Office of Transportation EPA420-D-06-008
United States and Air Quality September 2006
Environmental Protection
Agency
Renewable Fuel Standard
Program
Draft Regulatory Impact
Analysis
-------
EPA420-D-06-008
September 2006
Renewable Fuel Standard Program
Draft Regulatory Impact Analysis
Assessment and Standards Division
Office of Transportation and Air Quality
U.S. Environmental Protection Agency
NOTICE
This document 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 an exchange of
technical information and to inform the public of technical developments.
-------
Table of Contents
Overview ii
List of Acronyms and Abbreviations iv
Chapter 1: Industry Characterization 1
Chapter 2: Changes to Motor Vehicle Fuel Under the Renewable Fuel Standard Program 37
Chapter 3: Impacts on Emissions from Vehicles, Nonroad Equipment, and Fuel Production
Facilities 83
Chapter 4: National Emission Inventory Impacts 124
Chapter 5: Air Quality Impacts 152
Chapter 6: Lifecycle Impacts on Fossil Energy and Greenhouse Gases 172
Chapter 7: Estimated Costs of Renewable Fuels, Gasoline and Diesel 200
ChapterS: Agricultural Sector Impacts 286
Chapter 9: Small-Business Flexibility Analysis 294
Endnotes 299
-------
Overview
EPA is proposing standards which would implement a renewable fuel program as
required by the Energy Policy Act of 2005 (the Act). The Act specifies the total volume of
renewable fuel that is required to be used each year, and directs EPA to adjust this amount under
certain circumstances. The resulting standards represent a level of renewable fuel that each
refinery or importer must account for relative to its annual volume of gasoline produced or
imported. In reality, however, renewable fuel use is forecast to exceed the RFS standards due to
market forces. The analyses of the impacts associated with this increase in renewable fuel use
are discussed in this Draft Regulatory Impact Analysis (DRIA).
Chapter 1: Industry Characterization
This chapter discusses current gasoline, diesel and renewable fuel production, importation,
marketing and distribution, as well as likely future changes as a result of increased renewable
fuel use.
Chapter 2: Changes to Motor Vehicle Fuel Under the RFS Program
This chapter discusses our gasoline and renewable fuel consumption predictions (compared to a
2004 base year), and the expected impacts of various ethanol blends on gasoline properties.
Chapter 3: Impacts on Emissions from Vehicles, Nonroad Equipment and Fuel Production
Facilities
This chapter evaluates the impacts on vehicle and nonroad equipment emissions under various
oxygenate assumptions, specifically increasing ethanol and decreasing MTBE, and different
modeling techniques. The effect of biodiesel use on diesel-powered vehicle emissions is also
presented. Finally, emissions from ethanol and biodiesel production facilities are discussed.
Chapter 4: National Emissions Inventory Impacts
This chapter discusses the methods used to develop the national emissions inventories, and
quantifies the impact of expanded ethanol and biodiesel use on those inventories.
Chapter 5: Air Quality Impacts
This chapter discusses the impacts of expanded renewable fuel use on ozone and particulate
matter formation.
Chapter 6: Lifecycle Impacts on Fossil Energy and Greenhouse Gases
This chapter discusses our fuel lifecycle modeling, that is, analysis which accounts for all energy
and emissions of the fuel production process. A description of the model we used, how we used
it, and the results are presented. Impacts on greenhouse gases, including CC>2, fossil fuel use, and
petroleum use are presented. The effects on petroleum imports, import expenditures, and
domestic energy security are also discussed.
11
-------
Chapter 7: Estimated Costs of Renewable Fuels, Gasoline and Diesel
This chapter contains our analysis of the cost of corn and cellulosic ethanol. We also discuss
biodiesel and renewable diesel production costs. Costs associated with distributing the volumes
of ethanol necessary to meet the requirements of the proposed program, and the costs to prepare
gasoline and diesel blendstocks (for blending with renewable fuels) are also presented. Finally,
we present the overall fuel cost impacts of expanded renewable fuel use.
ChapterS: Agricultural Sector Impacts
This chapter discusses the likely economic impacts on the agricultural sector that may occur as a
result of the large expansion of renewable fuel production and use expected in the future. On-
going work using the FASOM model is also described.
Chapter 9: Small Business Flexibility Analysis
This chapter discusses our Small Business Flexibility Analysis (SBFA) which evaluates the
proposed rule to ensure that concerns regarding small businesses, which would be affected by the
rule, are sufficiently considered.
in
-------
List of Acronyms and Abbreviations
AAM
ABT
ACE
The Act
ADM
AEO
ANL
AQIRP
ARMS
BO, B5, B20, etc
Bbl
BEA
Bgal, bgal, bilgal, billgal, bg
BGY
BPCD
BPSD
bpd, bbls/day
BTU
BU
Bu/acre
BZ
CA
CAA
CAIR
CARB
CaRFGS
CBG
CBI
CD
CFEIS
CFR
c/gal
CG
CHP
CO
CO2
Co-op
CRC
DOS
DDGS
DOE
DRIA
E&C
EO
E10
E85
E200
E300
EIA
Energy Act
EO
Alliance of Automobile Manufacturers
Averaging, Banking, and Trading
American Coalition for Ethanol
Energy Policy Act of 2005 (also the Energy Act)
Archer Daniels Midland
Annual Energy Outlook (an EIA publication)
Argonne National Laboratory
Auto/Oil Air Quality Improvement Research Program
Agricultural Resource Management Survey
Percent of biodiesel, e.g., B5= 5%biodiesel, 95% diesel
Barrel
Bureau of Economic Analysis
Billion gallons
Billion gallons per year
Barrels Per calendar day
Barrels per stream day
Barrels Per Day
British Thermal Unit
Bushel
Bushels per acre
Benzene
California
Clean Air Act
Clean Air Interstate Rule
California Air Resources Board
California Phase 3 RFC
Cleaner Burning Gasoline
Caribbean Basin Initiative
Census Division
EPA's Certification and Fuel Economy Information System
Code of Federal Regulations
Cents per gallon
Conventional Gasoline
Combined Heat and Power Technology
Carbon Monoxide
Carbon Dioxide
Cooperative
Coordinating Research Council
Distillers' grains with solubles
Dried distillers' grains with solubles
Department of Energy
Draft Regulatory Impact Analysis
Engineering and Construction
Gasoline Blend which does not contain ethanol
Gasoline Blend containing a nominal 10 percent ethanol by volume
Gasoline Blend containing 85 percent ethanol by volume
Percent of Fuel Evaporated at 200 Degrees F (ASTM D 86)
Percent of Fuel Evaporated at 300 Degrees F (ASTM D 86)
Energy Information Administration (part of the U.S. Department of Energy)
Energy Policy Act of 2005 (also the Act)
Executive Order
IV
-------
EPA
EPAct
ETBE
ETOH
exCA
F, °F
FAPRI
FASOM
FBP
FCC
FCCU
FHWA
FOEB
FR
FRM
FRTP
FFV
FTP
GAL
g/Btu
g/day
GDP
GHG
GPA
GREET
GWP
HC
HCO
HDN
HSR
HVGO
IBP
IFQC
k
kbbl
kwh
Lb
LCD
LEV
LLE
LNS
LP
LSR
mg/m3
MM
MMBTU
MMbbls/cd
MMgal
MGY, MMGal/yr
MOBILE (5, 6, 6.2)
MON
MOVES2006
MSAT
Environmental Protection Agency
Energy Policy Act of 2005 (also 'the Energy Act' or 'the Act')
Ethyl Tertiary Butyl Ether
Ethanol
Excluding California
Fahrenheit
Farm and Agricultural Policy Research Institute
Forestry and Agriculture Sector Optimization Model
Feed Boiling Point (also Final Boiling Point)
Fluidized Catalytic Cracker
Fluidized Catalytic Cracking Unit
Federal Highway Administration
Fuel Oil Equivalent Barrel
Federal Register
Final Rulemaking
Fixed Reduction Trigger Point
Flexible Fuel Vehicle
Federal test procedure
Gallon
Grams per Btu
Grams per day
Gross Domestic Product
Greenhouse Gases
Geographic Phase-in Area
Greenhouse Gas, Regulated Emissions, and Energy Use in Transportation model
Global warming potentials
Hydrocarbon(s)
Heavy Cycle Oil (a refinery stream)
Naphtha Hydrotreater (also Hydro-Denitrogenation Unit)
Heavy Straight Run (a refinery stream)
Heavy Vacuum Gas Oil (a refinery stream)
Initial Boiling Point
International Fuel Quality Center
Thousand
Thousand barrels
Kilowatt Hour
Pound
Light Cycle Oil (a refinery stream)
Low emission vehicle
Liquid-Liquid Extraction
Light Naphtha Splitter
Linear Programming (a type of refinery model)
Light Straight Run (a refinery stream)
Milligrams per cubic meter
Million
Million British Thermal Units
Million barrels per calendar day
Million gallons
Millions of gallons per year
EPA's Motor Vehicle Emission Inventory Model (versions)
Motor Octane Number
EPA's Next Generation Highway Vehicle Emission Model
Mobile Source Air Toxics
-------
MSAT1
MSAT2
MTBE
NAAQS
NAICS
NASS
NBB
NCSU
NGL
NMHC
NMIM
NMOG
NONROAD
NONROAD2005
NOX
NPRM
NREL
OMB
OMHCE
ORNL
OTAQ
Oxy-fuel, oxyfuel
PADD
PM
PM10
PM25
PMA
POM
PONA
ppb
ppm
PRTP
PSI
QBtu
Quadrillion
(R+M)/2
RBOB
RFA (Chapter 9 Only)
RFA
RFC
RFS
RIA
RIMS
RIN
RON
RPMG
RSM
RVP
S
SBA
SBAR Panel, or 'the Panel'
SBFA
SBREFA
2001 Mobile Source Air Toxics Rule
2006 Proposed Mobile Source Air Toxics Rule
Methyl Tertiary-Butyl Ether
National Ambient Air Quality Standards
North American Industrial Classification System
National Agricultural Statistics Service
National Biodiesel Board
North Carolina State University
Natural gas plant liquids
Non-Methane Hydrocarbons
National Mobile Inventory Model (EPA software tool)
Non-methane organic gases
EPA's Non-road Engine Emission Model
EPA's Non-road Engine Emission Model Released in 2005
Oxides of nitrogen
Notice of Proposed Rulemaking
National Renewable Energy Laboratory
Office of Management and Budget
Organic Material Hydrocarbon Equivalent
Oak Ridge National Laboratory
Office of Transportation and Air Quality
Winter oxygenated fuel program
Petroleum Administration for Defense District
Paniculate Matter
Coarse Particle
Fine Particle
Petroleum Marketing Annual (an EIA publication)
Fob/cyclic Organic Matter
Paraffin, Olefin, Naphthene, Aromatic
Parts per billion
Parts Per million
Percentage Reduction Trigger Point
Pounds per Square Inch
Quadrillion btu
1015
Octane calculation (RON+MON)/2
Reformulated Blendstock for Oxygenate Blending
Regulatory Flexibility Act
Renewable Fuels Association
Reformulated Gasoline
Renewable Fuels Standard
Regulatory Impact Analysis
Regional Input-Output Modeling System
Renewable Identification Number
Research octane number
Renewable Products Marketing Group
Response Surface Model
Reid Vapor Pressure
Sulfur
Small Business Administration
Small Business Advocacy Review Panel
Small Business Flexibility Analysis
Small Business Regulatory Enforcement Fairness Act (of 1996)
VI
-------
scf
SOA
SOx
SULEV
T50
T90
TAME
ULEV
U.S.
U.S.C.
