Evaporative Emissions from
Onroad Vehicles in MOVES3
gPk United States
Environmental Protection
^1 Agency
-------
Evaporative Emissions from
Onroad Vehicles in MOVES3
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.
£% United States
Environmental Protection
^1 Agency
EPA-420-R-20-012
November 2020
-------
Table of Contents
1 Background 1
2 Test Programs and Data Collection 6
3 Design, Analysis and MOVES Inputs 7
3.1 Fuel Tank Temperature Generator 7
3.1.1 Fuel Tank Temperature for Hot and Cold Soaks 9
3.1.2 Fuel Tank Temperature while Running 11
3.2 Permeation 12
3.2.1 Temperature Adjustment 15
3.2.2 Fuel Adjustment 16
3.3 Tank Vapor Venting 17
3.3.1 Altitude 18
3.3.2 Cold Soak 19
3.3.3 Hot Soak 33
3.3.4 Running Loss 41
3.4 Inspection/Maintenance (I/M) Program Effects 44
3.4.1 I/M Factor (Relative Program Effectiveness) 45
3.4.2 Leak Prevalence 47
3.5 Liquid Leaks 48
3.6 Refueling 50
Appendix A Notes on Evaporative Emission Data 54
Appendix B Tank Fuel Generator 56
Appendix C List of Acronyms 60
Appendix D Glossary 61
References 63
-------
1 Background
EPA's Office of Transportation and Air Quality (OTAQ) has developed the Motor Vehicle
Emission Simulator (MOVES). The MOVES model estimates emissions for mobile sources
covering a broad range of pollutants and allows multiple scale analysis. MOVES currently
estimates emissions from cars, trucks and motorcycles.
Evaporative processes can account for a significant portion of gaseous hydrocarbon emissions
from gasoline vehicles. Volatile hydrocarbons evaporate from the fuel system while a vehicle
is refueling, parked or driving. MOVES does not include estimates for emissions from non-
fuel sources such as window washer fluid, paint, plastics, and rubber. Evaporative processes
differ from exhaust emissions because they don't directly involve combustion, which is the
main process driving exhaust emissions. For this reason, evaporative emissions require a
different modeling approach. In the previous MOBILE models and in certification test
procedures, evaporative emissions were quantified by the test procedures used to measure
them:
Running Loss - Vapor lost during vehicle operation.
Hot Soak - Vapor lost after turning off a vehicle.
Diurnal Cold Soak - Vapor lost while parked at ambient temperature.
Refueling Loss - Vapor lost and spillage occurring during refueling.
For MOVES, we instead model the underlying physical processes involved in evaporation of
fuels. This "modal" approach characterizes the emissions by different emissions generation
processes, each having its own engineering design characteristics and failure rates. This way,
certain physical processes can be isolated, for example, ethanol has a unique effect on
permeation, which occurs in all the above modes. The approach used in MOVES categorizes
evaporative emissions based on the evaporative mechanism, using the following processes:
Permeation - The migration of hydrocarbons through materials in the fuel system.
Tank Vapor Venting (TVV) - Vapor generated in fuel system lost to the
atmosphere, when not contained by evaporative emissions control system.
Liquid Leaks - Liquid fuel leaking from the fuel system, ultimately evaporating.
Refueling Emissions - Spillage and vapor displacement as a result of refueling.
Figure 1 illustrates the evaporative emission processes. Permeation occurs continuously
through the tank walls, hoses, and seals. It is affected by fuel tank temperature and fuel
properties.
1
-------
Figure I: Illustration of Evaporative Processes
VaporVenting
Vapor Leaks
%xCap T Charcoal
Ty\ ¦ Canister
I
Vapor
¦\
r'~
Fuel
V
Permeation
*
Liquid Leaks
These processes occur in each operating mode (Running Loss, Hot Soak, Cold Soak) used in
the MOVES model. Each emission process can be modeled over a user-defined mix of
operating modes, shown in Table 1. This makes for more accurate modeling of scenarios that
do not replicate test procedures. The processed values for the evaporative emission processes
used by MOVES are shown in Table 2.
Table 1: MOVES Evaporative Process Operating Modes
opModelD
Operating mode description
150
Hot Soaking
151
Cold Soaking
300
Engine Operation
Table 2: MOVES Evaporative Emission Processes
processID
Emission process description
11
Evap permeation
12
Evap vapor venting losses
13
Evap liquid leaks
18
Refueling displacement vapor losses
19
Refueling fuel spillage
2
-------
Evaporative emissions are a function of many variables. In MOVES, these variables
include:
Ambient temperature
Fuel Tank temperature
Model year group (as a surrogate for technology and certification standard)
Vehicle age
Vehicle class
- Passenger Vehicle
- Motorcycle
- Short/Long-haul Trucks
Fuel Properties
- Ethanol content
- Reid Vapor Pressure (RVP)a
Failure Modes
Presence of inspection and maintenance (I/M) programs
Both ambient temperature and engine operation cause increases in fuel tank temperature. An
increase in fuel tank temperature generates more vapor in the tank. Activated charcoal
canisters are a control technology commonly used to adsorb the generated vapor. During
engine operation, the canister is purged periodically and the captured vapor is diverted to the
engine and burned as fuel. The emission certification standards for a vehicle (associated with
model year and vehicle class) influence the capacity of the canister system. When the
generated vapor exceeds the capacity of the canister, the vapor is vented to the atmosphere.
This can occur when a fuel undergoes a large ambient temperature increase, or if a fuel with
higher volatility is used, or when a vehicle canister collects vapor for many days without
purging. In calculating vapor venting, MOVES accounts for co-mingling ethanol and non-
ethanol gasoline, and for RVP weathering of in-use fuel. Details on these Tank Fuel Generator
calculations are provided in Appendix B.
Fuel systems can develop liquid and vapor leaks that circumvent the vehicle emissions control
system. Some inspection and maintenance (I/M) programs explicitly intend to identify
vehicles in need of evaporative system repairs. Some states also implement Stage II programs
at gas stations to capture the vapors released during refueling. These programs capture
refueling vapor with technology installed at the pump rather than internal to the vehicle.
The model year groups for evaporative emissions are shown in Table 3. They reflect
evaporative emission standards and related technological improvements. Early controls
included the introduction of activated charcoal canisters for controlling fuel vapor emissions.
Later controls included fuel tanks and hoses built with more advanced materials less prone to
a The MOVES fuel supply table provides the characteristics of gasoline sold in each county and month. For vapor venting
calculations, the MOVES Tank Fuel Generator uses the fuel supply information to account for the effects of "comingling" ethanol
with non-ethanol gasoline and for the "weathering" effect on RVP for in-use fuel. See appendix for details.
3
-------
permeation. Also, reduction of fittings and connections became an important consideration for
vapor mitigation.
Table 3: Model Year Groups for Evaporative Emissions in MOVES
Model year group
Evaporative emissions standard or technology level
1971-1977
Pre-control
1978-1995
Early control
1996
80% early control, 20% enhanced evap
1997
60% early control, 40% enhanced evap
1998
10% early control, 90% enhanced evap
1999-2003
100% Enhanced evap
2004-2015
Tier 2, LEV II
2016-2017
40% Tier 3
2018-2019
60% Tier 3
2020-2021
80% Tier 3
2022 +
Tier 3
This report documents the evaporative emission rates measured in terms of total hydrocarbons
(THC). Total hydrocarbon gases are defined as the measurement of gaseous hydrocarbons by a
flame ionization detector (FID). Evaporative emissions also contain oxygenated hydrocarbons
such as alcohols and aldehydes.
MOVES estimates organic gas aggregate species (e.g., Volatile Organic Compounds, Total
Organic Gases) from the THC emissions as documented in the speciation report.1 MOVES
estimates specific hydrocarbon species as fractions of VOC and TOG emissions. Eight
important mobile source air toxics (MSATs), including benzene and ethanol, are calculated
from evaporative VOC emissions as documented in the air toxics report.2 Evaporative
emissions are not directly affected by the combustion process, and does not estimate any
emissions from combustion products. Table 4 contains a list of the evaporative pollutants
calculated by MOVES. MOVES calculates additional chemical mechanism species from
evaporative emissions used for air quality modeling as documented in the speciation report.1
The data used for this evaporative analysis was collected on light-duty gasoline vehicles but
were also applied to heavy-duty gasoline vehicles since heavy-duty gasoline data was not
available at time of analysis.
4
-------
Table 4: MOVES Evaporative Pollutants
pollutantID
pollutantName
NEIPollutantCod
shortName
1
Total FID Hydrocarbons
HC
THC
20
Benzene
71432
Benzene
21
Ethanol
ETOH
40
2,2,4-Trimethylpentane
540841
2,2,4-Trimethylpentane
41
Ethyl Benzene
218019
Ethyl Benzene
42
Hexane
206440
Hexane
45
Toluene
85018
Toluene
46
Xylene
123386
Xylene
79
Non-Methane Hydrocarbons
NMHC
NMHC
80
Non-Methane Organic Gases
NMOG
NMOG
86
Total Organic Gases
TOG
TOG
87
Volatile Organic Compounds
VOC
VOC
185
Naphthalene gas
91203
Naphthalene Gas
For diesel vehicles, it is assumed that there are no evaporative emission losses except for
refueling spillage. Due to the low vapor pressure of diesel fuel, diesel evaporative losses are
considered negligible.
At the time of this analysis, there was no relevant evaporative emissions data for compressed
natural gas (CNG) vehicles. CNG fuel systems and refueling procedures are significantly
different from those of liquid petroleum-based fuels. For the current release of MOVES, all
evaporative emission rates for CNG vehicles are set at zero.
We significantly updated the evaporative emission calculations and rates in MOVES2014
based on updated emissions data, failure rates, and vehicle activity in MOVES2014. Because
of the significant updates, the MOVES2014 version of this report was subject to peer review
under EPA's peer review guidelines. More information about this peer review, including peer
reviewer comments and EPA response is available on the web.3
Evaporative emission inputs for MOVES2014 were also reviewed by the Coordinating
Research Council.4 Based on our evaluation, most of the issues pointed out in the CRC report
are expected to have very little impact on the magnitude of the evaporative emissions
computed by MOVES. However, we continue to look for opportunities to improve how
MOVES estimates evaporative emissions.
Updates for MOVES3 were limited to an update to the onboard vapor recovery (ORVR)
phase-in values for heavy-duty gasoline as explained in Section 3.6 below.
As explained in the MOVES Population and Activity report5, the activity associated with
evaporative emissions remains the same as in MOVES2014.
5
-------
2 Test Programs and Data Collection
The modeling of evaporative emissions in MOVES is based on data from a large number of
studies (Table 5). Over a decade of research greatly modernized evaporative emissions
modeling. New test procedures provided modal emissions data that greatly advanced the state
of the science. For example, the CRC E-77 test programs6'7'8'9 measured permeation emissions
separately from vapor emissions. Implanted leak testing from these studies along with further
field research provided the first large database regarding the prevalence and severity of
evaporative leaks and other mal- functions. The studies applied an innovative sampling design
which preferentially recruited high emissions vehicles with the aid of infrared ultraviolet
remote sensing devices. The field studies used a portable test cell (PSHED) to measure in-use
hot soak emissions on a large number of vehicles. Findings from these studies were introduced
in MOVES2014 with the explicit modeling of vapor leaks.10'11.
Appendix A has a more detailed summary of these test programs.
Table 5: Evaporative Emission Research Programs
Program
# of Vehicles
CRC E-9
Measurement of Diurnal Emissions from In-Use Vehicles12
151
CRCE-35
Measurement of Running Loss Emissions in In-Use Vehicles13
150
CRCE-41
Evaporative Emissions from Late-Model In-Use Vehicles1415
50
CRCE-65
Fuel Permeation from Automotive Systems16
10
CRC E-65-3
Fuel Permeation from Automotive Systems: EO, E6, E10, and E8517
10
CRC E-77
Vehicle Evaporative Emission Mechanisms: A Pilot Study6
8
CRC E-77-2
Enhanced Evaporative Emission Vehicles7
8
CRC E-77-2b
Aging Enhanced Evaporative Emission Vehicles8
16
CRC E-77-2c
Aging Enhanced Evaporative Emission Vehicles with E20 Fuel9
16
High Evap field studies1011
Thousands
Fourteen Day Diurnal study18
5
Running Loss Testing with Implanted Leaks19
5
API Leakage Study20
Not Avail.
API Gas Cap Study21
Not Avail.
EPA Compliance Testing22
Thousands
6
-------
3 Design, Analysis and MOVES Inputs
This chapter provides detailed information on how evaporative emissions in MOVES are
calculated, and the analysis used to determine appropriate default inputs for the model.
As emission standards have tightened, fuel system materials and connections have become
more efficient at containing fuel vapors. Purge systems and canister technologies have also
advanced, resulting in less vented emissions. Fuel tank temperature is an important
consideration in modeling permeation and vapor emissions. However, liquid leaks occur
regardless and are not dependent on temperature.
3.1 Fuel Tank Temperature Generator
Fuel tank temperature is closely correlated with permeation and vapor venting as observed in
the CRC E-77 pilot testing program 6. This program tested ten vehicles in model years 1992
through 2007. The results showed that fuel temperature strongly influences evaporative
emissions in all testing regimes. Fuel tank temperature is dependent on the daily ambient
temperature profile and vehicle operation patterns. Modern vehicles (enhanced-evap, 1996 &
later) do not recirculate fuel from the engine to the fuel tank and therefore have a lower
temperature rise than older vehicles during operation. In Figure 2, the permeation emissions
are plotted over a 3-day California diurnal tests with (65-105°F) as the low temperature range and
85-120°F as the high temperature range. Both the effects of temperature and fuel volatility can
be observed.
7
-------
Figure 2: Permeation Temperature and RVP effects
Test Time hours
The MOVES "Fuel Tank Temperature Generator" calculates fuel temperature (also referred to
as fuel tank temperature) for a given ambient temperature profile and vehicle trip schedule
based on the vehicle type and model year. Different equations are used depending on the
operating mode of the vehicle: running, hot soak, or cold soak. Fuel tanks are warmer during
running operation than the ambient temperature. The routing of hot exhaust, vehicle speed,
and airflow can all affect tank temperature. Immediately after the engine is turned off, the
vehicle is in a hot-soak condition, and the fuel tank begins to cool to ambient temperature. In
cold soak mode, the vehicle has reached ambient temperature.
