Using MOVES to Generate Inventories for the RFS2 NPRM

Memorandum to Docket from Amanda Valente, Megan Beardsley, David
Brzezinksi, 

Ed Glover & Prashanth Gururaja

U.S. EPA/OTAQ/ASD/AQMC

November 2008

 I.  Introduction

The Energy Independence and Security Act (EISA) was enacted by Congress
in December of 2007. This Act requires 36 billion gallons of renewable
fuel be used in the United States fuel supply by the year 2022.  The
Environmental Protection Agency (EPA) proposes to implement EISA through
the Renewable Fuel Standard 2 (RFS2).  To evaluate the likely impact of
the RFS2 rule, EPA estimates that most of this renewable fuel
(approximately 34 billion gallons) will be ethanol.

This memo outlines the methods for estimating the impact of the RFS2
proposal on nationwide emissions of criteria pollutants, (NOx, CO, HC,
PM2.5), air toxics (Benzene, 1-3 Butadiene, Acrolein, Acetaldehyde,
Formaldehyde, Naphthalene) and ethanol from gasoline fueled highway
vehicles in 2005 and 2022.  The renewable fuels proposed by RFS2 will
also impact other mobile source sectors, including heavy-duty diesel
trucks and offroad equipment; these effects are not presented in this
memo, but are detailed in the RFS2 draft Regulatory Impact Analysis
(RIA).   

a.	The MOVES model

For this analysis, we used a preliminary draft of The EPA’s Motor
Vehicle Emissions Simulator (MOVES) to generate national inventories
based on the fuel supply scenarios determined for the RFS2. EPA is
developing the MOVES model to replace EPA’s MOBILE6 and National
Mobile Inventory Model (NMIM).  A draft version is slated for public
release in 2009.

MOVES incorporates EPA's most up-to-date information on gasoline vehicle
emissions.  Exhaust emission rates for HC, CO and NOx were developed
based on an analysis of state inspection/maintenance and roadside remote
sensing data from millions of vehicles.  Emissions of particulate matter
are based on EPA's recent Kansas City gasoline PM study.,  Evaporative
emission rates have been updated based on extensive evaporative testing
conducted by EPA and the Coordinating Research Council (CRC) since the
release of MOBILE6, including investigations quantifying the effects of
ethanol on permeation emissions.   To estimate emissions of mobile
source air toxics, MOVES applies toxic ratios from the MOBILE6.2 model
to updated MOVES HC estimates within the model.   

MOVES analysis is continuing and the emission rates used in MOVES may
change prior to the release of the draft version, but we felt it was
important to use an early version of MOVES for this analysis to reflect
the significant updates in gasoline emission rates and fuel emission
effects since the release of MOBILE6.   

Table 1.  Summary of Emission Rates used in RFS2 NPRM MOVES Runs

Vehicle Types	Pollutants	Emission Rate File Name	Description

LDG	Exhaust HC, CO, NOx	exhLD_run_MY01_21_080918

exhLD_run_MY80_00_080918	Rates based on Arizona inspection & maintenance
(I/M) program and EPA's In-use Verification Program data.  Assumes no
deterioration for start emissions.

HDG	Exhaust HC, CO, NOx	MOVESDB20080828	Rates based on CRC E-55 heavy
duty testing and additional EPA data.  See Appendix A.

LDG	Exhaust PM	NormalDetGasPM	Rates based on Kansas City data.  
Deterioration rates based on the KC data in comparison to new vehicle
emissions from past studies. Future deterioration rates are proportional
to past; assumes deterioration of start emissions is same as running;
I/M effects are not modeled.  

HDG	Exhaust PM	heavydutyPM	Rates based on LDG PM emissions.  See
Appendix A.

All	Permeation HC	MOVESDB20080828	Rates based on CRC E65 permeation
testing 

All	Air Toxics	varies with scenario	Ratios to HC by fuel formulation. 
See fuel section below.



b.	Scenario Overview

Our analysis focused on the projected impact of the RFS2 rule in 2022,
the first full year of implementation.  While the impacts of the program
were quantified as the difference in emissions between the proposed
volumes and several reference cases, we ran the MOVES model for only one
reference case and, as explained in the RIA, scaled the results to
estimate emissions for other reference cases.   Thus, the reference case
described in this memo refers to the roughly 14 billion gallons of
renewable fuels projected by AEO (summarized in Table V.A. 1-1 of the
Preamble of the NPRM).

	In addition to the 2022 reference case, we also ran MOVES for a 2005
"base case," intended to represent the actual fuels sold in 2005, and a
2022 "control case" intended to represent the full impact of the RFS2
rule.

In the RFS1 rule, two sets of fuel effects scenarios were analyzed in
order to understand the uncertainty in ethanol effects on on-road
gasoline engines.  For the RFS2 NPRM analysis, we carried forward these
scenarios. The differences between the fuel effect cases are summarized
below: 

“Primary”:  No exhaust VOC or NOx emission impact on Tier 1 and
later vehicles due to E10, and no impact due to E85.  This was the
“primary” case analyzed in the RFS1 rule, and results in only
pre-Tier 1 vehicles having an effect from E10.  

“Sensitivity”:  VOC and NOx E10 emission impacts based on test data
on newer technology vehicles, as presented in Table 3.1-4, and
statistically significant E85 effects presented in Table 3.1-7.  

 In the draft RIA and the Preamble to the RFS2 NPRM, the "Primary" and
"Sensitivity" fuel effect cases were renamed as the "Less Sensitive" and
"More Sensitive" cases, respectively.  This name change was meant to
reflect the current status of fuel effects testing, in which it is not
currently clear which set of fuel effects best describes how
Tier-1-and-later vehicles respond to change in fuel quality, and thus,
which set of fuel effects should be "primary."  However, since this memo
describes MOVES runspecifications, user input tables, output tables and
output summaries that were named with the RFS1 "Primary" and
"Sensitivity" nomenclature, this memo will also use that older
terminology.  

Our analysis also considered the potential emission effects of increased
use of E85 fuel.  Because of uncertainty in the E85 fuel effects, we
analyzed emissions both with no E85 effect, and with an E85 effect
combined with the "Sensitivity" effect described below.

Combining the different fuel effects cases and the different fuel supply
cases, we developed a list of seven scenarios that we needed to run
through the MOVES model, and an eighth scenario that was developed by
post-processing MOVES results.  These are summarized in Table 2, below,
and described in detail in the next sections.  Each scenario has an
associated MOVES fuel database.

Table 2.   Scenarios for RFS2 NPRM MOVES Runs

Scenario Name	Calendar Year	Fuel Supply	Fuel Effects

2005 Primary	2005	2005 Base	Primary (Less Sensitive)

2005 Sensitivity	2005	2005 Base	Sensitivity (More Sensitive)

2022 Primary Reference	2022	2022 Reference	Primary (Less Sensitive)

2022 Sensitivity Reference	2022	2022 Reference	Sensitivity (More
Sensitive)

2022 Primary Control	2022	2022 Control	Primary (Less Sensitive)

2022 Sensitivity Control	2022	2022 Control	Sensitivity (More Sensitive)

2022 Sensitivity E85	2022	2022 E85	E85

2022 Sensitivity Weighted	2022	2022 E85 + 2022 Control	Sensitivity (More
Sensitive) + E85



II. Fuel Supply Cases and Fuel Adjustment Factors

2005 Base Fuel Supply

The 2005 base case fuel properties were derived from 2005 historical
data.  These data included national summer and winter fuels surveys,
studies that tracked the total amount of ethanol produced for use in
gasoline each year, and Reformulated Gasoline (RFG) measurements. 
Additional data were available on the fuel properties of all gasoline
produced and imported annually by refiners, and on the distribution of
gasoline to and from Petroleum Administration for Defense Districts
(PADDs).  Where survey data was available, it was used to determine a
county's fuel properties for summer and winter.  Where survey data was
not available, fuel properties were set to equal the average fuel
properties in that PADD.  Special adjustments were made to some counties
to account for local gasoline volatility control programs and winter
oxygenated gasoline programs.   See Appendix C for details.

2022 Reference and Control Fuel Supplies

	

For the 2022 reference and control cases, the 2005 base case fuel
properties were adjusted to account for implementation of  other fuel
regulations and to account for increased ethanol use.  There is a
greater percentage of ethanol in the 2022 AEO reference case than in the
2005 base case because MTBE has been replaced with ethanol and because
of AEO projected growth in ethanol production for 2022.  As described in
the draft RIA, ethanol was allocated to the state and county level based
on the economics of distribution and blending, as well as other factors.
 The 2022 control case models the requirements of EISA, in which
approximately 34 bgal of ethanol are present in the fuel supply for
2022.  The control case models this by assigning E10 gasoline to all
U.S. counties.

In general, future fuel properties in both the reference and control
cases were adjusted to account for widespread increases in ethanol under
the assumption that, while historically ethanol has been splash blended
in conventional gasoline (CG), it will be match-blended (i.e., the
changes associated with ethanol addition will be accounted for by
refiners when producing the base gasoline) by the timeframe of our
analyses.  We believe this is reasonable given that there will be a
large (and thus more geographically predictable) volume of ethanol used
in gasoline, and that certain property changes taking place when ethanol
is blended (such as octane increase) could be financially beneficial to
refiners if able to be assumed when producing the base gasoline.  Thus,
we adjusted aromatics, olefins, T50, and T90 fuel parameters by first
backing out the effects of any existing oxygenate (by reverse dilution),
and then re-adjusting the properties for ethanol blends based on
refinery modeling done for the RFS1 rulemaking that projected how
gasoline properties were likely to change given widespread use of
ethanol.  Table 3.1-1 shows the adjustment factors used per volume
percent ethanol blended.  RVP was increased 1.0 psi wherever ethanol was
present.