USDA
VGO
VMT
voc
vol%
WDGS
wt%
yr, y
Standard cubic feet
Secondary Organic Aerosol
Oxides of Sulfur
Super ultra low emission vehicle
Temperature at which 50% (by volume)
Temperature at which 90% (by volume)
of fuel evaporates (ASTM D 86)
of fuel evaporates (ASTM D 86)
Tertiary Amyl Methyl Ether
Ultra low emission vehicle
United States
United States Code
U.S. Department of Agriculture
Vacuum Gas Oil (a refinery stream)
Vehicle Miles Traveled
Volatile Organic Compound
Percent by volume, volume percent
Wet distillers' grains with solubles
Percent by weight, weight percent
Year
Vll
-------
Chapter 1: Industry Characterization
1.1 Transportation Fuel Providers
1.1.1 Petroleum Refiners
As of the end of 2005, there were 142 crude oil refineries operating in the United States,
representing a total of 16.4 million barrels/day of refining capacity. (These refineries produce
gasoline and other products and are a separate category than "blender refiners" that do not
process crude oil, but make gasoline from blendstocks.) The greatest number of refineries per
PADD is in PADD 3 (the Gulf Coast region) which has 52 operating refineries as of the end of
2005. This PADD also has the greatest refining capacity, at 7.9 million barrels per day. Table
1.1-1 presents the refineries and their crude oil production capacity, and identifies the PADD
each is in.
Table 1.1-1.
Refining Capacity by Individual Refinery
(crude oil processing basis)
Company
Conoco Phillips
Wood River, IL
Belle Chasse, LA
Sweeny, TX
Westlake LA
Linden, NJ
Ponca City OK
Trainer, PA
Borger TX
Wilmington CA
Ferndale WA
Rodeo CA
Billings MT
Valero Energy Corp.
Port Arthur TX
Memphis TN
Lima OH
Texas City TX
Corpus Christ! TX
Houston TX
Sunray TX
Three Rivers TX
Norco LA
Paulsboro NJ
Benecia CA
Wilmington CA
Ardmore OK
Capacity
(MMbbls/cd)
2.2
0.31
0.25
0.25
0.24
0.24
0.19
0.19
0.15
0.14
0.10
0.08
0.06
2.0
0.26
0.18
0.15
0.21
0.14
0.08
0.16
0.09
0.19
0.16
0.14
0.01
0.08
PADD
2
3
3
3
1
2
1
3
5
5
5
4
3
2
2
3
3
3
3
3
3
1
5
5
2
-------
Company
Wilmington CA
Krotz Springs LA
Exxon Mobil Corp.
Baytown TX
Baton Rouge LA
Beaumont TX
Joliet IL
Torrance CA
Billings MT
Chalmette, LA
BPPLC
Texas City TX
Whiting IN
Toledo OH
Los Angeles CA
Ferndale WA
Chevron Corp.
Pascagoula MS
ElSegundo CA
Richmond CA
Honolulu HI
Salt Lake City UT
Marathon Oil Corp.
Garyville LA
Cattlettsburg KY
Robinson IL
Detroit Ml
Canton OH
Texas City TX
Saint Paul Park MN
Sunoco, Inc.
Marcos Hook PA
Toledo OH
Westville NJ
Tulsa OK
PDV America, Inc.
Citgo; Lake Charles LA
Citgo, Lemont IL
Citgo; Corpus Christi TX
Koch Industries
Corpus Christi TX
Saint Paul MN
Motiva Enterprises LLC
Port Arthur TX
Convent LA
Norco LA
Tesoro Corp.
Anacortes WA
Salt Lake City UT
Capacity
(MMbbls/cd)
0.08
0.08
2.0
0.56
0.50
0.34
0.24
0.15
0.06
0.19
1.5
0.44
0.41
0.13
0.26
0.23
0.9
0.33
0.26
0.24
0.05
0.05
1.0
0.25
0.22
0.19
0.10
0.07
0.07
0.07
0.58
0.78
0.16
0.15
0.09
0.81
0.43
0.17
0.16
0.57
0.29
0.28
0.76
0.29
0.24
0.23
0.51
0.12
0.06
PADD
5
3
3
3
3
2
5
4
3
3
2
2
5
5
3
5
5
5
4
3
2
2
2
2
3
2
2
2
1
2
3
2
3
3
2
3
3
3
5
4
-------
Company
Martinez CA
Kapolei HI
Kenai AK
Royal Dutch/Shell Group
Martinez CA
Anacortes WA
Wilmington CA
Saraland AL
Deer Park, TX
Lyondell Chem. Co.
(Houston)
Total SA (Port Arthur, TX)
Sinclair Oil
Tulsa OK
Sinclair WY
Evansville WY
Murphy Oil
Meraux LA
Superior I/I//
Frontier Oil
El Dorado KS
Cheyenne WY
Cenex Harvest States, Inc.
McPherson KS
Laurel MT
Coffeyville Acquisitions
(Coffeyville KS)
Navajo Refining Corp.
Artesia NM
Woods Cross UT
Great Falls MT
Pasadena Refining Systems
(Pasadena TX)
Giant Industries, Inc.
Yorktown VA
Gallup NM
Bloomfield NM
Big West Oil (North Salt
Lake UT)
Capacity
(MMbbls/cd)
0.17
0.09
0.072
0.82
0.16
0.15
0.10
0.08
0.33
0.27
0.23
0.17
0.07
0.07
0.03
0.15
0.12
0.03
0.15
0.11
0.04
0.14
0.08
0.06
0.11
0.11
0.07
0.03
0.07
0.10
0.10
0.06
0.02
0.02
0.10
PADD
5
5
5
5
5
5
3
3
3
3
2
4
4
3
2
2
4
2
4
2
3
4
4
3
1
3
3
4
Source: Table 5 in Energy Information Administration, Refinery Capacity 2006 found at
http://www.eia.doe.gov/pub/oil gas/petroleum/data_publications/refinery capacity data/current/table5.pdf
Refining capacity has steadily increased in the U.S. due to increased demand for
petroleum products, with gasoline representing approximately 45 percent of product demand.
Refining capacity (crude oil input) was about 14 million bbls/day in 1973 and 17 million
bbls/day in 2005. While refining capacity has increased, however, the number of refineries has
decreased as less economical refineries have been forced to close. (Many of these came into
-------
existence for a very short time due to oil price supports in the 1970's.) In the 1970's, the number
of refineries in the U.S. was approximately 270 and has decreased by 47 percent. Figure 1.1-1
shows the number of refineries and total capacity in the U.S. from 1973 through 2004.
Figure 1.1-1.
Number of Refineries and Total Capacity in the U.S. from 1973-2004
350
300 - -
E 200 -
E
E 150 -
50 -
- - 14.0
20.0
- - 2.0
0.0
Source: EIA; Annual Energy Report, 2005 (Table 5.9)
The increase in capacity combined with the decrease in amount of refineries and the
increased demand for gasoline and diesel fuels, has resulted in an increase in the average
utilization rate of refineries. In the 1970's, the utilization rate ranged from 84 to 94 percent. In
the last ten years, however, the utilization rate has ranged from 91 to 96 percent. Refineries
therefore have to produce more with less overall capacity. The amount of gasoline and diesel
produced by U.S. refiners has steadily increased. Since 1973 through 2004, gasoline and diesel
production has increased 27 and 36 percent, respectively. Figure 1.1-2 shows the change in
gasoline and diesel production from 1973 through 2004
-------
Figure 1.1-2.
Amount of Gasoline and Diesel Fuels Produced in the U.S.
140.0 -,
120.0
100.0
•a so.o
c
£
"re
at
c
o
= 60.0
m
40.0
20.0
0.0
/A
O,''
Source: EIA Annual Energy Report, 2005; Table 5.8
1.1.2 Petroleum Imports
The decrease in U.S. refining capacity discussed in Section 1.1.3, has resulted in
increases in the amount of gasoline and diesel fuels imported into the U.S. As of 2004, 5.4 and
11.5 percent of the total respective volumes of gasoline and diesel consumed in the U.S. were
imported.
Today, the United States imports approximately 70 percent of all petroleum products
used, with two-thirds of these products being used for transportation. From 1973 to 2004, the
amount of crude oil imported has increased from 1.2 to 3.7 billion barrels per year, a tripling of
volume, representing an average annual increase of about 6 percent. Over the same time period,
the amount of gasoline imported has increased from 2 to 7.4 billion gallons per year, more than
three times the amount of volume. The amount of diesel imported in the same time period
decreased slightly from 6 to 5 percent. Figures 1.1-12 and 1.1-2 show the increase in crude oil
and gasoline/diesel fuel imports, respectively, from 1973 to 2004.
-------
Figure 1.1-3.
Increase in Crude Oil Imports from 1973-2004
4.00
3.50
3.00
3 2.50
2.00
1.50
1.00
0.50
0.00
Years
(Source: Annual Energy Outlook, 2005; Energy Information Administration)
-------
Figure 1.1-4.
Change in Volumes of Imported Gasoline and Diesel fuels (1973-2004)
14.00
12.00
Years
Source: Annual Energy Outlook, 2005; Energy Information Administration
Twenty seven percent of our trade deficit is from imported petroleum products, a deficit
which reached $782 billion in 2005. Approximately 5 percent of the petroleum-related deficit is
due to imports of gasoline and diesel fuels. (Figure 1.1-5 shows the trade deficits since 1993 and
the portions due to petroleum products and crude imports). Over the last 25 years, the
cumulative cost of imported crude oil has reached $2.0 trillion in 2005 dollars.
-------
Figure 1.1-5.
U.S. Trade Deficit and Portions Due to Petroleum and Crude Imports
1993-2005
ann nn
ynn nn -
finn nn
_>,
")
«5
= cnn nn -
•a
"o
) 4nn nn
O
m ?nn nn
onn nn
mn nn
n nn -
D Trade deficit
• Cost of imported
petroleum products
DCost of imported crude oil
-
Tl
Tl
-
I
A
r
PI
r
i-i
r
r-|
-
I
_
\\
\\
-
i-i
-|
-
93 94 95 96 97 98 99 2000 2001 2002 2003 2004 2005
Years
Source: Energy Information Administration, Annual Energy Report, 2005
The amount of import facilities in the U.S. has stayed relatively constant since the U.S.
EPA has been requiring such facilities to register. In 1995 there was a total of 39 such facilities
in the U.S. The amount has remained relatively constant, in the 50's since that time and as of
2004 there were 53 such facilities registered with U.S. EPA. The great majority of such facilities
are located in PADD 1; as of 2004, 35 facilities were in PADD 1, and a total of 18 in the other
four PADDs.
1.2 Renewable Fuel Production
While the definition of renewable fuel in the Act does not limit compliance with the
standard to any one particular type of renewable fuel, ethanol is currently the most prevalent
renewable fuel blended into gasoline today. Biodiesel represents another renewable fuel, which
-------
while not as widespread as ethanol use (in terms of volume), has been increasing in production
capacity and use over the last several years. Both ethanol and biodiesel are likely to continue to
dominate renewable fuel use in the foreseeable future.
1.2.1 Current U.S. Ethanol Production
1.2.1.1 Overview
There are currently 102 ethanol production facilities in the United States with a combined
production capacity of 4.9 billion gallons per year. This baseline, or starting point for this
regulatory impact analysis is based on U.S. ethanol production facilities operational as of June
2006.A123
Of the current ethanol production capacity, 93 percent is produced exclusively from corn,
mainly from a dry-milling process. The remainder is derived from corn/grain blends, cheese
whey, and other starches. The majority of ethanol plants are located in Midwest where the bulk
of corn is produced. PADD 2 accounts for 4.7 billion gallons (or almost 97 percent) of the total
U.S. ethanol production. Leading the Midwest in ethanol production are Iowa, Illinois,
Nebraska, Minnesota, and North Dakota which together represent 80 percent of the total
domestic product. In addition to the concentration of facilities located in PADD 2, there is a
sprinkling of ethanol plants located outside the corn belt ranging from California to Tennessee to
Georgia.
