Fuel Supply Defaults:
Regional Fuels and the Fuel Wizard
in MOVES3
United States
Environmental Protection
^1 Agency
-------
Fuel Supply Defaults:
Regional Fuels and the Fuel Wizard
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.
in MOVES3
Assessment and Standards Division
Office of Transportation and Air Quality
U.S. Environmental Protection Agency
NOTICE
&EPA
United States
Environmental Protection
Agency
EPA-420-R-20-017
November 2020
-------
Table of Contents
1. Introduction 3
2. Development of Base Fuel Regions 5
3. Incorporation of Local Fuel Programs 6
4. Background on Fuel Property Data 10
5. Regional Gasolines for 2013 and Earlier 11
6. Regional Gasolines for 2014 and Later 15
7. Diesel, CNG, and E85 19
8. Nonroad Fuel Supply 22
9. Updates to Fuel Wizard Factors for Ethanol Blending 24
10. References 28
APPENDIX A: MOVES3 FUEL REGION MAPS, 1990-2020 30
APPENDIX B: AVERAGE GASOLINE FORMULATION TRENDS IN MOVES, 1990-2020
54
APPENDIX C: REFINERY MODELING RESULTS USED TO DEVELOP FUEL WIZARD
FACTORS 55
2
-------
1. Introduction
This document describes the background on the fuel supply components of MOVES3 and
methodology used to develop and update them.
MOVES estimates emissions of vehicles and nonroad equipment using the "fuel types" of
gasoline, diesel, compressed natural gas (CNG), ethanol (E85) and electricity. MOVES also
models nonroad equipment using liquified petroleum gas (LPG). Within the gasoline and diesel
fuel types, the model considers different fuel subtypes, as shown in Table 1-1. Each of these
subtypes has a number of specific formulations that vary with region, season, and calendar year.
Given their small market share and the lack of emissions data, MOVES does not currently model
emissions of vehicles or equipment using fuels such as liquified natural gas (LNG) or renewable
diesel.
Table 1-1. Fuel Subtypes in MOVES3.
Subljpe
Description
10
Conventional Gasoline
11
Reformulated Gasoline (RFG)
12
Gasohol (E10)
13
Gasohol (E8)
14
Gasohol (E5)
15
Gasohol (E15)
20
Conventional Diesel Fuel
21
Biodiesel (BD20)
22
Fischer-Tropsch Diesel (FTD100)
23
Nonroad Diesel Fuel
24
Marine Diesel Fuel
30
Compressed Natural Gas (CNG)
40
Liquefied Petroleum Gas (LPG)
50
Ethanol
51
Ethanol (E85)
52
Ethanol (E70)
90
Electricity
The MOVES fuel supply is primarily comprised of three tables that reference each other. The
regionCounty table assigns a fuel region to each county for each calendar year. This allows a
given county to change fuel regions over time, which may occur if a local area adopts a volatility
control program, for example. The fuelFormulation table assigns specific properties, such as
ethanol, aromatics, and sulfur level, to each fuelFormulationlD. The fuelSupply table designates
a fuelFormulationlD and marketShare for each fuel region by month and year, which allows
formulations to change by season as well as over the years as federal regulations or market
blending practices change.
3
-------
The default fuel supply in M0VES3 generally uses the same regional fuel structure and data
sources as in MOVES2014b, but with significant updates based on the latest available data.
Diesel sulfur was adjusted downward to be more consistent with in-use levels observed after the
final phase-in of current regulatory programs. Biodiesel levels were also revised to better reflect
data published by the Energy Information Administration.
With this release, gasoline formulations for calendar years 2014 and later have been replaced
with fuel property data derived from updated refinery production batch data. These updates
include phase-down of sulfur to 10 ppm in accordance with the Tier 3 standards.
Market shares of ethanol blends have been simplified to reflect 100 percent E10 in all regions for
2012 and later. Given the difficulty of estimating the sales volumes and locations of E0 (non-
ethanol fuel) and E15 ethanol blends, both of which are very small relative to E10, their market
shares have been set to zero. However, the fuel supply provides E0 and El 5 properties for all
regions to allow users to add market share for these fuels as needed.
Distillation values were also updated for all E10 blends. Since the refinery batch data underlying
the fuel formulations in MOVES reports distillation as percent evaporated at a particular
temperature (e.g., E200 = vol% evaporated at 200F), and some of MOVES' fuel adjustment
models work in terms of fractionation temperature (e.g., T50 = temperature at which 50 vol% is
distilled), a correlation is used to convert from E to T values. In this release, the correlation used
for E10 has been updated using retail survey data, and the results applied to all ElOs in the fuel
supply.
Finally, the Fuel Wizard includes results of updated refinery modeling examining the effects of
ethanol blending on gasoline aromatics, olefins, and distillation of the fuel supply.
The report is organized into the following sections plus three appendices.
Sections 2 and 3 describe development of the geographical areas used in the fuel supply
and how the regionCounty table was updated for MOVES3.
Sections 4 through 6 describe updates to data sources and analysis of fuel properties as
well as changes to ethanol market shares.
Sections 7 and 8 describe the onroad diesel and nonroad components of the fuel supply.
Section 9 describes the Fuel Wizard tool used to estimate secondary fuel property
changes for gasoline blending and updates that were made for MOVES3.
Appendix A provides color-coded maps showing which counties correspond with which
regionlDs for years 1990-2020.
Appendix B presents average gasoline formulation trends over years 1990-2020.
Appendix C describes the refinery modeling work used to update the Fuel Wizard.
The fuel supply report underwent two peer-reviews since the release of MOVES2014 consistent
with EPA's peer review guidance.1 In 2017, we peer-reviewed an internal version referred to as
"MOVES201X" that included initial updates made from MOVES2014. 2 In 2020, we conducted
another peer-review to review further updates made for MOVES3.3 Updates to the MOVES3
fuel supply were made in response to peer-review comments from both reviews. Materials from
each peer review, including peer-review comments and EPA responses are located on the EPA's
science inventory webpage.
4
-------
2. Development of Base Fuel Regions
The base fuel regions in MOVES were developed starting from Petroleum Area for Defense
District boundaries (a historic division of fuel supply areas developed in the 1950s) along with
locations of fuel terminals and pipelines using data from Oil Price Information Service (OPIS)
and Energy Information Administration (EIA), which allowed grouping of areas sharing
connections to similar refined product delivery networks.4 " A dividing line between base regions
1 and 2 can be seen along the Appalachians, with the Colonial pipeline forming a major
distribution corridor running from Houston, Texas, to the east side of the mountain range and up
into New England. Meanwhile, the Magellan pipeline network runs north from the Houston area
into the midwest and plains states that comprise base regions 2 and 3. A high-level depiction of
these pipelines overlaid onto state boundaries is shown in Figure 2-1. This analysis led to the
base fuel regions described in Table 2-1 and shown in Figure 3-1.
Figure 2-1 Illustration of petroleum product pipelines in the continental United States1
5
-------
Table 2-1. Base fuel region ID numbers and general descriptions.
Base
Region
ID#
liase Region
Name
General Description
1
East Coast,
Caribbean
East coast states, west to Appalachians; Florida; and Gulf Coast region;
Puerto Rico and US Virgin Islands
2
Midwest
Midwest states, east to Appalachians; Tennessee; Kentucky
3
South
Iowa to Texas (North to South); Alabama to New Mexico (East to West); does not
include counties along the Gulf Coast
4
North
North and South Dakota, Minnesota, Wisconsin
5
Rocky Mtns.
Pacific Northwest, Rocky Mountain states
6
AZ/NV/HI
Arizona besides Phoenix area; Nevada; Hawaii
7
Alaska
All Alaska counties
11
East Coast RFG
East Coast states and regions using reformulated gasoline (RFG)
12
MD/VA RFG
Maryland and Virginia regions using RFG
13
Texas RFG
Texas regions using RFG
14
Midwest RFG
Midwest regions using RFG
15
California RFG
California and Phoenix-area counties using
fuel produced by California refineries
3. Incorporation of Local Fuel Programs
After developing the base fuel regions, areas of local fuel controls were added. The
regionCounty table assigns a region [regionID] to each county [countylD] in a given year
|fuelYearlD], including regions with state or local fuel control programs. Each county also has an
identifier [regionCodeID\ which allows separate fuel regions for onroad and nonroad
applications (though this is not exercised in the default supply).
The regionID field contains formatted information regarding key parameters in the fuel region
and can be decoded as AABBCCDDXXwhere:
AA = base region ID
BB = maximum summer region RVP value, or 00 for ASTM
CC = absence of RVP waiver, where 01 indicates no waiver
DD = minimum ethanol level in vol%
XX = reservedfor future use
The full set of the 21 regionID values used in MOVES3 are shown in Table 3-1. The maximum
summer RVP refers to the maximum Reid Vapor Pressure, which is a measure of the volatility of
6
-------
the fuel. Local air agencies may set a limit on vapor pressure of the gasoline fuel to reduce
evaporative emissions of volatile organic compounds. A value of "00" indicates that the region is
using the federal RVP limit.
The RVP waiver refers to the "1.0 psi RVP allowance for gasoline containing ethanol at 9 to 10
volume percent."6 For example, for a region with an RVP waiver and a summer RVP limit of 7.8
psi, the RVP of an E10 fuel in the market is assumed to be 8.8 psi. Not all fuel regions allow for
the 1.0 psi fuel waiver; for example, a State Implementation Plan (SIP) may withhold the 1.0 psi
waiver to help meet air quality goals. Thus, the presence or absence of an RVP waiver is part of
each region definition.
The minimum ethanol content indicates the minimum ethanol level required either by RFG or a
local fuel program. While the RFG oxygenate requirement was lifted by the Energy Policy Act
of 2005, the regionlDs for RFG areas continue to show the minimum ethanol flag for all years,
something that has no impact on the emission computations.7
Table 3-1. RegionID values in MOVES3.
Region II)
Base
Region
II)#
Base Region Name
Maximum
Summer RYP
(psi) or 00 lor
ASTM
1.10 RYP
Waiter
Sialus
(0l=\o 1 -psi
Waiter)
Minimum
I'.lhanol
Yolunie.