Table 3.  Adjustment factors for 2022 Conventional Gasoline

	Aromatics	Olefins	E200	E300

Change per vol% ethanol added	-0.464	0.019	0.909	0.063



For RFG areas, refiners already account for the blending of ethanol when
producing the base gasoline, and therefore the properties are not
predicted to change in the same ways as for CG.  We used refinery
modeling results for each PADD (produced along with those above for CG)
to project the properties of fuel in RFG areas accounting for increased
ethanol in the national gasoline pool.  

2022 E85 Fuel Supply

For the control case, we also modeled a scenario with Flexible-Fueled
Vehicles running on E85.  For most pollutants, fuel adjustments were the
same for each E85 fuel as described in section 3.1.1.2.2 of the draft
RIA.  However, we considered local differences in E85 formulations for
two pollutants.  Because MOVES vapor venting emissions and sulfate
emissions are calculated directly from the RVP and sulfur level of the
fuel, it was important to estimate these properties.  The E85 RVP and
sulfur were derived from the properties of the same county's E10 fuel in
the Control Case, with the RVP adjusted to account for the lower vapor
pressure of ethanol and the sulfur level adjusted to 15% of that of the
original E10 fuel in that county.  The predicted sulfur adjustment
underestimates the actual sulfur content of the fuel because it neglects
the sulfur content of ethanol (about 1.5 ppm) and it does not account
for any refinery increases in sulfur in the base gasoline; thus the
benefit attributed to sulfate reductions in the E85 case may be
overestimated.  

Primary Case Fuel Effects

	For the Primary Case ("Less Sensitive") analyses, fuel adjustments for
exhaust HC and NOx were set to zero for Tier-1-and-later vehicles.  
Tier 0 vehicles used HC and NOx adjustments as calculated using the EPA
Predictive Model.  PM emissions had no fuel adjustment.  CO emissions
had the same fuel adjustments as in MOBILE6. Air toxic ratios to base
hydrocarbons were computed using MOBILE6.  The permeation emissions fuel
adjustment varied with fuel ethanol level.

Sensitivity Case Fuel Effects

	For the Senstitivity Case ("More Sensitive") analyses, fuel adjustments
were the same as in the Primary Case, except that all Tier 1-and-later
vehicles also used the HC and NOx fuel adjustments calculated using the
EPA Predictive Model.

E85 Fuel Effects

	

	The effects of E85 fuels on emissions are detailed in the draft RIA,
Section 3.1.1.2.2 and not repeated here.

Fuel Database Development

For each of the modeled scenarios, fuel information was input into an
NMIM database and used for NMIM runs.  For the MOVES runs described in
this memo, the NMIM databases were converted into MOVES databases using
a conversion program.  To reduce time needed for MOVES runs, we reduced
the size of the MOVES fuel database by processing the database with a
"binner" program that grouped fuels with similar properties and assigned
each group to a single fuel formulation identification number and a
single set of fuel properties.   

Each MOVES fuel database consists of seven tables, described in Table 5,
below.  Additional detail on how the databases were constructed is
available in Appendix D of this memo.  The fuels databases are listed in
Appendix B.

Table 5.  Tables in Each Fuel Database

MOVES Table 	Content

FuelSupply	Market share for each fuel formulation by county and month

FuelFormulation	Fuel properties for each fuel formulation

FuelAdjustment	Multiplicative fuel adjustments for each fuel formulation

ATRatioGas1	Ratios for computing emissions for most air toxics; these
vary by fuelformulationid and other vehicle characteristics.

ATRatioGas2	Ratios for computing emissions for ethanol, napthalene and
acrolein; these vary by fuelformulationid only.

ATRatioNoGas	Ratios for computing emissions for air toxics for fuels
other than gasoline.  These were not used in this analysis.

Year	Mapping of calendar year to "fuel year"  (allows the same fuels to
be used for mutiple calendar years)



Note that each Fuel Database is independent, with its own set of
fuelformulationids.  Mixing a fueladjustment table from one database
with a fuelsupply or fuelformulation table from another could result in
a mismatch of fuelformulationids, and thus a misapplication of
fueladjustments.

III.  Inventory Calculation Methods

To develop highway vehicle emissions, we ran MOVES for each of the fuel
databases described in the previous section.

Running the MOVES model for the entire nation with the level of detail
desired for this analysis requires several hours and creates very large
result databases.  To better manage runtime and output, we did separate
MOVES runs for (a) non-PM exhaust,  (b) PM and Napthalene, and (c)
evaporative emissions.   For PM and Evaporative emissions, the fuel
databases were identical for the primary and sensitivity cases; thus
only one of the two cases was run for these emissions. To reduce run
time, all runs were for January and July only, and all runs were
pre-aggregated to the state level.   Runs for non-PM exhaust and PM were
pre-aggregated to the month level. MOVES requires that evaporative
emissions be run at the hourly level. 

	We ran the MOVES model for seven sourcetypes:  passenger cars, light
passenger trucks, light commercial trucks, school buses, short-haul
single-unit trucks, long-haul single unit trucks, and motor homes.  For
the purposes of this rulemaking, the emissions of gasoline-powered
motorcycles, refuse trucks, transit buses, intercity buses and
combination trucks was determined to be negligible, and they were
omitted from the MOVES runs to reduce run time.  For better integration
with the NMIM results for diesel vehicles, output was produced by Source
Classification Code (SCC).  Details on the MOVES runspecs are listed in
Appendix B.

	

MOVES results were post-processed to generate annual inventories and to
account for some anomalies in the pre-release MOVES draft.

MOVES evaporative results for January and July were summed and then
multiplied by a factor of 5.7781 to adjust from monthly inventories to
annual.  The 5.7781 factor is the ratio of days in the modeled months to
the days in the year ((31+31)/365).

MOVES non-PM exhaust results for January and July were summed and then
multiplied by a factor of 6.04456 to adjust from monthly inventories to
annual.  The 6.044656 factor is the ratio of VMT in the modeled months
to annual VMT using data in the MOVESdb20080828 MonthlyVMTFraction
table.

MOVES PM and Napthalene exhaust results for January & July were
multiplied by 4.4 and 7.6 respectively to account for the monthly
distribution of PM emissions as determined from a national run by month.
 PM emissions are much more sensitive to temperatures than other
emissions and thus weighting January emissions by days or VMT would
overestimate total annual emissions.

To correct an error in some MOVES input tables, PM10 emissions were
recalculated as 1.086 times the sum of PM2.5 emissions.

Light Duty Napthalene emissions were calculated as (PM2.5 EC+PM2.5 OC)
*0.088.  

Heavy Duty Napthalene was calculated by MOVES.

There was a bug in the start calculator when output was requested by SCC
that caused errors in the start air toxics results.  To better estimate
start toxics (other than napthalene), we applied the running HC ratio,
as follows:

a) Computed MOVES HC & air toxics output results for start & running
emission processes.

b) For each air toxic, computed the average airtoxic/HC ratio for
running emissions

c) For each air toxic, multiplied total HC start & running emissions by
the air toxic running ratio.

The 2022 Sensitivity Reference scenario was run with incorrect air toxic
ratio tables.  To better estimate air toxics for this scenario, we took
the HC ratios from the 2022 Primary Reference case and applied them to
the HC emissions from the 2022 Sensitivity Reference case.   Because the
air toxic ratios are the same in the Primary and Sensitivity Case this
provided the results we intended.

For E85, we ran the model using a fuels database that had 100% of
gasoline vehicles using 100% E85 and fueladjustments from the E85
analysis.  However the model applied these ratios to HC rather than VOC
emissions.  Because the adjustments were the same for all sourcetypes,
we do not believe the results were not affected by the start calculator
bug described above.  However, because the difference between HC & VOC
is high for E85 fuels, we multiplied the toxics results by  1/(THC/VOC)
= 1/1.6161 (ie 0.62) to account for this. 

E85 emissions were weighted with 2022 Sensitivity Control case emissions
using MYSQL scripts that developed weighting factors based on the
expected fraction of FFVs in that model year and the expected E85 use in
each state. (See draft RIA, Section 1.7.1.2.) The 2022 Sensitivity
Control and E85 results were weighted together by applying the weighting
factors to each model year, SCC vehicle class and state, and summing the
results.  

The resulting inventory values were used to calculate the difference in
emissions between control and reference scenarios for both the "primary"
and "sensitivity" cases.

Appendix A:  Emission Rates for Heavy Duty Gasoline Vehicles in RFS2
NPRM

I.  Heavy-duty Gas PM Ratios for Start and Running Emissions

The heavy-duty gas PM2.5 emission rates were calculated by multiplying
the light-duty gasoline truck PM2.5 emission rates with ‘normal’
deterioration rates by a factor of 1.4.   

The heavy-duty gas factor of 1.40 is based on analysis of four gas
trucks tested in the CRC E55-E59 test program.  Two UDDS tests done at
different test weights were performed on each truck.  Other tests were
also available but were not used so that a consistent cycle could be
evaluated.  The trucks and tests are described as:

Vehicle 	MS_ID	MY	Age	Test_cycle	GVWR	PM2.5 g/mi

1	1GDJ7H1E81J901195	2001	3	UDDS	12975	1.8115942

1	1GDJ7H1E81J901195	2001	3	UDDS	19463	3.6101083

2	1FDNF60H9EVAVA045	1983	21	UDDS	9850	43.3212996

2	1FDNF60H9EVAVA045	1983	21	UDDS	14775	54.3478261

3	1FDNF70J1NVA37957	1993	12	UDDS	13000	67.0949972

3	1FDNF70J1NVA37957	1993	12	UDDS	19500	108.336719

4	1GBG7D1B2HV108608	1987	18	UDDS	10600	96.7349998

4	1GBG7D1B2HV108608	1987	18	UDDS	15900	21.5098555



Examination of the heavy-duty data shows two distinct strata of data –
vehicle #1 (MY=2001) and the other three vehicles.  Because of its lower
age (tests done in CY2004) and newer model year status, vehicle # 1 had
dramatically lower PM emission levles than the others, and was separated
in the analysis.  The emissions of the other three vehicles were
averaged together to produce the mean result of 

Mean for Vehicles 2 through 4:	0.06522 g/mi

Mean for Vehicle 1:			0.00271 g/mi

For comparison with the heavy-duty gas PM rates tested over the UDDS
cycle, simulated UDDS cycle emission rates were developed based on MOVES
light-duty gas PM2.5 emission rates for light-duty gasoline trucks.   
The UDDS cycle represents typical operation for the heavy-duty vehicles.
 Only selected light-duty gas PM emission rate bins were used.  All of
these emission rate bins match the heavy-duty vehicles in terms of model
year and age.  MOVES emission rates for the following sourcebinIDs were
used:

SourcebinID = 1010130620000000000 and AgeGroupID = 2099  (my 1983-84 age
20+)

SourcebinID = 1010130630000000000 and AgeGroupID = 1519  (my 1986-87 age
15-19) 

SourcebinID = 1010130650000000000 and AgeGroupID = 1014  (my 1991-93 age
10-14)

SourcebinID = 1010130210000000000 and AgeGroupID = 3	(my 2001 age 0 –
3)

The ‘30’ in the sourcebinID indicated light-duty truck.  The
‘62’, ‘63’, ‘65’ and ‘21’ indicate model year group. 