1.2.1.2 How is Ethanol Produced?
All of the ethanol currently produced comes from grain or starch-based feedstocks that
can easily be broken down into ethanol via traditional fermentation processes. The primary
feedstock is corn, although grain sorghum (milo), wheat, barley, beverage waste, cheese whey,
and sugars/starches are also fermented to make fuel-grade ethanol.
The majority of ethanol (almost 93 percent by volume) is produced exclusively from
corn. Most of the corn originates from the Midwest, and not surprisingly, most of the ethanol is
produced in PADD 2 close to where the corn is grown. However, corn-ethanol plants are also
found outside the traditional "corn belt". In Colorado, New Mexico, Colorado, and Wyoming,
corn is shipped in from the Midwest to supplement locally grown grains or in some cases, serve
as the sole feedstock. As for the remaining ethanol, almost 7 percent is produced from a blend of
corn and/or similarly processed grains (milo, wheat, or barley) and less than 1 percent is
produced from waste beverages, cheese whey, and sugars/starches combined. A summary of
ethanol production by feedstock is presented in Table 1.2-1.
A The June 2006 ethanol production baseline (plant locations, ownership, capacities, configurations, feedstocks,
energy sources, marketing agreements) was generated from a variety of data sources including Renewable Fuels
Association (RFA), Ethanol Producer Magazine, and International Fuel Quality Center (IFQC) publications as well
as ethanol producer/marketer websites. The production baseline includes small-scale ethanol production facilities as
well as former food-grade ethanol plants that have since transitioned into the fuel-grade ethanol market. Where
applicable, current ethanol plant production levels were used to represent plant capacity, as nameplate capacities are
often underestimated.
-------
Table 1.2-1. 2006 U.S. Ethanol Production by Feedstock
Plant Feedstock
Corn3
Corn/Milo
Corn/Wheat
Corn/Barley
Milo/Wheat
Waste Beverageb
Cheese Whey
Sugars & Starches
Total
Includes seed corn
Includes brewery waste
Capacity
MMGal/yr
4,516
162
90
40
40
16
8
2
4,872
%of
Capacity
92.7%
3.3%
1 .8%
0.8%
0.8%
0.3%
0.2%
0.0%
100.0%
No. of
Plants
85
5
2
1
1
5
2
1
102
%of
Plants
83.3%
4.9%
2.0%
1 .0%
1 .0%
4.9%
2.0%
1 .0%
100.0%
There are two primary plant configurations for processing grains (mainly corn) into
ethanol: dry mill and wet mill.
Dry mill plants simply grind the entire kernel and feed the flour into the fermentation
process to produce ethanol. At the end, the unfermentable parts are recovered as distillers'
grains along with a soluble liquid containing vitamins, minerals, fat and protein. The distillers'
grains are concentrated with the solubles stream to make a single co-product, referred to as
distillers' grains with solubles (DOS). The co-product is either sold wet (WDGS) or more
commonly dried (DDGS) to the agricultural market as animal feed. If the animal feed is going to
be used by local markets, it's usually sold wet precluding the need for process dryers. However,
if the feed is going to be shipped (usually by train) to more distant locations, the product is
usually dried to facilitate storage and transportation. Carbon dioxide is also produced during the
ethanol fermentation process and may be recovered as a saleable product.
Wet mill plants typically separate the kernel into four products: starch, gluten feed, gluten
meal, and oil. The starch is used in a fermentation process the same as in dry mill plants, while the
gluten, oil, and other possible co-products are sold into food and agricultural markets. Production of
these multiple streams is more capital-intensive than the dry mill process, and thus wet mill plants
are generally more expensive to build and tend to be larger in size.
Dry milling is the most predominant production process implemented by today's ethanol
plants. Of the 94 plants processing corn (and/or other similarly processed grains), 84 utilize dry
milling technologies and the remaining 10 plants rely on wet milling processes. Additionally, all
under construction or "planned" plants (defined in Section 1.2.2.1) are scheduled to be dry mill.
A list of the existing wet mill facilities is provided in Table 1.2-2.
10
-------
Table 1.2-2. 2006 U.S. Ethanol Production - Wet Mill Plants
Ethanol Plant
Archer Daniels Midland (ADM)a
Archer Daniels Midland (ADM)a
Archer Daniels Midland (ADM)
Archer Daniels Midland (ADM)a
Archer Daniels Midland (ADM)
Aventine Renewable Energy, Inc.
Cargill, Inc.
Cargill, Inc.
Grain Processing Corp
Tate & Lyle
City
Cedar Rapids
Clinton
Columbus
Decatur
Marshall
Pekin
Eddyville
Blair
Muscatine
Loudon
State
IA
IA
NE
IL
MN
IL
IA
NE
IA
TN
Capacity
MMgal/yr
300
150
90
250
40
100
35
85
20
67
Total 1,137
a
Estimated ADM plant capacities
The remaining 8 plants which process waste beverages, cheese whey, or sugars/starches,
operate differently than their grain-based counterparts. These facilities do not require milling
and instead operate a more simplistic enzymatic fermentation process.
In addition to grain and starch-to-ethanol production, another method exists for producing
ethanol from a more diverse feedstock base. This process involves converting cellulosic
feedstocks such as bagasse, wood, straw, switchgrass, and other biomass into ethanol. Cellulose
consists of tightly-linked polymers of starch, and production of ethanol from it requires
additional steps to convert these polymers into fermentable sugars. Scientists are actively
pursuing acid and enzyme hydrolysis to achieve this goal, but the technologies are still not fully
developed for large-scale commercial production. As of June 2006, there were no U.S ethanol
plants processing cellulosic feedstocks. Currently, the only known cellulose-to-ethanol plant in
North America is logen in Canada, which produces approximately one million gallons of ethanol
per year from wood chips. For more a more detailed discussion on cellulosic ethanol
production/technologies, refer to Section 7.1.2.
The ethanol production process is relatively resource-intensive and requires the use of
water, electricity and steam. Steam needed to heat the process is generally produced onsite or by
other dedicated boilers. Of today's 102 ethanol production facilities, 98 burn natural gas, 2 burn
coal, 1 burns coal and biomass, and 1 burns syrup from the process to produce steam. A
summary of ethanol production by plant energy source is found below in Table 1.2-3.
11
-------
Table 1.2-3. 2006 U.S. Ethanol Production by Plant Energy Source
Capacity % of No. of % of
Energy Source MMGal/yr Capacity Plants Plants
Natural Gasa
Coal
Coal & Biomass
Syrup
Total
4,671
102
50
49
4,872
95.9%
2.1%
1 .0%
1 .0%
100.0%
98
2
1
1
102
96.1%
2.0%
1 .0%
1 .0%
100.0%
Includes a natural gas facility which is considering transitioning to coal
Currently, 7 of the 102 ethanol plants utilize co-generation or combined heat and power
(CHP) technology. CHP is a mechanism for improving overall plant efficiency. CHP facilities
produce their own electricity (or coordinate with the local municipality) and use otherwise-
wasted exhaust gases to help heat their process, reducing the overall demand for boiler fuel.
1.2.1.3 How Much Ethanol is Produced?
Grain-to-ethanol fermentation technologies are well-known and have been used to
produce motor vehicle fuel since the 1860s. However, alcohol-based motor vehicle fuels have
had a hard time competing with their fossil fuel counter-parts until recently. Over the past 25
years, domestic fuel ethanol production has steadily increased due to technological advances,
environmental regulation (oxygenate requirements in ozone non-attainment areas, carbon
monoxide non-attainment areas, etc.), and the rising cost of crude oil. More recently, ethanol
production has soared due to state MTBE bans, steep increases in crude oil prices, and tax
incentives. Over the past three years, domestic ethanol production has nearly doubled from 2.1
billion gallons in 2002 to 4.0 billion gallons in 2005 as shown in Figure 1.2-14. Current ethanol
production capacity as of June 2006 was approximately 4.9 billion gallons per year. This
upward trend in ethanol production is expected to continue on into the future as discussed in
Section 1.2.2.1.
12
-------
Figure 1.2-1. U.S. Ethanol Production Over Time
4,500 -
4,000
3,500
O) 3,000
O 2,500
|
,? 2,000
ro
£ 1,500
LU
1,000
500
4,000
2810
2,130
1,770
1,630
Year
Source: Renewable Fuels Association, From Niche to Nation: Ethanol Industry Outlook 2006
1.2.1.4 Where is the Ethanol Produced?
Currently, the majority of ethanol is produced in the Midwest within PADD 2 - not
surprisingly, where most of the corn is grown. Of the 102 U.S. ethanol production facilities, 93
are located in Midwest. As a region, PADD 2 accounts for about 97 percent (or 4.7 billion
gallons per year) of domestic ethanol production, as shown in Table 1.2-4.
Table 1.2-4. 2006 U.S. Ethanol Production by PADD
PADD
PADD1
PADD 2
PADD 3
PADD 4
PADD 5
Capacity
MMgal/yr
0.4
4,710
30
98
34
%of
Capacity
0.0%
96.7%
0.6%
2.0%
0.7%
No. of
Plants
1
93
1
4
3
%of
Plants
1 .0%
91 .2%
1 .0%
3.9%
2.9%
Total
4,872
100.0%
102
100.0%
13
-------
Leading the Midwest in ethanol production are Iowa, Illinois, Nebraska, Minnesota, and
South Dakota with capacities of 1.61, 0.71, 0.57, 0.55, and 0.48 billion gallons per year,
respectively. Together, these five states' 69 ethanol plants account for 80 percent of the total
domestic product. Although the majority of ethanol production comes from the Midwest, there
is a sprinkling of plants situated outside the corn belt ranging from California to Tennessee all
the way down to Georgia. As of June 2006, 19 states contributed to the total domestic ethanol
production. A summary of these states' ethanol production capacities is found in Table 1.2-5.
Table 1.2-5. 2006 U.S. Ethanol Production by State
State
Iowa
Illinois
Nebraska
Minnesota
South Dakota
Wisconsin
Kansas
Indiana
Missouri
Colorado
Tennessee
North Dakota
Michigan
Kentucky
California
New Mexico
Wyoming
Ohio
Georgia
Total
Capacity
MMGal/yr
1,606
706
566
546
475
193
179
122
110
93
67
51
50
38
34
30
5
3
0.4
4,872
%of
Capacity
33.0%
14.5%
1 1 .6%
1 1 .2%
9.7%
4.0%
3.7%
2.5%
2.3%
1 .9%
1 .4%
1 .0%
1 .0%
0.8%
0.7%
0.6%
0.1%
0.1%
0.0%
100.0%
No. of
Plants
25
6
11
16
11
5
7
2
3
3
1
2
1
2
3
1
1
1
1
102
%of
Plants
24.5%
5.9%
10.8%
15.7%
10.8%
4.9%
6.9%
2.0%
2.9%
2.9%
1 .0%
2.0%
1 .0%
2.0%
2.9%
1 .0%
1 .0%
1 .0%
1 .0%
100.0%
In addition to the domestic ethanol production described above, the U.S. also receives a
small amount of ethanol imports from other countries. A discussion on ethanol imports is found
in Section 1.5
1.2.1.5 Who are the Ethanol Producers?
The U.S. ethanol industry is currently comprised of a mixture of corporations and farmer-
owned cooperatives (co-ops). More than half of the plants (55) are owned by corporations and
the remainder (47 plants) are farmer owned co-ops. On average, a U.S. ethanol production
facility has a mean plant capacity of about 48 million gallons per year. As shown below in Table
1.2-6, plants owned by corporations (company-owned) are above average in size and farmer-
owned co-ops are below average. Similarly, company-owned plants have a much broader range
in production levels than farmer-owned co-ops.