»/
¦ O
0
0
Nalionw idc iv;',ion
0 0
01
51
100000000
1
East Coast,
Caribbean
0.0
00
0
100010000
0.0
01
0
170000000
7.0
00
0
178000000
7.8
00
0
178010000
7.8
01
0
200000000
2
Midwest
0.0
00
0
270000000
7.0
00
0
278000000
7.8
00
0
278010000
7.8
01
0
300000000
3
South
0.0
00
0
370000000
7.0
00
0
370010000
7.0
01
0
400000000
4
North
0.0
00
0
500000000
5
Rocky Mtns
0.0
00
0
578000000
7.8
00
0
600000000
6
AZ, NV, HI
0.0
00
0
678000000
7.8
00
0
700000000
7
Alaska
11.5
00
0
1170011000
11
East Coast RFG
7.0
01
10
1270011000
12
MD/VA RFG
7.0
01
10
1370011000
13
Texas RFG
7.0
01
10
1470011000
14
Midwest RFG
7.0
01
10
1570011000
15
California RFG
7.0
01
10
Note:
'Region 0 is only used in estimating the emissions from vehicles running on high-level ethanol blends (e.g., E85). For
additional details, see MOVES fuel effects report.8
7
-------
Figure 3-1 shows the mapping of all fuel regions at a county level for the continental U.S. in
July, 2020. These region assignments are carried forward through 2060. Additional maps are
available in Appendix A.
Fuel Regions (fuelRegionID)
Figure 3-1. Map of MOVES fuel regions in the continental US for fuelYearlD 2020.
As mentioned above, the regionCounty table includes a fuelYearlD dimension, spanning 1990
through 2060. This allows the model to account for changes over time in the region to which a
county is assigned, for example when local volatility programs start or end. For MOVES3, this
table has been updated to include year-by-year changes in volatility designations from 1990 to
2020 based on information available in the Federal Register and other sources. Table 3-2
summarizes the updates and information sources that were included.
8
-------
Table 3-2. Local fuel programs in MOVES3 regionCounty table."
^ i-;i lJ'
Sl;ik-
('li;iii
-------
4. Background on Fuel Property Data
For gasoline, fuelSupply and fuelFormulation tables are based on the nationwide refinery gate
compliance data collected by EPA on all batches of gasoline entering the distribution system
from refiners, blenders, and importers. This includes several fuel properties that can be used to
predict emissions. For diesel, MOVES fuel adjustments are simpler than for gasoline, only using
inputs of sulfur level and, for engines 2006 and older, biodiesel content.8
The rest of this discussion of fuel properties pertains to gasoline, unless otherwise noted. The
data reported for each gasoline batch includes the properties and other values required to
determine compliance with EPA requirements. These are summarized in Table 4-1. These
properties are reported for approximately thirty-thousand batches per year, a database which
forms the basis for the fuel properties used in the regional fuel methodology. These compliance
reports are considered Confidential Business Information (CBI) and cannot be provided to the
public in its raw form. The data analysis and aggregation are discussed in more detail in the next
section.
Table 4-1. Fuel data used in MOVES regional fuel property methodology.
Batch descriptors: Product Type, Grade, Volume (gallons), Production Date.
Property data in vol%: Aromatics, Benzene, E200, E300, ETBE, Ethanol, MTBE, Olefins, TAME.
Other property data: Oxygen (wt%), RVP (psi), Sulfur (ppm wt), T50 (°F), T90 (°F).
The Product Type value specifies whether the fuel is designed by the producer as conventional
gasoline (CG), reformulated gasoline (RFG), or blendstock for oxygenate blending (BOB). Fuel
property data reported for RFG accounts for the downstream blending of ethanol, while the other
types generally do not, and therefore, we made the required adjustments mathematically when
producing the fuel supply. More detail is given on this in the next section.
While we have access to some retail survey data, for example hundreds of samples per year from
the Alliance of Automobile Manufacturers,17 review of the compliance data reveals a large
batch-to-batch variation. Figure 4-1 provides an example of the variation in gasoline aromatics
level occurring over the course of a year for each refinery. This suggests that trying to rely on
seasonal retail samples in a couple dozen key markets may provide results unrepresentative of
the mean. Further plots and analysis for other fuel properties are available in EPA's Fuel Trends
Reports.9'10 These documents also compare retail survey data to refinery production data,
generally validating the approach of using compliance data on a regional basis.
10
-------
70
60
O
>
50
c
oj 40
c
o
u
30
re
£
o
20
10
0
20
40
60
80
100
Refinery
Figure 4-1. Aromatics maximum, minimum, and mean by US refinery in 2016 for RFG (includes ethanol
blending) and CG (not adjusted for ethanol blending).9
5. Regional Gasolines for 2013 and Earlier
For MOVES3, we chose to focus our updates on the years 2014 and later where we expect most
model runs to occur. For years 2013 and earlier, the fuelSupply and fuelFormulation tables were
imported directly from MOVES2014b except for adjustments described in Section 5.2.
5.1 Information Carried Over from MOVES2014
For calendar years 1990-2006, refinery batch data was unavailable for use in MOVES2014
development. Thus, the MOVES20 i 4 fuel supply for these years was produced by aggregating
the county-level fuel properties in MOVES2010 into the new fuel regions using county VMT
weightings. For years 2007 and later, a large amount of gasoline batch data was available. Before
aggregating this data into fuel regions, we processed the dataset to exclude duplicate reporting
(i.e., a refinery and independent results may both report the same fuel). We also repaired or
excluded batches with missing or inappropriate data. For example, in many cases the T-number
distillation values were missing, since compliance reports require E-number values. This was
repaired using correlations in Equation 5-1 and Equation 5-2 (derived from E0 gasoline data).
T50 = 2.0408 x (147.91 - E200) Equation 5-1
T90 = 4.5454 a (155.47 - E300) Equation 5-2
Finally, we separated differing types of fuel batches for further processing. In these steps, non-
ethanol and pre-blended gasolines were included in the dataset without adjustment, blendstocks
for oxygenate blending (BOBs) were adjusted to account for ethanol added downstream from the
refinery gate. This was generally done by a dilution calculation because the properties reported
11
-------
were for the sub-grade hydrocarbon, which would simply be "splash blended" with ethanol at the
destination terminal prior to local distribution. This adjustment is described in more detail in the
introduction section of EPA's Fuel Trends Report.9
After these steps, we had between twenty and thirty-five thousand usable batch records,
depending on year, with no fuel region being represented by less than one thousand batches. The
fuel property data were then aggregated by fuel region (see Section 2), using fuel batch size as a
weighting factor. Initially, fuels were aggregated into four seasonal categories, including
summer, winter, and two transitionary 'shoulder' seasons. After reviewing the results of these
categories, we determined that there was not adequate data (<100 batches for some regions) on
fuel properties other than RVP and distillation to determine specific shoulder season values.
Thus, two aggregation seasons were used for this dataset: summer (May through September) and
winter (January, February, March, April, October, November, and December). The RVP values
for shoulder season (April and October) were set to intermediate values between summer and
winter, and distillation values adjusted using the factors described in Section 9, Table 9-2. Other
fuel properties were set to winter values for the shoulder months for calendar years 2013 and
earlier.
To determine ethanol content of fuels in years prior to 2014, we used information from the
Annual Energy Outlook report (AEO2014) generated by the U.S. Energy Information
Administration (EIA).11'12 The AEO report provides year-by-year projections for biofuel energy
consumption by fuel type for ethanol-gasoline blends, flexible- fuel vehicle (FFV) blends (E70-
E85), and biodiesel blends far into the future. We used these projections in conjunction with the
overall fuel energy requirements computed by the model to calculate the El0 market share for
years 2013 and earlier.
Gasoline compliance data suggests that over 80 percent of batches in the 2011 database consist
of blendstock (CBOB/RBOB) for downstream ethanol blending. Therefore, for calendar years
beyond 2011, when making adjustments to include the addition of ethanol, we did not include
dilution effects on sulfur or benzene, which are controlled to specific levels in finished gasoline
by federal or state regulations. For sulfur in calendar years prior to 2011, we used the batch fuel
data as the source of the fuel sulfur content. In MOVES2014, 2011 and later gasoline had sulfur
set to 30 ppm consistent with full phase-in of Tier 2 sulfur program. For MOVES3, since the
updates were made starting in 2014, the sulfur content of 30 ppm remains in place for 2011-
2013, and sulfur then begins to follow the declining batch data trend as the Tier 3 sulfur
requirement is phased in.13
The EPAct model implemented in MOVES2014 for fuel adjustments in 2001 and newer
vehicles, was built specifically around ethanol blends and cannot be used to compute emission
impacts of MTBE. Thus, in MOVES2014, MTBE was replaced with ethanol in oxygenate blends
in order to provide an approximation of fuel effects across the range of model years (no other
fuel properties were adjusted besides the oxygenate). More details on the EPAct models are
available in the MOVES2014 Fuel Effects report.14
As a final step, the production-based fuel properties were weighted together with incoming
regional transfers of gasoline produced elsewhere, using data from EIA's Petroleum Supply
Annual reports, in an effort to make them more representative of fuel being consumed in a
12
-------
particular area.15 This technique was developed for benzene in the MSAT2 rulemaking analysis
and is described in more detail in Chapter 6.10.1.3 of the regulatory impact analysis.15
5.2 Updates for MOVE S3
The underlying gasoline batch data used in generating the fuel supply contains E200/E300 values
for the vast majority of batches, consistent with reporting requirements, and T50/T90 for only a
subset. Since MOVES fuel adjustments rely on T50/T90, the missing data were estimated from
the E200/E300 values using correlations developed during a period when there was little ethanol
in the fuel supply. A review of recent market survey data suggested that the correlation between
E and T values should be updated for E10 fuels. Figure 5-1 shows a correlation analysis
developed from 2017 and 2018 Alliance of Automobile Manufacturers (AAM) surveys for E10
regular grade gasoline for both conventional and reformulated gasolines.17
230
220
210
200
^190
T3
o 180
170
160
150
140
370
350
330
L.
ib
2.310
3
n
290
y = -5.733X + 485.230
R2 = 0.931
CG E200 >=58 CG E200<58
a MOVES2014 - - Linear (CG E200 >=58)
Linear (CG E200<58)
T50
~
~
CbDo ~
a y = -0.9602X + 210.94
R2 = 0.7224
Conventional Gasoline (CG)
50 55 60
E200 (vol%)
T9i
Do
y =-4.216x + 679.615
R2 = 0.940
,t3 D
0
270
Conventional Gasoline (CG)
S+WCG a MOVES2014 Linear (S+W CG)
230
220
210
200
^190
o 180
LD
\-
170
160
150
140
350
340
330
320
£>310
"O
o 300
cn
290
280
270
260
RFG E200 > 55 RFG E200 <55
~ MOVES2014 - - Linear (RFG E200 > 55)
Linear (RFG E200 <55) Linear (MOVES2014)
T5C
y=-3.478x + 373.707 ?
R2 = 0.979
%-B3c0 .,
o
'a .
y = -1.1881x +224.56
R2 = 0.8777
Reformulated Gasoline (RFG)
50 55 60
E200 (vol%)
Alliance of Automotive
Innovators 2017-18 North
American Fuel Survey
T90
v = -4.736x + 726.529
R* 0.949
Reformulated Gasoline (RFG)
S+W RFG ~ MOVES2014
Linear (S+W RFG) Linear (MOVES2014)
75
80
85
E300 (vol%)
90
95
85
E300 (vol%)
90
95
Figure 5-1. Derivation of updated T50 and T90 correlations for regular grade E10 gasoline. Conventional
gasoline are the left plots and reformulated are on the right.