The simulated UDDS emission factors for the older model year light duty
gas truck group are 0.0362 g/mi for MOVES organic carbon PM2.5 emissions
and 0.002641 g/mi for elemental carbon.  Ignoring sulfate emissions
(which are in the order of 10e-07 for low sulfur fuels), the PM
emissions sum to 0.03884 g/mi and leads to the ratio 0.06522 g/mi /
0.03884 g/mi  = 1.679

The simulated UDDS emission factors for the newer model year light duty
gas truck group are 0.004368 g/mi for MOVES organic carbon PM2.5
emissions and 0.0003187 g/mi for elemental carbon.  Ignoring sulfate
emissions (which are in the order of 10e-08 for low sulfur fuels), the
PM emissions sum to 0.004687 g/mi and leads to the ratio 	0. 00271 g/mi
/ 0.004687 g/mi= 0.578

The newer model year group produces a ratio which is less than one and
implies that small trucks produce more PM2.5 emissions than larger
trucks.  This is intuitively inconsistent, and is the likely result of a
very small sample and a large natural variability in emission test
results.

All four data points were retained and averaged together by giving the
older model year group a 75 percent weighting and the newer model year
group (MY 2001) a 25 percent weighting.  This is consistent with the
underlying data sample.  It produces a final ratio of:

	1.679 * 0.75 + 0.578* 0.25	=	1.4

II.  Heavy-duty Gas  HC, CO & NOx Running Emissions

HC, CO and NOx running emissions for heavy-duty gasoline vehicles were
calculated based on data available from CRC E-55/59 and from the EPA's
Mobile Source Observation Database (MSOD).  Emissions were calculated
with the following methodology:

Calculate VSP

Calculate power then divide by survey-average weight for each regClass

Use certification data to group by model year

1960-1989

1990-1997

1998-2006

2007+

Apply deterioration/age effects as present in the data

Fill missing rates

Where data was missing, kept emission rate ratios between age groups and
model year groups consistent

2007+ : Assumed that tighter HD NOx standards brought all pollutants
down by 70% (3-way catalyst)

III.  Start Emissions of Heavy–Duty Gasoline–Fueled Vehicles

	The standard temperature range for the Federal Test Procedure (FTP) is
between 68 and 86 degrees Fahrenheit (with a nominal temperature of 75°
F).  A total of 78 FTPs were performed (with temperatures in that range)
on HDVs in 1997 through 1999.  Since this testing program was performed
during a short 26 month period, the age of the vehicles and their model
years are highly correlated (correlation coefficient of  -0.9848). 

	The differences in emissions of Bag-1 minus Bag-3 (HC, CO, and NOx)
were all correlated with:

	 -	fuel metering system (higher with TBI than with PFI),

	 -	engine displacement (increasing with increasing CID),

	 -	test weight (increasing with increasing test weight), and

	 -	model year (increasing with older model years).

However, none of the four parameters/variables was statistically
significant.

	Therefore, for this version of MOVES, we used the mean emissions
(differences in emissions of Bag-1 minus Bag-3) for those heavy-duty
trucks to applied it to all HDGVs, specifically:

	

Pollutant	Bag-1 minus Bag-3 (grams)	Standard

Error	

COV

	THC	    6.748246	   0.678834	0.100594

	CO	  135.4165	  10.88515	0.080383

	NOx	    3.160521	   0.423576	0.134021



	All of those values in this table represent the difference of Bag-1
minus Bag-3 and are adjusted to estimate the cold-start emissions.

Appendix B: MOVES Runspecs by Case

The MOVES runs for exhaust emissions were made with the 20080909 version
of the MOVES model, updated with specific scripts that added hydrocarbon
speciation capabilities.  This version of the model was run with the
MOVESDB20080828 default database.  An automated script merged the
required fuel database with the MOVES default database.

Because the state-by-state, hour-by-hour evaporative runs were extremely
time consuming and because the expected change in evaporative emissions
was quite small, the evaporative runs were not repeated when the fuel
databases were updated during the course of the NPRM analysis.  This
means that the MOVES runs for evaporative emissions were done with an
earlier 20080715 version of the model and the MOVESDB20080623 default
database.  The fuel databases for these runs also differed from the
databases used for the exhaust runs:  they lacked some corrections to
county-by-county benzene levels.  Because these runs were done with an
earlier default database, they also required additional user input files
that were later incoporated into the default database.  

At the time of this analysis, MOVES exhaust emission rates were
available only for conventional internal combustion engines; thus it was
necessary to run MOVES with an Alternate Vehicle and Fuel Technology
(AVFT) specification that set all vehicles to "conventional".  At the
time of this analysis, proper functioning of the AVFT required that both
gasoline and diesel vehicles be present in the run spec.  Therefore each
of the exhaust run specs includes an "all_conv" AVFT input file and
includes diesel vehicles in the run spec.

Table B-1 lists the databases used for exhaust-non PM, exhaust PM and
evaporative emissions runs.  Table B-2 provides additional detail on the
fuel databases used for each scenario.

Table B-1.   Input Databases for RFS2 NPRM MOVES Runs

Run	User Database	Tables	Description	Default MOVES DB

Exh	exhLD_run_MY01_21_080918 and exhLD_run_MY80_00_080918	emissionrate,
emissionratebyage	Newest HC, CO & NOx emission rates	MOVESDB20080828

	cleardefaultfuels	fuelsupply	Sets marketshare for default fuels=0 for
calendar years 2011 and earlier.



zonemonthhour2005	zonemonthhour	2005 Temperatures for NPRM runs



One of 7 fuel databases 	atratiogas1, atratiogas2, atrationongas,
fueladjustment, fuelformulation, fuelsupply, year	Sets fuels,
adjustments, toxics.  Each scenario has its own database.  See Table
B-2.

	PM	NormalDetGasPM	datasource

emissionratebyage

opmodepolprocassoc

pm10emissionratio	Newest light duty PM emission rates



zonemonthhour2005	zonemonthhour	2005 Temperatures for NPRM runs;



One of 4 fuel databases 	atratiogas1, atratiogas2, atrationongas,
fueladjustment, fuelformulation, fuelsupply, year	Sets fuels,
adjustments, toxics.  Primary and Sensitivity cases use the same inputs.
 See Table B-2.



heavydutyPM	emissionratebyage	Derived from light duty PM rates

	Evap	cleardefaultfuels_20080724	fuelsupply	Sets marketshare for default
fuels=0	MOVESDB20080623

	zonemonthhour2005 	zonemonthhour	2005 Temperatures for NPRM runs



Enhanced_Evap_Phase_In	emissionratebyage	Updated tank vapor venting
emission rates for model years 1996, 1997 and 1998



Perm_fuelAdj_RFS2_20080724	fueladjustment	updates effect of ethanol fuel
content on permeation emissions



SourcetypeModelYearGroup_pg	sourcetypemodelyeargroup	Model year groups
to work with Enhanced_Evap_Phase_In  rates



Evap_tvv_update_pg

	cumtvvcoeffs

emissionratebyage	Updated tank vapor venting emission rates for all
model years



One of 4 fuel databases  (Note, these databases were created earlier and
differ slightly from those used for exhaust runs)	atratiogas1,
atratiogas2, atrationongas, fueladjustment, fuelformulation, fuelsupply,
year	Sets fuels, adjustments, toxics.  Primary and Sensitivity cases use
the same inputs.  See Table B-2.

	

Table B-2.  Fuel Databases for Each Scenario

Scenario Name	Non-PM Exhaust Fuel Database	PM Fuel Database	Evap Fuel
Database

2005 Primary	rfscomplexpredictive_baseNOPRED
rfscomplexpredictive_baseNOPRED	newBaseSENS

2005 Sensitivity	rfscomplexpredictive_base2005



2022 Primary Reference	rfscomplexpredictive_reference2022 NOPRED
rfscomplexpredictive_reference2022 NOPRED	newRefSENS

2022 Sensitivity Reference	rfscomplexpredictive_reference2022



2022 Primary Control	rfscomplexpredictive_control2022
rfscomplexpredictive_control2022NOPRED	newControlSENS

2022 Sensitivity Control	rfscomplexpredictive_control2022NOPRED



2022 Sensitivity E85	newE85ControlSENS	newE85ControlSENS
newE85ControlSENS

2022 Sensitivity Weighted	na (post-processed)



Appendix C:  County-specific Fuel Properties for 2005 National Emission
Inventory

Fuel properties used to estimate emissions for the 2005 National
Emission Inventory (NEI) formed the foundation for the fuel properties
used in the RFS2 NPRM.  This appendix summarizes the methods and data
used to develop the 2005 NEI fuel properties.