Input parameters for the fuel tank temperature generator are:
o Hourly ambient temperature profile (zoneMonthHour table)
o Key on and key off times (sampleVehicleTrip table)
o Day and hour of first KeyON (hourDay table)
o Vehicle Type (Light-duty vehicle, Light-duty truck, Heavy-duty gas truck and
Motorcycles)
o Pre-enhanced or enhanced evaporative emissions control system
The MOVES algorithm iterates through a set of "typical" vehicle trips based on information
from instrumented vehicles. This data is stored in two tables in the MOVES default database.
8
-------
SampleVehicleDay lists a sample population of vehicles, each with an identifier (vehID), an
indication of vehicle type (sourceTypelD), and an indication (daylD) of whether the vehicle is
part of the weekend or weekday vehicle population. To represent vehicles that may sit unused
for multiple days, some vehicles in this table do not have any trips. The second table,
SampleVehicleTrip, lists the trips in a day made by each of the vehicles in the
SampleVehicleDay table. It records the vehID, daylD, a trip number (tripID), the hour of the
trip(hourlD), the trip number of the prior trip (priorTripID), and the times at which the engine
was turned on and off for the trip. The keyOnTime and keyOffTime are recorded in minutes
since midnight of the day of the trip. For more information on the activity data used to
determine the time of keyOn and keyOff events, see the MOVES technical report on vehicle
populations and activity 5 and supporting contractor reports 23, 24.
Coefficients for the Fuel Tank Temperature Generator are recorded in MOVES in the
tankTemperatureGroups and tankTemperatureRise tables.
3.1.1 Fuel Tank Temperature for Hot and Cold Soaks
Equation 1 is used to model tank temperature as a function of ambient temperature.
dTTank _ w,j, _j, x ^
i V air Tank)
at
TTank is the fuel tank temperature, Tair is the ambient temperature, and k is a constant
proportionality factor (k = 1.4 hr ). The value of k was established from EPA compliance
data on 77 vehicles that underwent a 2-day diurnal test and had a 1-hour hot soak (See
Appendix A). No distinction was made between hot and cold soak for this derivation. We
assumed that during any soak, the only factor driving change in the fuel tank temperature was
the difference between the tank temperature and the ambient temperature.
This equation only applies during parked conditions, which include the following time
intervals:
From the start of the day (midnight) until the first trip (keyOnTime)
From a keyOffTime until the next keyOnTime
From the final keyOffTime time until the end of the day
Mathematical steps:
1. At time to = 0 or keyOffTime (start of soak), TTank = TThis value will either be the
ambient temperature at the start of the day, or the fuel tank temperature at the end of a
9
-------
trip.
2. Then, for all t >0 and keyOffTime, the next tank temperature is calculated by integrating
numerically13 over the function for temperature change, using Equation 2
(TTcink)n+\ = TTank + k(Tajr Tfank)At (2)
where:
TTank = Tank temperature
Tair = Ambient air temperature
t = Time
k = Temperature constant (1.4 hr )
Figure 3 demonstrates the Euler approximation for calculating the tank temperature based on
ambient temperature.
Figure 3: Example Day Modeled with Euler Method
Hourof Day
b Numerical integration is used to perform this step using the Euler method, one of the simplest methods of integration. The
smaller the time step At, the more accurate the solution. MOVES uses a At of 15 minutes, which is accurate enough for our
modeling purposes without causing tremendous strain on computing resources.
10
-------
3.1.2 Fuel Tank Temperature while Running
Vehicle trips are short compared to the length of the day. Therefore, we assume a linear
temperature increase during a trip to improve model performance with minimal compromise to
accuracy.
In this algorithm, to determine AT tank, the increase in tank temperature for an arbitrary trip
length with arbitrary starting tank temperature, we must first find ATtank95, the average
increase in tank temperature during a standard 4300 second, 95°F running loss test. The
algorithm models the increase in fuel tank temperature using the tank temperature at
KeyOnTime, the amount of running time and the vehicle type and technology. Newer
technologies reduce the heat transferred to the fuel tank. The MOVES ATia/ik95
temperatures are as follows:
If the vehicle is pre-enhanced (model year pre-1996), vehicle type affects ATtank95'- 13
LDV ATtank95 = 35°F
LDT ATtank95 = 29°F
If the vehicle is evap-enhanced (model year 1996+):
ATtank95 =24°F
These values are used to calculate the change in temperature for a 4300 second test
(ATtank4300) for arbitrary starting fuel tank temperatures using Equation 3.
^ l Tank43()() = 0.352(95 71 auk, KeyON ) + ATjank95 (3)
The parameters in Equation 3 are derived from regression analyses of light-duty vehicles
driving the running loss drive cycle with varied starting temperatures 25. The lower the initial
tank temperature, the larger the increase over a given drive cycle.
The average ratio of fuel temperature increase to initial fuel temperature is -0.352. This gives
us the increase in tank temperature, so we can create a linear function that models fuel tank
temperature for each trip.
Trunk 4300/3600 ^ ~~ ^key°N) + ^Tank,KeyON (4)
Where:
TTank = Tank temperature
t = Time
tkeyON = Time of engine start
11
-------
The 4300/3600 in the Equation 4 denominator converts seconds to hours (4300 seconds in the
running loss certification test), maintaining temporal consistency in the algorithm. The
resultant tank temperatures for an example temperature cycle are illustrated in Figure 4.
Running operation is shown as a red line, and hot soak operation is shown as a blue line
Figure 4: Modeled Vehicle Tank Temperature During a Day of Operation
120
110
100
3 90
80
70
60-
-1
Ambient Temperature
Hot Soak Operation
Running Operation
9 14
Hour of Day
19
24
Assumptions:
The first trip is assumed to start halfway into the hour stated in the first trip's
HourDaylD.
The effect of a change in ambient temperature during a trip is negligible compared to
the temperature change caused by operation.
The KeyOn tank temperature is known from calculation of tank temperature from the
previous soak.
3.2 Permeation
Permeation emissions are specific hydrocarbon compounds that escape through micro-pores in
pipes, fittings, fuel tanks, and other vehicle components (typically made of plastic or rubber).
They differ from leaks in that they occur on the molecular level and do not represent a
mechanical/material failure in a specific location. In MOVES, base permeation rates are
estimated and then adjusted to the modelled tank temperatures and fuel properties.
12
-------
The base rates in MOVES represent rates for gasoline with no ethanol. As detailed in Section
3.2.2, while keeping other fuel properties constant, the presence of ethanol increases
permeation emissions approximately two-fold.
For model years before 1999, permeation base rates are developed using the mg/hour emission
data from the last six hours of a 72-96-72T diurnal test (also known as cold soak/resting loss)
The diurnal tests were measured on the federal cycle (72F-96T) for the CRC E-9 and E-41
programs 12'14'15. Together, these two programs tested a total of 151 vehicles with model
years ranging from 1971 to 1997. The final six hours of the diurnal are an appropriate
surrogate for a permeation-only test because the emission rate, ambient temperature, and fuel
temperature are relatively stable or constant. Permeation should be the only evaporative
process occurring. The rates were developed for distinct model year and age groups. Model
years 1996-1998 are represented individually to reflect the phase-in of Enhanced Evaporative
Emissions Standards (20 percent Enhanced Evap in 1996, 40 percent in 1997, 90 percent in
1998).
For the 1999 - 2003 model years (full implementation of the Enhanced Evaporative Emissions
Standards) and the 2004 - 2015 model years (Tier 2 Evaporative Emissions Standards) the E-
77-2 Static test data was analyzed using the 86° F tests for non-leaking vehicles, corrected to
72°F. We used the Static test for these vehicles because the last six hours of Diurnal test was a
surrogate to represent permeation rates in the older vehicle data, but the static test is intended
to measure permeation and only permeation. The values in MOVES are based on 54 tests in
the 1999-2003 model year group taken from EPA's Compliance Testing Program. Later data
from the E-77 programs served to validate the Tier 2 permeation base rates already used in
MOVES. The E-77-2 data points expanded the range of the age groups but the data was not
sufficient to differentiate the estimates for the age and model year groupings, thus, we kept the
0.0102 rate for all ages of both 1999-2003 and 2004+. Table 6 summarizes the data, analysis
and resulting emission rates for model years 1999-and-later for the number of tests within each
age group to understand how these decisions were made. There were three vehicles in the 0-3
age group for model year 2004+ that had an average of 0.003 but ten vehicles in the next age
group had essentially the same average emissions rate of 0.01. Assuming there is deterioration
for these vehicles, averaging all ages into one group made sense to characterize the fleet
without data to support finer age groups.
The Tier 3 evap standards apply starting in model year 2017, and phase in over model years
2016-2022, with early allowances. The Tier 3 permeation standard reflects a 40 percent
reduction from the previous standard and the introduction of 10 percent ethanol to the
certification fuel discussed earlier.
Permeation base rates are presented in Table 7. These rates are recorded in the MOVES
emissionRateByAge table.
13
-------
Table 6: Permeation Analysis for 1999 - 2003 and 2004 and later Model Years
Previous Data Decision for
(Compliance data) MOVES2014
Model Year
Age
# of tests
MOVES2010
# of E-77 tests E-77 avg rate
weighted avg of ALL
Enhanced Evap
0-3
52
0.0102
0.0102
4-5
0
0.0102
0.0102
6-7
2
0.0102
7 0.021
0.0186
1999-2003
8-9
0
0.0102
2 0.014
0.0140
0.0102
10-14
0
0.0102
0.0102
15-19
0
0.0102
0.0102
20+
0
0.0102
0.0102
Model Year
Age
# of tests
MOVES2010
# of E-77 tests E-77 avg rate
weighted avg of ALL
Tier 2 / LEV II
0-3
0
0.0102
3 0.003
0.003
4-5
0
0.0102
10 0.01
0.01
6-7
0
0.0102
0
2004-2015
8-9
0
0.0102
0
0.0102
10-14
0
0.0102
0
15-19
0
0.0102
0
20+
0
0.0102
0
Model Year
Age
# of tests
MOVES2010
weighted avg of ALL
40% of Tier 3
0-3
0
0.0102
0
4-5
0
0.0102
0
6-7
0
0.0102
0
2016-2017
8-9
0
0.0102
0
0.007
10-14
0
0.0102
0
15-19
0
0.0102
0
20+
0
0.0102
0
Model Year
Age
# of tests
MOVES2010
weighted avg of ALL
60% of Tier 3
0-3
0
0.0102
0
4-5
0
0.0102
0
6-7
0
0.0102
0
2018-2019
8-9
0
0.0102
0
0.006
10-14
0
0.0102
0
15-19
0
0.0102
0
20+
0
0.0102
0
Model Year
Age
# of tests
MOVES2010
weighted avg of ALL
80% of Tier 3
0-3
0
0.0102
0
4-5
0
0.0102
0
6-7
0
0.0102
0
2020 - 2021
8-9
0
0.0102
0
0.004
10-14
0
0.0102
0
15-19
0
0.0102
0
20+
0
0.0102
0
Model Year
Age
# of tests
MOVES2010
# of E-77 tests E-77 avg rate
weighted avg of ALL
100% of Tier 3
0-3
0
0.0102
1 0.005
0.005
4-5
0
0.0102
1 0.002
0.002
6-7
0
0.0102
0
2022 and later
8-9
0
0.0102
0
0.003
10-14
0
0.0102
0
15-19
0
0.0102
0
20+
0
0.0102
0
14
-------
Table 7: MOVES Base Permeation Rates at 72°F
First Model Year
Last Model Year
Age Group
Base Rate
1960
1970
20+
0.311
1971
1977
10-14
0.192
15-19
0.229
20+
0.311
1978
1995
0-5
0.0554
6-9
0.0913
10-14
0.124
15-19
0.148
20+
0.201
1996
1996
0-5
0.0464
6-9
0.0751
8-9
0.0751
15-19
0.12
20+
0.163
1997
1997
0-5
0.0373
6-9
0.0589
10-14
0.0784
15-19
0.0929
20+
0.125
1998
1998
0-5
0.0147
6-9
0.0183
10-14
0.0216
15-19
0.024
20+
0.0293
1999
2015
all ages
0.0102
2016
2017
all ages
0.0072
2018
2019
all ages
0.0056
2020
2021
all ages
0.0041
2022
2060
all ages
0.0026
3.2.1 Temperature Adjustment
The E-65 permeation study found that permeation rates, on average, double for every 18°F
increase in temperature. 16 This study tested 10 vehicle fuel systems (the vehicle body was cut
away from the fuel system, which remained intact on a frame) at 85°F and 105°F. The vehicles
ranged in model year from 1978-2001. In MOVES, the base permeation rates are calculated at
72°F, the same temperature as the certification test. Equation 5 is derived from this study and
15
-------
used to adjust the base permeation rate to account for the tank temperature described in
Section 3.1.
p _ p ^0.03S5(TTank
^''base )
adj base
Where:
Pbase Base Permeation Rate
Tiank = Tank Temperature
Tbase = Base Temperature for a given cycle (e.g., 72°F for a federal diurnal test)
3.2.2 Fuel Adjustment
Ethanol affects evaporative emissions from gasoline vehicles due to the increased permeation
of specific hydrocarbon compounds through tanks and hoses. Modeling permeation emissions
separately from vapor venting emissions allows us to apply these effects only on permeation
where the complex chemistry of ethanol-gasoline blends increases permeation through the
tanks and hoses of the fuel system.
Permeation fuel effects were developed from the CRC E-6516 and E-65-317 programs, which
measured evaporative emissions from ten fuel systems that were removed from the vehicles
and filled withEO, E5.7, and E10 fuels. This method assures that the emissions measured are
purely from permeation (assuming the systems were not leaking). Additional data was
provided from the CRC E-77-27 and E-77-2b8 programs, which measured evaporative
emissions from sixteen intact vehicles. For this analysis, vehicles certified to enhanced-
evaporative and Tier 2 standards were analyzed separately from vehicles certified to earlier
standards. Enhanced evaporative standards were phased in from 1996-1999 and imposed a 2.0
gram standard over a 24-hour diurnal test. Standards previously in effect applied a 2.0 gram
standard to a 1-hour simulated diurnal.