14
-------
Table 1.2-6. 2006 U.S. Ethanol Production by Plant Ownership
Plant Ownership
Company-Owned
Farmer-Owned
Total
Total No.
of Plants
55
47
102
Production Capacity, MMGal/yr
Total Avg Min Max
3,124
1,748
4,872
57
37
48
0.4
2.6
0.4
300
60
300
Based on the dominating number of company-owned plants and their above-average
production size, company-owned plants account for nearly 65 percent of the total U.S. ethanol
production capacity. Additionally, as of June 2006, 45 percent of the total capacity originated
from 22 plants owned by just 8 different companies. A list of the top eight ethanol producing
companies and their respective capacities is found in Table 1.2-7.
Table 1.2-7. 2006 U.S. Ethanol Production - Top Eight Producers
Capacity No. of
Company MMGal/yr Plants
Archer Daniels Midland (ADM)
VeraSun Energy
Hawkeye Renewables, LLC
MGP Ingredients, Inc.
Aventine Renewable Energy, Inc.
Cargill Inc.
Abengoa Bioenergy Corporation
New Energy Corp.
Total
1,070
230
200
190
150
120
110
102
2,172
7
2
2
3
2
2
3
1
22
1.2.1.6 Who are the Ethanol Marketers?
Over 90 percent of today's U.S. ethanol production is sold to the gasoline industry by 8
primary marketing companies. The remaining ethanol is marketed by other small marketers. A
summary of the top eight ethanol marketers and their respective volumes is found in Table 1.2-8.
15
-------
Table 1.2-8. 2006 U.S. Ethanol Production - Top Eight Marketers5
Ethanol Marketer
Archer Daniels Midland (ADM)
Ethanol Products
Renewable Products Marketing Group (RPMG)b
Aventine Renewable Energy
Eco-Energy
United Bio Energy
Cargill, Inc.
Abengoa Bioenergy
Total
Marketing
Volume3
MMgal/yr
1,172
906
850
648
325
287
120
110
4,417
No. of
Plants
9
20
14
14
5
8
2
3
75
aVolume based on marketing agreements and respective ethanol plant capacities
bEstimated RPMG marketing volume/plants.
1.2.2 Expected Growth in U.S. Ethanol Production
The Act requires 7.5 billion gallons of renewable fuel to be used in gasoline by 2012. Of
that, a large percentage (or 7.2 billion gallons, explained further in DRIA Section 2.1.4.1) is
expected to be ethanol. In addition to the Act's renewable fuel requirements, record-high crude
oil prices coupled with a growing number of state ethanol mandates and MTBE bans is strongly
driving the U.S. ethanol industry. Ethanol production technologies continue to improve making
fuel-grade ethanol production economically-favorable and profitable in most cases.
Accordingly, EPA predicts that ethanol production capacity will exceed the Act's renewable fuel
requirements in 2012 and beyond. The forecasted ethanol production, presented below, supports
this prediction.
1.2.2.1 Increases in Ethanol Plant Capacity
Today's U.S. ethanol production capacity (4.9 billion gallons) is already exceeding the
2006 renewable fuel requirement (4.0 billion gallons). In addition, there is another 2.5 billion
gallons of ethanol production capacity currently under construction.8678 A summary of the new
construction and expansion projects currently underway (as of June 2006) is found in Table 1.2-9
T3
Under construction plant locations, capacities, feedstocks, and energy sources as well as planned/proposed plant
locations and capacities were derived from a variety of data sources including Renewable Fuels Association (RFA),
Ethanol Producer Magazine, and International Fuel Quality Center (IFQC) publications as well as ethanol producer
websites.
16
-------
Table
1.2-9. Under Construction
2006 ETOH Baseline
PADD
PADD
PADD
PADD
PADD
Total
1
2
3
4
5
MMGal/yr
0.4
4,710
30
98
34
4,872
Plants
1
93
1
4
3
102
New Construction
MMGal/yr Plants
0
2,048
30
50
90
2,218
U.S.
Ethanol
Plant
Ca
Plant Expansions
MMGal/yr
0
35
1
1
2
39
0
252
0
7
0
259
Plants
0
8
0
1
0
9
pacity
2006 Baseline + UCa
MMGal/yr
0.4
7,010
60
155
124
7,349
Plants
1
128
2
5
5
141
allnder Construction
A select group of builders, technology providers, and construction contractors are
completing the majority of the construction projects described in Table 1.2-9. As such, the
completion dates of these projects are staggered over approximately 18 months, resulting in the
gradual phase-in of ethanol production shown in Figure 1.2-2.
Figure 1.2-2. Estimated Phase-In of Under Construction U.S. Plant Capacity
8,000
1,000
Jun-06
Dec-07
DAdditional Capacity •Total ETOH Capacity
Source: April 6, 2006 Biofuels Journal: Ethanol Plants Under Construction in the United States and Canada
(supplemented by ethanol producer website information)
As shown in Table 1.2-9 and Figure 1.2-2, once all the construction projects currently
underway are complete (estimated by December 2007), the resulting U.S. ethanol production
capacity would be over 7.3 billion gallons. Together with estimated biodiesel production (300
17
-------
million gallons by 2012), this would be more than enough renewable fuel to satisfy the 2012
renewable fuel requirement (7.5 billion gallons). However, ethanol production is not expected to
stop here. There are more and more ethanol projects being announced each day. The potential
projects are at various stages of planning from conducting feasibility studies to gaining
city/county approval to applying for permits to fmancing/fundraising to obtaining contractor
agreements. If all these plants were to come to fruition, the combined domestic ethanol
production could exceed 20 billion gallons as shown in Table 1.2-10.
Table 1.2-10. Potential U.S. Ethanol Production Projects
2006 Baseline + UCa
MMGal/yr Plants
PADD
PADD
PADD
PADD
PADD
Total
1
2
3
4
5
0.4
7,010
60
155
124
7,349
1
128
2
5
5
141
Planned
MMGal/yr Plants
1,
2,
250
940
108
0
128
426
3
15
1
0
2
21
Proposed
MMGal/yr Plants
1,005
7,508
599
815
676
10,603
21
90
9
14
18
152
Total ETOH
MMGal/yr
1,255
16,458
767
970
928
20,378
Potential
Plants
25
233
12
19
25
314
allnder Construction
However, although there is clearly a great potential for growth in ethanol production, it is
unlikely that all the announced projects would actually reach completion in a reasonable amount
of time. There is no precise way to know exactly which plants would come to fruition in the
future; however, we've chosen to focus our further discussions on only those plants which are
under construction or in the final planning stages (denoted as "planned" above in Table 1.2-10).
The distinction between "planned" versus "proposed" is that as of June 2006 planned projects
had completed permitting, fundraising/financing, and had builders assigned with definitive
construction timelines whereas proposed projects did not.
As shown in Table 1.2-10, once all the under construction and planned projects are
complete (by 2012 or sooner), the resulting U.S. ethanol production capacity would be 9.8 billion
gallons, exceeding the 2012 EIA demand estimate (9.6 billion gallons, discussed in DRIA
Section 2.1.4.1). This forecasted growth would double today's production capacity and greatly
exceed the 2012 renewable fuel requirement (7.5 billion gallons). In addition, domestic ethanol
production would be supplemented by imports, which are also expected to increase in the future.
A more detailed discussion on future ethanol imports is found in Section 1.5.
1.2.2.2 Changes in Ethanol Production
Of the 60 forecasted new ethanol plants (39 under construction and 21 planned), all
would (at least initially) rely on grain-based feedstocks. Of the plants, 56 would rely exclusively
on corn as a feedstock. As for the remaining plants: two would rely on both corn and milo, one
would process molasses and sweet sorghum, and the last would start off processing corn and then
transition into processing bagasse, rice hulls, and wood. A summary of the resulting overall
feedstock usage is found in Table 1.2-11.
18
-------
Table 1.2-11. Forecasted U.S. Ethanol Production by Feedstock
Plant Feedstock
Corn3
Corn/Milo
Corn then bagasse, rice hulls, wood
Corn/Wheat
Corn/Barley
Milo/Wheat
Milo
Waste Beverageb
Molasses, sweet sorghum
Cheese Whey
Sugars & Starches
Total
Capacity
MMGal/yr
9,226
202
108
90
40
40
30
16
15
8
2
9,775
%of
Capacity
94.4%
2.1%
1.1%
0.9%
0.4%
0.4%
0.3%
0.2%
0.2%
0.1%
0.0%
100.0%
No. of
Plants
141
6
1
2
1
1
1
5
1
2
1
162
%of
Plants
87.0%
3.7%
0.6%
1 .2%
0.6%
0.6%
0.6%
3.1%
0.6%
1 .2%
0.6%
100.0%
Includes seed corn
Includes brewery waste
The Act requires 250 million gallons of the renewable fuel consumed in 2013 and beyond
to meet the definition of cellulosic biomass ethanol. The Act defines cellulosic biomass ethanol
as ethanol derived from any lignocellulosic or hemicellulosic matter that is available on a
renewable or recurring basis including dedicated energy crops and trees, wood and wood
residues, plants, grasses, agricultural residues, fibers, animal wastes and other waste materials,
and municipal solid waste. The term also includes any ethanol produced in facilities where
animal or other waste materials are digested or otherwise used to displace 90 percent or more of
the fossil fuel normally used in the production of ethanol.
As of June 2006, there were zero cellulosic ethanol plants (as discussed above in 1.2.1.2).
Of the forecasted plants, only one is expected to meet the definition of "cellulosic biomass
ethanol" based on feedstocks.0 The 108 MMgal/yr Bionol facility slated for East Carroll Parish,
LA is proposing to start off processing corn and then transition into processing bagasse, rice
hulls, and wood (cellulosic feedstocks).9 It is unclear as to whether this facility would be
processing cellulosic material by 2013, however there are several other facilities that could
potentially meet the Act's definition of cellulosic biomass ethanol based on plant energy sources.
There are 7 ethanol production plants with a combined ethanol production capacity of 461
MMgal/yr that burn or plan to burn renewable fuels to generate steam for their processes. A brief
description of each potentially-cellulosic facility is provided in Table 1.2-12.
c At the time of this analysis (June 2006) there were other plants proposing cellulosic ethanol production
technologies. However, they are not included in this in-depth discussion of forecasted plants because they were not
under construction or in the final stages of planning.
19
-------
Table 1.2-12. Potential U.S. Cellulosic Ethanol Plants (Based on Energy Source)
Capacity
Ethanol Plant City state Plant Energy Source MMGal/yr Status
Corn Plus, LLP Winnebago
Central Iowa Renewable Energy Goldfield
E Caruso Ethanol Goodland
Central Minnesota Ethanol Co-op Little Falls
E3 Biofuels Mead
Harrison Ethanol, LLC Cadiz
Archer Daniels Midland (ADM) Columbus
MN Syrup
IA Coal & Biomass
KS Coal & Biomass
MN Natural Gas then Biomass
NE Manure/Syngas
OH Manure/Syngas
NE Coal, Tires & Biomass
Total cellulosic ethanol potential based on plant energy source
49
50
25
22
20
20
275
461
Existing
Existing
Under Construction
Under Construction
Under Construction
Under Construction
Planned
Depending on how much fossil fuel is displaced by these renewable feedstocks (on a
plant-by-plant basis), a portion or all of the aforementioned ethanol production (up to 461
MMgal/yr) could potentially qualify as "cellulosic biomass ethanol" under the Act. Combined
with the 108 MMgal/yr Bionol plant planning to process renewable feedstocks, the total
cellulosic potential could be as high as 569 MMgal/yr in 2013. Even if only half of this ethanol
were to end up qualifying as cellulosic biomass ethanol, it would still be more than enough to
satisfy the Act's cellulosic requirement (250 million gallons).0
Including the above-mentioned facilities, a summary of the resulting overall ethanol plant
energy usage is found below in Table 1.2-13.