Visible in the plots is a clear seasonal difference in the correlation for T50/E200 but none for
T90/E300. Thus, two correlations (orange and blue points) were fit around an E200 cutpoint of
58% for conventional gasoline and 55% for reformulated. While there was little change in the
T90/E300 correlation from the historical EO-based relationship, new values were computed for
13
-------
M0VES3 regardless. These updated correlations were used to generate new T-numbers for all
E10 formulations in the fuel supply, including historical fuels. The updated correlations for E10
are shown in Equation 5-3 through Equation 5-8.
Another change to the 2013 and earlier fuel supply in MOVES3 was to replace the
MOVES2014b California fuel properties with a set derived from AAM surveys collected in Los
Angeles and San Francisco annually between 2000 and 2016.17 This applies to all years, where
1990 and 1999 values duplicated the 2000 properties, and 2017 and later duplicated the 2016
properties. After review of available data sources, we felt this would produce the most accurate
and consistent results.
Market shares of ethanol blends have been simplified to reflect 100% E10 in all regions for 2012
and later. This required replacing very small market shares of E15 in year 2013 with E10 (this
was the earliest year E15 had been included in the MOVES2014b supply). Given the difficulty of
estimating the sales volumes and locations of E0 (non-ethanol fuel) and El 5 ethanol blends, both
of which are very small relative to E10, their market shares have been set to zero. However, the
MOVES default fuel supply provides E0 and E15 properties in 2014 and later for all regions to
allow users to model these fuels by adding market share for these fuels as needed. Note that
summer El 5 was assumed to be made with sub-RVP blendstocks prior to 2019 when the 1 psi
waiver was extended to this blend level.
Finally, we removed several near-duplicate fuel formulations. Fuel formulations in MOVES2014
were generated algorithmically from the compliance batch data, which resulted in a number of
formulations that differed by only a tiny variance in one property which were unlikely to
represent real-world differences in in-use fuel supplies.
Specifically, a series of sorting operations were carried out on the properties ETOHVolume,
aromaticContent, RVP, E200, and sulfurContent and then neighboring fuels were compared for
differences smaller than some margin deemed insignificant (0.5 ppm for sulfur and 0.05 vol% or
psi for the others). Where all five properties were a match within the margin, the lowest
fuelformulationID of the set was recorded as the representative fuel for the others. The sort was
repeated for each property until a robust list of representative fuels had been assigned throughout
the fuelFormulation table. Though some properties were ignored in the process (e.g.,
benzeneContent and E300), a review of the results found the five properties to be sufficiently
precise.
RFG T50 = 373.707 - 3.478 x E200 (where E200 < 55)
RFGT50 = 224.56- 1.1881 x E200 (where E200 > 55)
RFG T90 = 726.529 - 4.736 x E300
CG T50 = 485.230 - 5.733 x E200 (where E200 < 58)
CG T50 = 210.94 - 0.9602 x E200 (where E200 > 58)
CG T90 = 679.615 - 4.216 x E300
Equation 5-3
Equation 5-4
Equation 5-5
Equation 5-6
Equation 5-7
Equation 5-8
14
-------
Similarly, a number of fuel regions that were not referenced in any year or county were removed
from the model, along with their associated formulations if those were not used elsewhere, to
reduce the size of the fuel supply database in MOVES3.
6. Regional Gasolines for 2014 and Later
With MOVES3, we replaced the MOVES2014b gasoline fuel supply entirely for calendar years
2014 and later, where we expect most model runs to occur. For years 2013 and earlier, the
fuelSupply and fuelFormulation values were imported directly from MOVES2014b except for
the adjustments described in Section 5.2.
6.1 Updates to Calendar Years 2014 and Later
For years 2014 and later (non-California) fuels, the MOVES3 fuelSupply and fuelFormulation
tables were updated based on batch data from 2015 and 2016, using the MOVES2014
methodology described above. It also assumed 100 percent E10 because the data indicated usage
of other ethanol blend levels (including E0) was too small to account for on a national or regional
basis. One notable difference was that shoulder season gasoline properties were set to values
intermediate between summer and winter in each region. To produce 2014, 2015, 2017, and 2018
fuel supplies from the 2015-16 dataset, adjustments were made forward and backward year-by-
year using relative differences computed from publicly-available batch data summaries on EPA's
website.18 The analysis computed separate adjustments by season, year, CG/RFG, and fuel
property, but did not attempt to adjust for the ethanol level of the batch data. These adjustment
factors were generally small (e.g., a few percent) and the ratio of BOB to finished gasoline was
unlikely to be changing greatly from year to year. Since 2018 was the latest dataset available,
fuel formulations in 2019 and later were duplicated from 2018 with the exception of sulfur.
Considering the Tier 3 Sulfur Standard, any sulfur level above 10 ppm in 2018 was set to 10 ppm
for 2020, with 2019 set to a mid-point interpolation. Other fuel properties were adjusted per the
Fuel Wizard factors described in Section 9. Fuel formulations in 2020 were duplicated for 2021-
2060.
Finally, E0 and E15 formulations were produced for calendar years 2014 and later (E15 summer
blends begin in 2019). While their market share was set to zero in the fuelSupply table, the
formulations were computed to give users a simple and consistent way to model these fuels if
desired. Table 6-1 depicts the overall scheme for updating the gasoline fuel supply.
15
-------
Table 6-1. Scheme for updating the MOVES3 default gasoline fuel supply.
Calendar year >
1990-2011 | 2012 | 2013
2014 I 2015 I 2016 | 2017 | 2018 | 2019 | 2020+
EO formulations
MOVES2014
Computed EO properties from local E10 formulation using
fuel wizard factors (i.e., reverse match blending)
E10 formulations
MOVES2014 with updated
distillation correlation
Adjusted 2016values
using batch data and
any local RVP changes
Refinery
batch data
Adjusted 2016values
using batch data and
any local RVP changes
Derived from 2018 using
Fuel Wizard sulfur effects
assuming 10 ppm in all
fuels by 2020, and any
local RVP changes
E15 formulations
None
Computed splash-blends from local ElOfor
winter and shoulder seasons only
(Summer RVP waiver was extended to E15 in 2019)
Computed splash-blends
from local E10
Market share
MOVES 2014
100% E10
E15 properties were computed assuming splash blending from the local E10, consistent with the
extension of the 1 psi waiver to E15. This is the most likely scenario until E15 volumes are large
enough to warrant their own regional sub-grade blendstocks. Thus, the aromatics, sulfur,
benzene, and olefin values for El 5 were computed as 0.95 times the E10 property value based on
mathematical dilution. Effects on distillation were estimated by comparing regular grade E10 and
E15 values from Appendix A of the 2010 API blending study19 (presented in Figure 6-1 and
Figure 6-2).
T50 Effect of Splashing Blending E15 from E10
170
168
166
tr
§? 164
2 162
UJ
o 160
i/i
i-
158
156
154
140 160 180 200 220 240
T50 of E10 Base (deg.F)
y = 0.1284x +
137.15
R2 = 0.6658
0
*F
Figure 6-1. T50 effect of splash blending E15 from E10 in 2010 API blending study.
16
-------
ao
-------
60
59
g58
o
^57
o
o
2 56
55
54
53
150 155 160 165 170 175 180 185
T50 (deg.F)
>1
.
y = -0.
1909x + 87.784
-! "
R
2 = 0.602f
I s
p~
.
%\
:
,
;
s
: ' : ¦
^ S
: 1 : -
¦ g
Figure 6-3. E200 to T50 correlation for E15 fuels based on AAM 2017-18 market surveys.
The earlier E300/T90 relationship was simply inverted, but for E200/T50 the correlation was
refit to a limited T50 range of 55-60 vol% expected to be most applicable to E15 fuels. Results of
this analysis are shown in Equation 6-3 and Equation 6-4 and Figure 7-3.
E200 = 87.784 - 0.1909 x T50 Equation 6-3
E300 = 156.69 + 0.2229 x T90 Equation 6-4
Producing E0 properties for each region based on the local E10 required a reverse match-
blending computation to make up for ethanol's octane. The updated Fuel Wizard factors
(discussed in more detail in Section 9) were applied for all properties except RVP and T50/T90.
RVP was adjusted downward by 1 psi except in areas without the 1-psi waiver. The distilation
values of T50/T90 were computed using the E0 correlations to E200/E300 shown in Equation
5-1 and Equation 5-2.
Table 6-2 summarizes the assignment of fuelFormulationID (FFID) values after these updates.
The fuels for years 2014 and later were assigned FFIDs in bands of 122 values according to their
calendar year and ethanol blend level. For example, E15 formulations in year 2019 fall between
8600 and 8722. This scheme leaves gaps for user-created fuels to follow the same scheme.
Calendar year 2013 and earlier gasolines imported from MOVES2014b span the FFID range
1000-2801 after elimination of historical El 5 formulations and near-duplicates as described in
Section 5.2.
18
-------
Table 6-2. FuelFormulationID numbering scheme.
Shirling liiellorimihilionl 1)
(silemlsir
Yesir
K10
i:o
i:i5
1990-2013
1000, 2000
2014
3000
3300
3600
2015
4000
4300
4600
2016
5000
5300
5600
2017
6000
6300
6600
2018
7000
7300
7600
2019
8000
8300
8600
2020+
9000
9300
9600
7. Diesel, CNG, and E85
For these fuels, we do not use regulatory compliance data as the source of the fuel properties.
Data sources for each fuel are discussed in the subsections below.
7.1 Diesel
MOVES3 uses two fuel properties when computing emissions from diesel: sulfur and biodiesel
(methyl ester) content. These values are summarized by calendar year in Table 7-1. The model
applies the same formulation across all regions for a given fuel type and calendar year.
Table 7-1. Onroad diesel sulfur and biodiesel contents in MOVES across all fuel regions.
Year
SuH'iir level, ppm
liioriiesel. v»l%
1990
1000
0
1999-2006
130
0
2007-2010
6
0
2011+
6
3.4
National average diesel sulfur levels in survey data collected by the Alliance for Automotive
Innovation17 indicate that an average level of 6 ppm is representative of years 2007 and later
(Figure 7-1). A review of individual locations suggests there is very little regional variation, with
10th-90th percentile ranges spanning 4-10 ppm during the same period.
19
-------
7
5
CL
I 6
M
00
5
4
2007 2009 2011 2013 2015 2017 2019
Winter Summer
f #
0
t
Avg = 6
ppm
Alliance of Automotive Innovators North American Fuel Survey
Figure 7-1. National average diesel sulfur level from market survey data.