The ethanol effect is estimated with a mixed model shown in Table 8. The evaporative
certification level, ethanol content, and RVP were modeled as fixed effects and the vehicle was
modeled as a random effect. The natural logarithm of the emission rates over the 65-105-65°F
diurnal cycle provided a normally distributed dataset to the model. The dataset was not large
enough to distinguish the three ethanol levels within each evaporative certification bin.
Therefore, E5.7 and E10 test results were combined into a single bin of ethanol-containing
fuel which had a significant effect compared to E0 fuel.
16
-------
Table 8 Mixed Model for Ethanol Effects on Permeation
Mixed Model for ETOH Effects on Permeation
logHC-1 ~ Standard Group + ETOHYN * Standard Group
Standard
Value
Std. Error
DF
t-value
p-value
3.0279465
0.1887947
59
16.038305
0
Tier 2
-0.761386
0.3095233
15
-2.459867
0.0265
Zero Evap
-1.050358
0.6297706
15
-2.938146
0.0102
The percent difference between the ethanol rate and the EO rate is used in MOVES as the fuel
adjustment. Due to the enhanced-evaporative certification standards phase in from 1996-1999
(20/40/90/100 percent), the two fuel adjustments must also be phased in for those model years.
Table 9 lists the fuel adjustments used for E5 through E85 for the model year groupings used
in MOVES. These values are recorded in the MOVES hcPermeationCoeff table as
multiplicative factors.
Table 9: Ethanol effect for Permeation Emissions
Model Years
Percent increase due to
ethanol (5-85%)
1995 andearlier
65.9
1996
75.5
1997-2000
107.3
2001 and later
113.8
3.3 Tank Vapor Venting
Vapor generated in the tank can escape to the atmosphere during a process labeled "Evap
Vapor Venting" or "Tank Vapor Venting" (TVV). Hydrocarbons emitted by this process
originate from a variety of sources. As tank temperature rises and vapor is generated within the
tank, the vapors are forced out of the tank from increased pressure. Fully sealed gas tanks are
rare as they must be constructed with metal to prevent bloating. Using metal as a tank
material can be expensive, heavy, and difficult to shape for tightly packed modern vehicles.
Instead, most vehicles are equipped with an activated charcoal canister to adsorb the vapors as
they are generated. Later, the vapors are consumed as they purge to the engine (through the
intake manifold) during vehicle operation. The canister is open (or vented) to the atmosphere
to prevent pressure from building within the fuel system. Consequently, if the engine is not
operated for several days, fuel vapors can diffuse through the charcoal or even freely pass
17
-------
through a completely saturated canister. Tampering, mal-maintenance, and system failure can
result in excess evaporative emissions. Inspection and maintenance (I/M) programs can also
influence how leaks and other problems are controlled over the life of a vehicle.
Integral to the understanding of Tank Vapor Venting (TVV) is the calculation of Tank Vapor
Generated (TVG). Tank vapor generated depends on the rise in fuel tank temperature (F),
ethanol content (vol. percent), Reid vapor pressure (RVP, psi) and altitude. Calculations in
MOVES use the Wade-Reddy equation for vapor generation (Equation 6).26
TVG = AeB*RVP (eCT* - eCT1) (6)
Where:
T i = Initial temperature
Ta = Temperature at timex
In Equation 6, coefficients A, B, C vary by altitude and fuel ethanol content. These coefficients
are shown in Table 10.
Table 10: TVG Constants for Equation 6
E0 Gasoline
E10 Gasoline
Constant
Sea Level
Denver alt.
Sea Level
Denver alt.
A
0.00817
0.00518
0.00875
0.00665
B
0.2357
0.2649
0.2056
0.2228
C
0.0409
0.0461
0.0430
0.0474
The vapor venting emission process occurs during all three operation modes: running, hot soak
and cold soak. While running, vapors are generated as the fuel system is warming and active.
During hot soak, vapor generation is caused by latent heat transfer due to fuel recirculation and
other convective processes. Cold soak vapor generation is concurrent with ambient
temperature increases. MOVES modeling of fuel system warming is detailed in Section 3.1.
3.3.1 Altitude
Evaporative vapor generation is affected by the lower ambient pressure at high altitudes.
MOVES accounts for this effect during the calculation of tank vapor generated. This process
relies on the values in Table 10 for high altitude (Denver, CO) and a low altitude (Sea Level).
MOVES applies linear interpolation/extrapolation based on the barometric pressure in each
county.
18
-------
3.3.2 Cold Soak
Cold soak vapor emissions occur while a vehicle is not operating and the engine and fuel
system have cooled to ambient temperature. Emissions occurring under these conditions are
also referred to as diurnal emissions.
Emission quantification started with the measurement of emissions based solely on standard
regulatory test cycles. As the emissions levels over the test cycles became more controlled,
concern grew about "off-cycle emissions"- real-world emissions that occur outside of the test
procedure constraints on ambient temperatures, fuel RVP and soak time.
As a vehicle sits through multiple diurnal cycles, the carbon canister accumulates vapor every
day. It can only adsorb vapor until it reaches its capacity; then it begins to vent to the
atmosphere. A canister with degraded/damaged carbon may have reduced capacity, and
eventually every canister will vent to the atmosphere once it reaches saturation. During cooling
hours, a canister back purges to the fuel tank and regains some capacity. Then, during the
subsequent wanning period the canister is re-filled with vapor and any vapor generated
beyond capacity will escape to the atmosphere.
Figure 5: Multiday Vapor Accumulation in Charcoal Canister
Day 2
breakthrough
VaP°r Vapor
generated I after
I backpurge
Day 3
breakthrough
Empty warming cooling warming cooling warming cooling
Canister Day 1 ^ay ^ Day 3
Figure 5 illustrates the dynamic behavior of vapor within a charcoal canister over three days of
continuous cold soaking. During the first day, vapor accumulates within the canister but does
not exceed the canister capacity. During the cooling period of day 1, we observe backpurge
when some of the fuel vapors that were previously adsorbed to the charcoal flow back into the
19
-------
cooling tank. The fresh air is drawn in through the canister vent while the vapor condenses in
the tank during the cooling portion of the cycle. During warming on day 2, we see generated
fuel vapors that exceed the canister capacity (though some canisters may be constructed to
hold more than 2 days of vapor). These emissions are lost to the atmosphere, and only what
remains in the canister can be backpurged during the subsequent cooling cycle. In day 3, more
vapor is generated and consequently lost to the atmosphere. Any additional days without
engine purge during normal driving (i.e. inactivity) will exhibit the same behavior as day 3. It
should be mentioned that plug-in hybrid electric vehicles that are mainly driven on short
(electric only) trips, may also exhibit similar breakthrough over time. However, modeling of
these vehicles is currently beyond the scope of MOVES since the penetration rates of these
technologies are low, and we are not aware of any multi-day diurnal data collected on PHEVs.
Modeling a fleet of vehicles involves a diverse population of canisters with differing capacities.
A given amount of vapor will be fully contained by some vehicles but exceed the canister
capacity in others. Figure 6 illustrates the approach for calculating the tank vapor vented
(TVV) as a function of the tank vapor generated (TVG).
Figure 6: Vapor Vented Curve
Several factors accommodate this modeling approach. The following variables, explained in
more detail below, are included in the MOVES default database in the 'cumTvvCoeffs'
table:
Back Purge Factor
Average Canister Capacity
Tank Size
Tank Fill Fraction
Leak Fraction
Leak Fraction IM
TVV Equation
20
-------
Leak Equation
3.3.2.1 B ack Purge F actor
The back purge factor is the percent of hydrocarbon vapor that is desorbed from a vehicle's
canister during cooling hours when pressure decreases within the tank, drawing ambient air in
through the canister vent. In the real-world, this process occurs nightly as temperatures cool
and restore some canister capacity. In the Multiday Diurnal Study 18, test vehicles soaked for
14 consecutive 72°F-96°F diurnals (the Federal Test Procedure temperature cycle). During this
time, the vehicle canister mass was measured continuously. During the cooling period, the
measured mass of the vehicle canisters decreased. This cyclical effect can be observed in
Figure 7.
An average value of 23.8 percent backpurge was developed from these results and is used in
the MOVES model. For example, a vehicle canister with 100 grams of hydrocarbons will
backpurge 23.8 grams and begin the next day with 76.2 grams. A more complex model for
backpurge was considered (similar to tank vapor generation), but it would require significant
computational resources and potentially slow model performance considerably. As diurnal
temperatures are relatively symmetrical, detailed modeling of tank vapor generation has
already provided a high level of precision, justifying a simpler model here.
21
-------
Figure 7: Vehicle Canister Mass, 14-day Diurnal Test
an
xfl
Eh
03
15
Ph
5s an
CD
^ 3-
<5 ai
10g
0 Oh
-5 S
IS H
;>
5
Legend
lORVPfuel
9 RVP fuel
SHED Temperature
5000 10000
Test Time (Minutes)
15000
20000
3.3.2.2 A verage Canister Capacity
The canister capacity reflects how much vapor generated in the tank can be contained by the
canister before breaking through. Note that while average canister capacity is stored in the
MOVES database, it is not actually used in the MOVES model. Instead, it was used in the
"Delta" pre-processor
27
To calculate a sales-weighted average canister size, we used sales data28 and EPA evaporation
certification data.22 Certification data includes the evaporative family code which contains the
Butane Working Capacity (BWC) of the canister; it is found in digits 7, 8 and 9 for enhanced
evap vehicles, and in digits 5, 6 and 7 for pre-enhanced vehicles. The BWC represents the
ability of a canister to capture butane vapor, rather than gasoline vapor, so it must be adjusted
by a factor of 0.9229.
Evaporative control was introduced in 1971, so canisters are not modelled for pre-1971
vehicles. For model years beyond 2010, the 2010 average canister capacity was used. The
calculated average canister capacities for cars and trucks combined are listed in Table 11.
A peak in average canister size at model year 2005 corresponds to greater sales of cars with
larger fuel tanks. Motorcycles are modeled without canisters, because they were not being
used at the time of the analysis. We hope to review and revise in future versions of the model.
Heavy-duty gasoline vehicles are modeled with the same canister capacity as light-duty
22
-------
vehicles.
Table 11: Average Canister Capacity by Model Year
Model Year Group
Average Canister Capacity (grams)
1960-1970
0
1971-1977
64.7
1978-1995
72.8
1996
78.7
1997
83
1998
115.4
1999-2003
122.9
2004
145
2005
150.7
2006
145.3
2007
142.9
2008
138.6
2009
136.2
2010+
137.5
3.3.2.3 Tank Size
The average tank size for a given model year is an important facet of the vapor generation
calculation because a larger tank will have more space in which vapor can accumulate. Both
sales data28 and tank size information 30 were required to calculated a sales-weighted average
tank size for model years 1990-2010. For this analysis, car and truck sales, and tank sizes were
combined. For vehicles with multiple styles (i.e. different cab sizes on pick-up trucks) with
different tank sizes, the average available tank size was used as sales information was
unavailable by style. Data sources only span from 1990-2010, so past and future values were
projected. Vehicles in the 1990-2010 range have tanks with an average capacity of 1.25 times
greater than a calculated 300-mile range, so this ratio was applied using fuel economy data
going back to model year 1975. 31. Pre-1975 vehicles use the 1975 fuel tank size. For future
vehicles, tank size is assumed to stay constant from 2010 on. It is also possible that
manufacturers will maintain range with a shrinking fuel tank. In future versions of MOVES,
we will reexamine this assumption. The calculated sales-weighted tank sizes are in Table 12.
23
-------
Table 12: Sales-Weighted Average Fuel Tank Size
Model Year Group
Tank Size (gal)
1960-1970
28
1971-1977
27.3
1978-1995
18.6
1996-1997
19.1
1998
19.5
1999-2003
19.9
2004
20.5
2005
20.3
2006
20
2007
19.7
2008
19
2009-2030
19.1
HD Vehicles
38
Motorcycles
3
3.3.2.4 Tank Fill Fraction
The tank fill fraction is an important input used in calculating tank vapor generation. The more
vapor space above the liquid fuel, the more capacity there is for vapors to accumulate. The
average tank fill fraction used in the model is 40 percent fill. This is a typical fill level for
certification procedures and many of the test programs from which our data originates. It is
also a figure supported by existing research on tank filling behavior by consumers 32.
3.3.2.5 Vapor Leak Prevalence
In order to accurately quantify emissions from leaking vehicles, one must not only estimate
emission rates from leaks of various sizes, but also prevalence of leaks in the fleet. This
corresponds to an emissions rate and its corresponding activity. Our estimates of leak
prevalence are informed by the analysis of a field study which took place at the Ken Caryl IM
Station in Denver, CO during the summer of 2009 u. In this study, a remote sensing device
(RSD) was used to recruit high emitting vehicles which were then tested in a Portable Sealed
Housing for Evaporative Detection (PSHED). The vehicle's hydrocarbon emissions were
measured over 15 minutes during hot-soak conditions, and vehicles were inspected to identify
the cause/source of the leaks when possible. The set of hot-soak measurement from individual
vehicles, with inverse-probability sampling weights and solicitation response weights applied
to all vehicles, allows the prevalence of leaks in the fleet to be estimated.
We have defined a vapor leaker as any vehicle that would fail the enhanced evaporative
standard of 2 grams. The standard sums the emissions from the worst day of a 3-day diurnal
test and the hot soak. To develop a surrogate standard for a 15-minute hot soak test, we used
knowledge of certification testing to attribute 0.4 grams (g) of the 2 g standard to the hot soak
24
-------
portion, and 76 percent of 0.4 g to the first 15 minutes of the hour-long hot soak test. This
approach suggests that 0.3 g can be taken as a surrogate standard for a 15-minute hot soak.
Figure 8: Prevalence of Vapor Leaks Above a Given Threshold in the 2009 Ken Caryl
Fleet
Vapor Leak
Size (g/
15min)
Table 13 (plotted in Figure 8) displays leak prevalence at various emission thresholds for what
constitutes a "leak". Observing the difference between any two points determines how many
vehicles fall into a particular range. Looking at Table 13, in model year group 1981-1995, 2.6
percent of vehicles are leaking at more than 20 grams and 4.2 percent of vehicles are leaking at
more than 10 grams. Subtracting these two values yields that 1.6 percent of vehicles in the
model year group have a leak between 10 and 20 grams.