Table 1.2-13. Forecasted U.S. Ethanol Production by Energy Source
Capacity % of No. of % of
Energy Source MMGal/yr Capacity Plants Plants
Biomass
Coal3
Coal & Biomass
Coal, Tires & Biomass
Manure/Syngas
Natural Gas
Syrup
Total
22
729
75
275
40
8,586
49
9,775
0.2%
7.5%
0.8%
2.8%
0.4%
87.8%
0.5%
100.0%
1
12
2
1
2
143
1
162
0.6%
7.4%
1 .2%
0.6%
1 .2%
88.3%
0.6%
100.0%
Includes one existing and three under construction plants that plan on transitioning
from natural gas to coal.
Of the 60 new forecasted plants, 2 plan to utilize co-generation or combined heat and
power (CHP) technology. This would increase the number of energy efficient CHP ethanol
plants from 8 to 10.
D We anticipate a ramp-up in cellulosic ethanol production in the years to come so that capacity exists to satisfy the
Act's 2013 requirement (250 million gallons of cellulosic biomass ethanol). Therefore, for subsequent analysis
purposes, we have assumed that 250 million gallons of ethanol would come from cellulosic biomass sources by
2012.
20
-------
1.2.2.3 Changes in Where Ethanol is Produced
In 2012, the majority of ethanol production is still expected to originate from PADD 2.
Once all the under construction and planned projects are complete, approximately 92 percent of
the U.S. ethanol production capacity would come from PADD 2, as shown below in Table 1.2-
14. This is a slight decrease from the Midwest marketshare held in June 2006 (97 percent as
described in Section 1.2.1.4).
Table 1.2-14. Forecasted U.S. Ethanol Production by PADD
Capacity % of No. of % of
PADD MMgal/yr Capacity Plants Plants
PADD1
PADD 2
PADD 3
PADD 4
PADD 5
Total
250
8,950
168
155
252
9,775
2.6%
91 .6%
1 .7%
1 .6%
2.6%
100.0%
4
143
3
5
7
162
2.5%
88.3%
1.9%
3.1%
4.3%
100.0%
Despite the growth in PADD 2 ethanol production, the shift in marketshare is attributed
to the growing number of ethanol plants outside the cornbelt. In particular, New York,
Louisiana, Texas, Arizona, Hawaii, and Oregon are scheduled to join the 19 ethanol producing
states described in Table 1.2-5. A summary of the forecasted ethanol production by state is
found below in Table 1.2-13.
21
-------
Table 1.2-15.
State
Iowa
Nebraska
Illinois
South Dakota
Minnesota
Indiana
Kansas
Wisconsin
North Dakota
Michigan
Missouri
Ohio
New York
Colorado
Oregon
Louisiana
Tennessee
Georgia
California
Arizona
Kentucky
New Mexico
Texas
Hawaii
Wyoming
Total
Forecasted
Capacity
MMGal/yr
2,418
1,790
1,200
678
659
622
299
283
261
212
195
193
150
143
113
108
104
100
69
55
38
30
30
15
12
9,775
U.S. Ethanol
%of
Capacity
24.7%
18.3%
12.3%
6.9%
6.7%
6.4%
3.1%
2.9%
2.7%
2.2%
2.0%
2.0%
1 .5%
1 .5%
1 .2%
1.1%
1.1%
1 .0%
0.7%
0.6%
0.4%
0.3%
0.3%
0.2%
0.1%
100.0%
Production by
No. of
Plants
30
24
12
13
18
8
10
7
5
4
5
4
2
4
1
1
1
2
4
1
2
1
1
1
1
162
State
%of
Plants
18.5%
14.8%
7.4%
8.0%
11.1%
4.9%
6.2%
4.3%
3.1%
2.5%
3.1%
2.5%
1 .2%
2.5%
0.6%
0.6%
0.6%
1 .2%
2.5%
0.6%
1 .2%
0.6%
0.6%
0.6%
0.6%
100.0%
1.2.3 Current Biodiesel Production
Biodiesel is a diesel fuel substitute produced by combining virgin plant or animal oils
with alcohol through a transesterification process, yielding esters of the fat (biodiesel) and a
glycerine byproduct. The resulting biodiesel product can be used as a fuel for diesel engines
with minor modifications and is commonly blended with refinery produced diesel fuel.
Biodiesel is defined in several sections of the Act, which we have used in formulating our
definition for the regulations. Biodiesel is registered with the EPA for commercial sale and is
legal for use at any blend level in both highway and nonroad diesel engines although most engine
manufacturers will only honor the warranty if biodiesel is used in blends of 2, 5 or 20 percent.
Biodiesel can be made from almost any vegetable or animal fat, with most of the world's
production coming from plants oils, notably soy bean and rapeseed (canola) oil. Biodiesel fuel
22
-------
production is rapidly increasing in many regions of the world. The choice of the feedstock oil
used to make it is dependent upon the vegetable oils and fat supplies that are economically
available. For the U.S. market, there are many potential plant oil feedstocks that can be used to
make biodiesel, including soybean, peanut, canola, cottonseed and corn oil. Biodiesel can also
be made from animal fats such used restaurant grease (yellow grease) and tallow. Though,
typically for the U.S. market, soybean oil has been the primary major feed stock supply,
followed by use of yellow grease and animal tallow.
Raw vegetable and animal oils consist of fatty acids and glycerine products. Though
these oils can directly be used in engines and give good short term performance, this is highly
discouraged as their use can cause severe engine problems. This is primarily due to the raw oils
forming engines deposits, with coking and plugging in engine injectors nozzles, piston rings,
lubricating oil, etc. This happens due to polymerization of the triglycerides in the raw oils as the
fuel is combusted. Therefore, it is necessary to convert the raw oils into a form of esters or
biodiesel which prevents theses issues. The biodiesel production process converts the raw
vegetable and animal oils into esters, though the virgin oils themselves are sometimes
(inappropriately) referred to as biodiesel. The production process called transesterification
consists of adding methanol or ethanol to the virgin vegetable oil and animal oil, in the presence
of a catalyst such as sodium or potassium hydroxide, resulting in esters or biodiesel and a
byproduct glycerol.
Biodiesel blends such as B2, B5 and in some cases B20, can be used in existing engines
without modification, and most engines exhibit no performance problems with the use of
biodiesel, though this depends on the blend and the season. However, engine fuel filters may
need to be changed more often, and there may be cold start problems due to biodiesel's higher
cloud point. As a result most engine manufacturers will only recognize their warranties if
biodiesel is used in low concentrations. Biodiesel produced from vegetable oil has practically
zero amounts of sulfur and aromatics and a high cetane value, thus making it a good for blending
into 15 ppm highway and offroad diesel fuel, though biodiesel made from yellow grease and
animal fat may contain about 24 ppm of sulfur10. Biodiesel also has good lubricity qualities
and can be used in concentration (~2 vol%) as a lubricity-enhancing additive for conventional
diesel.
1.2.4 Forecasted Biodiesel Production
Biodiesel production has been increasing rapidly over the past five years and is projected
to continue at a high rate in part because of the Renewable Fuel Standard (RFS) program. This
expansion has primarily been driven by better economics, due to the recent large increase in
diesel prices associated with the run up in crude prices, along with the Biodiesel Blenders Tax
Credit programs and the Commodity Credit Commission Bio-energy Program, both of which
subsidize producers and offset production costs. The Act extended the Biodiesel Blenders Tax
Credit program to year 2008, which provides about one dollar per gallon in the form of a federal
excise tax credit to biodiesel blenders from virgin vegetable oil feedstocks and 50 cents per
gallon to biodiesel produced from recycled grease and animal fats. This program was started in
2004 under the American Jobs Act. The existing Commodity Credit Commission Bio-energy
Program also pays biodiesel producers grants when the economics to produce biodiesel are poor;
23
-------
the program averaged about one dollar per gallon in 2004. Recent payments through the
Commodity Credit program have been reduced, however, and the program is expiring in fiscal
year 2006. Historically, the cost to make biodiesel was an inhibiting factor to production. The
cost to produce biodiesel was high compared to the price of petroleum derived diesel fuel, even
with consideration of the benefits of subsidies and credits provided by federal and state
programs. Another factor which is expanding the use of biodiesel are the mandates from states
and local municipalities, which require the use of biodiesel in transport fuels.
In 2005 approximately 91 million gallons of biodiesel were produced in the U.S. based
on program payments to biodiesel producers under USDA's Bio-energy Program. This volume
represents approximately 0.15 percent of all diesel fuel consumed in the domestic market. EIA
projects the future production volume to expand to 414 million gallons per year in 2007 and then
decrease to about 303 MM gallons per year in 2012, assuming that the biodiesel blender tax
credits program expires in 2007 (see Table 1.2-16).
Table 1.2-16. Estimated Biodiesel Production11
Year
2001
2002
2002
2003
2004
2005
2006
2007
2012
Million Gallons per Year
5
15
15
20
25
91
150
414
303
a Historical data from 2001-2004 obtained from estimates from John Baize " The Outlook and Impact of
Biodiesel on the Oilseeds Sector" USDA Outlook Conference 06. Year 2005 data from USDA Bioenergy Program.
Year 2006 data from verbal quote based on projection by NBB in June of 06. Production data for years 2007 and
higher are from EIA's AEO 2006.
With the increase in biodiesel production, there has also been a corresponding rapid
expansion in biodiesel production capacity. Presently, there are 65 biodiesel plants in operation
with an annual production capacity of 395 million gallons per year11. The majority of the current
production capacity was built in 2005, and was first available to produce fuel in the last quarter
of 2005. Though capacity has grown, historically the biodiesel production capacity has far
exceeded actual production with only 10-30 percent of this being utilized to make biodiesel, see
Table 1.2-17.
24
-------
Table 1.2-17. U.S. Production Capacity History
Plants
Capacity
(MM gals/yr)
2001
9
50
2002
11
54
2003
16
85
2004
22
157
2005
45
290
2006
53
354
Note: Capacity Data based on surveys conducted.
Excess production capacity is not easily quantified, though some of these plants may not
run at full rate all of the time and may be "idled" for certain days of the week, seasons, time of
day, etc. The capacity can be classified into two types of producers; capacity dedicated to
biodiesel production and capacity available from the ole-chemical industry. The plants that
primarily operate in the ole-chemical industry produce esters for use in the chemical industry.
These plants are swing producers of biodiesel, as when the economics are favorable they can
shift their operations and make biodiesel esters, instead of products for the ole-chemical market.E
The capacity from the ole-chemical industry, produces mono-alkyl esters using a similar
transesterification process, with the ester products being sold for to make plasticizers, soaps,
paints, solvents and other industrial uses. Additionally, the biodiesel production capacity
volumes may be optimistic, as this is not officially tracked. The capacities listed here are those
based on each company's self reported volumes to and may have some inaccuracies due to
informal reporting procedures.
We anticipate that future capacity additions will be geared more towards production of
biodiesel for use as transportation fuel, rather than serving primarily the oleochemicals markets.
In early 2006, there were 58 plants in the construction phase, which when completed would
provide 714 billion gallons per year of additional throughput capacity. Of these facilities, fifty
are new while eight are expanding capacity to their existing plants. This planned capacity is
likely to be built, since the equity has been raised and the new plants are actively being built at
the site of production. Also in early 2006, there were approximately 36 plants with a capacity of
754.7 billion gallons/year in the preconstruction phase (i.e. raising equity, permitting, conceptual
design, buying equipment) but had not started construction. For these plants, it is not as likely
that they will be completed since industry capacity, equity financing and other issues may alter
the economics for new plants. Table 1.2-18 presents the data for the biodiesel plant capacities
per the categories discussed.