The fuel properties of gasoline are currently measured regularly in
national surveys in both summer and winter.  There are also studies that
track the total amount of ethanol produced for use in gasoline each
year.  The properties of Reformulated Gasoline sold in areas which are
federally required to have them are measured each year.  In addition,
the fuel properties of all gasoline produced and imported by refineries
are reported annually to EPA.  This information can be used to estimate
the expected average gasoline properties in a given historical calendar
year.  Using this information combined with information about the
distribution of gasoline to and from Petroleum Administration for
Defense Districts (PADDs) in combination with the city specific surveys,
the average gasoline properties in each county can be estimated.

Data Sources

1.	The U.S. EPA Office of Transportation & Air Quality 2005 Reformulated
Gasoline (RFG) Properties Survey Data.  These surveys are conducted by
the RFG Survey Association, an association of refiners, importers and
blenders, as a requirement of EPA regulations.  All areas participating
in the federal RFG program (except areas in California) are measured in
both the summer and winter each year.  The results of these surveys are
posted on the web.

	http://www.epa.gov/otaq/regs/fuels/rfg/properf/rfgperf.htm

2.	2005 North American Fuel Survey fuel survey data conducted by the
Alliance of Automobile Manufacturers (AAM), excluding data from Canada
and Mexico.  The AAM data reflect the properties of gasoline from summer
and winter in many geographic locations.  This survey information is
sold to the public.

	http://autoalliance.org/fuel/fuel_surveys.php

3.	Reformulated Gasoline and Anti-Dumping Regulation batch reports.
Producers and importers of reformulated gasoline, Reformulated Gasoline
Blendstock for Oxygenate Blending  (RBOB), conventional gasoline, or
applicable blendstocks must fill out and submit these forms to EPA. 
This data is confidential business information and cannot be accessed by
the public.

	http://www.epa.gov/oms/regs/fuels/rfgforms.htm

4.	US Department of Energy, Energy Information Administration Petroleum
Supply Annual 2005.  This report (Table 34) describes the movements of
gasoline between PAD Districts.

	http://www.eia.doe.gov/oil_gas/petroleum/data_publications/petroleum_su
pply_annual/psa_volume1/psa_volume1.html

5.	2005 National Emission Inventory.  VMT estimates by vehicle type and
county developed for this analysis are used to allocate gasoline
consumption to individual counties.

	http://www.epa.gov/ttn/chief/net/2005inventory.html

6.	Core Based Statistical Areas (CBSAs) defined by the U.S. Census
Bureau are used to define which counties are included for city specific
surveys.  The counties are listed in the Counties with metropolitan and
micropolitan statistical area codes (List3.xls, December 2005).

	http://www.census.gov/population/www/estimates/metrodef.html

Overview of the Methodology

The goal of the method is to produce a full set of average gasoline
properties for each county in the nation (including Puerto Rico and the
Virgin Islands) for each month of the year.  In RFG areas and
conventional gasoline areas, these properties will be the same as the
properties observed in the surveys done in those areas.  In non-survey
areas, the properties will match the average properties of gasoline
produced (and imported) in each PADD.  Special adjustments are made to
some counties to account for local gasoline volatility control programs
and winter oxygenated gasoline programs.  The basic steps are: 

1.	Where surveys exist, apply their average gasoline properties to all
counties associated with that survey area.  Since all RFG areas have
surveys, only conventional gasoline properties are unknown.  

2.	Determine the volumes and average gasoline properties of conventional
gasoline surveyed in each PADD.  Volumes are determined by distributing
total PADD consumption volumes by gasoline vehicle VMT to each county.

3.	Determine the volumes and average gasoline properties of conventional
gasoline produced in each PADD.

4.	Using the PADD movements, adjust the average properties in each PADD
to reflect gasoline imported and exported from other PADDs.    

5.	Subtract the total volumes of the properties of conventional gasoline
from surveys from the total volumes of properties in each PADD to
determine the remaining volumes of properties from counties without
surveys.

6.	Assume that all counties within a PADD that do not have survey data
will have the average PADD gasoline properties in both summer and winter
(except for gasoline Reid Vapor Pressure, E200 and E300).  

7.	Using the refinery batch dates, determine the winter and summer RVP,
E200 and E300 values for the PADD averages.  

8.	Adjust summer RVP values in specific counties to reflect local RVP
control programs. 

9.	Determine the total volume of ethanol consumed in each state. 

10.	Add ethanol (using market share with a 10% by volume blend) to every
non-survey winter and summer gasoline in each state to match with the
state totals.  Adjust other gasoline properties (RVP, E200, E300,
aromatics, olefins) to reflect the addition of ethanol.

11.	Add ethanol (using 100% market share) to non-survey winter gasoline
in each county with a winter ethanol program.  Adjust other gasoline
properties (RVP, E200, E300, aromatics, olefins) to reflect the addition
of ethanol.

12.	Use winter gasoline in January and summer gasoline in July.  Use
ASTM limits to weight winter and summer gasoline properties together in
each county to estimate gasoline properties in the other months.

The following sections will address the details of each of the steps
outlined above.

1.0	Average Gasoline in Survey Areas

All areas participating in the federal Reformulated Gasoline (RFG)
program (except areas in California) are measured in both the summer and
winter each year.  The average gasoline properties reported for these
areas were used for the counties listed for each area.  The gasoline
properties reported for the California RFG areas are incomplete and were
not used.  The RFG survey areas in 2005 were:

Atlantic City, NJ

Baltimore, MD

Boston-Worcester, MA

Chicago-Lake Co., IL, Gary, IN

Covington, KY

CT - remainder

Dallas-Fort Worth, TX

Hartford, CT

Houston-Galveston, TX

Louisville, KY

Manchester, NH

Milwaukee-Racine, WI

Norfolk-Virginia Beach, VA

NY-NJ-Long Is.-CT

Phila.-Wilm, DE-Trenton, NJ

Portsmouth-Dover, NH

Poughkeepsie, NY

Queen Anne Co.-Kent Co., MD

Rhode Island

Richmond, VA

Springfield, MA

St. Louis, MO

Sussex County, DE

Warren County, NJ

Washington, D.C.-area

The reported gasoline properties for oxygenates (ETOH, ETBE, MTBE and
TAME) in the RFG surveys are by weight percent of oxygenate, not volume
percent as needed for input into EPA OTAQ models (MOBILE6, NMIM).  Also
MOBILE6 and NMIM do not allow mixing of oxygenates, so that each
oxygenate must have their own market share and volume percent oxygenate.

The appropriate market shares were determined by assuming that the
oxygen content by weight is the same in each market share and that the
sum of the market shares is always one.  The oxygen content and market
shares when combined must equal the average volume percent observed in
the survey.

Oxygenate	Target Vol%	Maximum Vol%	Wt Fraction of Oxygen	Wt% Oxygen/Vol%

ETOH	10.0/6.0	10.7	0.3473	0.3488

MTBE	11.8	20.7	0.1815	0.1786

ETBE	13.7	24.1	0.1566	0.1533

TAME	22.6	22.6	0.1566	0.1636

 

For example, if survey results report average properties, ETOH (9%) and
MTBE (1%) oxygenate by weight, the average properties assume that 100%
of the gasoline has both ETOH and MTBE.  The two oxygenates must be
split into separate market shares.  We first convert to weight percent
oxygen and determine the distribution of oxygen by weight.

ETOH wt% O2 = 9.0 * 0.3473 = 3.1257%

MTBE wt% O2 = 1.0 * 0.1815 = 0.1815%

ETOH market share = 3.1257 / (3.1257+0.1815) = 0.945

MTBE market share = 0.1815 / (3.1257+0.1815) = 0.055

This market share assumes that 100% of the gasoline market contains
either ETOH or MTBE.  Once the market share has been determined, you
adjust the volume percent of oxygenates to reflect the new reduced
market share.

ETOH vol% = ( 3.1257 / 0.3488 ) / 0.945 = 9.48%

MTBE vol% = ( 0.1815 / 0.1786 ) / 0.055 = 18.52%

 

 The values must be checked against the maximum allowed values (10.7%
and 20.7% for ETOH and MTBE).  If any calculated oxygenate volume
percents exceed the maximum allowed values, the market shares may need
to be adjusted manually (increasing some volumes percents to reduce
others) until all oxygenate values are under their limits.

Regardless of the mix of oxygenates, the average RVP, E200 and E300
values are not adjusted and the MOBILE6 flag should reflect "no RVP
waiver" case so that MOBILE6 does not adjust these value either.  The
survey average RVP, E200 and E300 values are measured and are assumed to
already reflect any impact of oxygenates on gasoline volatility. 

The Alliance of Automobile Manufacturers (AAM) conducts the North
American Fuel Survey in a variety of cities, with and without
Reformulated Gasoline programs.  Only the data from U.S. cities without
RFG were used.  EPA RFG survey data was used for RFG cities.  The 2005
AAM survey cities used in the analysis are:

Atlanta, GA

Miami, FL

Pittsburgh, PA

Cleveland, OH

Detroit, MI

Kansas City, MO

Minneapolis/St. Paul, MN

Albuquerque, NM

New Orleans, LA

San Antonio, TX

Billings, MT

Denver, CO

Fairbanks, AK

Las Vegas, NV

Los Angeles, CA

Phoenix, AZ

San Francisco, CA

Seattle, WA

Los Angeles, California, is a federal RFG area.  However, since the EPA
survey was incomplete, the AAM survey results were substituted for the
EPA survey results for that city.  California has statewide,
state-specified Reformulated Gasoline regulations.  These regulations
require that counties which are part of the mandatory federal RFG
program must meet the federal RFG gasoline specifications.  All federal
RFG counties in California were assigned the Los Angeles survey results
and all remaining California counties were assigned the San Francisco
survey results.

The counties that use the AAM survey results were determined by the Core
Based Statistical Areas (CBSAs) defined by the U.S. Census Bureau for
each city.  For Alaska, the entire state was assigned the results from
the Fairbanks survey.  Minnesota has statewide fuel control regulations.
 The Minneapolis/St. Paul survey results were applied to all counties in
that state.