The data only contain prevalence rates for PSHED measurements as low as 1.0g/15min.
Failure rates are extrapolated to 0.3g/15min. Using aggregate data from the Ken Caryl station,
it is found that 0.3g/15min PSHED measurements are 50 percent more prevalent than
1.0g/15min PSHED measurements.
25
-------
Table 13: Prevalence of Leaks Above a Given Threshold (g/15min)
Leak Threshold (g/15 min)
Denver
100
50
20
10
5
2
1
0.3
Sea Level (MOVES)
70.9
35.5
14.1
7
3.6
1.4
.7
.2
Prevalence by MY Group
1961 - 1970
0
0
0.53
0.53
0.68
0.68
1
1
1971 - 1980
0
0
0
0.3
0.85
1
1
1
1981 - 1995
0.004
0.004
0.026
0.042
0.083
0.22
0.26
0.39
1996- 2003
0
0
0
0.02
0.021
0.029
0.033
0.064
2004- 2010
0
0
0
0
0
0
0
0
Because the data used to estimate leak prevalence was collected in Denver, Colorado at an
altitude of 5,280 feet above sea level, measurements must be adjusted to sea level. At sea
level, the amount of vapor generated will be less due to higher atmospheric pressure. To
determine the appropriate correction factor, we performed the Wade-Reddy calculation and
found that under identical conditions, the higher altitude will generate 41 percent more vapor.
Colorado is a strategic location to perform a leak quantification program because a given
vapor leak will produce higher levels of emissions at a higher altitude, therefore making it
easier to detect. Each of the leak magnitude bins have been corrected for altitude by this factor.
For example, the prevalence of leaks at lg-2g levels in Denver will be the same prevalence of
leaks at ,71g-1.42g levels at sea level.
Because this was a cross-sectional study, we must populate the model for many model year and
age group combinations that were not measured. A set of linear regressions was used to model
vapor leak prevalence for ages and model years where data is not available. We divided model
year groups in years when new technologies or standards were introduced. Modeling was
based on the assumption that newer cars will have lower leak prevalence than older cars due to
the advancing technology and use of more durable materials. Therefore, data from the 1996-
2003 model year group was used as a surrogate for new vehicles in the 1971-1980 and 1981-
1995 model year groups. However, because vapor leaks also occur due to tampering and mal-
maintenance, deterioration is not the only factor involved in occurrence of vapor leaks. The
regressions from the older model year show more rapid vehicle deterioration rates than newer
model years.
Figure 9 shows the vapor leak prevalence as the percent of the vehicle fleet with a leak larger
than 0.3g/15min. For model years 1996 and later, the estimate for leak prevalence at ages 0-3
was developed with I/M data from five states. The analysis revealed that 1-2 percent of
vehicles consistently arrived at I/M stations with an evap Diagnostic Trouble Code (DTC) set.
The vast majority of the DTCs set specifically indicated a vapor leak detected. The green
diamonds in the 1971-1980 and 1981-1995 model year groups are an assumption made based on
the 1996-2003 data to describe these vehicles' leak rates when they were new. The slope of the
26
-------
2004-2010 prevalence rates was developed by applying the 5-10 year-old 1996-2003 data point
to the 10-15 year old 2004-2010 point.
Figure 9: Non-IM Vapor Leak Prevalence, Extrapolated from data
Tier 3 and LEY III Leak Prevalence
To model the leak prevalence rates of LEY III and Tier 3 vehicles, the effectiveness of
improved OBD systems and the efficacy of vehicle leak testing were quantified. In the
above mentioned field study performed in Colorado, it was found that 70 percent of
evaporative leaks were due to deterioration of the evaporative system (e.g. corroded fuel lines,
filler neck, cracked hoses etc.) that could be improved with new design and material
considerations. The remaining 30 percent of evaporative leaks were beyond manufacturer
control, (e.g. Improper maintenance, tampering, missing gas caps, etc). See Table 14.
OBD effectiveness and OBD readiness are also important factors in the detection and repair of
leaks after they occur. OBD effectiveness refers to the ability of diagnostic systems to identify
leaks within the fuel system and alert the driver by illuminating a warning light. OBD
readiness refers to the time during which vehicle diagnostics are actively assessing the
integrity of the vehicle fuel system.
Our reference case assumed 40 percent OBD effectiveness and 95 percent OBD readiness.
These numbers were based on an assessment of vehicles with OBD-detectable leaks and
whether or not the leak was identified by the vehicle and the driver alerted via a check engine
light. 33
27
-------
We estimated the implementation of LEV III would immediately reduce the 70 percent of
deterioration-caused leaks by 33 percent simply due to the lower emissions standard.
Longitudinally, we see reductions in leak prevalence associated with lower emissions
standards. We also estimated that due to improved vehicle diagnostic systems, 80 percent of
detectable leaks will be discovered and reported by the vehicle. In addition, we assumed that
with the increased rigor of requirements the readiness will increase to 99 percent.
We estimated that the implementation of Tier 3 would immediately reduce the 70 percent of
deterioration-caused leaks by 66 percent due to the additional benefit of the Tier 3 leak
standard. As in LEV III estimates, we also estimated that 80 percent of detectable leaks will
be discovered and reported by Tier 3 vehicles, as well as an increase of 99 percent readiness.
These estimates result in an overall reduction of leak frequency of 26 percent for the LEV III
program and 49 percent for the Tier 3 program
Table 14: Summary of Tier 3 and LEV Assumptions
Base Inputs
# of Leaks>0.020"
100
% Mai-Rep air
30%
% Durability
70%
Tier 2 Case
OBD Ready %
95%
OBD Effectiveness
40%
LEV III Control Case
% of "durability" leaks prevented
33%
OBD Ready %
99%
OBD Effectiveness
80%
Tier 3 Control Case
% of "durability" leaks prevented
66%
OBD Ready %
99%
OBD Effectiveness
80%
28
-------
Figure 10: LEV III, Tier 3 Leak Prevalence Estimates
u
J=
«
o
C
O
tJ
TO
100
t/i
TO
a)
80
60
j 40
'20
0
~ Leaks Prevented
¦ Leaks w/ MIL on
¦ Leaks w/ MIL off
Reference LEV III Case With Tier 3
3.3.2.6 Vapor Leak Emissions
In MOVES2014, vapor leak emissions are a distinct emissions mode, separate from vapor
emissions vented from the canister during normal operation. It is important to characterize
leaking emissions separately because they can potentially be orders of magnitude higher than
the other emission modes described above. Unlike non-leak emissions, leak emissions can be
modeled as a linear function with vapor generation. In Figure 11, measured vapor emissions
are plotted on the y-axis against the calculated tank vapor generated. The average for four
vehicles is overlaid and is used as the representative leak emission rate in MOVES.
Figure 11: SUED Leak Emissions for one Severity Bin
14
Vapor generated in the tank (TVG) is calculated using the Wade-Reddy equation, thus
29
-------
requiring fuel RVP, fuel ethanol content, and temperature data. Two datasets containing this
information were used in developing leak emission rates. The E-77 suite of programs 6' 7'
8' 9 measured high-emitting vehicles, with known fuel properties and artificially implanted
leaks, on the California (65°F-105°F) diurnal cycle. In another effort, the Colorado Department
of Public Health and Environment (CDPHE) carried out a repair effectiveness program
during the summer of 2010 in collaboration with the Regional Air Quality Council (RAQC).
This program 34 measured 16 vehicles with identified leaks. A 6-hour test was performed with a
temperature increase of 72°F- 96°F. This effort was less resource-intensive than the full diurnal
procedure and still provides the necessary information to calculate TVG. The SHED
measurements of Tank Vapor Vented (TVV) and calculated TVG form the basis for a linear
regression of TVV vs. TVG for each vehicle. The resulting slope represents the mass of vapor
vented per mass of vapor generated. The average of the regressions becomes the leak rate for
that severity bin. This approach can be observed in Figure 11. This test procedure could not
distinguish permeation and leak vapor emissions. However, permeation for these vehicles is
assumed to be negligible during the 6-hour test given the severity of the leak emissions. In the
E-77 program, TVV emissions were collected in a canister external to the SHED. The external
canister was connected to the vent on the vehicle canister. No permeation was included in the
measurement.
Because the emissions measured were highly variable, spanning several orders of magnitude,
the emissions data for leaking vehicles was binned by magnitude. Accordingly, both emission
rates and prevalence were calculated within these bins. As the leak prevalence estimates were
measured at high altitude in Denver, it is essential to develop adjustments to apply the binning
process at lower altitudes, such as sea level. Application of the Wade-Reddy equation
(Equation 6) suggests that an E10 fuel in Denver generates 1.41 times as much vapor as at sea
level. For example, a vapor leak at 0.3g/15min in Denver would have an equivalent rate of
0.21g/15min at sea level. The bins used to categorize leak severity as well as the average leak
emission rate for that bin are listed in Table 15.
Table 15: Leak Emission Rates by Bin
Denver bins
(g/15min)
Sea Level bins
(g/15min)
Average Ratio of Grams vented /
Grams generated
0.3 - 2
0.2 - 1.4
0.12
2 - 5
1.4- 3.6
0.27
5 - 10
3.6- 7.1
0.65
>10
>7.1
1.33
Each data point was binned by its hot soak measurement from the E-77 programs or PSHED.
(Portable SHED) measurement from the Denver program. The PSHED tests are 15-minute hot
soak measurements.
30
-------
Figure 12 illustrates the leak emission rates for each leak severity bin. The average emission
rate for vehicles with 15-min hot soak measurements greater than lOg exceeds 1. It is possible
to measure more fuel vapor in the SHED than is calculated with Equation 6. It is known that
the equation is less reliable at higher temperatures. Also, complicated factors such as fuel
sloshing and tank geometry can influence vapor generation beyond the estimation capabilities
of the Wade-Reddy equation.
Figure 12: Leak Emission Rates by Leak Severity Bin
5 10 15 20 25 30 35
Tank VaporGenerated
3.3.2.7 Estimation of Tank Vapor Vented
For normally operating non-leaking vehicles, tank vapor vented (TVV) from the canister was
calculated. This quantity of vapor is calculated with Equation 6 in g/gal-headspace. The
model uses tank size and tank fill to calculate the headspace volume for a given vehicle. This
information allows calculating the total vapor generated inside the tank. Equation 7 is the
final calculation of TVG, where a, b, and c are the appropriate coefficients, as defined in
Table 10.
TVG = [aeb(RVP\ect2 - ecf1)) * (tankSize * (1 - tankFill)) (7)
With TVG as an input, the TVV equation estimates the amount of vapor vented. During a
model run, MOVES calculates vapor vented for consecutive days. The algorithm accounts for
average canister capacity (ACC) and backpurge factor. Daily backpurge removes fuel vapors
from the canister, increasing capacity to store vapor generated during successive days. Vapor
generated above the ACC is lost to the atmosphere, therefore backpurge only applies to what
remains in the canister.
31
-------
lfXn ACC, thenXn+i = ((1 - backpurgeF actor) * ACC) + TV G (8b)
In Equation 8a, Xn represents the TVG on Day n. The conditions in Equation 8a will
determine the vapor generated for each day until n=5. To maintain model performance,
emissions are calculated for a maximum of five successive days. Beyond five days, the
algorithm assumes that breakthrough has occurred and that behavior over additional days has
stabilized. The vapor emissions are fleet averages by model year group. Vapor venting
emissions are expected to rise as more vehicles exceed their canister capacities and begin venting
fuel vapors. The development of the emission rates is covered in greater detail in the DELTA
report. 27
3.3.2.8 Activity
In order to properly account for off-cycle emissions, MOVES must account for the different
emissions rates of short (several hours) and long (multiple day) soaks. For any modeled day,
there is a sub-population of vehicles exhibiting 1st, 2nd, 3rd, nth day diurnal emissions. The
fractional allocations for 1st, 2nd, 3rd, and nth day diurnals are calculated from the
sampleVehicleTrip and sampleVehicleDays tables in MOVES. SampleVehicleTrip assigns
numbers of first starts during each hour of the day. For the fraction of vehicles having soaked
since at least midnight, the first engine start ends the cold soak episode. SampleVehicleDay
contains the population of vehicles for each sourceTypelD. Combining information for both
tables, it is simple to calculate the fraction of vehicles having soaked since midnight at any
given hour. For example, at 1:00AM, some fraction of vehicles less than 100 percent have not
yet started. The fraction continuously decreases throughout the day as more and more vehicles
start. At 12:00AM, the fraction only represents vehicles that were not driven.
Once the fraction of vehicles soaking at a given hour has been calculated, it must be estimated
how many prior days each has been soaking. We classify vehicles as 1st day, 2nd day, 3rd day, 4th
day, or 5+ days. We assume that after the 5th day, vehicles will exhibit repeat emissions since
the evaporative canister will either have broken through or be in conditions that will never
cause breakthrough. An activity study performed by Georgia Technological University 35
suggest that 16 percent of vehicles drive less than 3,000 miles per year. The MOVES inputs
are based on the conservative estimate that 50 percent of these low-mileage vehicles, or 8
percent of all vehicles, have been soaking for more than 5 days on any given day.
The sampleVehicleSoakingDayBasis table establishes the fraction of vehicles soaking for 5+
days. It contains five values, one for each soak day. The value for SoakDaylD 1 is the
percentage of vehicles soaking at the final hour of day 1. The product of SoakDayID=l and
SoakDayID=2 is the percent of vehicles soaking at the final hour of day 2. The product of all
five values is the percent of vehicles soaking for five days or longer.
Figure 13 presents the fraction of soaking vehicles throughout the day. The majority of
vehicles were driven the previous day and are on their first day soaking. The fractions of
vehicles on 2nd through 4th day soaking are developed from the remainder of 1st day soaking
32
-------
vehicles at hour 24. The fraction of vehicles soaking for 5 days or longer is 8 percent at hour
24. This method models bimodal vehicle usage, with most vehicles being driven almost daily
and the remaining vehicles being driven more intermittently.