E Oleochemicals are derived from biological fats and oils using hydrolysis or alcoholysis with products of fatty acid
esters and glycerol.
25
-------
Table 1.2-18. Biodiesel Plant Capacities
Total Plant Capacity,
(MM Gallon/year)
Existing Plants
(53 total)
354
Construction Phase
(58 total)
714
Pre-Construction
Phase
(36 total)
754.7
Considering that it takes 12 to 18 months to construct a biodiesel plant (from project
feasibility analysis to startup), a large portion of the capacity in the construction phase in early
2006 will be available to produce fuel in early 2007.12 Data on biodiesel plant construction
reveal that most of the new capacity that is currently being constructed is expected to be online
and producing fuel in 2006 or by early 2007. Therefore, the existing capacity plus the capacity
in the construction phase totals an aggregate amount of about one billion gallons per year.
Though there is no volume mandate for biodiesel fuel under the RFS program, the total capacity
available from new and existing plants exceeds EIA's projected biodiesel volume of 414 MM in
2007 and 303 MM in 2012 by a wide margin.
The plants in the construction and preconstruction phase are larger than existing biodiesel
plants. The average capacity of existing plants is 6.7 MM gallons per year, while plants in
construction phase are averaging 7.7 MM gallons per year, and plants in the preconstruction
phase are averaging 22.1 MM gallons per year and are presented in Table 1.2-19. The
distribution of biodiesel plants by size and number of companies within each size range are
presented in Table 1.2-20.
26
-------
Table 1.2-19. Average Plant Capacity by Feedstock (MM gallons per year)
Feedstock
Canola
Multi Feedstock
Other Vegetable
Recycled Cooking Oil
Soybean Oil
Tallow
Existing*
5.7
3.0
0.5
8.7
5.0
Construction*
6.5
8.6
30.0
0.3
9.1
Pre-Construction*
50.0
18.4
8.3
30.3
Table 1.2-20. Biodiesel Plant Size versus Number of Companies
Plant Size (MM
gallons per year)a
<1.00
1.0-5.0
5.0-10.0
10.0 to 15 .0
15.0 to 20.0
20.0+
Average Plant Size
Existing Plants
(58 total)
12
26
O
6
1
5
6.7
Construction Phase
(36 total)
12
15
8
1
1
5
7.7
Pre-Construction
Phase (22 total)
1
3
5
3
1
9
22.1
"Total capacity of plants in each category; existing plants is 354 MM gal/yr, construction phase is 324 MM
gal/yr, and pre-construction is 485 MM gal/yr.
Because newer plants are likely to be larger than existing plants, have better technology
and may have greater alignment with feedstock and feed sources, some of the older plants may
operate at an economic disadvantage once the new plants come on line. At the moment, it is not
possible to predict actual biodiesel production based on capacity, since in the past the capacity
was used at rates less than maximum. Thus, how excess production capacity evolves will be
dictated by economics, profitability, and fuel demand.
The majority of existing biodiesel plant capacity is located in the middle and midwestern
parts of the country and use soy bean oil as the feedstock. The other plants are scattered with
locations based on the east and west coasts, with feedstocks based on use of soybean, canola and
other oils as well as yellow grease as the feedstock. The new plants in the construction and pre-
construction phase, are being built to process a wider variety of feedstocks, with multi feedstock
and recycle grease capability. The feedstocks for these plants are listed in Table 1.2-21.
27
-------
Table 1.2-21. Feedstock Selection for Biodiesel Producers
Feedstock
Camelia
Canola
Cottonseed
Multi Feedstock
Palm Oil
Recycled Cooking Oil
Soybean oil
Tallow
Unknown
Existing
1
13
1
5
30
1
2
Construction
1
2
8
8
23
Pre-Construction
1
10
3
7
1
1.2.5 Baseline and Projected Biodiesel Volumes for Analysis
For cost and emission analysis purposes, three biodiesel usage cases were considered: a 2004
base case, a 2012 reference case, and a 2012 control case. The 2004 base case was formed based on
historical biodiesel usage (25 million gallons as summarized in Table 1.2-16). The reference case
was computed by taking the 2004 base case and growing it out to 2012 in a manner consistent with
the growth of gasoline (described in Section 2.1.3). The resulting 2012 reference case consisted of
approximately 28 million gallons of biodiesel. Finally, for the 2012 control case, forecasted
biodiesel use was assumed to be 300 million gallons based on EIA's AEO 2006 report (rounded
value from Table 1.2-16). Unlike forecasted ethanol use (described in 2.1.4), biodiesel use was
assumed to be constant at 300 million gallons under both the statutory and higher projected
renewable fuel consumption scenarios.
1.3 Renewable Fuel Distribution
1.3.1 Current Renewable Fuel Distribution System
Ethanol and biodiesel blended fuels are not currently shipped by petroleum product
pipeline due to operational issues and additional cost factors.13 The ability to ship by pipeline is
also limited because the sources of ethanol and biodiesel are frequently not in the same locations
as the sources of gasoline and petroleum-based diesel fuel. Hence, a separate distribution system
is needed for ethanol and biodiesel up to the point where they are blended into petroleum-based
fuel as it is loaded into tank trucks for delivery to retail and fleet operators. Ethanol and
biodiesel can either be added by "splash blending" where the renewable is added separately to
the tank truck, or by in-line injection where the renewable is injected into the petroleum fuel
stream as it is being dispensed into the tank truck. Ethanol and biodiesel are sometimes added to
petroleum-based fuels downstream of the terminal, but this is accounts for little of the total
volume of used.
28
-------
In cases where ethanol and biodiesel are produced within 200 miles of a terminal,
trucking is often the preferred means of distribution. However, most renewable fuel volumes are
produced at greater distances from potential centers of demand. For longer shipping distances,
the preferred method of bringing renewable fuels to terminals is by rail and barge. Dedicated
pipelines have not been used to distribute renewable fuels to terminals due to the high cost of
installing new pipelines, the relatively large shipping volumes that would be needed to justify
such expenditures, and the fact that renewable fuel production facilities tend to be relatively
numerous and dispersed.
The relatively low volumes of ethanol used prior to 2002 constrained the ability of the
distribution system to efficiently move ethanol to distant markets. Ethanol shipments by rail
were typically made on an individual car basis. Under such an approach, small groups of rail
cars travel to market as part of trains that carry other goods. This approach results in relatively
high transportation costs, longer transit times, and potential delays in delivery. Substantial
improvements in the efficiency of distributing ethanol by rail are being made due the need to
move large volumes of ethanol over long distances as a consequence of the elimination of MTBE
in California, New York, and Connecticut beginning in 2004. The use of unit trains, sometimes
referred to as "virtual pipelines" reduces delivery costs, shortens delivery times, and improves
reliability. Unit trains are composed entirely of approximately 100 rail cars containing ethanol.
Ethanol shipped by unit trains is delivered to hub terminals for further distribution to other
terminals by barge and tank truck.
Substantial volumes of ethanol can potentially be shipped down the Mississippi river by
barge for temporary storage in New Orleans.14 From New Orleans, ethanol can be loaded onto
ocean transport for delivery to the East and West Coast. There is also potential to move ethanol
via the Missouri and Ohio as well as other river systems and the Great Lakes. Marine shipments
of ethanol require a relatively large minimum shipment size, determined by the minimum size of
the marine tank compartment.F Similar to the case for "unit trains", there are also efficiencies in
dedicating whole barges, barge tows, or marine tankers to ethanol distribution. The increased
demand for ethanol has made it possible to better benefit from these efficiencies of scale.
The use of inland barges to transport ethanol from production facilities is in large part
driven by whether there is river access at such facilities. Historically, corn prices tend to be
higher near river systems that serve as arteries for the export of corn than at inland locations
distant from these river systems. To take advantage of lower corn prices at inland locations and
to avoid competing for corn with grain elevators that serve the export market, all of the new
ethanol production facilities that have been built since 1999 have been built at inland locations.15
Consequently, the majority of the growth in ethanol freight volumes since 1999 has been in the
rail sector.
1.3.2 Changes to the Renewable Fuel Distribution System Due to Increased Demand
This section addresses that changes that we expect will take place in the renewable fuel
distribution system in response to the anticipated increase in demand for such fuels through
F River barges typically have a capacity of 10,000 barrels. Ocean barges typically have a capacity of 20,000 barrels.
Barges are sometimes subdivided into 2 or 3 compartments.
29
-------
2012. There may be some limited opportunity to ship renewable fuels by pipeline in the future
as demand increases. However, because of the constraints discussed previously (see section
1.3.1), we believe that rail and barge are likely to remain the predominant means of
transportation. The 2002 DOE Study also reached this conclusion.16 While this constraint on
the ability to ship ethanol and biodiesel by pipeline presents logistical challenges that result in
additional transportation costs, the need to transport these alternative fuels by other means may
work to the overall advantage of the fuel distribution system. Petroleum product pipelines are
nearing capacity. Thus, it seems likely that the pipeline distribution system will find it
increasingly difficult to keep pace with annual increases in the demand for transportation fuels.
Displacing some of the volume of transportation fuels from the pipeline distribution system
through the use of ethanol and biodiesel will relieve some of this strain.
Small volume rail shipments made on by-car basis are likely to remain an important
feature in supplying markets that demand limited volumes. However, as the demand for ethanol
increases we anticipate that the expansion of the use of unit trains will continue, and that this will
be a significant means of bringing ethanol to distant markets. There has been some expansion of
capacity at existing ethanol plants with river access and some new plants are projected to be built
with river access. However, we anticipate that most new ethanol capacity will not have river
access. In addition, at least one new ethanol plant slated for production that does have river
access is planning to move its ethanol to market via rail. Nevertheless, in cases where rail is the
means to transporting ethanol to hub terminals, marine transport can play an important role in
further distribution to satellite terminals.
Substantial improvements to the rail, barge, tank truck, and terminal distribution systems
will be needed to support the transport of the volumes of renewable fuels necessary to meet the
requirements of the RFS program. These improvements include the addition of a significant
number of additional rail cars, and tank trucks. Additional marine barges will also be needed.
To facilitate the increased use of unit trains, new rail spurs will be needed at terminals.
Terminals will also need to add facilities to store and blend ethanol. In addition, those terminals
and retail facilities that had not previously handled ethanol blended fuel will need to make
certain one-time upgrades to ensure the compatibility of their systems with ethanol. These type
of changes have been occurring as demand for ethanol and biodiesel has grown rapidly over the
last several years, and there is no reason to suspect that they would not continue as demand
continues to warrant it. The costs associated with these changes are discussed in Chapter 7.3 of
this DRIA.
In the past, the refining industry raised concerns regarding whether the distribution
infrastructure can expand rapidly enough to accommodate the increased demand for ethanol.
The most comprehensive study of the infrastructure requirements for an expanded fuel ethanol
industry was conducted for the Department of Energy (DOE) in 2002 .17 The conclusions
reached in this study indicate that the changes needed to handle the increased volume of ethanol
required under the RFS will not represent a major obstacle to industry.0 While some changes
have taken place since this report was issued (as discussed below), we continue to believe that
the rail and marine transportation industries can manage the increased growth in an orderly
fashion. This belief is supported by the demonstrated ability of the industry to handle the rapid
' See section 7.3 of this DRIA regarding the projected costs of the necessary infrastructure improvements.
30
-------
increases and redistribution of ethanol use across the country over the last several years as
MTBE was removed. Given that future growth in ethanol use is expected to take place in an
orderly fashion in response to economic drivers, we anticipate that the distribution system will be
able to respond appropriately.