1. Determine appropriate county groupings by PADD and sub-PADD. 
Counties with RFG, Oxy or RVP controls would be sub-PADD groupings.

	a) Determine Core Based Statistical Area (CBSA) counties for each
survey city.  Survey results will apply to all CBSA counties in a survey
city.

	b) Determine counties that are affected by RFG or Oxy programs.

2. Use gasoline survey data from survey sources to determine average
uncontrolled conventional gasoline properties in each PADD.  Weight
survey Regular/Premium results by 2005 PADD average volume values.

Department of Energy (DOE) Energy Information Administration (EIA)
Petroleum Marketing Annual 2005 Table 43 Refiner Motor Gasoline Volumes
by Grades, Sales Type, PAD District, and State (Thousands of Gallons per
Day).

Only Regular and Premium gasoline was surveyed, so only Regular and
Premium gasoline volumes were used to determine the weighting factors.

All Alliance survey ETOH, MTBE, ETBE, and TAME oxygen weight percents
(except California) are assumed to be 3.5%, 2.1%, 2.1% and 2.1%,
respectively.  These are used, then to determine market share. 
California assumes 100% market share.

	b) Use gasoline VMT as a surrogate for gasoline sales for weighting
multiple survey areas within a PADD.

	c) Make adjustments to RVP to account for lack of surveys in
uncontrolled counties.

3. Use survey results (not PADD averages) in conventional gasoline
survey CBSA counties.

4. Use gasoline survey data to determine average gasoline properties for
each Reformulated Gasoline (RFG) area.

	a) Some areas may have multiple survey results.  Use VMT as surrogate
for gasoline sales.

	b) Apply survey results to all RFG counties associated with each survey
city.

Survey states with more ETOH gallons than predicted (negative additional
gallons) are assumed to have no more ETOH gallons added.  Survey states
with over 90% of the predicted gallons (additional gallons less than
10%) are also assumed to have no more ETOH gallons.  The remaining
additional ETOH gallons needed does not match the total national
difference in survey ETOH gallons versus predicted ETOH gallons.  So the
remaining additional ETOH gallons are used to determine a proportional
distribution that is used to distribute the remaining national predicted
gallons of ETOH (total ETOH minus the survey gallons).  The distributed
ETOH gallons are then assumed to be used only in 10% ETOH gasoline
blends, so that using the remaining state wide gasoline gallon totals
(minus survey gallons) and 10% ETOH, the state wide average market share
can be calculated for each state needed additional ETOH gallons.

None of the RFG surveys has RVP for the winter survey.  The RVP for
January in the counties with RFG surveys from the NCD20070123 NMIM
database will be used for all RFG survey winter gasolines.

Pittsburgh and Seattle did not have Winter gasoline surveys.  All of the
gasoline properties will be the same as Summer gasoline, except RVP,
which will be taken from the NCD20070123 NMIM database.

Use refinery "batch" reports, rather than "facility averages".  Include
imported gasoline as well as domestic production reports.

Average refinery RVP by PADD by month will be used to populate the RVP
for the PADD average gasolines, using just January and July.

Refinery data for RBOB gasoline reflect the properties for the final
blend, but the volumes only show the gasoline portion.  All gasoline
gallons for RBOB were adjusted upward to reflect the actual blend
gallons (gasoline plus oxygenate).

Delete records (gallons) that are OBO-EEP or CNV-EEP that contain only
ethanol (oxygen>34).

Delete all RBOB and RFG batches from the analysis and all RFG survey
results.  California is not included in the analysis.

Determine the PADD average fuel properties accounting for all imports
and exports before subtracting the survey totals.

Both winter and summer gasoline will use the PADD average values.

PADD average market shares for MTBE/ETBE/TAME are based on volume
percent.  Market shares for survey data is based on weight percent. 
Both are arbitrary and differ little.

5. Determine national average gasoline properties based on refinery
data.

	a) Make adjustments to national totals to account for added oxygenates.

	b) Determine national average gasoline properties based on survey based
gasoline allocations using VMT as a surrogate for gasoline sales.

	c) Make adjustments, as necessary to the conventional gasoline
allocations until there is reasonable agreement between refinery and
survey allocation results.

6. Determine gasoline properties for other months using existing ASTM
process.

Assumptions:

User RFG survey results for counties included in Federal RFG.  Do not
use other survey results from RFG areas.

Include Cecil County (MD) and New Castle County (DE) in the
Philadelphia/Trenton RFG survey area.

Use Fairbanks (AK) survey for all Alaska counties.

Use the Minneapolis/St. Paul (MN) survey for all Minnesota counties.

Include Lake and Porter Counties (IN) in the Chicago RFG survey area.

Include Kent County (DE) in the Sussex County (DE) RFG survey area.

Include all Atlanta RVP control counties in the Atlanta survey area.

Include all Detroit RVP control counties in the Detroit survey area.

Include all New Orleans RVP control counties in the New Orleans survey
area.

Use San Francisco (CA) survey for all California counties not covered by
Federal RFG.

Use Miami (FL) survey for all Florida RVP control counties.

Counties with partial control (RVP, ethanol or RFG) use controlled
gasoline.

Partial month (September) coverage for control (RVP) use controlled
gasoline.

Appendix D  

Development of the MOVES Fuel Formulations, Fuel Adjustment Factors

 and County-Year Fuel Supply Distributions

Introduction

This document describes the MOVES fuel parameters (version 20080623) and
the analysis / programming process used to develop the parameters
“MOVES Fuel Algorithm”.  The first section describes the structure
and design of the MOVES model, and its accompanying data structures. 
The second section gives a brief overview of the individual emission
fuel models, their use in the MOVES context, and the process used to
generate fuel formulations, fuel adjustment factors and fuel supply
distributions.  This document contains only limited and generalized
results from the model.   It does not contain the specific results from
regulatory analysis runs or sensitivities.

Structure of the MOVES Fuel Algorithm Database

The MOVES model is a data driven model that consists primarily of a
central database that is manipulated using a Java based GUI and a series
of complex MySQL scripts.  The MOVES program also allows the use of
alternative user defined data tables in the program.

The MOVES Fuel Algorithm consists of four primary data tables.   These
are:

FuelFormulation

FuelSupply

FuelAdjustment

HCSpeciation

a.	Fuel Formulation Table

The FuelFormulation table contains a table of fuel formulations
‘keyed’ using an arbitrary fuelformulationID variable, and a set of
fuel properties that includes fuel subtype, RVP, sulfur level, ethanol
volume percent, etc.  The fuel subtype designates fuel classifications
such as conventional gas, reformulated gas, gasohol (i.e, 10% ethanol),
and E-85/E70 ethanol-gasoline blends.  The model contains several
hundred individual fuelformulationIDs which cover the range of all
important fuel properties.

MOVES Fuel Formulation Table Structure

fuelFormulationID	key field

fuelSubtypeID

RVP

sulfurLevel

ETOHVolume

MTBEVolume

ETBEVolume

TAMEVolume

aromaticContent

olefinContent

benzeneContent

e200

e300

volToWtPercentOxy

b.	Fuel Supply Table

The Fuel Supply table contains the market share data for each fuel
formulation as a function of calendar year, month and county.    This
information varies by considerably by calendar year and county with new
fuel formulations being phased in over time.   The MOVES model contains
data from the 1999 through 2022 calendar years, and allows the user to
input specific data for a county / year / month combination.  Market
share data for gasoline, diesel and ethanol based fuels are available. 
The data typically varies by season or summer / winter, although MOVES
is designed for monthly variation.   Fuel market share data is provided
for all 3,222 counties and is based on in-use fuel surveys and
econometric modeling projections for future calendar years.  (See
Appendix C.) The market share data is used to weight the effects of
different fuelformulationID together to obtain an overall average fuel
effect / fuel adjustment.  The market share information is a function of
county, calendar year (i.e., fuelYearID) and month.

MOVES Fuel Supply Table Structure

CountyID		key field

fuelYearID		key field

monthGroupID	key field

fuelFormulationID	key field

marketShare

marketShareCV

c.	Fuel Adjustment Table

The Fuel Adjustment table contains all of the MOVES fuel adjustments. 
It is a function of pollutant (Total HC, CO and NOx) – process (start,
running, etc), fuel formulation, model year group and source type.  
Fuel Adjustment factors were developed for each of the following model
year groups.

1974

19751986

19871989

19901993

1994

1995

1996

19972000

20012003

2004

2005

2006

2007

20082009

20102050

Fuel Adjustment factors were developed for the following sourcetypeIDs:

11	Motorcycle

21	Passenger Car

31	Passenger Truck

32	Light Commercial Truck

41	Intercity Bus

42	Transit Bus

43	School Bus

51	Refuse Truck

52	Single Unit Short-haul Truck

53	Single Unit Long-haul Truck

54	Motor Home

61	Combination Short-haul Truck

62	Combination Long-haul Truck

The MOVES model fuel adjustment factor is also a function of specific
fuel properties.  Most of these effects are non-linear and were derived
from the EPA’s Complex model, Predictive model and MOBILE6 Sulfur
model (labeled as Primary EPA Fuel Models in the remainder of this
document).  These effects are applied in MOVES by linking the fuel
formulation table and the fuel adjustment table using the variable
fuelFormulationID.

All of the MOVES fuel adjustment factors are multiplicative correction
factors to the basic emission factor in MOVES.  This is shown
mathematically in the Equation 1 below.

Fuel Corrected Emissions =  Fuel Adjustment Factor * Base Emissions
Factor	 Eq 1

The Base Emission Factor variable in Eq 1 represents the base emission
rates computed by MOVES.  These were computed from Arizona IM240 lane
data, for pre-2001 model years, and computed from EPA in-use vehicle
surveillance testing for model years 2001 through 2006.  Base emission
factors for current and future model years were computed from
multiplicative adjustment factors that ratio the existing post-2000
calendar year emission and the future Tier2 emission standard bins.

The Fuel Adjustment Factors are created from emission results obtained
by running the Primary EPA Fuel Models for all combinations of in-use
fuel formulations. The actual Fuel Adjustment Factor for a given fuel
(i) is the ratio of the emissions using fuel (i) and the emissions using
the reference fuel (see Eq 2).