Figure 13. Passenger Car Soak Distribution on a Weekday
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 Day
3.3.3 Hot Soak
Hot-soak vapor emissions begin immediately after a car ceases operation and continue until
the fuel tank reaches ambient temperature. In MOVES, the process of calculating hot-soak
vapor emissions is simpler than that for cold soak. Base rates exist for each model year and
age group and are expressed in units of grams per hour. They represent emissions at sea
level with RVP assumed at 9.0 psi. In developing the rates, leak and non-leak rates are
weighted together to form the base rate, similar to cold soak.
3.3.3.1 Hot Soak Data
Hot soak data comes from several programs with diverse testing procedures, vehicle model
years and technology, fuel parameters, and altitude. These programs include three summer
programs in Colorado and the E-77-2 programs in Arizona. See Table 16.
33
-------
Table 16: Hot Soak Evaporative Test Programs
Program
Location
Hot Soak Length
Fuel RVP
Altitude (ft)
No. Obs.
High Evap
Lipan IM station, CO
15 min
Fuel Supply
5130
100
High Evap
Ken Caryl IM station, CO
15 min
Fuel Supply
5130
175
High Evap
Denver IM station, CO
15 min, 1 hour
Fuel Supply
5130
100
E-77-2
Mesa, AZ
1 hour
7, 9, 10
1243
100
As explained below, the data collected in these programs was adjusted to the MOVES baseline
of a one-hour rate on 9.0 RVP fuel at sea level.
In addition, some tests were removed from our analysis. The vehicles in Colorado that
participated in the studies were recruited in-situ and therefore were subject to a wide range of
leak mechanisms. It was observed that some vehicles emitting more than 50 grams in 15
minutes in the PSHED had liquid leaks present. All vehicles with a calculated 15-minute
measurement greater than 50g/15min were removed from vapor leak analysis.
Furthermore, vehicles in the E-77 program were tested multiple times with different fuels,
whereas each vehicle in the Colorado population was tested once. In order to not over-
represent the E-77 vehicles in our sample, one measurement from each vehicle was selected
with preference given to the measurements on 9 RVP, E10 fuels (where available).
3.3.3.2 Test Duration Conversion
Every datum required a 15-minute mass and a one-hour mass because base rates in the
MOVES input table must be expressed in grams per hour; however, our method for
distinguishing leaks from non-leaks uses the 15-minute rate. Furthermore, if a measurement is
designated as from a leaking vehicle, the 15-minute measurement is used to project its rate of
occurrence in the fleet.
Because engines and fuel systems do not cool at a uniform rate, existing data was used to
develop this test-length conversion factor. In the E-77 suite, the cumulative time series data
for hot-soak tests on a minute-by-minute scale was readily available, enabling estimation of
vapor emissions over 15 minutes. Each set of vehicle data also contained a permeation rate.
The permeation rate was subtracted from the 15-minute hot soak measurement. The result is
the assumed vapor emissions during 15-minutes of hot soak. Similarly, hourly permeation
was subtracted from the 1-hour hot soak measurement. After compiling the 15-minute and 1-
hour values, the fraction of emissions occurring in the first 15 minutes can be calculated.
All of the Denver testing programs provided similar vehicle measurements to augment the E-
77 dataset. A subset of the vehicles were transported to a lab to receive a Hot Soak test.
Readings were taken at both 15 and 60 minutes.
Figure 14 illustrates the evaporative emissions occurring during a Hot Soak test. Vapor
emitted by permeation is assumed to accumulate at a linear rate while vapor emissions
attributed to the hot soak accumulate rapidly following engine shutoff but more slowly as the
engine cools.
34
-------
Figure 14: Cumulative Hot Soak and Permeation Emi ssions
~ HolSoak
S Permeation
C
o
w
0
15
30
Time (m}
45
60
Using the combined data from E-77 and Denver testing, we estimated the average fraction of
emissions in the first fifteen minutes following engine shutoff At first, it was thought that this
fraction would vary among groups of vehicles certified to different evaporative standards.
However, analysis of test results by certification groups did not seem to yield notably different
results. Instead a single fraction developed from all available data was applied fleet-wide. It
was estimated that 54 percent of emissions from a one-hour hot soak occur in the first 15
minutes. Conversely, emissions from a 15-minute hot soak must be multiplied by 1.85 to
estimate a full hour's emissions.
3.3.3.3 Correction for RVP and Altitude
MOVES base emission rates are intended to represent emissions on 9.0 RVP gasoline at sea
level so, the hot soak test data must also be corrected to account for the RVP and altitude of
each test.
Emissions in the available datasets were measured at varying levels of RVP. Some programs
recorded RVP, while other data has no explicit RVP information. Our first step is to estimate
the RVP for all measurements that do not contain this information.
The majority of the data with unknown RVP was gathered in the summer months in locations
with available fuel survey data. The mean RVP for June through August 2010 in Denver was
35
-------
8.40 RVP (standard deviation 0.20 RVP), and this value was assumed for all vehicles tested
from May through September. For non-summer months, RVP information was collected with a
small subset of the vehicle measurements. In the case of a non-summer measurement without
RVP information, the mean of all non-summer months is assumed. The mean RVP for non-
summer vehicles is 10.67 (standard deviation 1.75 RVP). The testing at the Lipan station was
all performed in the summer, so the RVP of the Lipan dataset is assumed to be 8.4.
Associating an RVP value with every measurement enables calculation of corrections for
altitude. All vehicles were tested either in Colorado (Elev. = 5,130 ft) or Mesa, AZ (Elev. =
1,243 ft). Both locations are far enough above sea-level that it would be erroneous to assume
their emissions are representative of sea-level emissions. Our approach is to solve the Wade-
Reddy equation for RVP (Equation 9) and calculate the equivalent RVP at sea level that would
generate the same amount of emissions. The E10 coefficients were used for this analysis.
Equation (9) Wade-Reddy Vapor Generation
TVGhigh = AhigheBhi3h*RVPmeas^eChigh*T1 _ eChlgh*T0)
Wade-Reddy Tank Vapor Generation (Solved for RVP)
DT/p _ / 1 \ , / TVGhigh \
SeaLevel ~ * U \ASeaLevel * (e^eaLevel*^ _ eCseaLevel*^)
This approach requires the assumption that vapor emissions will increase/decrease
proportionally to vapor generation. As a rule, to generate the same amount of vapor at high
altitude as generated at sea level, a fuel will have a lower RVP. Also, after a Monte-Carlo
analysis of varying starting and ending temperatures, the effect of either was found to be
negligible within the conditions these vehicles are likely to experience during testing.
Therefore, temperatures TO = 60°F and T1 = 65°F were chosen for this analysis.
The Wade-Reddy equation provides no coefficients for Mesa, AZ elevation so the adjustment
is a simple linear interpolation between Sea Level and Denver elevations. For example, to
solve for the TVGhigh used in Equation 9 corresponding to Mesa, Equation 10 was used.
Equation 10 Interpolation of Vapor Generation for elevations between Sea Level and Denver
(ElevationMpqn \
(TVG - TVGd . F... """
Thus, every measurement was paired with an RVP value that would generate the same
emissions at sea level. The next step was to estimate the emission result on fuel with an RVP
of 9.0 psi.
36
-------
In order to calculate an adjustment for each measurement, the same assumptions were
employed as above. Using the same temperature values, vapor generated at the sea level RVP
and at 9.0 RVP was calculated. Using equation 11, the ratio between these two values was
applied to the original emission measurements, and becomes the base MOVES emission rate.
Equation 11 Wade-Reddy Equation as applied to calculate emissions at adjusted RVP
TV G __rT .pBseaLevel*RVPmeas (pCseaLevel*Ti p^SeaLevel
1 v umeasRVP nSeaLevelc * )
TVGmoves =AhigheBW9-°(ecWTi - ecWo)
f i ,'V ^1Tl(3n?RI/P
HotSoakMOVES = HotS oak Measured
tvgmoves J
Thus, for each measurement we have an estimated emission rate for both 15 minutes and 60
minutes, at sea level, with 9 RVP fuel. As a quality check, the results of our 15-minute
emissions to 60-minute conversion and the results for data at both durations are plotted in
Figure 17. As expected, the estimated hourly emissions (red circles) from the 15-minute
measurements closely match the measurements (blue triangles) where data at both test lengths
were available.
Figure 17: Hot Soak Measurement Test Length
10
3
c
0
2 101
=5
(/)
10
10
10"'
Legend
~ 1 hour estimated ,±
Data available for both lengths A
A
¦ S
A# ~
~
~
¦A 1 1 r
10"' 10"1 10° 101
15 minute Measurement (g)
Quality assurance checks were also performed on the emissions values before and after
calculating their equivalences at Sea Level and 9.0 psi fuel. As expected, the tests measured
with higher RVP fuels at high altitude were reduced by wider margins under the influence of
the two corrections.
37
-------
Figure 18:
Hot Soak Measurement Normalization to 9.0 RVP
RVP
1 1 1 1 1 1 1 1
0 5 10 15 20 25 30 35
Normalized Emissions (g)
3.3.3.4 Extrapolation to Missing Model Years and Ages
After normalizing the complete dataset, the results were incorporated into the MOVES
database. In the MOVES emission rates tables, emission rates must exist for all model year
and age group combinations. As with most cross-sectional datasets, this required additional
modeling. For example, there is no data for 20-year-old, model year 2010 vehicles, or brand
new 1980 vehicles. To address this problem, we extrapolated the emission rate values. Table
17 describes the data.
Table 17: Hot Soak Measurements by Model Year and Age
Age Group
0-3
4-5
6-7
8-9
10-14
15-19
20+
Leak?
N
Y
N
Y
N
Y
N
Y
N
Y
N
Y
N
Y
Total
1961-1970
5
5
re
1971-1980
8
8
> %
1981-1995
6
15
46
55
8
39
169
« £
"O r>
1996-2003
1
26
6
36
6
53
30
158
o v-
2004-2010
12
3
26
2
5
48
Total
12
3
27
2
31
6
36
6
59
45
46
55
8
52
388
In ranges where no data was collected, leak and non-leak measurements are extrapolated from
similar MY/age groups. In MY/age groups where very small amounts of data were collected,
the measurements are combined with similar MY/age groups. Figure 18 illustrates how we
populated model year and age group emission rates where there was no data.
38
-------
Figure 19: Measurement Averaging
Age Groups Age Groups
0-3 4-5 6-7 8-9 10-14 15-19 20-99 0-3 4-5 6-7 8-9 10-14 15-19 20-99
1961-1970 F
S £ 1971-1980
"S o 1981-1995
| 15 1996-2003
2004-2010
(a) Non-Leak
(b) Leak
A darker shaded cell represents a bin where data is present.
An enclosed area represents one rate. The rate is calculated by averaging all enclosed
data.
For example, one non-leak rate exists for model years 1996-2003, ages 0-7. The rate is
calculated by averaging available data, which only exists at age 6-7. For every model year and
age group, there is a leaking rate and non-leaking rate. The two rates, weighted by leak
prevalence, form the average hourly hot soak emission rate for a given bin. Figure 20
demonstrates how leak rates and non-leak rates are combined to form a final weighted rate for
a given model year and age combination.
39
-------
Figure 20: Calculate Weighted Evaporative Emissions
> lOg / Qhr >5g/Qhr >2g/Qhr >0.3g/Qhr
|
mean
Leak Measurements
x
mean
mean
I
mean
Prevalence Prevalence Prevalence Prevalence
Leak Prevalence
< 0.3g / Qhr
Non-leak
measurem
ents
5 weighted
5 leak rate
mean non-
leak rate
Non-Leak Prevalence
I ¦
MOVES
-> Emission
Rate
For every model year and age group combination, the calculation outlined in Figure 20 is
performed. Figure 21 shows the Flot Soak rates. The inclusion of leaking vehi cle resulted in
higher emissions, particularly for older model years where leaks are more prevalent.
40
-------
Figure 21: Hot Soak Emission Base Rates (9.0 RVP at Sea Level)
20
15'
E 1°
LU
MYG
10 15 20 0
10 15 20
0 5 10 15 20
Vehicle Age
10 15 20 0
10 15 20
1961-1970
1971-1980
1981-1995
1996-2003
2004-2010
3.3.4 Running Loss
3.3.4.1 Pre-Tier 2 Emission Rates
Running Loss emissions consist of vapor venting during vehicle operation. Data used to
develop running loss emission rates for Pre-Tier 2 vehicles is from CRC E-3513 and CRC E-
41.14-15 These two programs tested 200 vehicles with model years ranging from 1971-1997.
For each vehicle, fuel tank temperature was calculated at the end of the running loss test using
the fuel tank temperature algorithm (See Section 3.1). The running loss test performed in E-35
consisted of a HDDS, a two minute idle, a New York City Cycle (NYCC), a two minute idle, a
second NYCC, a two minute idle, a HDDS, and a final two minute idle.
The data was filtered/reduced such that each test meets the following requirements:
Non-liquid-leakers (emissions <137.2 g/hourc)
As received vehicles (no retests)
Fuel system pressure test result must be pass, fail, or blank
The average tank temperature was calculated by assuming a linear increase in temperature.
c Converted from 7.0 g/mile used in MOBILE6 31
41
-------
Thus, the average is calculated by averaging the start temperature of the test and the final
temperature. The average temperature is used to estimate the permeation rate using default
permeation rates and the permeation temperature adjustment.
Gram/hour rates were calculated by dividing total emissions by the duration of the running loss
test (4300 seconds). Permeation was subtracted for each hour to segregate tank vapor venting
(TVV) emissions. After analysis of the TVV data, running loss TVV rates were distinguished
by model year only. Table 18 shows the results of the analysis.
An I/M effect was not observable from this data, so the MOVES running loss TVV rates for
I/M and non-I/M are the same.
Table 18: Pre-Tier 2 Running Loss Emission Rates by Model Year and Age
Model year group
TW mean [g/hr]
Pre-1971
12.59
1971-1977
12.59
1978-1995
11.6
1996-2003
0.72
3.3.4.2 Tier 2 & Later Emission Rates
Running loss emission rates for Tier 2 and later vehicles were developed from a 2014 study of
five Tier 2 vehicles.25 In this study, vehicles were tested at two fuel RVP levels (7.51 psi and
10.33 psi) with and without implanted vapor leaks. Vapor leaks were installed at either the
canister or top of fuel tank, and with either 0.020" or 0.040" diameters, for a total of 4
possible leak configurations. The canister and fuel tank locations were chosen due to their high
rate of occurrence in the fleet. 34
MOVES running loss emission rates are expressed in grams per hour and with a fuel vapor
pressure of 9 psi. Results from this testing are expressed in grams per test (4300 seconds) and
at two fuel vapor pressures (7.51 and 10.33). Therefore, the reported results must be
normalized to MOVES dimensions.