For onroad diesel, we assume that conventional non-ester diesel (fuelSubtypelD 20) constitutes
100 percent of the market share in all fuel regions for 1990 through 2010 calendar years.
Beginning in 2011, MOVES3 assumes B3.4 (3.4 vol% biodiesel; fuelSubtypelD 21) constitutes
100 percent market share for the nation. The national average biodiesel blend levels were
computed using EIA data for biodiesel and transportation distillate consumption.20 Figure 7-2
shows the national blend level trend over the past decade, where blend levels have varied
according to blending mandates and market forces.
20
-------
3.4 vol%for 2011+
Figure 7-2. National average biodiesel blend level computed from EIA Monthly
Energy Review data. Shows MOVES3 blend level for years 2011 and later.
The year 2011 was selected as this was when the national average blend level first surpassed 1
vol%. As we lack consistent and reliable data on biodiesel across the country, using a national
average blend level is a reasonable simplifying assumption that is consistent with aggregate
usage figures from EIA and other sources. The fuelFormulation table also contains biodiesel
blend levels of 0, 5, and 20 vol%, which users may specify as alternatives.
The energy content of biodiesel is set at 43.061 (KJ/g) to represent B5 fuel. The density, energy
and carbon content of the fuels are based on aggregate values that are constant across fuel types
and fuel subtypes as documented in the MOVES GHG and energy report.21
7.2 Compressed Natural Gas (CNG)
MOVES assumes that CNG (fuelSubtypelD 30) used in onroad vehicles has a sulfur content of
7.6 ppm based on a CNG transit bus study documented in the MOVES fuel effects report.8
7.3 High-level Ethanol Blends (E85)
As described in the Population and Activity technical report, MOVES models a subset of the
vehicle fleet as "flexible fueled vehicles" (FFVs).22 These vehicles may operate on gasoline or
E85 fuel as indicated in the FuelUsageFraction table. For MOVES3, we set the E85 usage
fraction to 1.78% for all years and regions starting 2010 and later. This value was derived from
AEO2014 during the analysis described above in Section 5.1, and represents the fraction of total
fuel used by FFVs that is E85.
21
-------
The sulfur level of E85 is set to 8 ppm in all years, based on dilution of 30 ppm Tier 2 gasoline
with 74 vol% ethanol, consistent with the ethanol value in the fuelFormulation table. Benzene
was set to 0.16 vol% using the same computation with the national average limit of 0.62 vol%.
The MOVES algorithms used to model the emissions from FFVs using E85 are described in the
Fuel Effects technical report.8 The E85 algorithm uses the ElOFuelProperties table, which
includes representative E10 fuel formulations for every fuel region, calendar year, and month.
The table also stores national average E10 fuel properties as RegionID 0, which are used for
national scale runs using pre-aggregation in the advanced features. The E10 fuel properties for
RegionID 0 are calculated as an average of the E10 fuel formulations stored in the fuelSupply
table by calculating a weighted average of the E10 fuels assigned to different regions using their
respective share of light-duty vehicles VMT (stored in the zoneRoadType table) and E10 fuel
marketshare (in the fuelSupply table).
8. Nonroad Fuel Supply
In MOVES3, the nonroad gasoline fuel supply is identical to the onroad fuel supply except that it
contains no fuels with ethanol content over 10.5 vol%.
For nonroad diesel, MOVES includes two types of diesel: nonroad (fuelTypelD 23) and marine
diesel (fuelTypelD 24). The only difference between them is sulfur content, as shown in Table
8-1. These values represent a walk-down in sulfur to meet regulatory requirements based on what
was known about refining, consumption, and credit trading patterns.23 The final sulfur level of 6
ppm in nonroad reflects the end of the phase-in period and merger of onroad and nonroad
refinery products, thus nonroad fuel attains the sulfur level observed in onroad fuel. Survey data
suggests there is little variation in diesel sulfur levels across the continental US, so the 6 ppm
level is applied nationwide to nonroad as in onroad fuel. The locomotive/marine fuel type is
assumed to remain separate from onroad/nonroad, and thus, the final sulfur level is left at the 15
ppm regulatory level. Note that MOVES does not model locomotives or commercial marine
vessels, so these fuels apply to railroad support equipment and recreational marine.
Nonroad CNG and liquified petroleum gas (LPG, fuelSubtypelD 40) sulfur levels are 7.6 ppm
for all years, consistent with the onroad CNG sulfur level. Other properties of nonroad fuels are
shown in Table 8-2.
22
-------
Table 8-1. Nonroad diesel sulfur content in MOVES3 (ppm wf).
Calendar Year
Nonroad
Marine
1999 and earlier
2284
2640
2000
2284
2640
2001
2284
2635
2002
2284
2637
2003
2284
2637
2004
2284
2637
2005
2284
2637
2006
2242
2588
2007
1139
1332
2008
351
435
2009
351
435
2010
165
319
2011
32
236
2012
6
124
2013
6
44
2014 and later
6
15
23
-------
Table 8-2. Onroad and Nonroad fuel properties by SubtypelD in MOVES3.
Subl> pell)
Ttpell)
Siib(\pel)ese
Pel i'o leu in
l-'raelion
l-'ossil
l-'raelion
(arbon
C on lent
(li/k.l)
r.nertij
( onlenl
(M.l/kii)
l-uel
l)ensil\
(ii/iiah
10
1
Conventional Gasoline
0.95
1
0.0196
43.488
2819
11
1
Reformulated Gasoline
(RFG)
0.95
1
0.0196
42.358
2819
12
1
Gasohol (E10)
0.94
0.94
0.0196
41.762
2819
13
1
Gasohol (E8)
0.945
1
0.0196
42.1
2819
14
1
Gasohol (E5)
0.945
1
0.0196
42.605
2819
15
1
Gasohol (El5)
0.94
0.94
0.0196
40.92
2819
20
2
Conventional Diesel
Fuel
1
1
0.02
42.791
NULL
21
2
Biodiesel (BD20)
0.81
0.81
0.0199
41.738
NULL
22
2
Fischer-Tropsch Diesel
(FTD100)
0
1
0.0205
43.247
NULL
23
23
Nonroad Diesel Fuel
1
1
0.02
43.306
3167
24
24
Marine Diesel Fuel
1
1
0.02
43.306
3167
30
3
Compressed Natural
Gas (CNG)
0
1
0.0161
48.632
500
40
4
Liquefied Petroleum
Gas (LPG)
0
1
0.0161
46.607
1923
50
5
Ethanol
0.26
0.26
0.0194
26.592
NULL
51
5
Ethanol (E85)
0.26
0.26
0.0194
29.12
NULL
52
5
Ethanol (E70)
0.46
0.46
0.0194
31.649
NULL
90
9
Electricity
0.02
0.87
0
NULL
NULL
9. Updates to Fuel Wizard Factors for Ethanol Blending
Since MOVES2014a, the model software has included a "Fuel Wizard" tool to help users create
realistic gasoline blends that are not present in the default fuel supply. In the real world,
specifying a limit or range for one fuel property can produce collateral changes in other
properties as a result of impacts on refining and blending processes. For example, reducing sulfur
may affect aromatics and olefins, or allowing higher RVP may affect distillation points.
The Fuel Wizard allows a user to input a change in one of three gasoline properties (RVP,
ethanol level, or sulfur level) and the tool estimates secondary fuel property changes using data
from refinery modeling runs. This allows for the full emissions impact of proposed fuel changes
24
-------
(as part of state or local programs) to be estimated properly, including the subsequent effects of
non-regulated fuel property changes. The Fuel Wizard is currently capable of creating fuels with
ethanol variations between EO and E15, sulfur from 5-80 ppm, and RVP from 5-14 psi.
The Fuel Wizard is used in conjunction with the county data manager in the MOVES graphical
user interface (GUI). More information on when users should use the Fuel Wizard is provided in
our technical guidance.24
The Fuel Wizard adjustments are stored in the fuelWizardFactors table and are applied in an
additive way (not multiplicative factors). In MOVES2014 (a and b), the adjustments used in the
Fuel Wizard were derived from refinery modeling done as part of the Tier 3 rulemaking analysis,
and are the same as those used in developing a portion of the default fuel supply (see Section
5.1).
However, in MOVES3, we updated the factors for ethanol blending using the results of recent
refinery modeling work conducted by MathPro, Inc. Data on petroleum and biofuels markets
from EIA and other sources was used to develop assumptions and market inputs to the model.
The refinery modeling analysis was conducted at the PADD level for summer and winter seasons
after developing and validating a calibration case to ensure key outputs aligned with actual
observed performance of the refining sector. A business-as-usual scenario was run for calendar
year 2016 as a reference case, followed by two "low-biofuel" scenarios where the only ethanol
use was in E10 in RFG areas. The two low-biofuel scenarios differed in the price and availability
of alkylation feedstocks, which affected the relative proportions of aromatics and iso-paraffins
used to make up octane if a large volume of ethanol were removed from the gasoline supply.
The updated Fuel Wizard factors in MOVES3 for ethanol blends, shown in Table 9-1, were
derived from the "Low-Biofuel #2" case based on the favorable economics of that scenario as
described by MathPro. The increase in refinery demand for isobutane and butane in this scenario
can easily be met by the large increases in U.S. isobutane and butane production associated with
increasing U.S. light crude oil production.25 The MathPro report is available in Appendix B, and
additional details on the refinery model development and calibration are available in the docket
of the 2020 Renewable Volumes rule.26 The specific factors shown in Table 9-1 were computed
by taking the difference between the "Low-Biofuel #2" and "Reference" cases in Table 1 of
Appendix B, except for RVP, which were derived from the API blending study as discussed in
Section 6.1.19
An alternative method suggested by some stakeholders for estimating the effect of ethanol
blending on gasoline properties involves looking back over a period when ethanol use was
increasing steeply, for example comparing 2006 to 2016. While this approach may seem to have
the advantage of utilizing direct observations, the problem is that other market changes were also
at play and their impacts are overlaid upon and confound the effects of increasing ethanol use.
Such changes include: phase-in of Tier 2 sulfur and MSAT2 benzene control regulations; the
phase-out of MTBE; rising crude oil prices and the associated expansion of domestic oil and gas
production, including light hydrocarbons, which favored the use of alkylation as an octane source
over reforming; and reduced demand for gasoline hydrocarbons relative to diesel, which caused
many refiners to swing a portion of the heavier end of gasoline into distillate products. Given
these complexities, we chose to utilize a well-established refinery model where the impacts of
ethanol blending could be isolated from several other changes in petroleum markets.
25
-------
Table 9-1. Updated Fuel Wizard factors for ethanol blends (additive changes for ETOH Change shown).