Fuel Adjustment Factor (fuel i) = emission result(fuel i) / emission
result (reference fuel)  Eq2

Two reference fuels were used in the development of the MOVES Fuel
Adjustment Factors, and their properties are shown in Table 1.  The
reference fuels are intended to represent the fuels used to generated
the base emission rates for light-duty gasoline vehicles.

Table 1

Reference Fuel Properties

Fuel Property	Pre-2001 Reference Fuel	2001+ Reference Fuel

Fuel Subtype	Conventional Gasoline	Conventional Gasoline

RVP (psi)	6.9	6.9

Sulfur Level (ppm)	90	30

Ethanol Volume (vol%)	0	0

MTBE Volume (vol%)	0	0

TAME Volume (vol%)	0	0

Aromatic Content (vol%)	26.1	26.1

Olefin Content (vol%)	5.6	5.6

Benzene Content (vol%)	1.0	1.0

E200 (F)	41.1	41.1

E300 (F)	83.1	83.1



The light-duty gasoline base rate derivations relied heavily on data
from in-use Arizona Inspection / Maintenance (I/M) testing, and, as is
typical in I/M testing, the exact fuel formulations were not controlled
or measured.  The two fuels chosen as reference fuels represent typical
fuels used Phoenix, Arizona in the summer months during the calendar
year 1995 through 2002.   The two reference fuels are identical except
for the fuel sulfur levels of 30 ppms sulfur (fuelformulationid=98) an
90 ppm sulfur (fuelformulationid=99).  The reference fuel with a sulfur
level of 30 ppm is used to create Fuel Adjustment Factors for all 2001
and later model years, and the value of 90 ppm is used for all 2000 and
earlier model years.  The value of 30 ppm was chosen for the late model
vehicles because these model years and future ones are more likely to be
exposed to the lower sulfur fuel than the older model years.  Using a 90
ppm sulfur reference fuel on nLEV and Tier2 vehicles which were
generally certified or operated on lower sulfur fuels would
over-emphasize the effects of sulfur on new and future vehicles.

The Fuel Adjustment table contains both the fuelAdjustment and the
fuelAdjustmentGPA variables.   The "GPA" (Geographic Phase-in Area)
refers to a group of counties in the western U.S. that had special
gasoline sulfur requirements under Tier 2 regulations. See 65 Fed.
Reg.6755-6759 for more information on the GPA.  The values are identical
except for specific counties and fuelMYGroups 2004, 2005 and 2006.  In
those cases, the fuelAdjustmentGPA is slightly higher than the
fuelAdjustment because it accounts for the irreversible effects of
higher fuel sulfur levels.

In MOVES, if a particular county / year / month bin has more than one
fuel formulation (a common situation), the fuel adjustment factors are
weighted linearly and individually using the market share data. See
Equation 3 below.   

Fuel Adjustment Factor = Adjustment1 * market share1 + Adjustment2 *
market share2 + … + Adjusment i * market share i	    Eq 3

MOVES Fuel Adjustment Table Structure

polProcessID				key field

fuelMYGroupID			key field

sourceTypeID				key field

fuelFormulationID			key field

fuelAdjustment

fuelAdjustmentCV

fuelAdjustmentGPA

fuelAdjustmentGPACV

d.	Hydrocarbon Speciation Table

The HCSpeciation table provides factors which convert the base emission
factors which are in terms of total hydrocarbon(THC) emissions into
volatile organic compounds (VOC), non methane hydrocarbons (NMHC), total
organic gases (TOG) and non methane organic gases (NMOG).   See Table 2
below for definitions of the various hydrocarbon species.  

Table 2

Hydrocarbon Speciation Types

PollutantID	PollutantName	Flame Ionization Detector (FID)  HC	Methane
Ethane	Aldehydes



1	Total Hydrocarbons	

Yes	

Yes	

Yes	

No



79	Non Methane Hydrocarbons	

Yes	

No	

Yes	

No



87	Volatile Organic Compounds	

Yes	

No	

No	

Yes



86	Total Organic Gases	

Yes	

Yes	

Yes	

Yes



80	Non Methane Organic Gases	

Yes	

No	

Yes	

Yes



Methane emissions are calculated in MOVES using independent emission
factors derived from methane emission data.   The other hydrocarbon
species are calculated as functions of THC and methane emissions. 
Coefficients for these chained calculations are stored in the MOVES
HCSpeciation table.

MOVES HCSpeciation Table Structure

polProcessID				key field

fuelMYGroupID			key field

fuelFormulationID			key field

speciationConstant

oxySpeciation

The fuelMYGroups and the fuelFormulationID keys have the same
definitions as those used in the  Fuel Adjustment table.   The
speciationConstant and the oxySpeciation coefficient are applied in the
equations below.

NMHC	=	THC  - Methane								Eq 3

VOC	=	NMHC *

(speciationConstant + oxySpeciation* volToWtPercentOxy*ETOHVolume)		Eq
4a

For fuels containing ethanol (fuelsubtypeID = 12, 13, 14, 51, 52, or 53)

VOC	=	NMHC *

(speciationConstant + oxySpeciation* volToWtPercentOxy*MTBEVolume) 	Eq
4b

For fuels containing MTBE

VOC	=	NMHC *

(speciationConstant + oxySpeciation* volToWtPercentOxy*ETBEVolume) 		Eq
4c

For fuels containing ETBE

VOC	=	NMHC *

(speciationConstant + oxySpeciation* volToWtPercentOxy*TAMEVolume) 	Eq
4d

For fuels containing TAME

NMOG	=	NMHC *

(speciationConstant + oxySpeciation* volToWtPercentOxy*ETOHVolume)		Eq
5a

For fuels containing ethanol (fuelsubtypeID = 12, 13, 14, 51, 52, or 53)

NMOG	=	NMHC *

(speciationConstant + oxySpeciation* volToWtPercentOxy*MTBEVolume) 	Eq
5b

For fuels containing MTBE

NMOG	=	NMHC *

(speciationConstant + oxySpeciation* volToWtPercentOxy*ETBEVolume)		Eq
5c

For fuels containing ETBE

NMOG	=	NMHC *

(speciationConstant + oxySpeciation* volToWtPercentOxy*TAMEVolume)		Eq
5d

For fuels containing TAME

TOG	=	NMOG + Methane							Eq 6

Process Used to Create MOVES Fuel Adjustments

This section an overview of the process and algorithms used to develop
the MOVES fuel formulations, fuel supply information and fuel
adjustments.  

a.	Extracting the Primary Data from NMIM

The primary fuel formulation and fuel supply data were obtained from
EPA’s NMIM model and national emission inventory processs.  The NMIM
model is currently used to develop national emission inventories every
three years.  NMIM currently contains fuel formulation and fuel supply
data and projections for all calendars years from 1999 through 2030. 
For support of EPA’s RFS2 Rulemaking, only fuel data from calendar
years 2005 and 2022 were extracted and used.

For calendar year 2005, the database NMIMRFS2Fuels2005Base was created. 
It contains detailed fuel formulation and fuel supply data for calendar
year 2005 collected from in-use fuel surveys.  It contains
fuelformulation data for several thousand individual in-use fuels and
fuel supply data from all 3,222 US counties on a seasonal basis.  The
analysis process used to create this database was done under EPA
contract by Eastern Research Group and has been fully documented in the
EPA 2007 Renewable Fuels Standard (RFS1) Rulemaking documentation.

Two additional databases  were created for calendar year 2022.  These
are:

NMIMRFS2Fuels2022ReferenceREV

NMIMRFS2Fuels2022ControlREV

The 2022 Reference fuel database containing fuel supply projections by
county, month and fuel formulation was created from two sources:  (a)
the output of EPA refinery modeling under contract, which provided the
bulk of the information, and (b) Department of Energy projections from
the Advanced Energy Outlook (AEO).   The AEO projections were the volume
of E10 (a 90% gasoline / 10% ethanol mixture) penetration into the
market.  Ethanol market share information is important because the focus
of the RFS2 rulemaking is ethanol production, marketing and emission
effects.  The 2022 Control fuel database is virtually the same as the
Reference fuel database except it assumes that E10 fuels have 100
percent market share in 2022.

MOVES Fuel Binner

A real in-use gasoline fuel can vary continuously according to all of
the fuel properties listed in Table 1.  For example, a fuel can have any
benzene level or RVP or ethanol level or E300 temperature ranging from
level X to level Y.   However, the MOVES model design does not currently
process such fuel formulation information.  Instead, the MOVES
FuelBinner organizes specific NMIM fuels into bins according to the
eleven fuel properties listed in Table 1, and assigns average values. 
The fuel property values for each bin are then processed outside the
MOVES model in the Predictive, Complex and EPA Sulfur models.  This
process creates fuel adjustment factors and air toxic pollutant factors
which are used in MOVES.

The MOVES fuel binning approach was created as a compromise between
utilizing detailed fuel information and the need to streamline
processing in the MOVES model.  Some pollutants are sensitive to certain
fuel properties and detail is required for accuracy.  In these cases the
MOVES Fuel binner was designed to capture these effects.  On the other
hand, every possible combination (i.e., greater than 10,000) of eleven
fuel properties, twelve source / vehicle types, three fuel types, nine
pollutants and ten model year groups quickly overwhelms the MOVES model
making some aggregation necessary.

The final version of MOVES will likely be re-designed to eliminate the
discrete fuelformulationID and fuel adjustment and perform the
predictive / complex model calculations directly in the model.   This
will eliminate the performance issues associated with the current
version of MOVES and allow for accurate determination of fuel factors. 
However, the current version of MOVES (i.e., October, 2008) does not
contain this capability.

All 10,000+ NMIM fuel formulations were binned according to the eleven
fuel properties listed in Tables 1 and 3.  Table 3 shows the resultant
bins which were created, and the definition of such bins.   A particular
MOVES fuelformulationID has one and only one value for each of the fuel
property bins.  The "Bin Value" represents EPA's judgment of the most
typical value for fuels in the bin, and is the value used in the
computation of fuel adjustments.