As in the development of Pre-Tier 2 emission rates, gram/hour rates were calculated by
dividing total emissions by the duration of the running loss test (4300 seconds). The
measurements were then adjusted to a 9-RVP equivalent emissions measurement using the
equations and coefficients described in Section 3.3.4.3.
Because our determination of a given vapor leak's rate of occurrence among all vapor leaks is
based on its hot soak emissions, each running loss test was immediately followed by a standard
one-hour hot soak procedure. Using the same process as in Section 3.3.3.2, the one-hour hot
soak results were multiplied by .54 to estimate the emissions at the 15 minute point. With this
result, each measurement was binned as in Table 18 and the weighted average leak emissions
rate determined.
42
-------
Using the average non-leak value, the weighted average leak value, and the leak prevalences
from Section 3.3.2.5, an average emissions rate is calculated. Tier 2 and later running loss
emission rates are the first running loss rates in MOVES to account for vapor leak emissions
Tier 3 rates were estimated using ratios to Tier 2 standards. The resulting running rates are
shown in Figure 22.
Figure 22: Tier 2 & Tier 3 Running Loss Rates
Vehicle Age
3.3.4.3 Running Loss Fuel & Temperature Effects
Running Losses are affected by both temperature and fuel Reid Vapor Pressure (RVP). The
adjustments used in MOVES3 are taken from MOBILE6 and are applied to all model years
and source types. MOBILE6 was run for a series of temperatures and RVP levels for passenger
cars. A linear model was fit to the MOBILE6 results. The mean base emission rates for
running losses in MOVES are recorded in the 'EmissionRciteByAge ' table. Running loss
rates were assumed to be measured at 9 RVP and 95°F. The results from MOBILE6 were
normalized to the MOVES emission rates as multiplicative adjustments to the mean base rates.
For example, a multiplicative adjustment of 1 would be applied to a 9 RVP fuel at 95°F.
The running loss adjustments:
Are multiplicative adjustments.
Apply to all gasoline source types and model years.
Are the same at temperatures at or below 40 F.
43
-------
Are applied as a function of both RVP and ambient temperature.
Use the 7 RVP coefficients for RVP values below 7 psi.
Use the 10 RVP coefficients for RVP values above 10 psi.
Are not adjusted for RVP at temperatures of 40 F or below.
AdjustedRunningLoss = RunningLoss * Adjustment[Temperature, RVP) (12)
The adjustment coefficients are recorded in the evapRvpTemperatureAdjustment table in the
default MOVES database. The RVP adjustment range is dynamic; if new sets of coefficients
for RVP values greater than 10 or less than 7 are added to the table, MOVES will use those
values and set new minimum and maximum RVP values. Figure 23 shows the calculated
adjustment values.
Fi gure 23: Running Loss Temperature and RVP Effect
0.6
40 50 60 70 80 90 100
Ambient Temperature (degrees Fahrenheit)
3.4 Inspection/Maintenance (I/M) Program Effects
Inspection and Maintenance program efforts vary in their procedures for testing evaporative
emissions. Some locations use a fill pipe pressure check and gas cap check, others use just a
scan of the onboard diagnostics (OBD), and others will use all three approaches. These types
of tests are not expected to reduce permeation or liquid leaks and do not guarantee the
detection of a vapor leak within a vehicle. MOVES assumes tank vapor venting is the only
evaporative process where I/M benefits are realized.
44
-------
3.4.1 I/M Factor (Relative Program Effectiveness)
In MOVES, the I/M factor describes the relative effectiveness of an I/M program; a higher I/M
factor indicates a more effective I/M program. Data from four I/M programs were used in the
development of MOVES I/M factors. The Phoenix, AZ program contained the most extensive
data, so we have used it to represent a reference condition relative to which other programs can
be assessed. Data from the programs in Tucson, AZ, Colorado, and North Carolina were used
to determine the effectiveness of other I/M program designs. See Table 19.
Table 19: Description of I/M Programs 36
Location
Gas Cap
Test
OBD
Pressure
test
Frequency
Network
Years
Colorado
Y
Advisory
N
Biennial
Hybrid
Annual
2003-2006
N.Carolina
N
Y
N
Annual
Annual
2002-2006
Phoenix
Y
Y
Y
Biennial
Centralized
2002-2006
Tucson
Y
Y
N
Annual
Centralized
2002-2006
In order to develop I/M factors, failure data was used from I/M programs. These failure
frequencies were only used to estimate the relative effectiveness of differing evaporative I/M
programs. They were not used to model the actual prevalence of evaporative leaks. For
information on the modeling of vapor leak prevalence please see Section 3.3.2.5. For
information on liquid leaks, see Section 3.5.
The Phoenix evaporative I/M program performed gas-cap tests on all vehicles, OBD scans on
OBD-equipped vehicles, and fill-pipe pressure tests on pre-OBD vehicles. The OBD codes
used to assign evaporative failures are listed in Table 20 for all vehicle makes and additionally
P1456 and P1457 for Honda and Acura vehicles. Vehicles with one or more of these faults
were flagged as failing vehicles, analogous to pre-OBD vehicles that failed the pressure test.
Very few vehicles failed both the gas cap test and the pressure/OBD test. Therefore, the total
number of failures is the sum of gas cap and pressure/OBD failures.
Table 20: OBD Evaporative Emission Trouble Codes
OBD Code
Description
P0440
Evaporative Emission Control System Malfunction
P0442
Evaporative Emission Control System Leak Detected (small leak)
P0445
Evaporative Emission Control System Purge Control Valve Circuit Shorted
P0446
Evaporative Emission Control System Vent Control Circuit Malfunction
P0447
Evaporative Emission Control System Vent Control Circuit Open
P1456
EVAP Emission Control System Leak Detected (Fuel Tank System)
P1457
EVAP Emission Control System Leak Detected (Control Canister System)
45
-------
The I/M failure frequencies were developed from the Phoenix data using initial and final
results for a vehicle in a given I/M cycle. For passing vehicles, the initial and final tests are the
same. The initial and final failure frequencies were averaged to develop an I/M failure
frequency for each model year and age group. Using the initial failure frequencies alone would
neglect the required repairs occurring on most failing vehicles, while using only final failure
frequencies would neglect the prior existence of failing vehicles.
To develop non-I/M failure frequencies, the sample was restricted to vehicles registered in
states that do not have any I/M programs.
The Tucson data was used to determine the effect of I/M program frequency (annual vs.
biennial). For OBD-equipped vehicles, Tucson performs gas-cap and OBD tests annually,
while Phoenix performs them biennially. Therefore, we were able to develop an the
effectiveness ratio for Annual/Biennial programs by analyzing the Tucson data.
The North Carolina data was used to estimate the effectiveness of using the OBD scan as the
sole test in a program. In North Carolina, expansion of I/M program boundaries has led to
many vehicles being tested for the first time. These vehicles were effectively non-I/M until
their first test. Vehicles were flagged as non-I/M tests if they were tested before the official
start of the I/M program or were registered in a new I/M county. Failure frequencies of the
non-I/M vehicles were compared to vehicles tested in I/M areas. The I/M effectiveness of an
OBD only I/M program is estimated to be a 63 percent reduction in failures or a non-I/M to
I/M failure ratio of 1.6. This ratio was then applied to Phoenix OBD and pressure test failure
frequencies to determine non-I/M failure frequencies.
The Colorado data was used to determine the effectiveness of gas cap tests. In Colorado, the
I/M data is primarily from the Denver and Boulder metropolitan areas. However, some
residents are new to this area, having moved from non-I/M counties and states. These vehicles
were effectively non-I/M until their first test. Vehicles were flagged as non-I/M if they were
registered in a state without an I/M program, or in a non-I/M county within Colorado.
Colorado OBD data was not used, because OBD in Colorado is only advisory and does not
pass or fail a vehicle. The failure rates of the non-I/M vehicles were compared to those in the
I/M fleet. The effectiveness of a gas-cap-only I/M program is estimated to be a 45 percent
reduction in failures or a non-I/M to I/M failure ratio of 1.2. This ratio was then applied to gas
cap failure frequencies to determine non-I/M failure frequencies.
The I/M factor in MOVES adjusts emission rates depending on the characteristics of a given
county's I/M program. Our reference program, Phoenix, has an IM factor of 1. Non-I/M areas
have an IM factor of 0. The failure frequencies from the other counties are used to calculate
I/M factors for the diverse types of evaporative I/M procedures. The I/M factor is assumed to
have a linear relationship with failure frequency. Figure 24 illustrates how the I/M factor varies
with different I/M programs. Different programs fall on the line as determined by the analysis
above, based on specific evaporative tests performed. For the vehicles in Figure 24, Tucson's
OBD and gas cap tests are annual, compared to Phoenix's biennial requirement, which gives
Tucson a lower failure frequency, thus a higher I/M factor. Colorado's frequency is biennial,
but their OBD test is non-enforcing. As a result, their data shows a higher failure frequency,
46
-------
resulting in a lower I/M factor.
Figure 24: I/M Factor, MY 1999-2003, Age 4-5
l/M Factor
3.4.2 Leak Prevalence
To estimate the impact of an I/M program, the appropriate I/M factor is applied to the
estimated vapor leak prevalence rates. However, because the leak prevalence rates were
developed from a test program in the Denver, CO area, and because, as explained above, the
Denver program is not the MOVES reference program, the Denver vapor leak prevalence rates,
developed in Section 3.3.2.5 must be adjusted for use in MOVES.
As illustrated in Figure 24, the with-I/M failure frequency in Denver is about 30 percent less
than non-IM (I/M factor = 0) and 30 percent higher than Phoenix (I/M factor = 1) so the leak
prevalence rates developed from Denver data were adjusted accordingly before being added to
the MOVES database. This adjustment reflects the analysis described in the previous section
and can be observed in Figure 25. During a MOVES run for the Denver area, the Denver I/M
factor will be applied to the adjusted leak prevalence rates and emissions will be modeled with
the same prevalence rates originally estimated for Denver.
47
-------
Figure 25: Adjusting Denver Leak Prevalence Data
Denver Leak Frequency
Phoenix (I/M) non-I/M
3.5 Liquid Leaks
Liquid leaks include any non-vapor fuel escaping the fuel system. The average liquid leaking
rate is determined using the leaking vehicles excluded from the analysis above. Because the
testing methods used did not distinguish the different evaporative emission processes,
permeation and tank vapor venting are estimated using the calculation methods described in
Section 3.2 and Section 3.3 and subtracted from the total measurement. The remaining
emissions after permeation and vapor venting are subtracted are assumed to be caused by
liquid leaks. Due to limitations in the data quality and quantity, the measurements are
averaged across all vehicles by the three different modes and shown in Table 21.
Table 21: Liquid Leak Emission Rates (g/hr)
Operating Mode
Liquid leak rate
Cold Soak
9.85
Hot Soak
19.0
Operating
178
48
-------
The liquid leak emission rates must be multiplied by the percentage of leakers in the fleet to
get an average liquid leaking emission rate. The studies by BAR 20 and API 21 provided
this data. The estimates of liquid leak prevalence are shown in Table 22. It is assumed that
most leaks do not occur until vehicles are 15 years or older.
Table 22: Percentage of Liquid Leaks by Age
Age group
Percentage of leakers in fleet
0-9
0.09%
10-14
0.25 %
15-19
0.77%
20+
2.38%
Table 23 contains the fleet-weighted liquid leak rate. There is insufficient data to conclude that
these rates change with model year or are affected by I/M programs.
Table 23: Weighted Liquid Leak Emissions(g/hr)
Age group
Cold soak
Hot soak
Operating
0-9
0.009
0.017
0.158
10-14
0.025
0.048
0.450
15-19
0.075
0.145
1.360
20+
0.235
0.452
4.230
As with vapor leaks, we expect a reduction in liquid leak prevalence due to improved system
design and integrity under the Tier 3 regulations. However, liquid leaks in advanced evaporative
systems are primarily caused by tampering and mal-maintenance. Therefore, we estimate Tier
3 will prevent half as many liquid leaks as vapor leaks, shown in Table 24.
Table 24: Weighted Tier 3 Liquid Leak Emissions (g/hr)
Age group
Cold soak
Hot soak
Operating
0-9
0.007
0.013
0.123
10-14
0.019
0.037
0.342
15-19
0.058
0.113
1.054
20+
0.180
0.348
3.258
49
-------
3.6 Refueling
Refueling emissions are the displaced fuel vapors when liquid fuel is added to the tank. The
calculation of vapor losses includes any liquid fuel that is spilled during refueling and
subsequently evaporates. Refueling emissions are estimated from the total volume of fuel
dispensed (gallons). This volume is estimated from the average daily distance travelled (VMT)
and the estimated fuel consumption. Both the spillage and the vapor displacement associated
with refueling events are in terms of grams per gallon of fuel dispensed. Diesel vehicles are
assumed to have negligible vapor displacement, but fuel spillage is included in the diesel
refueling emissions.
Uncontrolled and unadjusted refueling emissions are simply the displaced grams of fuel vapor
per gallon of liquid fuel, plus the grams per gallon for spillage. AP-42 Volume I Section 5.2.2.3
20 lists the spillage as 0.7 lb/1000 gallons, which is 0.31g/gallon of dispensed fuel. MOVES
uses this value for gasoline, E85, and diesel fueled-vehicles.
The vapor displaced by refueling gasoline and E85 vehicles is a function of temperature and
gasoline Reid Vapor Pressure (RVP)37:
E= - 5.909 -0.0949dT+ 0.08847/^+ 0.485RVP (12)
Where:
E
= Displaced Vapor (non-methane grams)
RVP
= Reid Vapor Pressure (psi)
Tdf
= Dispensed gasoline temperature (degF)d, 38
dr
= Temperature difference between tank and dispensed
dr
= 0.418*7nF -16.6
Dispensed fuel temperature is the temperature of the fuel flowing from the pump. For
MOVES, the temperature of the dispensed fuel in the equation is assumed to be the same as
the monthly average ambient temperature found in the ZoneMonthHour table. The monthly
average temperature assumed in MOVES are between 20 and 90 degrees Fahrenheit. For
ambient temperatures beyond those limits, the dispensed fuel temperature is set to the value
calculated at the limit. Furthermore, the dT value cannot be greater than 20 degrees. The dT
equation is developed in an Amoco study.39 In that study, the difference in temperature was
never greater than 20 degrees.