The use of unit trains has accelerated beyond that anticipated in the 2002 DOE report,
leading to the more efficient distribution of ethanol by rail. As a result, rail has taken a relatively
greater role in the transportation of new ethanol volumes as compared to shipment by barge than
was projected in the report. Thus, there is likely to be a relatively greater demand on the rail
distribution system and somewhat less demand on the marine distribution system than was
projected in the study.
The 2002 DOE study estimated that the increase in the volume of ethanol shipped by rail
needed to facilitate the use of 10 billion gallons of ethanol annually would represent an increase
in total tank car loadings of 0.33 percent. The increase in tank car loadings for Class I railroads
was estimated at 4.75 percent. The DOE report concluded that this increase is relatively modest
by railroad industry standards and could be accommodated given the available lead time. The
DOE study estimated that the increase in demand on barge movements due to the need to carry
an increased volume of ethanol would equate to a one percent increase in the total tonnage
moved by barge. Given that on the one hand relatively few new ethanol plants are projected to
be cited with river access, and that on the other hand barge is expected to play an important role
in redistributing ethanol from rail hub terminals, we estimate that the increase in barge
movements may be 30 percent less than that projected in the 2002 DOE study. This equates to
an increase in total tank car loadings of 0.44 percent rather than the 0.33 percent projected in the
DOE study. We believe that this relatively modest potential increase in the demand on the rail
distribution system can be accommodated without major difficulty given the available lead time.
Although, the 2002 DOE study generally concluded that the projected one percent
increase in the demand on the river barge industry could be accommodated without major
difficulty, it highlighted two potential concerns. The report noted that delays are already being
experienced at locks on the Mississippi river. The question was raised regarding how the
projected increase of one percent in river traffic due to increased ethanol shipments might be
accommodated at these locks. The report also raised concerns regarding the availability of
sufficient marine vessels capable of traveling between two ports in the United States (Jones Act
compliant vessels). Given that it appears that there will be less demand placed on the river
barge industry to transport ethanol than was projected in the 2002 DOE study, the concerns
raised in the study regarding the capability of the inland waterway system to cope with the
increased traffic associated with shipping the anticipated new volume of ethanol will be less
pronounced.
At the present time, the industry is experiencing a shortage of tractor trailers and drivers
to transport ethanol. The boom in demand for truck transport is due to a number of factors,
including the precipitous removal of MTBE from gasoline and its replacement by ethanolH
H MTBE is typically blended with gasoline at the refinery. MTBE production plants are often located nearby to
refineries allowing transport to the refinery by dedicated pipeline. In cases where, the sources of MTBE are more
-------
which has taken place when the demand for truck transport was already growing at a rapid place
due to the increased imports. The implementation of EPA's ultra-low sulfur diesel (ULSD)
program this summer may also cause an increase in the demand for tank trucks if more trucks
must be dedicated to ULSD service. Given the gradual increase expected from year to year in
ethanol production, we anticipate that the industry will be able to add sufficient additional tank
truck service in an orderly fashion with out undue burden.
The necessary facility changes at terminals and at retail stations to dispense ethanol
containing fuels have been occurring at a record pace due to the removal of MTBE from
gasoline. The use of ethanol has also become more economically attractive due to higher
gasoline prices. Now that MTBE has been removed, a more steady increase in the use of ethanol
is anticipated over time. This will also allow for a smooth transition for terminals and retail
operators.
The volumes of biodiesel that are expected to be used by 2012 to comply with the RFS
will be relatively modest (approximately 300,000,000 gallons). Consequently, we anticipate that
biodiesel will continue to be distributed to terminals by tank truck and by individual rail car
shipments. One hundred percent biodiesel (B100)1 forms wax crystals when the temperature
falls to 35 to 45 degrees Fahrenheit/ Thus, storage tanks for B100 need to be heated to maintain
flow-ability during the cold seasons. Shipping vessels used to transport B100 such as barges, rail
cars, and tank truck containers also typically must either be insulated (and sometimes heated)
during the cold season or alternately facilities can be provided at the terminal to reheat the vessel
prior to delivery. Biodiesel that is blended with diesel fuel and enhanced with cold flow
additives (if needed) can have comparable cold flow performance to petroleum based diesel
fuel.K
As temperatures fall during the cold seasons, some terminals currently avoid the need for
heated B100 tanks and facilities to heat shipping vessels by accepting progressively less
concentrated biodiesel blends (for final blending to produce fuels for use in vehicles). During
the warm seasons, such terminals typically accept B100 or B99.9. As the weather grows colder,
the terminal might switch to accepting B80 and during the coldest parts of the year might accept
B50 (that is blended with 50 percent number one diesel fuel). The need for insulated tank trucks
and tank cars is also sometimes avoided if transit times are brief by shipping warmed biodiesel.
We believe that as the volume of biodiesel grows, most terminals will opt to receive B100 (or
B99.9) year round for blending into diesel fuel for the consistency in operations which this
distant from the refinery, barge and rail are the preferred means of transport and relatively little MTBE is transported
by truck.
1 The concentration of biodiesel in a biodiesel blend is indicated by the number following the "B" designation. For
example, B99.9 indicates a biodiesel blend containing 99.9 percent biodiesel, and B80 indicates a blend containing
80 percent biodiesel. Manufactures of biodiesel sometimes blend in one tenth of one percent diesel fuel into
biodiesel to create B99.9 prior to shipping the fuel to terminals to create more dilute biodiesel blends so that the
producer can claim the biodiesel tax credit (pursuant to Internal Revenue Service requirements).
1 The point at which wax crystals form is referred to as the cloud point. The cloud point of B100 varies depending
on the feed stock used in its production.
K The relatively low concentration biodiesel blends that are typically used in vehicles (up to 20% biodiesel) can be
formulated to have comparable cold flow performance to petroleum based diesel fuel. Thus, there is no need to heat
such biodiesel blends in vehicle fuel tanks.
32
-------
practice offers. A number of terminals are already following this practice. These terminals have
installed heated storage tanks for biodiesel and insist that biodiesel be delivered in insulated tank
trucks (or rail cars) so that it may be pumped into the terminal storage tank without concern
about the potential need for reheating. The cost of the necessary heated and/or insulated
equipment is not insignificant. However, the modest additional volumes that will need to be
shipped via rail and tank due to the use of biodiesel do not materially affect the conclusions
reached above regarding the ability of the fuel distribution system to cope with the increased
volumes of renewable fuels.
1.4 Blenders
1.4.1 Ethanol Blending
Ethanol is miscible with water, and thus can introduce water into the distribution system
causing corrosion and durability problems as well as fuel quality problems. For this reason,
ethanol is blended downstream at terminals or into tank trucks.
The distribution of ethanol to be blended is described in more detail in Section 1.3.
Briefly, ethanol producers provide ethanol either directly to terminals, to marketers or to
terminals that are owned by refiners. In the first case, ethanol is provided to terminals that are
owned entities other than refining companies. They receive ethanol from the ethanol producer,
and gasoline from any number of refiners. The blenders than add ethanol to the gasoline at the
terminal. For RFG, the terminals receive the blendstock for RFG, called Reformulated
Blendstock for Oxygenate Blending or RBOB, to which they add the amount of ethanol called
for on the Product Transfer Document that accompanies such shipments. Once the ethanol is
added to the RBOB, the product becomes a finished gasoline (RFG) and is sent via truck to
retailers. For conventional gasoline (CG) ethanol is also added and shipped to retailers. The
tracking mechanism for CG is not as detailed as it is for RFG, however. The majority of ethanol
that is blended into CG is "splash-blended" although an increasing volume of ethanol is being
blended into special blends of conventional gasoline (e.g. sub-octane). Finally, a very small
amount is blended as E85 or made into ETBE.
1.4.2 Biodiesel Blending
Biodiesel generally leaves the production facility in its neat form and is shipped by truck
to locations where it can be blended with conventional diesel fuel. The blending generally
occurs at centralized distribution points such as terminals, although it also sometimes occurs
within tank trucks themselves. Biodiesel is only rarely used in its neat (unblended) form.
1.5 Imports/Exports of Renewable Fuel
Since the early 1980s, the U.S. has maintained a 54 cent per gallon tariff on imported
ethanol, primarily to offset the blending tax subsidy of the same magnitude that had been put in
place to support alternative energy production and domestic agriculture. Legislation and
agreements implemented since then have waived or significantly reduced the tariff on imports
33
-------
from Canada, Mexico, and about two dozen Central American and Caribbean nations covered by
the Caribbean Basin Initiative (CBI). Under the Caribbean Basin Economic Recovery Act,
which created the CBI, these countries can export ethanol duty free to the U.S. at a rate up to 7%
of the U.S. fuel alcohol market; quantities above this limit have additional stipulations for
feedstocks being grown within the supplying country.
Historically, the CBI nations have had little ethanol production capacity of their own but
have supplemented it by importing Brazilian ethanol and re-exporting it to the U.S. duty free.
More recently, with the rapid phase-out of MTBE and the high price of ethanol, it has become
economically viable to import significant quantities of ethanol directly from other nations despite
the tariff. Brazil, currently the largest ethanol producing nation in the world, has become the
largest single country supplier to the U.S. market. As shown in Figure 1.5-1, total imports have
increased more than 30% in 2004-5 over the previous three-year average.
Figure 1.5-1. Historic U.S. Ethanol Import Volumes and Origins11
a
o
=3
cd
O
a
o
2001
2002
2003
2004
2005
a P.O. Licht, "World Ethanol Markets, The Outlook to 2015" (2006). Gross imports (does not account for
export volumes) including hydrous, dehydrated, and denatured volumes.
Going forward, as domestic ethanol production capacity increases rapidly, its price is
expected to fall back into the historic range of 30-40 cents per gallon above gasoline (before
blending subsidy). This is expected to once again make direct imports from Brazil and other
full-tariff producers less attractive, and to decrease total imports. According to a current report
by P.O. Licht, U.S. net import demand is estimated to be around 300 million gallons per year by
2012, being supplied primarily through the CBI, with some direct imports from Brazil during
times of shortfall or high price.18
34
-------
Changes in the production and trade climate may influence this however. The Caribbean
countries with duty free status are seeing both internal and foreign investment to increase ethanol
production capacity significantly over the next several years, making more cheap imports
available. It is unclear at this point what volume of ethanol will be supplied through these
channels.
On the export side, the U.S. has averaged about 100 million gallons per year since 2000,
mostly to Canada, Mexico, and the E.U. Figure 1.5-2 shows historical U.S. exports. There is a
trend over the past five years of exporting larger quantities to fewer countries, with declining
volumes to Asia and increasing volumes to the E.U. and India. The demand for ethanol in all
these areas remains strong, and it appears that Asian imports from Brazil and China are making
up for the decrease in U.S. ethanol moving into the region.
Figure 1.5-2. Historic U.S. Ethanol Export Volumes and Origins21
Canada H Mexico
Japan D EU
India ffl Others
2001
2002
2003
2004
2005
a P.O. Licht, "World Ethanol Markets, The Outlook to 2015" (2006). Gross exports (does not account for
import volumes), includes hydrous, dehydrated, and denatured volumes.
These numbers are expected to increase modestly as more production comes online, with
more dramatic increases possible during periods of depressed domestic prices or stock surges.
Looking out over the next decade, the E.U. has a biofuels directive in place that will bolster
demand, and Japan and South Korea are expected to increase their use of biofuels steadily as
well. World ethanol production is projected to grow from the current 10 billion gallons per year
35
-------
to more than 25 in 2015, and the international biofuels markets are just beginning to take shape.
During this period we can expect significant changes in who is supplying and who is demanding
as the players determine their places and forge agreements on subsidies and tariffs. As of 2005,
the U.S. became largest ethanol producing nation, eclipsing Brazil, and ample foreign markets
will be available if conditions are right.