Table 3

Gasoline Fuel Property Bin Definitions

Fuel Property	

Bin ID	Bin Value	

Lower Bound

(>)	

Upper Bound

(<=)

Fuel Subtype





Conventional Gas	10



	RFG	11



	Ethanol 10 vol%	12



	Ethanol  8 vol%	13



	Ethanol  5 vol%	14



	Diesel	20









	RVP (psi)	1	6.9	0	7.2

RVP (psi)	2	7.5	7.2	8.2

RVP (psi)	3	8.7	8.2	9.0

RVP (psi)	4	9.2	9.0	10.0

RVP (psi)	5	10.0	10.0	11.4

RVP (psi)	6	11.5	11.4	13.4

RVP (psi)	7	13.5	13.4	14.9

RVP (psi)	8	15.0	14.9+







	Sulfur Level (ppm)	1	5.0	0	10

Sulfur Level (ppm)	2	15.0	10	25

Sulfur Level (ppm)	3	30.0	25	35

Sulfur Level (ppm)	4	50.0	35	70

Sulfur Level (ppm)	5	90.0	70	120

Sulfur Level (ppm)	6	180.0	120	210

Sulfur Level (ppm)	7	280.0	210	300

Sulfur Level (ppm)	8	400.0	300	500

Sulfur Level (ppm)	9	600.0	500+







	Ethanol Volume (vol%)	1	0	0	0

Ethanol Volume (vol%)	2	5.0	0	6.0

Ethanol Volume (vol%)	3	8.0	6.0	8.0

Ethanol Volume (vol%)	4	10.0	8.0+







	MTBE Volume (vol%)	1	0	0	0

MTBE Volume (vol%)	2	2.5	0	5.0

MTBE Volume (vol%)	3	8.0	5.0	10.0

MTBE Volume (vol%)	4	11.0	10.0+

	

Fuel Property	

Bin ID	Bin Value	

Lower Bound

(>)	

Upper Bound

(<=)

TAME Volume (vol%)	1	0	0	0

TAME Volume (vol%)	2	0.015	0	0.03

TAME Volume (vol%)	3	0.06	0.03+







	Aromatic Content (vol%)	1	17.5	0	20.0

Aromatic Content (vol%)	2	26.1	20.0	30.0

Aromatic Content (vol%)	3	32.0	30.0+







	Olefin Content (vol%)	1	5.6	0	8.0

Olefin Content (vol%)	2	9.2	8.0	10.0

Olefin Content (vol%)	3	11.9	10.0+







	Benzene Content (vol%)	1	0.35	0	0.40

Benzene Content (vol%)	2	0.65	0.40	0.80

Benzene Content (vol%)	3	1.50	0.80	2.00

Benzene Content (vol%)	4	3.50	2.00







	E200 (F)	1	41.0	0	47.0

E200 (F)	2	50.0	47.0+

	E200 (F)











E300 (F)	1	78.6	0	80.0

E300 (F)	2	83.0	80.0	87.0

E300 (F)	3	89.1	87.0+







	  

After binning the 10,000+ NMIM fuel formulations into a more manageable
set of about 500 MOVES fuel formulations, the MOVES fuel adjustment
factor process uses an algorithm to create the specific fuel adjustment
factors for each of the MOVES fuel / pollutant-process combinations. 
These fuel effects are contained in the EPA Predictive Model, the EPA
Complex Model and the EPA Sulfur Model.

Predictive Model Effects

The EPA Predictive Fuel model is one of EPA’s more recent fuel models
used to predict the impact on HC and NOx emissions from varying gasoline
fuel properties (i.e., six of the eleven fuel properties listed in Table
1).  This model was developed by EPA from vehicle and fuel testing done
prior to calendar year 2001.  See EPA420-R-01-016, “Analysis of
California’s Request for Waiver of theReformulated Gasoline Oxygen
Content Requirement for California Covered Areas” of June, 2001 for
background on the Predictive Model.  

In MOVES the EPA Predictive Fuel Model was used only to model fuel
effects of HC and NOx emissions.  The MOVES Predictive Fuel Model
differs from the standard model in that the effect of sulfur was removed
from the model by normalizing to 30 ppm sulfur for 2001 and later model
years and 90 ppm for pre-2001 model years.  The MOVES sulfur effects are
the MOBILE6.2 sulfur effects.  These include both the short term and
long term irreversibility effects.  For complete see EPA420-R-01-039
“Fuel Sulfur Effects on Exhaust Emission – Recommendations for
MOBILE6”.   The effect of High emitters is also different between the
EPA Predictive Model and the MOVES Predictive model.  The EPA Predictive
Model assumes a 50 percent weighting for High Emitters, and MOVES does
not use the concept of High and Normal emitters.  For the purposes of
this analysis a weighting of 20 percent High emitters was used to
generate the MOVES fuel effects.  This lower percentage of High emitters
better reflects the modern vehicle fleet.

Complex Model Effects

The MOVES fuel adjustment factors for CO emissions were developed from
the EPA Complex Fuel Model.  This model is an older model and is based
on data from late 1980’s and early 1990’s.  It was used only in
MOVES for non sulfur fuel effects for CO emissions.  It was selected for
use because the EPA Predictive Model does not contain CO emission
effects.  The sulfur CO emission effects were taken from MOBILE6
(EPA420-R-01-039 “Fuel Sulfur Effects on Exhaust Emission –
Recommendations for MOBILE6”).   For more details regarding EPA’s
complex model, the reader is referred to the website   HYPERLINK
"http://www.epa.gov/otaq/regs/fuels/rfg/58-11722.txt" 
http://www.epa.gov/otaq/regs/fuels/rfg/58-11722.txt  and the document
“Regulation of Fuels and Fuel Additives: Standards for Reformulated
Gasoline -  Proposed Rule”

Process Used to Create Air Toxic Adjustment Factors

The MOVES model reports air toxic emission inventories based on air
toxic emission adjustment factors which are built into the MOVES model. 
The air toxic pollutants which are reported by MOVES are:

Benzene

Ethanol

MTBE

1,3 Butadiene

Formaldehyde

Acetealdehyde

Naphthalene

Acrolein

Equation 7 shows the general formula which is used to compute all the
air toxic pollutants in MOVES except Naphthalene, which is a function of
total particulate matter emissions.  The formula is applied separately
for running and start emissions.   Currently, in MOVES (and for the RFS
2 Rulemaking), the air toxic pollutants are a function total hydrocarbon
emissions (i.e., the Hydrocarbon Emission Rate in Eq 7).  In the future,
all of them will be a function of volatile organic compounds (VOC).  The
Hydrocarbon Emission Rate used in Equation 7 is a fully adjusted
emission rate and includes fuel, temperature, I/M and all other
correction factors.  The Air Toxic Factors are applied last in the chain
of MOVES multiplicative calculations.

Air Toxic Pollutant Emission Rate = Hydrocarbon Emission Rate * Air
Toxic Factor   	Eq 7

In MOVES, the Air Toxic Factor differs by pollutant, process, model
year, sourcetype (vehicle type) and fuel formulation for benzene, MTBE,
1,3 Butadiene, Formaldehyde and Acetaldehyde.  All of the values used in
MOVES for these pollutants were taken from the MOBILE6.2 model. The
MOBILE6.2 air toxics model was developed from the 1993 EPA Complex
Model, and is based primarily on data from 1990 and earlier model year
vehicles. The extraction of data from the MOBILE6.2 model was an
empirical process that involved the generation of over 20,000 of
MOBILE6.2 full database runs where the ratio of the various air toxic
pollutant emissions and the hydrocarbon emissions for both start and
running were computed and averaged.  To make the analysis manageable,
all calculations were made for a single age only.  A typical age of
eight years was chosen for all source types (vehicle types) and model
years.  

The Air Toxic Factors for Naphthalene and Acrolein differ only by source
type (vehicle type) and fuel type.  These factors were also taken from
MOBILE6.2 but are not factors of model year and age.

The Air Toxic Factors for ethanol (shown in Table 4) are based on a
literature search of new data and reports.  The reports from three
studies were used in the search / analysis.  They are:

Southwest Research Institute.  2007.  Flex Fuel Vehicles (FFVs)  VOC/PM
Cold Temperature Characterization When Operating on Ethanol (E10, E70,
E85).  Prepared for U. S. Environmental Protection Agency.  Available in
Docket EPA-HQ-OAR-2005-0161

Graham, L. A.; Belisle, S. L. and C. Baas.  2008.  Emissions from light
duty gasoline vehicles operating on low blend ethanol gasoline and E85. 
Atmos. Environ. 42: 4498-4516.

Environment Canada.  2007.  Comparison of Emissions from Conventional
and Flexible Fuel Vehicles Operating on Gasoline and E85 Fuels.  ERM
Report No. 05-039, Emissions Research Division.  Available in Docket
EPA-HQ-OAR-2005-0161

Table 4 shows an average value for the air toxic to total hydrocarbon
emissions ratio and the range of values used in MOVES for calendar year
2005.  These are for gasoline vehicles.  The values represent the
average of all 360 fuel formulations used in the calendar year 2005 fuel
dataset.  The relatively large standard deviations for Benzene and MTBE
reflect the strong function of individual fuel formulation on the
average results.  