Two emission control strategies exist to limit fuel lost during refueling for gasoline and E85
vehicles. First, there are programs designed to capture refueling vapors at the pump. These are
often referred to as Stage II vapor control programs. Second, vehicles manufactured since
d While MOVES uses the monthly average temperature as the dispensed fuel temperature, a California study has
suggested an alternate calculation where dispensed gasoline temperature = 20.30 + 0.81*Tamj^&. We will
consider revising this in future versions of MOVES.
50
-------
1998 have onboard refueling vapor recovery (ORVR) systems that store refueling vapors in
the vehicle's evaporative emission canister.
The implementations of Stage II systems vary from area to area and affect the fuel vapor
displacement and the amount of spillage. MOVES uses two factors to adjust the refueling
losses and account for this variation.
1. The refueling vapor program adjustment is a value between zero and one indicating the
percent reduction of total potential vapor losses by state or local programs (such as Stage
II recovery programs).
2. The refueling spill program adjustment is a value between zero and one indicating the
percent reduction of refueling spillage losses by state or local programs (such as Stage II
recovery programs). These program adjustments in MOVES are applied by county. Each
county has a unique value for vapor and spillage program adjustments. The MOVES
default database contains information about all of the existing Stage II programs by
county based on the parameters used for the 2005 National Emission Inventory (NEI).
The estimated effects of Stage II programs can be altered by manually editing the
MOVES Stage II tables. The program adjustment values for each county and calendar
year are stored in the default MOVES 'CountyYear' table. We are aware that this table
does not capture some Stage II program terminations, and plan to update this table in a
future MOVES version.
MOVES uses a separate factor to address the on-board refueling vapor recovery (ORVR)
systems on gasoline and E85 vehicles. The effects of ORVR technology is phased in
beginning in model year 1998, as shown in Table 25. There is no ORVR for motorcycles.
51
-------
Table 25: Phase-In of Onboard Refueling Vapor Recovery
Model Year
Passenger
Light Trucks
Light Trucks
Heavy Duty
Heavy Duty
Cars
<6,000 lbs
6,000-
Trucks 8,500-
Trucks
GVWR
8,500 lbs
10,000 lbs
10,000-
GVWR
GVWR
14,000 lbs
GVWR
1998
40%
0%
0%
0%
0%
1999
80%
0%
0%
0%
0%
2000
100%
0%
0%
0%
0%
2001
100%
40%
0%
0%
0%
2002
100%
80%
0%
0%
0%
2003
100%
100%
0%
0%
0%
2004
100%
100%
40%
40%
0%
2005
100%
100%
80%
80%
0%
2006-2017
100%
100%
100%
100%
0%
2018 and
100%
100%
100%
100%
100%
newer
MOVES applies a 98 percent reduction in refueling vapor losses and 50 percent reduction in
refueling spillage losses for ORVR equipped vehicles, except for heavy duty vehicles with
GVWR > 14,000 lbs. The refueling tech adjustment is a number between zero and one which
indicates the reduction in full refueling vapor and spillage losses by each sourcetype and model
year that result from improvements in vehicle technology (such as the Onboard Refueling Vapor
Recovery rule). The technology adjustment is applied the same in all locations.
The technology adjustment values are stored in the MOVES 'SourceTypeTechAdjustment'
table, which are stored by source type and model year. The technology adjustment values
take into account the population fraction of the vehicles by fuel type and regulatory class
(GVWR for heavy-duty vehicles) within each source type as documented in the
Population and Activity Report.5 Note that the factors for heavy-duty trucks 8,500 lbs and
higher are reflected in single-unit short-haul and long-haul vehicles (source types 52 and
53). For example, LHD2b3 vehicles (GVWR < 14,000 lbs) compose 49% of the
population of gasoline source type 52 and 53 vehicles in model year 2014 and later in
MOVES3. Accordingly, in MOVES3, we updated the technology adjustment values for
refueling vapor losses to be consistent with the updated population information to be set
to 49% to account for 100% refueling vapor loss control of the LHD2b3 gasoline vehicles
in these model years.6 We did not update the technology adjustment factors for refueling
spillage losses, and they reflect the fuel type and regulatory class population fractions
from a previous version of MOVES, coupled with an assumption that the LHD2b3
e We assumed that the LHD2b3 vehicles followed the Class 2 vehicles (GVWR 8,500 to 10,000) phase-in schedule.
However, within the source type 52 and 53, these vehicles will be either Class 2 vehicles with 6 tires or Class 3
vehicles (GVWR 10,000 to 14,000).5 We plan to update these values to follow the Class 3 phase-in in a future
version of MOVES.
52
-------
gasoline vehicles have a 50% technology adjustment factor for spillage losses when
ORYR technology is fully implemented/
MOVES applies both the program and technology adjustment to all model years. This
means that Stage II programs are assumed to affect vehicles not equipped with ORYR and
additionally, any refueling emissions that are not captured by the ORYR systems.
MOVES does not account for any interaction between ORYR systems and gasoline
dispensing stations equipped with Stage II equipment.
f We plan to update these values with the latest population numbers in the next version of MOVES.
53
-------
Appendix A Notes on Evaporative Emission Data
Parameters: Vehicle Numbers, Test No., Ambient Temperature, RYP, Model Year, Fuel
System, Purge, Pressure, Canister, Gram HC, Retest
E-41 CRC Late Model In-Use Evap. Emission Hot Soak Study (1998)
50 vehicles (30 passenger cars and 20 light duty trucks)
Model years 1992 to 1997
Average RYP: 6.5 psi
Diurnal Temperature: 72 to96°F
Fuel System: Port Fuel Injection, Throttle Body Injection
Vehicle fuel tank drained and refilled to 40% of capacity with Federal Evaporative
Emission
Test Fuel
Driving schedule will be a full LA-4-NYCC-NYCC-LA4 sequence, with two minute idle
periods
following the first LA-4, the second NYCC, and the final LA-4.
Hydrocarbon readings will be taken continuously throughout the running loss test.
Cumulative mass emissions will be reported at one-minute intervals.
Ambient Temperature in running loss enclosure: 95 °F
E-9 CRC Real Time Diurnal Study (1996)
151 vehicles (51 vehicles MY 1971-1977, 50 vehicles MY 1980-1985, 50 vehicles MY 1986-
1991)
Odometers range from 39,000 to 439,000 miles
Fuel tank volume was 15% of the rated capacity
RVP: 6.62 psi (average sum of 47 vehicles)
Diurnal temperature: 72 to 96°F
Fuel System: Port Fuel Injection, Carburetor, Throttle Body Injection
CRC E-35 Running Loss Study (1997)
150 vehicles (50 vehicles MY 1971-1977, 50 vehicles MY 1980-1985, 50 vehicles MY 1986-
1991)
Ambient Temperature in running loss enclosure: 95 °F
54
-------
RYP: 6.8 psi
Fuel System: Port Fuel Injection, Carburetor, Throttle Body Injection
EPA Compliance Data
2-Day Test
Length of the hot soak: 1 hour
77 vehicles
RYP: average 8.81 psi
Ambient Temperature:
Federal Standard (72 to 96° F) Diurnal
Cal. (65 to 105°F)Diurnal
Hot Soak: 81.67'F
Fuel System: Port Fuel Injection
MSOD (Mobile Source Observation Database):
Hot Soak 1 hour hot soak evaporative test
FTP Federal test procedure (19.53 mph), also referred to as the UDDP schedule
NYCC New York City Cycle Test (7.04 mph)
BL1A 1 hour Breathing Loss Evap. Test Gas Cap left On BL1B 1 hour Breathing Loss
Evap. Test Canister as reed. ST01 Engine Start cycle test
4HD 4 hour Diurnal test
24RTD 24 Hour Real Time Diurnal 33RTD 33 Hour Real Time Diurnal 72RTD 72 Hour
Real Time Diurnal
3Rest 3 Hour Resting Loss Evap. Emission Test (follows 1 HR Hot Soak) CY6084 Real time
diurnal temperature pattern: range 60 to 84 F CY7296 Real time diurnal temperature
pattern: range 72 to 96 F CY8210 Real time diurnal temperature pattern: range 82 to
102 F
DIURBL Standard temperature rise for 1 hour diurnal or breathing loss evaporative emis-
sion test
F505 Bag 1 of federal test procedure (25.55 mph)
ASM Acceleration Simulation Mode Test Procedure
ATD Ambient Temperature diurnal evaporative Test, shed temp constant, vehicle begins 24
degree cooler
55
-------
Appendix B Tank Fuel Generator
In MOVES, we account for changes to the RVP of gasoline fuels due to weathering and comingling as a
preliminary step before calculating vapor venting emissions. These calculations are handled in the Tank
Fuel Generator (TFG).
TFG-la: Calculate Average Pump Gasoline and Ethanol Blend Type
Inputs:
marketShare from FuelSupply (county, fuelYear, monthGroup, fuelFormulationID)
ETOHvolume from FuelFormulation (fuelFormulationID)
RVP from FuelFormulation (fuelFormulationID)
fuelSubTypelD from FuelFormulation (fuelFormulationID)
fuelTypelD from FuelSubType (fuelsubtypelD)
Outputs:
averageRVP (county, fuelYear, monthGroup)
tankAverageETOHVolume (county, fuelYear, monthGroup)
Calculations:
averageRVP = For all FuelFormulations in county, fuel year & monthGroup where
fuelType = "gasoline" (ie fuelTypelD =1))
(Sum (RVP*marketshare)) / (Sum (marketshare))
tankAverageETOHVolume = For all Fuel Formulations in county, fuel year & monthgroup
where fuelType = "gasoline" (ie fuelTypelD = 1))
(Sum (ETOH Volume*marketshare)) / (Sum (marketshare))
TFG-lb: Calculate Ethanol Market Share and Ethanol BlendType
Inputs:
marketShare from FuelSupply (county, fuelYear, monthGroup, fuelFormulation)
fuelSubTypelD from FuelFormulation
fuelTypelD from FuelSubType
Outputs:
gasoholMarketShare (countyID, fuelYearlD, monthGroupID)
ethanolBlendType (county, fuelYear, monthGroup)
Calculation:
gasoholMarketShare: For all FuelFormulations in county, fuelyear & monthgroup where
ETOHVolume >= 4
gasoholMarketShare =Sum (marketShare)
lowETOHRVP: For all FuelFormulations in county, fuel year & monthgroup WHERE
56
-------
fuelType = "gasoline" (ie fuelTypelD = 1) and ETOHVolume <4
IF (sum (marketshare) = 0,
lowETOHRVP=AverageRVP
ELSE
lowETOHRVP=(Sum (RVP*marketshare)) / (Sum (marketshare))
highETOHRVP: For all FuelFormulations in county, fuel year & monthgroup WHERE
fuelType = "gasoline" (ie fuelTypelD = 1)) and ETOHVolume >= 4
IF gasoholMarketShare = 0,
highETOHRVP = AverageGasolineBlendRVP
ELSE
highETOHRVP =(Sum (RVP*marketshare)) / gasoholMarketShare
ethanolBlendType:
IF absolute value (highETOHRVP -lowETOHRVP) <= 0.2,
ethanolBlendType ="Match"
ELSE ethanolBlendType = "Splash"
TFG-lc: Calculate Commingled Tank Fuel RVP
Inputs:
gasoholMarketShare (countylD, fuelYearlD, monthGroupID) from TFG-lb
averageRVP (countylD, fuelYearlD, monthGroupID) from TFG-la
Commingling Lookup (stored in MOVES)
LookupMarketShare
Commingling
RVP Factor
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
1.000
1.016
1.028
1.035
1.039
1.040
1.038
1.034
1.027
1.018
1.000
0.8
0.9
1.0
Outputs:
commingledRVP (countylD, fuelYearlD, monthGroupID)
57
-------
Calculation:
comminglingFactor (countylD, fiielyearlD, monthgroupID) = lookup from table using
smallest value of "LookupMarketShare" that is greater than or equal to the
gasoholMarketShare.