ETON ( hanjic
RVP
SI I.I
AKOM
ou:i
m:\z
1:200
K300
150
T')0
Yol% lo \ ol%
psi
ppm
\ol%
\oi%
\oi%
\ ()l%
\ ol%
l)o«. 1
l)o«. 1
E0 to E10 Winter
0.80
-1.7
1.7
-0.01
6.4
0.2
-23.8
-0.63
E0 to E10 Summer
0.90
-2.2
1.6
7.0
-0.2
-26.0
0.63
E10 to E15 Winter
-0.15
-0.85
0.85
-0.01
3.2
0.1
-11.9
-0.32
E10 to E15 Summer
-0.15
-1.1
0.80
3.5
-0.1
-13.0
0.32
Table 9-2 and Table 9-3 show the Fuel Wizard factors for changes resulting from adjusting RVP
and sulfur level, respectively. These values were not updated in MOVES3 and are the same as in
MOVES2014b.
Table 0-2. Acl jiisimoiH faclors for lower RVP blends laddKixe adjiislnicnis per psi)
l)I.S( KIP I ION
KVP
SI I.I
AKOM
OI.I.I
M'.NZ
1.200
1.300
150
190
psi
ppm
Vol%
Vol%
Vol%
Vol%
Vol%
Deg.F
Deg.F
Volatility adj.
-1.00
0
0
0
0
-1.26
-0.50
2.57
2.27
Table (>-3. Adjustment factors for sulfur blends below 30 ppm (additive adjustments per ppm)
DESCRIPTION
KM'
SI IT
AKOM
Ol. IT
lil-'.N/.
1.200
1.300
150
190
psi
ppm
Vol%
Vol%
Vol%
Vol%
Vol%
Deg.F
Deg.F
Sulfur fuel adj
0
-1.00
-0.032
0
0
0
0
0
0
As with the compliance batch data, the distillation values in the refinery modeling output are in
terms of E200 and E300 and need conversion to T-values. Since the Fuel Wizard uses a single
entry for each parameter, the multi-region T50 correlations described in Section 5.2 for
conventional gasoline (CG) were simplified as shown in Figure 9-1, resulting in Equation 9-1.
For the E300 to T90 conversion, Equation 5-5 was used directly.
For simplicity, the E10 to E15 factors in Fuel Wizard were made by taking 50 percent of the E0
to E10 change except for RVP, where the value was derived from the 2010 API ethanol blending
study as described in Section 6.1. Note that the Fuel Wizard is not currently set up to perform
splash blends. Users wanting to model emissions on El 5 splash blends, a more likely scenario
for the foreseeable future, should use the formulations available in the default supply, as
explained in Section 6.1.
T50 = 377.34 - 3.7159 x E200 Equation 9-1
26
-------
Simplified T50 Correlation for Fuel Wizard (CG E10)
240
220
200
^180
DJD
CU
~o
° 160
H-
140
120
100
*i ufa*
a "v. .n/
y = -3.7159x +377.34
M* |#
R2 = 0.8571
40
45
50 55 60
E200 (vol%)
65
70
Figure 9-1. Simplified T50 by E200 correlation for use in MOVES3 Fuel Wizard.
27
-------
10. References
1 USEPA(2015). U.S. Environmental Protection Agency Peer Review Handbook. EPA/100/B-15/001.
Prepared for the U.S. Environmental Protection Agency under the direction of the EPA Peer Review
Advisory Group. Washington, D.C. 20460. October 2015.
https://www.epa.gov/sites/production/files/2020-
08/documents/epa_peer_review_handbook_4th_edition.pdf.
2 USEPA (2017). Fuel Supply Defaults for Regional Fuels and Fuel Wizard Tool in MOVES201X - Draft
Report. Draft report and peer-review documents. Record ID 328850. EPA Science Inventory. September
2017. https ://cfpub.epa.gov/si/si_public_record_report.cfm?dirEntryId=328 85 0.
3 USEPA (2020). Fuel Supply Defaults: Regional Fuels and the Fuel Wizard in MOVES3 - Draft Report.
Draft report and peer-review documents. Record ID 347139. EPA Science Inventory. July 2020.
https://cfpub.epa.gov/si/si_public_record_report.cfin?dirEntryId=347139.
4 OPIS/STALSBY Petroleum Terminal Encyclopedia. OPIS/STALSBY Wall, NJ: 2013.
5 US Energy Information Administration. US Energy Mapping System.
http://www.eia.gov/state/maps.cfm. 2020.
6 USEPA (2020). Gasoline Reid Vapor Pressure, https://www.epa.gov/gasoline-standards/gasoline-reid-
vapor- pressure.
7 Energy Policy Act of 2005, lifting of RFG oxygenate requirement
8 USEPA (2020). Fuel Effects on Exhaust Emissions from Onroad Vehicles in MOVES3. EPA-420-R-20-
016. Office of Transportation and Air Quality. US Environmental Protection Agency. Ann Arbor, MI.
November 2020. https://www.epa.gov/moves/moves-technical-reports.
9 USEPA. Fuel Trends Report: Gasoline 2006-2016. EPA-420-R-17-005. October 2017.
https://nepis.epa.gov/Exe/ZyPDF.cgi?Dockey=P100T5J6.pdf
10 USEPA. Fuel Trends Report: Gasoline 1995-2005. EPA420-R08-002, January 2008.
https://nepis.epa.gov/Exe/ZyPDF.cgi?Dockey=P100B3YI.pdf
11 US Energy Information Administration. Annual Energy Outlook 2014 (Early Release). DOE/EIA-
0383ER (2014). December 2013.
12 US Energy Information Administration. Annual Energy Outlook 2014. DOE/EIA-0383(2014). April
2014.
13 US EPA. Control of Air Pollution from Motor Vehicles: Tier 3 Motor Vehicle Emission and Fuel
Standards Final Rule. Regulatory Impact Analysis, Chapter 7.1 Impacts of the Rule on Emissions and Air
Quality-Criteria and Toxic Pollutant Emission Impacts. EPA-420-R-14-005. March 2014.
14 USEPA Office of Transportation and Air Quality. Assessing the Effect of Five Gasoline Properties on
Exhaust Emissions from Light-Duty Vehicles Certified to Tier 2 Standards: Analysis of Data from EPAct
Phase 3 (EPAct/V2/E-89): Final Report. EPA-420-R-13-002. Assessment and Standards Division, Ann
Arbor, MI. April 2013. http://nepis.epa.gov/Exe/ZyPDF.cgi?Dockey=P 100GA0V.txt
15 US Energy Information Administration. Petroleum Supply Annual 2004. Volume 1, Tables 4, 6, 8, 10,
12, 32. Volume 2, Table 20.
16 US EPA. Control of Hazardous Air Pollutants from Mobile Sources, Final Rule. Regulatory Impact
Analysis, Chapter 10. EPA420-R-07-002, February 2007.
https://nepis.epa.gov/Exe/ZyPdf.cgi?Dockey=P1004LNN.PDF
17 North American Fuel Survey, Alliance for Automotive Innovation, Washington, DC, 20001.
18 US EPA. Public Data on Gasoline Fuel Quality Properties. 2020. https://www.epa.gov/fiiels-
registration-reporting-and-compliance-help/public-data-gasoline-fiiel-quality-properties
19 American Petroleum Institute. Determination of the Potential Property Ranges of Mid-Level Ethanol
Blends, Final Report. April 2010.
28
-------
20 US Energy Information Administration, Monthly Energy Review, Tables 3.7c and 10.4. August 2020
edition.
21 USEPA (2020). Greenhouse Gas and Energy Consumption Rates for Onroad Vehicles in MOVES3.
EPA-420-R-20-015. Office of Transportation and Air Quality. US Environmental Protection Agency. Ann
Arbor, MI. November 2020. https://www.epa.gov/moves/moves-technical-reports.
22 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.
23 See 40 CFR Part 80, Subpart I: Motor Vehicle Diesel Fuel; Nonroad, Locomotive, and Marine Diesel
Fuel; and ECA Marine Fuel.
24 USEPA (2020). MOVES3 Technical Guidance: Using MOVES to Prepare Emission Inventories for
State Implementation Plans and Transportation Conformity. Ann Arbor, MI, Office of Transportation and
Air Quality. US Environmental Protection Agency. November 2020.
25 US Energy Information Administration. Petroleum and Other Liquids, U.S. Field Production of
Isobutane - 1990 to 2019. Retrieved October 2020.
https://www.eia.gov/dnav/pet/hist/LeafHandler.ashx?n=PET&s=M_EPLLBAI_FPF_NUS_MBBLD&f=A
26 Modeling aNo-RFS Case; ICF Incorporated; Work Assignment 0,1-11, EPA contract EP-C-16-020;
July 17, 2018. Included in the docket for Renewable Fuel Standard Program: Standards for 2020 and
Biomass-Based Diesel Volume for 2021, Response to the Remand of the 2016 Standards, and Other
Changes, document ID EPA-HQ-OAR-2019-0136-2147.
29
-------
APPENDIX A: MOVES3 FUEL REGION MAPS, 1990-2020
The following 23 pages present 50-state maps of summertime MOVES fuel regions by county. Puerto Rico and
US Virgin Islands aren't shown but have the same formulation as base region 1 (regionID = 100000000) in all
years. While RVPs are generally higher in winter than summer, the MOVES regionID values are identical for
both seasons.