Table 4

Typical Gasoline / Ethanol Air Toxic Ratios for Calendar Year 2005

	

Gasoline Vehicles

	Ethanol

Vehicles

	Min AT Ratio	Avg AT Ratio	Max AT Ratio	Std Dev	E-85 Ratios

Benzene	0.032	0.050	0.086	0.0082	0.0041

MTBE	0.00	0.0017	0.018	0.0048	0.00

Naphthalene	0.088	0.088	0.088	0.00	0.086

1,3 Butadiene	0.0038	0.0055	0.0066	0.00063	0.00062

Formaldehyde	0.0097	0.013	0.016	0.0012	0.010

Acetaldehyde	0.0036	0.0070	0.013	0.0032	0.075

Acrolein	0.00061	0.00061	0.00061	0.00	0.00027







	Ethanol  - E0

0.00



	Ethanol  - E10

0.024



	Ethanol  - E85

0.484













	Table 5 shows the air toxic to THC ratios for diesel vehicles as a
function of pollutant.  These were taken from MOBILE6.2.  They are
constants due to a lack of test data.  Specific diesel fuel properties
(i.e., sulfur level) are currently assumed not to affect air toxic
emission ratios.

Table 5

Diesel Vehicle Air Toxic Ratios

Pollutant	Air Toxic Ratio

Benzene	0.020

Ethanol	0.00

MTBE	0.00

Naphthalene	0.0037

1,3 Butadiene	0.0090

Formaldehyde	0.039

Acetaldehyde	0.012

Acrolein	0.0035



Use of the MOVES Fuel Algorithm in RFS2

	The MOVE Fuel Algorithm was used in the RFS2 inventory development
process to estimate the effects of fuel parameters on HC, CO and NOx
emissions from light and heavy-duty gasoline vehicle applications. The
fuel effects are a simple multiplicative adjustment factor applied to
the base emission factors that account for the effects of various fuel
properties and the distribution of particular gasoline fuel throughout
the United States.  Separate factors were generated for HC, CO, NOx and
the evaporative process of permeation.

	The analysis started with the development of three datasets of national
fuels (fuel supply and fuelformulation data) according to individual
county, year and month.  Thus, each of the 3,222 individual United
States counties had a set of specific fuels (with market shares) for
each of the twelve months and for the two years of analysis interest.
These years were the base year of 2005 and the control/reference year of
2022.  

	The base year of 2005 contained a set 261 fuel formulations that
represent every fuel used in the United States in calendar year 2005. 
Since the fuel adjustment factors are also a function of model year
group and source type (vehicle type), the base year analysis contained a
total of 51,285 different fuel adjustment factors.  For the
"Sensitivity" case, these had the following range of values by
pollutant.

Table 6

Average Fuel Adjustment for RFS2 Base Year 2005, Sensitivity Case

Pollutant / Process	Average Adjustment	Std deviation

Running HC	1.044	0.12

Start HC	1.041	0.11

Running CO	1.150	0.31

Start CO	1.128	0.28

Running NOx	1.211	0.20

Start NOx	1.211	0.20



	The fairly large standard deviations illustrate the fairly large range
of possible fuel adjustment factors.  These factors range so much
because in 2005 the individual fuel properties such as sulfur,
conventional gas versus E-10, aromatics, etc., varied widely. For
example, a low sulfur fuel has a lower adjustment factor versus a high
sulfur fuel. Also, the average fuel adjustments shown in Table 6 do not
include E-70 or E-85 fuel effects.

The Control fuel dataset consisted of a set of fuels that were composed
of all E-10 (10 vol% ethanol).  This dataset was developed to model the
effect of 100 percent penetration of E-10 in the fleet.  Other fuel
properties varied so there are 114 individual fuels in this dataset. 
The average fuel adjustments shown in Table 7 do not include E-70 or
E-85 fuel effects.

Table 7

Average Fuel Adjustment for RFS2 Control Year 2022, Sensitivity Case

Pollutant / Process	Average Adjustment	Std deviation

Running HC	0.94	0.05

Start HC	0.98	0.04

Running CO	0.97	0.25

Start CO	1.03	0.24

Running NOx	1.11	0.04

Start NOx	1.10	0.04



The Reference fuel dataset consisted of a set of fuels that were
composed of both E-10 (10 vol% ethanol) and conventional gasoline fuels
in 2022.   Other fuel properties varied so there are 139 individual
fuels in this dataset.  The majority (106 out of 139) of the fuels in
this dataset were E-10 fuels.  The fuel adjustments shown in Table 8 do
not include E-70 or E-85 fuel effects.

Table 8

Average Fuel Adjustment for RFS2 Reference Year 2022, Sensitivity Case

Pollutant / Process	Average Adjustment	Std deviation

Running HC	0.95	0.06

Start HC	0.99	0.04

Running CO	0.99	0.24

Start CO	1.05	0.23

Running NOx	1.10	0.05

Start NOx	1.09	0.05



As described in the main body of this memo, the RFS2 NPRM analysis also
included consideration of a "Primary" fuel effects case.  Table 9
summarizes the fuel adjustments for the Primary Case in the 2022 Control
and Reference scenarios.  The adjustments for the 2005 Primary Case are
not summarized here, but because Tier 0 vehicles are a larger portion of
the 2005 fleet, the Primary Case adjustments are closer to the
Senstivity Case in 2005 than in 2022. 

Table 9

Average Fuel Adjustment for RFS2 Control and Reference 2022 Primary Case

Pollutant / Process	Average Adjustment

Control	Average Adjustment

Reference

Running HC	0.95	0.95

Start HC	0.99	1.00

Running CO	0.97	0.99

Start CO	1.02	1.05

Running NOx	1.03	1.03

Start NOx	1.03	1.02



		

References

 PAGE   

 PAGE   15 

Page   PAGE  15  of   NUMPAGES  38 		  DATE \@ "M/d/yyyy"  5/12/2009 

Page   PAGE  14  of   NUMPAGES  38 		  DATE \@ "M/d/yyyy"  5/12/2009 

  U.S. EPA, 2008.  Development of MOVES Rates for HC, CO and NOx Exhaust
Emissions from Light-duty Vehicles for use in the RFS2 NRPM, Memorandum
from James Warila, U. S. EPA, November 2008.

 U.S. EPA.  2008.  Kansas City Characterization Study:  Final Report
(EPA420-R-08-009)  

  HYPERLINK
"http://www.epa.gov/otaq/emission-factors-research/420r08009.pdf" 
http://www.epa.gov/otaq/emission-factors-research/420r08009.pdf 

 U.S. EPA.  2008.  Analysis of Particulate Matter Emissions from
Light-Duty Vehicles in Kansas City (EPA420-R-08-010)     HYPERLINK
"http://www.epa.gov/otaq/emission-factors-research/420r08010.pdf" 
http://www.epa.gov/otaq/emission-factors-research/420r08010.pdf 

 U.S. EPA, 2008.  Modeling Evaporative Emissions in MOVES for the RFS2
NPRM.  Memorandum from Prashanth Gururaja, U.S. EPA, November 2008.

 U.S. EPA. 2002.  Technical Description of the Toxics Module for
MOBILE6.2 and Guidance on Its Use for Emission Inventory Preparation. 
(EPA420-R-02-029)    HYPERLINK
"http://www.epa.gov/otaq/models/mobile6/r02029.pdf" 
http://www.epa.gov/otaq/models/mobile6/r02029.pdf 

 U.S. EPA, 2008.  Development of MOVES Rates for HC, CO and NOx Exhaust
Emissions from Light-duty Vehicles for use in the RFS2 NRPM, Memorandum
from James Warila, U. S. EPA, November 2008.

 Clark, Nigel, et al., "Heavy-Duty Vehicle Chassis Dynamometer Testing
for Emissions Inventory, Air Quality Modeling, Source Apportionment and
Air Toxics Emissions Inventory," Final Report for the Coordinating
Research Council, and others, CRC Project E-55/59, August 2007. 
(Available at
http://www.crcao.com/reports/recentstudies2007/E-55-59/E-55_59_Final_Rep
ort_23AUG2007.pdf)

 U.S. EPA.  2008.  Kansas City Characterization Study:  Final Report
(EPA420-R-08-009)  

  HYPERLINK
"http://www.epa.gov/otaq/emission-factors-research/420r08009.pdf" 
http://www.epa.gov/otaq/emission-factors-research/420r08009.pdf 

 U.S. EPA.  2008.  Analysis of Particulate Matter Emissions from
Light-Duty Vehicles in Kansas City (EPA420-R-08-010)     HYPERLINK
"http://www.epa.gov/otaq/emission-factors-research/420r08010.pdf" 
http://www.epa.gov/otaq/emission-factors-research/420r08010.pdf 

 Haskew, Harold M., Thomas F. Liberty and Dennis McClement, “Fuel
Permeation from Automotive Systems,” Final Report, for the
Coordinating Research Council and the California Air Resources Board,
CRC Project E-65, September 2004.  

 U.S. EPA, 2008.  Modeling Evaporative Emissions in MOVES for the RFS2
NPRM.  Memorandum from Prashanth Gururaja, U.S. EPA, November 2008.

 Reddy, Sam R.  2007.  “A Model for Estimating Vapor Pressures of
Commingled Ethanol Fuels.” SAE 2007-01-4006.

 U.S. EPA, 2008.  Modeling Evaporative Emissions in MOVES for the RFS2
NPRM.  Memorandum from Prashanth Gururaja, U.S. EPA, November 2008.

 Clark, Nigel, et al., "Heavy-Duty Vehicle Chassis Dynamometer Testing
for Emissions Inventory, Air Quality Modeling, Source Apportionment and
Air Toxics Emissions Inventory," Final Report for the Coordinating
Research Council, and others, CRC Project E-55/59, August 2007. 
(Available at
http://www.crcao.com/reports/recentstudies2007/E-55-59/E-55_59_Final_Rep
ort_23AUG2007.pdf)

 Clark, Nigel, et al., "Heavy-Duty Vehicle Chassis Dynamometer Testing
for Emissions Inventory, Air Quality Modeling, Source Apportionment and
Air Toxics Emissions Inventory," Final Report for the Coordinating
Research Council, and others, CRC Project E-55/59, August 2007. 
(Available at
http://www.crcao.com/reports/recentstudies2007/E-55-59/E-55_59_Final_Rep
ort_23AUG2007.pdf)

 Mobile Source Observation Database  (Available at
http://www.epa.gov/otaq/models/msod/msodannc.htm)

 MOVES Versions and Inputs 