commingledRVP = averageRVP * comminglingFactor
TFG-2: WeatheredRVP
TFG-2a: Calculate "EvapTemp" by ZonelD, MonthGroupID
Inputs:
temperature (zonelD, hourlD monthgroupID)
zonelD from masterloopcontext
Outputs:
zoneEvapTemp (zonelD, monthgroupID)
Calculation :
zoneMin(zoneID, monthgroupID) = MIN (temperature(zoneID, monthgroupID, hourlD))
zoneMax (zonelD, monthgroupID) = MAX(temperature(zoneID, monthgroupID, hourlD)
zoneEvapTemp =
IF zoneMax <40 or zoneMax-zoneMin <=0, (zoneMin+zoneMax)/2
ELSE zoneEvapTemp(zoneID, monthGroupID) =
-1.7474+1.029*zoneMin+ 0.99202* (zoneMax-zoneMin)-0.0025173*zoneMin*
(zoneMax-zoneMin)
TFG-2b: Calculate ratio of weathering loss for gasoline by Zone, Year & Month at actual ambient
temperatures relative to a diurnal swing of 72-96 F
Inputs:
zoneEvapTemp (zonelD, monthgroupID) from previous step
commingledRVP (countylD, fuelYearlD, monthgroupID) from TFG-lc
zone(countyID, zonelD)
Outputs:
ratioGasolineRVPLoss(zoneID, fuelYearlD, monthgroupID)
Calculation :
ratioGasolineRVPLoss =MAX (0, [-2.4908 + 0.026196 * zoneEvapTemp + 0.00076898 *
zoneEvapTemp * commingledRVP]/[-0.0860 + 0.070592 * commingledRVP ] )
TFG-2c: Calculate weathering loss for average fuel for standard temperatures
58
-------
Inputs:
ethanolBlendType (county, fuelYear, monthGroup) from TFG-la
gasoholMarketShare (countylD, fuelYearlD, monthGroupID) from TFG-lb
Outputs:
avgWeatheringConstant (countylD, fuelYearlD, monthGroupID)
Calculations:
IF ethanolBlendType = "Match", avgWeatheringConstant = 0.049 - 0.0034 *
gasoholMarketShare
ELSE avgWeatheringConstant = 0.049 - 0.0116 * gasoholMarketShare
TFG-2d: Calculate weathered RVP for county-average fuel adjusted for zone temperatures
Inputs:
ratioGasolineRVPLoss (zonelD, fuelYearlD, monthgroupID) from TFG-2b
avgWeatheringConstant (countylD, fuelYearlD, monthGroupID)
from previous step
commingledRVP (countylD, fuelYearlD, monthGroupID) from TFG-lc
zone(countyID, zonelD)
Outputs:
tankAverageGasolineRVP(zoneID, fuelYearlD, monthgroupID)
Calculation :
tankAverageGasolineRVP (zoneID) = commingledRVP(countylD) * (1 -
ratioGasolineRVPLoss (zonelD) * avgWeatheringConstant (countylD))
59
-------
Appendix C List of Acronyms
API
American Petroleum Institute
BWC
Butane Working Capacity
CNG
Compressed Natural Gas
CRC
Coordinating Research Council
DTC
Diagnostic Trouble Code
E0
Gasoline containing 0 percent ethanol by volume
E10
Gasoline containing 10 percent ethanol by volume
E-65
CRC fuel permeation from automotive systems study
E-77-2
evaporative emission/permeation test program
EPA
U.S. Environmental Protection Agency
ETOH
Ethanol
FTP
Federal Test Procedure
HC
Hydrocarbons
HD
Heavy-Duty
I/M
Inspection and Maintenance program
LDGV
Light-Duty Gasoline Vehicle
LEV
Low emission vehicle
LEV III
California Tier 3 light-duty emission standards of 2012
MOBILE
Original Highway Vehicle Emission Factor Model pre-2004
MOBILE6
Versioned Highway Vehicle Emission Factor Model
MOVES
Motor Vehicle Emission Simulator Model
MSAT
Mobile Source Air Toxics
MTBE
Methyl tertiary-butyl ether
NMHC
Non-Methane Hydrocarbons
NMOG
Non-methane organic gases
OBD
Onboard Diagnostics
ORVR
Onboard Refueling Vapor Recovery
OTAQ
Office of Transportation and Air Quality
PI
Petroleum Institute
RVP
Reid Vapor Pressure
PSHED
Portable Sealed Housing for Evaporative Determination
SHED
Sealed Housing for Evaporative Determination
THC
Total Hydrocarbon
Tier 2
Vehicle emissions certification standards phased in from 2002
Tier 3
Vehicle emissions certification standards phased in from 2017
TOG
Total Organic Gases
TVG
Tank Vapor Generated
TVV
Tank Vapor Venting
voc
Volatile Organic Compound
VSP
Vehicle specific power
60
-------
Appendix D Glossary
backpurge - as the temperature decreases a vacuum is created in the fuel system which pulls
the hydrocarbons from the charcoal canister into the fuel tank, creating more space in the
canister for hydrocarbons to adhere during the next heating period
breakthrough - when the vapor generated by the fuel system overwhelms the charcoal can-
ister and uncontrolled hydrocarbons are released into the atmosphere
canister - the device in an evaporative emission control system that captures and stores
evaporative emissions generated within the vehicle for later combustion by the engine; a
canister typically contains activated carbon as a storage medium
CRC - Coordinating Research Council, a consortium of auto and oil industry members which
sponsors common research programs
diurnal cold soak - Vapor lost while vehicles are parked at ambient temperature.
HC - hydrocarbon, an organic compound consisting entirely of hydrogen and carbon; a
com- bustible fuel source which can be either gaseous or liquid
hot soak - Vapor lost in the time period immediately after turning off a vehicle.
I/M - Inspection and Maintenance program run by States to find and correct emissions
problems for vehicles registered in the State
light-duty vehicle/LDGV - passenger cars
MOVES - MOtor Vehicle Emissions Simulator; official US EPA model for estimating
emis- sions from national fleet of onroad vehicles
MSAT - Mobile Source Air Toxic rule which regulates toxic mobile source emissions such
as benzene and ethanol
permeation - the migration of hydrocarbons through materials in the fuel system
OBD - Onboard Diagnostics, an electronic automotive system with the ability to continually
track the functionality of emissions control and other components, and alerts the driver and/or
vehicle inspector when a problem is found
ORVR - Onboard refueling vapor recovery system which is designed to capture fuel vapors
at time of refueling
PSHED - portable SHED for evaporative emissions field measurements
purge - evaporative emissions control system that creates a vacuum in the fuel system to pull
the hydrocarbons from the charcoal canister while the engine is running for combustion
refueling loss - Vapor lost and spillage occurring during refueling
61
-------
running loss - Vapor lost during vehicle operation.
RVP - Reid Vapor Pressure, a measure of volatility in the gasoline at 100 degrees
Farenheit, as determined by the test method ASTM-D-323
SHED - Sealed Housing for Evaporative emissions Determination; structure for
evaporative testing in a laboratory
Stage II - vapor control programs at refueling stations to recover fuel vapor losses from fuel
displacement at the refueling pump
tank vapor generated (TVG) - vapor generated in the fuel system as temperature rises
tank vapor vented (TVV) - vapor generated in fuel system lost to the atmosphere, when
not contained by evaporative emissions control systems
Tier 2 - vehicle emissions certification standards phased in from 2004 through 2007
Tier 3 - vehicle emissions certification standards will phase in from 2017 through 2025
62
-------
References
1 USEPA (2020). Air Toxic Emissions from Onroad Vehicles inMOVESS. EPA-420-R-
20-022. Office of Transportation and Air Quality. US Environmental Protection
Agency. Ann Arbor, MI. November 2020. https://www.epa.gov/moves/moves-
technical-reports.
2 USEPA (2020). Speciation of Total Organic Gas and Particulate Matter Emissions from
Onroad Vehicles in MOVES3. EPA-420-R-20-021. Office of Transportation and Air
Quality. US Environmental Protection Agency. Ann Arbor, MI. November 2020.
https://www.epa.gov/moves/moves-technical-reports.
3 EPA Science Inventory, MOVES2014: Evaporative Emissions Report, Record Type:
DOCUMENT (INTERNAL REPORT)
Product Published Date: 09/30/2014, Record Last Revised: 02/26/2016,OMB Category:
Other, Record ID: 263651
https://cfpub.epa.gov/si/si public record report.cfm?dirEntryId=263651&Lab=OTAO&sea
rchAll=moves&simpleSearch=0&showCriteria=2&searchAll=evaporative&TIMSType=&
4 J. Heiken, T. Carlson, D. McClement and J. Lyons. CRC Project No. E-116: Assessment of
MO VES Model Evaporative Emission Inputs, 2017. http://crcsite.wpengine.com/wp-
content/uploads/2019/05/CRC E-116 MOVES Final-Report 2017-06-14.pdf
5 USEPA (2020). Population and Activity of Onroad Vehicles in MOVES3. EPA-420-R-20-
023. Office of Transportation and Air Quality. US Environmental Protection Agency. Ann
Arbor, MI. November 2020. https://www.epa.gov/moves/moves-technical-reports.
6 Harold M. Haskew and Thomas F. Liberty. CRC E-77 Vehicle Evaporative Emission
Mechanisms: A Pilot Study. http://www.crcao.com/reports/recentstudies2008/E-77%
20Pilot%20Study/E-77%20Pilot%20Study%20Final%20Report%206.24.08.pdf, June 2008.
7 Harold M. Haskew and Thomas F. Liberty. CRC E-77-2 Enhanced Evaporative Emission
Vehicles. http://www.crcao.com/reports/recentstudies2010/E-77-2/E-77-2_Final_Report
March_2010.pdf, March 2010.
8 Harold M. Haskew and Thomas F. Liberty. CRC E-77-2b Evaporative Emissions From In-use
Vehicles: Test Fleet Expansion, http://www.epa.gov/otaq/emission-factors-research/
420rl0025.pdf, October2010.
9 Harold M. Haskew and Thomas F. Liberty. CRC E-77-2c Study to
Determine Evaporative Emission Breakdown, Including Permeation Effects and
63
-------
Diurnal Emissions Using E20 Fuels on Aging Enhanced Evaporative Emissions Certified
Vehicles, http://www.crcao.com/reports/recentstudies201 l/E-77-2c/E-77-2c%20Final%
20Report%20for%20sure%201-28-ll.pdf, December 2010.
10 Timothy H. DeFries, J. Lindner, S. Kishan, and C. Palacios. Investigation of Techniques
for High Evaporative Emissions Vehicle Detection: Denver Summer 2008 Pilot Study
at Lipan Street Station, 2009.
11 Timothy H. DeFries. Estimated Summer Hot-Soak Distributions for Denver Ken Caryl
IM Station Fleet, 2013.
12 CRC E-9 Measurement ofDiurnal Emissions from In-Use Vehicles, http://www.crcao.com/
reports/emission/e9.htm, September 1998.
13 Harold M. Haskew and Associates Inc. CRC E-35 Measurement of Running Loss
Emissions from In-Use Vehicles, http://www.crcao.com/reports/emission/e35.htm, February
1998.
14 CRC E-41-1 Real World Evaporative Testing of Late-Model In-Use Vehicles. http://www.
crcao.com/reports/emission/e41.htm, October 1999.
15 CRC E-41-2 Evaporative Emissions from Late-Model In-Use Vehicles. http://www.crcao.
com/reports/emission/e41.htm, October 1999.
16 Harold M. Haskew, Thomas F. Liberty, and Dennis McClement. CRC E-65 Fuel
Permeation from Automotive Systems. http://www.crcao.com/reports/recentstudies2004/E65%
20Final%20Report%209%202%2004.pdf, September 2004.
17 Harold M. Haskew, Thomas F. Liberty, and Dennis McClement. CRC E-65-3 Fuel
Permeation from Automotive Systems: EO, E6, E10, E20, andE85.
http://www.crcao.com/reports/ recentstudies2006/E-65-3/CRC%20E-65-
3%20Final%20Report.pdf, December 2006.
18 J Linder and G Glinsky. Multi-Day Diurnal Testing, February 2012.
19 M. Sabisch, S. Kishan, Inc. Eastern Research Group, J. Stewart, G. Glinsky, and SGS
Environmental Testing Corporation. Running loss testing with implanted leaks, 2014.
64
-------
20 Transportation AndMarketing Of Petroleum Liquids, http://www.epa.gov/ttnchiel/ap42/
ch05/final/c05s02.pdf, June 2008.
21 E.M. Liston, American Petroleum Institute, and Stanford Research Institute. A Study of
Variables that Effect the Amount of Vapor Emitted During the Refueling of Automobiles.
http: //books, google. com/books?id=KW 2IGwAAC AAJ, 1975.
22 EPA Certification Data, http://www.epa.gov/otaq/certdata.htm.
23 Sierra Research. Development of Trip and Soak Activity Defaults for Passenger Cars and
Trucks in MOVES2006. http://www.sierraresearch.com/ReportListing.htm, March 2006.
24 Sierra Research. Development of Trip and Soak Activity Defaults for Passenger Cars
and Trucks in MOVES. http://www.sierraresearch.com/ReportListing.htm, June 2007.
25 T. Cam, K. Cullen, S. Baldus, and K. Sime. Running Loss Temperature Profiles.
http: //papers. sae. org/930078/, 1993.
26 S.R. Reddy and Society of Automotive Engineers. Prediction of Fuel Vapor
Generation from a Vehicle Fuel Tank as a Function of Fuel Reid Vapor Pressure
and Temperature. http://books.google.com/books?id=J7vEHAAACAAJ, 1989.
27 J. Brown. The DELTA Model: Improved Evaporative Emissions Modeling for
EPA MOVES, http://www.epa.gov/otaq/models/moves/documents/faca-meeting-jan2013/ 04-
delta.pdf, 2011.
28 Wards Automotive Group. Summary Sales Data. http://http://wardsauto.com/keydata/
hi storical/UsaSaO 1 summary.
29 H.R. Johnson and R.S. Williams. Performance of Activated Carbon in Evaporative Loss
Control Systems. http://papers.sae.org/902119/, 1990.
30 Edmund's. Vehicle Fuel Tank Data, http://www.edmunds.com/.
31 U.S. EPA. MOBILE6 Emissions Model. http://www.epa.gov/otaq/m6.htm.
32 P. Caffrey and P. Machiele. In-use Volatility Impact of Commingling Ethanol and Non-
Ethanol Fuels, http://papers.sae.org/940765/, 1994.
65
-------
33 M. Weatherby, M. Sabisch, and S. Kishan. Analysis of Evaporative On-Board
Diagnostic (OBD) Readiness andDTCs UsingI/MData, 2014.
34 Jim Kemper. High Evaporative Emitter Repairs, http://www.sae.org/servlets/works/
upcomingmeetingResources.do?eventGenNum= 10105, December 2010.
35 Randall Guensler and Georgia Institute of Technology. Atlanta Commute Vehicle Soak
and Start Distributions and Engine Starts per Day: Impact on Mobile Source Emission
Rates, http://transportation.ce.gatech.edu/sites/default/files/files/atlanta_commute_
vehucle_soak_and_start_time_distributions_and_engine_starts .pdf, April 2007.
36 Sierra Research. United States Motor Vehicle Inspection and Maintenance (I/M)
Programs. http://www.sierraresearch.com/ReportListing.htm. December2005.
37 P. Cingle and D. McClement. Uncontrolled Automotive Refueling Emissions. http://www.
ntis.gov/search/product.aspx?ABBR=PB88193099, January 1988.
38 G. Schremp and N. Janusch, Fuel Delivery Temperature Study, California Energy
Commission. CEC-600-2009-002-CMF, 2009.
https://ww2.energv.ca.gov/2009publications/CEC-600-2009-002/CEC-600-2009-0Q2-
CMF.PDF
39 "A Study of Variables that Effect the Amount of Vapor Emitted During the Refueling of
Automobiles," API Report CEA-21, May 16, 1975.
66
------- |