30
-------
1990 - July
Fuel Regions (fuelRegionID)
100000000
20COOOODO
300000000
400000000
-------
1999- July
U)
bo
Fuel Regions (fuelRegionID)
| 1ocoooooc 1
100010000 I
178000000
I 178010000 I
I 200000000 |
j 270000000 I
278000000
I 278010000 j
I 300000000 I
370000000 I
370010000
I 400000000
500000000
578000000
678000000
| 700000000
1170011000
I 1270011000
I 1370011000
1470011000
1570011000
-------
2000 - July
-------
2001 - July
Fuel Regions (fuelRegionID)
j 400000000
I 600000000
I 578000000
I 600000000
I 678000000
I 700000000
1170011000
1270011000
1370011000
1470011000
1570011000
-------
2002 - July
U)
Wr a
.^av-»v^4
Fuel Regions (fuelRegionID)
100000000 I
100010000 I
170000000 I
178000000 I
178010000
200000000 I
1 270000000 |
278000000 I
278010000 I
I 300000000 I
370000000
1 370010000
J 400000000
I 500000000
578000000
I 600000000
78000000
700000000
1170011000
1270011000
1370011000
1470011000
1570011000
-------
2003 - July
U)
On
mmmmt
Fuel Regions (fuelRegionID)
100000000 I
100010000 I
170000000
I 178000000 I
178010000
I 200000000 I
I 270000000
278000000 I
278010000 I
I 300000000 I
370000000
I 370010000
400000000
500000000
578000000
678000000
I 700000000
1170011000
I 1270011000
1370011000
1470011000
1570011000
-------
2004 - July
-------
2005 - July
U>
00
Fuel Regions (fuelRegionID)
100000000
100010000 I
170000000 [
| 178000000 I
178010000
I 200000000 I
270000000
278000000 I
] 278010000 I
300000000 I
| 370000000 I
1 370010000
400000000
I 500000000
578000000
| 600000000
678000000
700000000
1170011000
I 127001100Q
1370011000
I 1470011000
1570011000
-------
2006 - July
-------
2007 - July
JL ¦ ^
*
Fuel Regions (fuelRegionID)
I 1OOOOOOOO n
100010000 I
I 170000000
178000000 I
178010000
200000000 I
] 270000000 I
278000000 |
[ 278010000 I
I 300000000 I
370000000
370010000
I 400000000
I 500000000
578000000
I 600000000
678000000
I 700000000
1 1170011000
| 1270011000
1370011000
1470011000
1570011000
-------
2008 - July
Fuel Regions (fuelRegionID)
I 400000000
I 500000000
J 578000000
I 600000000
678000000
| 700000000
1170011000
I 1270011000
1370011000
I 1470011000
1570011000
-------
2009 - July
Fuel Regions (fuelRegionlD)
-------
2010 - July
IH
Fuel Regions (fuelRegionID)
loooooooo y
I 100010000 I
170000000
173000000 I
178010000
, 200000000 I
I 270000000 [
1 278000000 H
278010000 I
I 300000000 I
| 370000000
370010000
I 400000000
I 500000000
578000000
I 600000000
678000000
| 700000000
1170011000
I 1270011000
I 1370011000
| 1470011000
1570011000
-------
2011 - July
Fuel Regions (fuelRegionID)
-------
2012-July
IH
Fuel Regions (fuelRegionID)
loooooooo y
I 100010000 I
170000000
173000000 I
178010000
, 200000000 I
I 270000000 [
1 278000000 H
278010000 I
I 300000000 I
| 370000000
370010000
I 400000000
I 500000000
578000000
I 600000000
678000000
| 700000000
1170011000
I 1270011000
I 1370011000
| 1470011000
1570011000
-------
2013-July
Fuel Regions (fuelRegionlD)
-------
2014-July
-------
2015-July
00
msm
Fuel Regions (fuelRegionID)
I 400000000
I 100010000 500000000
170000000 I 578000000
178000000 I
I 178010000
200000000 I
j 270000000
778000000
278010000 I
300000000 I
370000000
370010000
678000000
700000000
1170011000
I 1270011000
1370011000
1470011000
1570011000
-------
2016-July
Fuel Regions (fuelRegionID)
100000000 ¦ 400000000
100010000 I
178000000
178010000 I
200000000 |
270000000 I
278000000
1 278010000 I
300000000
370000000 I
370010000
500000000
578000000
678000000
I 700000000
1170011000
I 1270011000
I 1370011000
I 1470011000
1570011000
-------
2017-July
-------
2018-July
Fuel Regions (fuelRegionID)
-------
2019- July
Ut
to
¦¦SliSnri^ ?*£«*&»!
Fuel Regions (fuelRegionID)
100000000 I
100010000
178000000 I
178010000 [
200000000 I
270000000
27SDOOOOO
300000000 I
370000000 I
370010000
400000000
I 500000000
578000000
678000000
| 700000000
1170011000
I 1270011000
1370011000
I 1470011000
1570011000
-------
2020 - July
Fuel Regions (fuelRegioniD)
-------
APPENDIX B: AVERAGE GASOLINE FORMULATION TRENDS IN
MOVES, 1990-2020
This plot series show average fuel property values across all E10 fuel formulations by calendar year.
RVP
sulfurLevel
200
150
100
50
ETOHVolume
aromaticContent
10.0
24
9.9
22
9.8
20
\
oiefinContent
benzeneContent
0.9
0.8
0.7
e200
e300
87
86
84
83
48
T50
T90
330
325
320
315
210
200
190
180
170
2000 2004 2008 2012 2016 2020
2000 2004 2008 2012 2016 2020
season April+Oct May-Sep Nov-March
54
-------
APPENDIX C: REFINERY MODELING RESULTS USED
TO DEVELOP FUEL WIZARD FACTORS
55
-------
Addendum to No-RFS Study: Analysis of Low-Biofuel Case
Final Report
MathPm
ANALYSIS OF THE EFFECTS OF LOW-BIOFUEL USE ON
GASOLINE PROPERTIES
AN ADDENDUM
TO
THE ICF "NO-RFS" STUDY
Prepared for
ICF Incorporated, LLC
Under
EPA Contract No. EP-C-16-020
By
MathPro Inc.
June 7, 2019
MathPro Inc.
P.O. Box 34404
West Bethesda, Maryland 20827-0404
301-951-9006
June 7, 2019
56
'MathPro
-------
Addendum to No-RFS Study: Analysis of Low-Biofuel Case
Final Report
1. Study Objectives and Scope
The work discussed in the main body of this report -- the No-RFS Study - was conducted to
estimate the likely volumes of biofuels (primarily ethanol, but also bio-diesel and renewable
diesel) that would be used if the RFS2 program were no longer in effect.
The analysis covered in this Addendum addresses a supplementary case - a "Low-Biofuels"
case - in which ethanol is used only in RFG and there is no use of bio/renewable diesel in
ULSD. This is similar to the situation in 2005/2006, when ethanol replaced MTBE in the RFG
pool, essentially eliminating its use in the CG pool, and the use of biodiesel was minimal. The
primary focus of the analysis is to assess the potential effects on finished gasoline properties
(which would have consequent effects on particulate, VOC, NOx and toxics emissions) that
could result from substantially reduced ethanol blending, by comparing a gasoline pool that is
nearly all E10 to one that is predominantly E0.
The analysis involved developing a new Low-Biofuel case using the same refinery LP model as
was used in the No-RFS study and comparing the results of the Low-Biofuel case to those of
the Reference case developed in the No-RFS study. That Reference case represents PADD-
level refinery operations modeled for 2020 based on projections reported in ElA's 2017 Annual
Energy Outlook and the RFS remaining in force.
2. Technical Approach
Pursuant to the Work Assignment, the refining analysis conducted to assess the new Low-
Biofuels case is an extension of the refinery modeling described in the main report. More
specifically, the Low-Biofuels cases (for each PADD and season) are based on the
corresponding 2020 Reference cases in the main study, but with the following modifications
regarding the use of biofuels (ethanol and bio/renewable diesel):
> Conventional and low-RVP finished gasolines no longer would be blended with ethanol, i.e.,
they would be E0. Conventional and low-RVP gasolines together account for about 70% of
gasoline production for the domestic market.
> Federal RFG and California RFG would continue to be blended with 10 vol% denatured
ethanol and would continue to remain in compliance with applicable standards (including the
requirements of California's Predictive Model). The Federal and California RFG programs,
which are separate from and were promulgated before the RFS program, are assumed to
remain in effect, along with their requirements for oxygenate blending.
> There would be no (1) post-refinery, splash-blending of ethanol into finished E0 gasoline; (2)
spill-over of RFG into non-RFG areas (not already accounted for in the volumes specified in
the Reference cases); or (3) use of higher ethanol blends, including E85. These
assumptions simplify the modeling in terms of setting gasoline volumes and ethanol use.
> No biodiesel or renewable diesel would be blended in ULSD (or other distillate products).
This is based on the assumption that, without the RFS mandates, such bio-fuels would be
too expensive to be used as fuel extenders.
We also assume that the MSAT2 and Tier 3 gasoline sulfur standards, and the impending
MARPOL sulfur standards, would have been implemented by 2020 in the absence of the RFS2
program.
June 7, 2019
57
"Math
-------
Addendum to No-RFS Study: Analysis of Low-Biofuel Case
Final Report
The assumptions regarding reduced use of ethanol and no production of E85 required some
adjustments be made to finished gasoline output in the Low-Biofuel cases. We assumed that
volumes of conventional and low-RVP EO in the Low-Biofuel cases would be roughly 3% lower
than the corresponding volumes of E10 in the Reference cases to account for the lower energy
density of the ethanol removed from the gasoline pool. We also adjusted the volume of
conventional, regular-grade EO in the Low-Biofuels cases to include small volumes of "clear"
gasoline and E85 (energy-adjusted) produced for the domestic market in the Reference cases.
These volume adjustments bring the total energy content (in BTU) of gasoline supplied in the
Low-Biofuel cases in closer alignment with that supplied in the Reference cases.
Refinery production of ULSD was set at the same levels in the Low-Biofuel cases as in the
corresponding Reference cases, because (1) the energy contents of hydrocarbon ULSD and
bio/renewable diesel are similar; and (2) we assumed in the Reference cases that
bio/renewable diesel was blended in ULSD at the refinery (rather than at terminals or large retail
outlets). In practice, with "no bio/renewable diesel blending", refineries would produce greater
volumes of hydrocarbon-only ULSD at the refinery, but end-use sales of ULSD would remain
about the same.
These are the major structural changes incorporated in the refinery modeling for all of the Low-
Biofuel cases. However, a number of other modeling issues surfaced when conducting the
refinery modeling. Some of the issues pertained to all regions (PADDs) modeled, while others
were specific to particular regions.
The most important of these issues are:
> Setting RVP constraints on finished conventional gasoline and low-RVP gasoline.
> Response of gasoline exports to a large reduction in ethanol use in the domestic gasoline
pool.
> Constraints on gasoline properties of RFG and export gasoline not "directly" affected by the
removal of ethanol from conventional and low-RVP gasoline.
> "Harmonization" across seasons of new investments in refining processes needed to
replace the gasoline volume and octane "lost" through displacement of ethanol.
> Setting the potential availability of butanes across regions to support investments in
additional alkylation capacity, an alternative to reforming capacity, as a means of replacing
the "lost" octane in the absence of ethanol.
The following sections discuss the resolution of these issues in the Low-Biofuel case.
2.1 RVP Constraints on Finished Conventional and Low-RVP Gasoline
The posited shift of E10 conventional gasoline and low-RVP gasoline to EO would affect the
RVP of finished gasoline in a number of ways.
In the summer, most conventional E10 gasoline (about 4.5 MM b/d) now and in the Reference
cases qualifies for the ethanol 1 psi RVP waiver. When this gasoline pool shifts to EO, the RVP
of finished gasoline would decline by about 1 psi - from about 9.7 psi to 8.7 psi. However, a
small volume of conventional gasoline sold in upstate New York and Maine (about 180 K b/d)
that does not qualify for the RVP waiver would continue to meet the 8.7 psi RVP standard. Low-
RVP gasoline now sold in parts of Pennsylvania, Eastern Texas, and El Paso (about 470 K b/d)
does not qualify for the ethanol RVP waiver and, therefore, would continue to meet RVP
standards of 7.8, 7.8, and 7.0 psi, respectively. The remaining low-RVP gasoline (about 720 K
June 7, 2019
58
"Math
-------
Addendum to No-RFS Study: Analysis of Low-Biofuel Case
Final Report
b/d or a little over 50% of all low-RVP gasoline sales) qualifies for the ethanol RVP waiver and
so, when converted from E10 to E0, would see a reduction in RVP of about 1 psi - from a
weighted average of about 8.6 psi to about 7.6 psi.
