountain View,
CA: 11 May 1998, http://nepis.epa.gov/Exe/ZyPDF.cgi?Dockey=P1001ZUK.PDF.
47 TL Enterprises, Inc., Good Sam Club, Highways Member Study 2000, Ventura, CA: 2000.
48 US EPA, "The 2011 National Emission Inventory (NEI)," Technology Transfer Network:
Clearinghouse for Inventories & Emission Factors, last updated 5 June 2015,
http://www.epa.gov/ttn/chief/net/201 linventory.html.
49 US FHWA, Tables VM-1 and VM-2, "Annual Vehicle Distance Traveled," and "Functional System
Travel," Highway Statistics 2011, Washington, DC: March 2014,
https://www.fhwa.dot.gov/policyinformation/statistics/2011.
50 Data from EPA's 2011 NEI Version 1 for source type VMT fractions by road type can be downloaded
directly at ftp://ftp.epa.gov/EmisInventory/2011/doc from the zip file labeled,
"201 lnei_supdata_or_VMT.zip," last updated 21 August 2013.
51 US EPA, MOVES2010 Highway Vehicle Population and Activity Data, EPA-420-R-10-026, Ann
Arbor, MI: November 2010. http://www.epa.gov/otaq/models/moves/420rl0026.pdf.
52 Brzezinski, D., Updates to MOVES Vehicle Activity, FACA MOVES Review Workgroup, 9 July 2013,
http://www.epa.gov/otaq/models/moves/faca.htm
53 Boriboonsomsin, K., Zhu, W., and Barth, M., "Statistical Approach to Estimating Truck Traffic Speed
and Its Application to Emission Inventory Modeling," Transportation Research Record (2233), pg. 110-
119, Washington, DC: January 2011, http://trrjournalonline.trb.org.
54 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.
55 Sierra Research, Inc., Development of Speed Correction Cycles, M6.SPD.001, EPA 68-C4-0056,
Sacramento, CA: 26 June 1997, http://www.epa.gov/oms/models/mobile6/r01042.pdf
56 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, http://epa.gov/oms/models/ngm/r02026.pdf
163
-------
57 DieselNet, "Emission Test Cycles: New York Bus," last accessed 8 June 2015,
https://www. dieselnet.com/standards/cycles/nybus.php.
58 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/36355.pdf.
59 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,
http://www.crcao.org/reports/recentstudies2005/E55-2%20FINAL%20REPORT%20071205.pdf.
60 Sensors, Inc., On-Road Emissions Testing of 18 Tier 1 Passenger Cars and 17 Diesel Powered Public
Transport Buses, QT-MI-01-000659, Saline, MI: 22 October 2002,
http://www.epa.gov/oms/models/ngm/r02030.pdf.
61 Eastern Research Group, Inc., Roadway-Specific Driving Schedules for Heavy-Duty Vehicles, EPA420-
R-03-018, Austin, TX: August 2003, http://nepis.epa.gov.
62 US EPA, Greenhouse Gas Emissions Model (GEM) for Medium- and Heavy-Duty Vehicle
Compliance, Simulation Model v2.0.1, May 2013, http://www.epa.gov/otaq/climate/gem.htm.
63 Systems Applications International, Inc., Development of Methodology for Estimating VMT Weighting
by Facility Type, M6.SPD.003, EPA420-R-01-009, Fairfax, VA: April 2001,
http://www.epa.gov/oms/models/mobile6/r01009.pdf.
64 US Government Publishing Office (GPO), Code of Federal Regulations, Title 49 - Transportation, Vol.
65, No. 85, CFR Parts 350, 390, 394, 395 and 398, 2 May 2000, http://www.gpo.gov/fdsys.
65 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.
66 Motorcycle crash data from NHTSA, Fatality Analysis Reporting System (PARS),
http://www.nhtsa.gov/FARS; raw 2010 data last updated 11 December 2012, ftp://ftp.nhtsa.dot.gov/fars.