In the winter, the average finished RVP of conventional gasoline of about 13.9 psi was
estimated using data from the Alliance gasoline surveys. This includes the uplift in RVP
associated with E10 ethanol blending at terminals. Backing out the estimated RVP uplift of
about 0.8 psi (based on the estimated relationship between finished gasoline RVP and ethanol's
RVP uplift), indicates that E10 BOBs produced by refineries had an average RVP of about 13.1
psi. Our understanding is that refineries currently produce winter BOBs with RVPs that are
compliant with ASTM winter RVP standards (before being blended with ethanol at terminals).
Hence, we assume that the "calculated" average BOB RVP of 13.1 psi (with ethanol "backed
out") is compliant with ASTM winter RVP standards and would be maintained for finished E0 in
the Low-Biofuel cases.
2.2 Volume of Gasoline Exports
A shift in conventional gasoline produced by U.S. refineries for the domestic market from E10 to
E0 would: (1) reduce ethanol use by about 600 K b/d (equivalent to about 400 K b/d of
hydrocarbon gasoline on an energy-adjusted basis); and (2) remove ethanol's octane (AKI)
contribution to the finished conventional gasoline pool (the octane of E10 hydrocarbon BOBs
averages about 31/4 octane numbers less than the octane of finished gasoline).
The U.S. refining sector potentially could make up the volume and octane loss by: (1) investing
in new, gasoline-related refining process capacity and increasing crude oil throughput; (2)
shifting some gasoline exports (which are forecast to average almost 800 K b/d annually in
2020) to the domestic market to offset the volume loss (but not the loss in octane); or (3)
pursuing some combination of the two. However, we elected to hold gasoline export volumes
constant in this analysis for a number of reasons.
First, PADD 3 is the primary source of gasoline exports - accounting for almost 90% of gasoline
exports forecast for the Reference case - with most of the remainder from PADD 5. This
indicates that shifting export gasoline into the domestic pool is an option available mostly to
refineries only in PADD 3 (and within that region to refineries situated along the Gulf Coast).
Second, positing certain volume reductions in gasoline exports, without pursuing supporting
refinery modeling, could be viewed as arbitrary. Third, conducting an extensive series of
supporting modeling runs to attempt to identify some sort of conceptual global and inter-PADD
"equilibrium" regarding gasoline production and supply is beyond the scope of the study and the
capabilities of the modeling tool used for the analysis.
Thus, in this analysis, the changes estimated in gasoline properties reflect the investments in
new refining process capacity and changes in refining operations undertaken in each PADD to
replace the volume and octane loss from reduced ethanol use ~ without recourse to shifting
export gasoline into the domestic market.
2.3 Constraints on Gasoline Properties of RFG and Export Gasoline
RFG (about 2.6 MM b/d) and export gasolines (almost 800 K b/d) are not directly affected by the
posited shift from E10 to E0 for conventional and low-RVP gasolines. Their properties could be
affected indirectly as refineries that produce a mix of these gasoline types adjust their
operations and make investments in new process capacity to offset the loss in volume and
octane associated with reductions in the use of ethanol. However, the extent to which their
properties might change is difficult to ascertain for a number of reasons.
June 7, 2019
59
"Math
-------
Addendum to No-RFS Study: Analysis of Low-Biofuel Case
Final Report
Aggregate refinery models can easily move gasoline blendstocks between different types and
grades of gasoline in ways that are not necessarily consistent with refinery blending practices.
This can result in gasoline properties returned by refinery models differing from reported
properties for specific grades or types of gasoline, even though properties for the "aggregate"
gasoline pool are reasonably close to reported properties. Thus, the "allocation" of gasoline
properties across gasoline types returned by refinery models can be misleading.
Most refineries in PADD 2 and refineries in the Pacific Northwest (which are included in the
aggregate refinery model representing PADD 5) do not produce RFG or export gasoline and do
not have the option of shifting gasoline blendstocks (and properties) across different types of
gasolines. (Only a handful of refineries in PADD 2, predominantly in Illinois and Indiana,
produce RFG, and California refineries produce relatively small volumes of conventional
gasoline.) About half of the refineries in PADD 3 do not produce RFG, but these tend to be the
smaller refineries, many of which are located inland. The remainder, mostly larger refineries
along the Gulf Coast, produce a mix of conventional gasoline and RFG. Refineries in PADD 1
also produce a mix of conventional gasoline and RFG. (No refineries in PADD 4 produce RFG
or export gasoline.) Thus, for PADD 2 and PADD 5, and partially for PADD 3, the options for
moving blendstocks between gasoline types are more constrained, in practice, than as
represented in aggregate refinery models.
Gasoline is exported primarily by refineries in PADD 3. Data are not available regarding which
refineries are active in the export market. It may well be that shifting blendstocks between
export gasoline and conventional gasoline is not an option available to many PADD 3 refineries,
particularly the smaller inland refineries.
In view of the inherent problems associated with "allocating" the effects on gasoline properties
and composition of a posited large reduction in ethanol use, we concluded that the most
appropriate analytic technique was to concentrate such effects in the combined conventional
and low-RVP gasoline pools - the gasoline types directly affected by the posited shift from E10
to EO. We did so by (1) constraining the aromatics and olefins content of RFG and export
gasoline in the Low-Biofuels cases to be the same as in the Reference cases1; and (2)
"reassigning" gasoline blendstocks and properties returned by the refinery modeling such that
changes in the gasoline composition and properties were completely concentrated in the EO
pool.2
2.4 Harmonization of Investments in Refining Process Capacity across
Seasons
1 The RVP, benzene content, and sulfur content of RFG and export gasoline already were specified. The refinery models
returned either no, or very low, "shadow values" for the constraints on aromatics and olefin content, which implies that such
constraints affected the "allocation" of properties among gasoline pools, but not materially the response (cost and investments in
process capacity) of the refinery models in moving from E10 to EO for conventional gasoline. On the other hand, constraining all
gasoline properties of RFG and export gasoline to their Reference case values, specifically E200 and E300 (the percent of
gasoline distilled off at 200°F and 300°F, respectively), generally resulted in the refining models returning small shadow values
for those properties, with associated minor increases in the cost of supplying gasoline. Elence, we allowed E200 and E300 to vary
in Low-Biofuel cases, which they did, to provide the refinery models with limited flexibility in producing multiple gasoline types
and grades.
2 Under this procedure, the EO pool meets the applicable standards (constraints) for octane, RVP, benzene content, and sulfur
content. Elowever, changes in aromatics content, olefins content, E200, and E300 are concentrated in the EO pool, as are changes
in refinery production of gasoline blendstocks.
June 7, 2019
60
"Math
-------
Addendum to No-RFS Study: Analysis of Low-Biofuel Case
Final Report
The refinery modeling for the Low-Biofuel case was conducted for the summer and winter
seasons, as in the main report, rather than annually, to account for seasonal differences in
gasoline RVP and ethanol's effects on gasoline RVP. The seasonal refinery models are "stand-
alone" models, in the sense that refinery process capacity added in one season in response to
reductions in ethanol blending (say for the summer) are not formally "linked" via computer code
to refinery operations in the other season (the winter). Thus, the initial profiles of investment in
new refining process capacity returned by the refinery models can differ substantially by season.
In most cases, initial investments in process capacity made for the summer season were only
partially adopted in the winter season. This implies that the models' investment profiles would
not be "optimal," as the required rate of return on investments in new process capacity would
not be met on an annual basis.
To address this issue, we conducted a series of modeling iterations in which we adjusted the
capital charges associated with the major gasoline upgrading processes (alkylation, pen/hex
isomerization, and reforming), generally up for the summer and down for the winter, until the
seasonal investment profiles were roughly consistent with the investments meeting the required
annual rate of return. This generally resulted in "final" investments in process capacity returned
by the models being (1) lower than "initial" investments for the summer and higher for the winter
and (2) fairly similar (with exceptions) across seasons. The procedure is imprecise, but it
moves the modeling results closer to what might result from a planning process that considers a
refinery's year round operations.3
2.5 Availability of Butanes to Support Alkylation Capacity
The projected growth in natural gas production and associated natural gas liquids production
forecast by EIA suggests that the supply of butanes (a feedstock in the production of alkylate)
could increase significantly for favorably situated refining centers, particularly those in PADDs 2
and 3, and possibly in PADD 4. This led us to consider two alternative modeling scenarios
regarding the potential response of the refining sector to a large scale reduction in the use of
ethanol.
In the Low-Biofuel #1 case, we: (1) limited the volume of purchased butanes to that forecast for
the Reference cases for each PADD; and (2) constrained FCC operations such that catalyst
use, which affects light olefin production (another feedstock to alkylation), was the same as in
the corresponding Reference case.4 With these constraints on refining operations, the volume
and octane deficit arising from the large reduction in ethanol use called out a mix of additional
refining capacity for alkylation, pen/hex isomerization, and reforming.
In the Low-Biofuel #2 case, for PADDs 2, 3, and 4 only, we: (1) priced iso-butane at $50/b5,
both in the summer and winter, and allowed unlimited purchases; and (2) allowed FCC
3 Initial modeling results for PADD 1 (for the summer only) indicated that moving from E10 to EO for conventional gasoline
would increase the incremental cost of producing gasoline beyond the likely delivered cost of gasoline from Europe or PADD 3.
Consequently, we reduced PADD 1 's production of premium gasoline for the summer, which reduced its incremental cost of
gasoline production, bringing it more in line with estimated delivered costs from PADD 3 (to which we assigned an additional
production volume of premium EO).
4 The refinery model is structured so that catalyst use can vary depending on internal refinery demand for light olefins (propylene
and butylenes that are inputs to alkylation).
5 This iso-butane price is roughly equal to the average of the spot prices reported by OPIS for PADDs 2 and 3 for iso-butane in
2016 times the ratio of the crude oil acquisition costs projected for 2020 to those reported for 2016. This likely overstates future
spot prices for iso-butane, because the recent increase in natural gas and LPG production appears to have reduced the prices of
LPGs relative to crude oil acquisition costs.
June 7, 2019
61
"Math
-------
Addendum to No-RFS Study: Analysis of Low-Biofuel Case
Final Report
operations to modify catalyst use and produce additional light olefins as needed by alkylation
operations. This resulted in significant increases in iso-butane purchases and new alkylation
process capacity. On the other hand, process capacity additions of pen/hex isomerization
declined somewhat and almost no new reforming capacity was added. This suggests there
could be an alternative, "less-reforming-heavy," path to replacing ethanol, at least in regions
situated to take advantage of forecast future increases in butane availability.