67 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/P100AE2E.pdf
68 US EPA, Evaporative Emissions from On-road Vehicles inMOVES2014, EPA-420-R-14-014, Ann
Arbor, MI: September 2014, http://www.epa.gov/otaq/models/moves/documents/420rl4014.pdf
69 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.
70 US EPA, Development of Evaporative Emissions Calculations for the Motor Vehicle Emissions
Simulator MOVES2010, EPA-420-R-12-027, Ann Arbor, MI: September 2012,
http://www.epa.gov/otaq/models/moves/documents/420rl2027.pdf.
164
-------
71 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, http://trrjournalonline.trb.org.
72 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.
73 US FHWA, Highway Performance Monitoring System Field Manual, OMB 2125-0028, Washington,
DC: December 2000.
74 Vehicle miles traveled obtained from Version 5 ("VMT_NEI_vl_2011_21aug2013_v5") of the VMT
estimates from 2011 NEIVI, http://www.epa.gov/ttn/chief/net/201 linventory.html.
75 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/fdsys.
76 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.
77 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, http://www.crcao.org/workshops/index.html.
78 US EPA, IM240 andEvap Technical Guidance, EPA420-R-00-007, Ann Arbor, MI: April 2000,
http://www.epa.gov/otaq/regs/im/r00007.pdf.
79 Petrushov, V., "Coast Down Method in Time-Distance Variables," SAEInternational, SAE 970408,
Detroit, MI: 24 February 1997, http://www.sae.org.
80 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, http://www.epa.gov/otaq/climate/documents/420rl 1901.pdf
81 US EPA, Emission Adjustments for Temperature, Humidity, Air Conditioning, and Inspection and
Maintenance for On-road Vehicles inMOVES2014, EPA-420-R-14-012, Ann Arbor, MI: December
2014, http://www.epa.gov/otaq/models/moves/documents/420rl4012.pdf.
82 Koupal, J., Air Conditioning Activity Effects inMOBILE6, M6.ACE.001, EPA420-R-01-054, Ann
Arbor, MI: November 2001, http://www.epa.gov/otaq/models/mobile6/r01054.pdf
83 Choi, D. and Koupal, J., US EPA, "MOVES Validation Efforts To Date," presentation to MOVES2014
Federal Advisory Committee Act (FACA) Review Work Group, Ann Arbor, MI: 12 July 2012,
http://www3.epa.gov/otaq/models/moves/documents/faca-meeting-july2012/03-moves-validation-
faca.pdf.
84 Ward's Automotive Inc., http://www.wardsauto.com.
165
-------
85 American Public Transportation Association (APTA), Public Transportation Fact Book, 58th Ed.,
Washington, DC: May 2007, http://www.apta.com.
86 Horie, Y., Tranby, C., and Sidawi, S., Valley Research Corporation, Tables 3-9 & 2-2, On-RoadMotor
Vehicle Activity Data: Volume I - Bus Population and Activity Pattern, Final Report, California Air
Resources Board Contract A132-182, Northridge, CA: September 1994,
http://o3.arb.ca.gov/research/apr/past/al32-182.pdf.
87 O'Rourke, L. and Churchill, C., ICF International, Peer Review of May 2015 Vehicle Population and
Activity Update Report, EPA Science Inventory, Record ID 309336, Cambridge, MA: 28 September
2015, http://cfpub.epa.gov/si/si_public_record_report.cfm?dirEntryId=309336.
88 Natural Gas Vehicles for America, 2014 NGVProduction and Sales Report, Washington, DC: 24
March 2015, https://www.ngvamerica.org/vehicles/2014-ngv-production-and-sales-report.
89 US EPA, UsingMOVES20l4 in Project-Level Carbon Monoxide Analyses, EPA-420-B-15-028, Ann
Arbor, MI: March 2015, http://www3.epa.gov/otaq/stateresources/transconf/documents/420bl5028.pdf
166
------- |