We did not pursue the second option (the Low-Biofuel #2 case) in: (1) PADD 1, because of the
lack of investment incentives to replace lost volume and octane in that region due to the
availability of low-cost gasoline supplies from Europe and the Gulf Coast; and (2) PADD 5,
because it is not clear that the region would benefit from increased natural gas production and
butane availability east of the Rockies.
June 7, 2019
62
"Math
-------
Addendum to No-RFS Study: Analysis of Low-Biofuel Case
Final Report
3. Results of the Analysis
The results of primary interest are the potential effects of reducing ethanol's use as a gasoline
blendstock on the properties of finished gasoline and the consequent effects on vehicle
emissions. Other analytical results, such as potential investments in new refining process
capacity, changes in refining operations, and changes in the composition of finished gasoline
indicate how the refining sector is likely to respond to large-scale changes in ethanol use.
Finally, effects on the incremental production cost of refined products and on the blending value
of ethanol indicate potential market effects stemming from reduced ethanol use.
Table 1, below, shows the estimated average properties of the U.S. conventional gasoline pool
(including low-RVP gasoline) for the Reference case and the two Low-Biofuel cases in which
ethanol no longer is blended in conventional gasoline. As discussed above, we held the
properties of RFG and export gasoline constant, so that the effects on gasoline properties of
moving from E10 to EO for conventional gasoline were concentrated in the conventional
gasoline pool. The most significant changes in gasoline properties occur for aromatics and
E200. In the Low-Biofuel #1 case (which features additional reforming), aromatics increased by
about six percent points in both the summer and winter, and E200 declined by about seven to
eight percent points. In the Low-Biofuel #2 case (which features more alkylation), aromatics
increased, but only by about two percent points, and E200 declined by about seven percent
points.6 More detailed, PADD-level estimates, of the effects on gasoline properties of large
scale removal of ethanol from the U.S. gasoline are shown in Table A-5 in the accompanying
Appendix.
Table 1: Properties of Finished Conventional Gasoline, U.S Average
Properties
Reference
Low-BioHiel #1
Low-Biofuel #2
Summer
Winter
Summer
Winter
Summer
Winter
RVP (psi)
9.4
13.9
8.5
13.1
8.5
13.1
Fuel Ethanol (vol%)
9.9
9.9
Aromatics (vol%)
19.0
15.6
25.1
22.3
21.2
17.3
Benzene (vol%)
0.57
0.55
0.57
0.59
0.57
0.56
Olefins (vol%)
6.4
6.5
5.7
7.0
4.8
4.8
Sulfur (ppm)
9
9
9
9
9
9
E200 (vol% off)
55.3
57.3
47.0
50.4
48.3
50.9
E300 (vol% off)
83.7
85.6
81.7
82.8
83.9
85.4
Energy Density1
4.663
4.600
4.854
4.780
4.821
4.750
Octane
(R+M)/2
87.9
88.0
87.9
88.0
87.9
88.0
MON
83.4
83.7
84.6
84.4
85.0
85.0
RON
92.5
92.3
91.3
91.6
90.9
90.9
Sensitivity
9.0
8.6
6.8
7.2
6.0
5.9
Note: Includes conventional and low-RVP gasoline; excludes RFG, exports, and imports.
1 Lower heating value (MM btu/b).
6 Octane Sensitivity (the difference between RON and MON) declined by about one and a half to two numbers for the Low-
Biofuel # 1 case and about three numbers for the Low-Biofuel #2 case, because alkylate has much lower Octane Sensitivity than
reformate, and both have lower Octane Sensitivity than ethanol.
June 7, 2019
63
"Math
-------
Addendum to No-RFS Study: Analysis of Low-Biofuel Case
Final Report
Table 2, below, shows refining capacity for gasoline upgrading processes present in the
Reference case and added in the Low-Biofuel cases. For the Low-Biofuel #1 case a mix of
alkylation, isomerization, and reforming capacity is added. As discussed above, the addition of
alkylation capacity was limited by constraints on purchases of iso-butane. In this case, relative
to the Reference case, reformer throughput increases by about 500 K b/d and reformer severity
increases by about two and a half numbers, averaged across seasons. Such increases are
consistent with the increase in the aromatics content of gasoline shown in Table 1.
In the Low-Biofuel #2 case, the loss in gasoline volume and octane from the removal of ethanol
is made up primarily by a large increase in alkylation process capacity (about a 75% increase
relative to existing capacity), along with additional isomerization capacity. The increase in
alkylation capacity was facilitated in the refinery modeling by setting a $50/b price for purchased
iso-butane and allowing unlimited purchases.
Table 2: Gasoline Upgrading Process Capacity and Reformer Operations, U.S.
Processes &
Operations
Reference
Low-BioHiel #1
Low-Biofuel #2
Summer
Winter
Summer
Winter
Summer
Winter
Capacity (K b/cd)
Existing
Alkylation
1,214
Pen/hex Isomerization
530
Reforming
3,438
New1
Alkylation
333
133
925
932
Pen/hex Isomerization
384
345
273
229
Reforming
316
280
10
10
Reformer Operations
Throughput (K b/d)
2,864
2,697
3,264
3,349
2,969
2,900
Severity (RON)
96.3
93.8
99.2
95.6
98.9
95.2
Feed (%)
Straight Run Naphtha
77%
72%
78%
77%
78%
77%
160-250° F
23%
20%
26%
32%
28%
33%
250-325° F
54%
52%
51%
45%
50%
44%
Hydrocrk. (200-325° F)
12%
16%
12%
13%
11%
12%
Coker Nap. (160-375° F;
12%
12%
11%
10%
11%
11%
1 Adjusted from K b/sd to K b/cd using a specified 90% utilization rate.
As indicated in Table 3, the Low Biofuel #2 case virtually eliminates investment in additional
reforming process capacity. However, existing reforming capacity is used more extensively
than in the Reference case - both reformer throughput and severity increase. This is consistent
with the aromatics content of the gasoline pool increasing, but not by as much as in the Low-
Biodiesel #1 case.
The extent to which alkylation, rather than reforming, would be the preferred approach for
replacing the lost volume and octane when moving from E10 to E0 would depend on the price
and availability of iso-butane in the various refining regions relative to the price of crude oil and
practical constraints that might limit expansion of alkylation capacity. If we had priced iso-
June 7, 2019
64
'MathPro
-------
Addendum to No-RFS Study: Analysis of Low-Biofuel Case
Final Report
butane higher than $50/b, the refinery models would have invested more heavily in additional
reforming capacity in the Low-Biofuel # 2 case, rather than alkylation capacity. However, recent
increases in natural gas and LPG production and observed prices suggest that iso-butane likely
would be priced lower, relative to crude oil acquisition costs, than is assumed in our analysis.
Table 3, below, shows how the composition of gasoline (the relative use of gasoline
blendstocks) changes in the Low-Biofuels cases. As indicated in the Table, ethanol's blending
volume declines from about 10% in the Reference case to zero in the Low-Biofuel cases. This
loss in volume (and octane) is made up by increases in higher-octane blendstocks (C5s and
isomerate, alkylates, and reformate), with some corresponding declines in lower octane
blendstocks (mostly raffinate, naphthas, and hydrocrackate). The use of FCC naphtha
blendstocks also declined in the Low-Biofuel #2 case. This results from a relative reduction in
FCC naphtha production and expansion in FCC olefin production to support the large increase
in alkylate production (olefins are a feedstock for alkylation). In effect, the Model indicates
refiners would increase alkylation capacity and alkylate production by modifying the FCC
operation to produce more butylenes (olefins) at the expense of FCC naphtha to increase the
octane of the pool.
Table 3: Composition of Finished Conventional Gasoline, U.S. Average
Gasoline
Blendstock
I Reference
Low-Biofuel #1
Low-Biofuel #2
(Summer
Winter
Summer
Winter
Summer
Winter
C4s
2.0%
10.1%
2.6%
10.9%
3.4%
12.4%
Natural Gas Liquids
2.0%
1.7%
0.6%
0.3%
1.7%
0.8%
C5s & Isomerate
5.9%
5.3%
11.5%
11.7%
9.3%
7.0%
Raffinate
2.8%
3.0%
2.2%
1.7%
1.4%
2.3%
Naphthas (C5-2500)
12.6%
13.3%
5.8%
4.1%
9.1%
7.5%
Hydrocrackate
7.2%
4.7%
8.0%
3.9%
4.1%
2.6%
Alkylate
8.8%
9.2%
17.0%
12.4%
27.4%
26.5%
Poly Gas
0.1%
0.1%
0.1%
0.0%
FCC Naphtha
27.2%
25.3%
23.4%
28.1%
18.4%
20.0%
Reformate & Aromatics
21.5%
17.6%
28.8%
26.8%
25.1%
20.9%
Ethanol
9.9%
9.9%
Table 4, below, shows the effects of large-scale removal of ethanol from the gasoline pool on
the incremental production costs of major refined products and on the blending value of ethanol.
The Table indicates that removing ethanol as a blendstock from conventional gasoline would
increase the incremental refining cost of producing all gasoline - both RFG and conventional
gasoline. Increases in incremental production costs for gasoline are highest: (1) in the Low-
Biofuels #1 case (the refining models are more constrained regarding investment options in this
case because investment in alkylation is constrained via limits imposed on butane purchases);
(2) in the summer; and (3) for conventional gasoline. On an annual basis, the increase in
incremental production costs averages about $3/b for RFG and $11/b for conventional gasoline
in the Low-Biofuel #1 case, and about $1/b for RFG and $5/b for conventional gasoline in the
Low-Biofuel #2 case. However, if the prices for purchased iso-butane in the Low-Biofuel #2
case were set at a higher level, incremental gasoline production costs also would be higher.
June 7, 2019
65
"Math
-------
Addendum to No-RFS Study: Analysis of Low-Biofuel Case
Final Report
Ethanol's implicit blending values increased in both Low-Biofuel cases, with more substantial
increases in the summer (about $34/b), the season in which octane is more constrained. Prices
for jet fuel and ULSD were relatively unaffected in both Low-Biofuel cases.
Table 4: Incremental Production Cost of Major Refined Products and Refinery Blending
Values of Ethanol, U.S. Average ($/b)
Reference
Low-Biofuel #1
Low-Biofuel #2
Summer
Winter
Summer |
Winter
Summer
Winter
Refined Product
Finished Gasoline
RFG
88.2
80.2
93.8
81.2
91.5
78.7
Conventional
80.9
78.2
97.2
84.0
92.0
77.8
Jet Fuel
86.8
84.6
86.7
85.4
87.5
85.5
ULSD
84.4
83.9
83.7
85.2
85.0
85.0
Ethanol
100.0
93.0
135.0
105.0
133.0
99.0
June 7, 2019
66
"Math
------- |