36
-------
Chapter 2: Changes to Motor Vehicle Fuel Under the Renewable
Fuel Standard Program
As described in the preamble, we developed two scenarios representing renewable fuel
volumes produced in 2012, the year when the Renewable Fuel Standard (RFS) program will be
fully phased in. The first scenario represented the statutorily required minimum volume of 7.5
billion gallons, while the second scenario reflected the 9.9 billion gallon volume estimated by the
Energy Information Administration (EIA). These two control cases were compared to a
reference case to determine the impacts of incremental use of renewable fuel.
In order to evaluate the economic and environmental impacts of the control and reference
cases, it was first necessary to evaluate the impacts of renewable fuels on the motor vehicle fuel
pool. In this context, we investigated a number of relevant issues for both current and future
renewable fuel use scenarios, with a particular focus on the use of ethanol in gasoline:
• What factors drive ethanol use?
Where is ethanol used (geographically and by fuel type)?
• When is ethanol blended into gasoline (seasonal differences)?
• How will other fuel properties change when ethanol is blended into gasoline?
Our analysis of these issues led us to estimate the amount of ethanol used in each state,
each fuel type (reformulated gasoline, oxygenated gasoline, and conventional gasoline) for each
season. These ethanol use estimates were then used as the basis for our emissions and air quality
analyses as well as our estimates of production, distribution, and blending costs.
In Section 2.1, we estimate the volumes of renewable fuels (namely ethanol) which are
currently being used in the U.S. as a whole and by state, and we forecast the volumes into the
future. We also project the geographical and seasonal distribution of ethanol use. These
analyses led us to our reference and control cases. In Section 2.2, we estimate the impact that
ethanol blending and the removal of methyl tertiary butyl ether (MTBE) will have on gasoline
properties. Section 2.3 summarizes our estimate of the effect of blending biodiesel into
conventional diesel fuel.
2.1 Gasoline/Oxygenate Use
Fuel ethanol use has steadily increased over the past decade due to its high gasoline
octane value, increasing availability, and more recently due to a series of state MTBE bans and
extremely favorable economics. Over the past four years, ethanol consumption has more than
doubled from 1.7 billion gallons in 2001 to 4.0 billion gallons in 2005.19 This growth in
domestic ethanol use shows no signs of stopping any time soon, especially given today's record-
high crude oil prices.
37
-------
In August 2005, the president signed into law the Energy Policy Act of 2005 (the Energy
Act), creating a national Renewable Fuels Standard (RFS). This RFS program institutes a
requirement for renewable fuel consumption beginning with 4.0 billion gallons in 2006 and
growing to 7.5 billion gallons in 2012. Despite the forecasted expansion in biodiesel, ethanol is
expected to continue to dominate U.S. renewable fuel consumption in the future. As such, the
nation is on track for meeting if not exceeding the RFS, and the use of ethanol is expected to
more than double again over the next six years.
To understand the impact of the increased ethanol use on gasoline properties and the
corresponding impact on overall air quality, we first need to gain a better understanding of where
ethanol is used today and how the picture is going to change in the future. We begin Section 2.1
of the draft regulatory impact analysis (DRIA) by discussing current ethanol use and go onto
examine four potential 2012 ethanol use/distribution scenarios (control cases). We arrive at four
different 2012 control cases based on the uncertainty of future ethanol use (EIA predicts ethanol
consumption will exceed the minimum RFS requirements) and the uncertainty of the distribution
of ethanol into reformulated gasoline (we predict that refiners and thus RFG areas may behave
differently in response to the recent removal of the RFG oxygenate requirement). An more in
depth discussion is described below.
2.1.1 Why are oxygenates currently blended into gasoline?
The blending of oxygenates into gasoline dates back to the 1970's. However, their use
expanded greatly in response to the Clean Air Act (CAA) amendments of 1990. Areas found to
be out of compliance (i.e., in non-attainment) with the National Ambient Air Quality Standards
(NAAQS) for carbon monoxide were required to use oxygenated fuel, and areas with the worst
ambient ozone levels were required to use reformulated gasoline (containing oxygenate).
Oxygenates have also been used in gasoline for other reasons, including state mandates and as a
volume extender. This section summarizes the current driving forces behind gasoline oxygenate
use in the U.S.
2.1.1.1 Federal Reformulated Gasoline Program
Areas found to be in non-attainment with the ozone standard are required to use
reformulated gasoline (RFG) year-round. In 2004, the Federal RFG program contained a
minimum oxygenate requirement as well as other fuel quality standards.L Adding oxygen to
gasoline is one way to reformulate gasoline to reduce the production of smog-forming pollutants
that contribute to unhealthy ground-level ozone. In addition to the ozone non-attainment areas
required to use oxygenate in gasoline, several states/areas also opted into the Federal RFG
program (otherwise known as "opt-in"). Additionally, some states/areas (namely California and
Arizona) have state-implemented programs which require or promote the use of oxygenated
gasoline.
L RFG minimum oxygenate requirement found at 40 CFR 80.41(f). This requirement was effective for 2004 but has
since been eliminated by the Act (Section 1504). Final rule promulgated on May 8, 2006 at 71 FR 26691.
38
-------
A list of the 2004 Federal RFG areas and their corresponding oxygenate(s) is provided in
Table 2.1-1. For the purpose of this analysis, only ethanol (ETOH) and methyl tertiary-butyl
ether (MTBE) have been considered as oxygenates.M
MOther low-usage oxygenates (e.g. ETBE, TAME, etc.) were assumed to be negligible for the purpose of this
analysis.
39
-------
Table 2.1-1. 2004 Federal RFG Areas by State
2021
RFG Area Location
State City
California
Connecticut3
Delaware3
District of Columbia
Illinois
Indiana
Kentucky
Maryland
Massachusetts3
Missouri
New Hampshire
New Jersey3
New York
Pennsylvania
Rhode Island3
Texas
Virginia
Wisconsin
Los Angeles
Sacramento
San Diego
San Joaquin Valley
Hartford
Long Island Area
Rest of State
Philadelphia Area
Sussex County
Washington DC Area
Chicago Area
Chicago Area
Covington
Louisville
Baltimore
Kent & Queen Anne's
Philadelphia Area
Washington DC Area
Boston Area
Springfield
St. Louis
Boston Area
Atlantic City
Philadelphia Area
Warren County
Long Island Area
Essex Area
Long Island Area
Philadelphia Area
Providence Area
Dallas/Fort Worth
Houston
Norfolk
Richmond
Washington DC Area
Milwaukee
Total (Required+Opt-ln)
No. of
Counties
5
6
1
8
6
3
6
2
1
1
8
2
3
3
6
2
1
5
10
4
5
4
2
6
1
11
2
11
5
5
4
8
11
7
10
6
181
Type of
RFG Area
Required
Required
Required
Required
Required
Required
Opt In
Required
Opt In
Opt In
Required
Required
Opt In
Opt In
Required
Opt In
Required
Opt In
Opt In
Opt In
Opt In
Opt In
Opt In
Required
Opt In
Required
Opt In
Required
Required
Opt In
Opt In
Required
Opt In
Opt In
Opt In
Required
Oxygenate
Usedb
ETOH
ETOH
ETOH
ETOH
ETOH
ETOH
ETOH
MTBE
MTBE
MTBE
ETOH
ETOH
ETOH
ETOH
MTBE
MTBE
MTBE
MTBE
MTBE*
MTBE*
ETOH*
MTBE*
MTBE
MTBE
MTBE
ETOH, MTBE
ETOH
ETOH
MTBE
MTBE*
MTBE
MTBE
MTBE
MTBE
MTBE
ETOH
3Entire state operates under the RFG program
bOxygenate determination based on 2004 EPA RFG fuel survey results. An asterisk next
to the oxygenate name denotes the predominant oxygenate, but also indicates that there
were trace amounts (<3% by vol) of the other oxygenate (either MTBE of ETOH) found.
All other RFG oxygenate usage was assumed to be exclusive within a given area with the
exception of the NJ Long Island area (57/43 percent volume ratio of MTBE to ETOH).
40
-------
As shown above in Table 2.1-1, in 2004 a little more than half of the Federal RFG areas
(on a county-by-county basis) used MTBE as an oxygenate as opposed to ethanol. However, on
a volumetric basis, more ethanol was consumed than MTBE (2.2 billion gallons compared to 1.9
billion gallons as shown below in Table 2.1-3).
2.1.1.2
State Oxygenated Fuel Programs
In addition to the RFG program, several states require oxygenated fuel (oxy-fuel) to be
used in the wintertime to address carbon monoxide (CO) non-attainment. CO is formed from the
incomplete combustion of hydrocarbons (found in all gasoline blends). Production of the
poisonous gas is more prevalent in oxygen-deficient environments and more harmful to human
health in the wintertime due to temperature inversions.N Together, the winter oxy-fuel program
coupled with improving vehicle emissions control systems help to reduce CO emissions. Many
areas have and are continuing to come into attainment with the CO national ambient air quality
standards (NAAQS). However, many former non-attainment areas continue to use winter oxy-
fuel as part of a maintenance strategy for remaining in compliance with the CO NAAQS. A list
of the 2004 oxy-fuel areas is provided in Table 2.1-2. According to regional fuel contacts, all
oxy-fuel areas used ethanol as an oxygenate in 2004.
Table 2.1-2. 2004 State Implemented Oxy-Fuel Programs
22
Oxy-Fuel Area Location
State City
Alaska
Arizona
California
Colorado
Montana
Nevada
New Mexico
Oregon
Texas
Utah
Washington
Anchorage
Tucson
Phoenix
Los Angeles
Denver/Boulder
Longmont
Missoula
Las Vegas
Reno
Albuquerque
Portland
El Paso
Provo/Orem
Spokane
Oxy-Fuel
Period
11/1-2/29
10/1-3/31
11/2-3/15
10/1-2/29
11/1-1/31
11/1-1/31
11/1-2/29
10/1-3/31
10/1-1/31
11/1-2/29
11/1-2/29
10/1-3/31
11/1-2/29
9/1 -2/29
Carbon Monoxide Status
Designation Working on RDa
Non-attainment0
Attainment
Non-attainment
Non-attainment
Attainment
Attainment
Non-attainment
Non-attainment
Non-attainment
Attainment
Attainment
Non-attainment
Non-attainment11
Non-attainment6
X
X
X
X
X
X
X
Winter Oxy-Fuel Program
Required Part of MPb
X
X
X
X
X
X
X
X
X
X
X
X
X
X
Currently working on redesignation to attainment
bOxy-fuel program is part of CO maintenance plan.
°Area was redesignated to attainment effective 7/23/04. Oxy-fuel program will be a contingency measure.
dEPA has been granted enforcement discretion during redesignation process.
eArea was redesignated to attainment effective 8/29/05. Oxy-fuel program will be a contingency measure.
2.1.1.3
Other Reasons to Blend Ethanol
N Temperature inversions in the lower atmosphere are relatively common, especially during winter months in cold
climates. A temperature inversion occurs when cold air close to the ground is trapped by a layer of warmer air,
creating stagnation and trapping pollution close to the ground.
41
-------
In addition to the RFG and oxy-fuel programs, gasoline refiners have several other
motivations for blending oxygenate (namely ethanol) into gasoline. First and foremost, the state
they provide gasoline to could be operating under a state ethanol mandate. In 2004, Hawaii
joined Minnesota in approving a state ethanol requirement (10 volume percent ethanol required
in each gallon of gasoline). Second, blending ethanol into gasoline could help them meet their
mobile source air toxics (MSATI) performance standards as determined by the Complex
Model.0 Additionally, adding ethanol increases both octane and total fuel volume, thus helping
refiners extend their gasoline production. Finally, and perhaps most importantly, with record-
high crude oil prices and the growing availability of grain-based ethanol (especially in PADD 2),
ethanol use has become extremely economical. The 1.1 billion gallons of ethanol used in PADD
2 conventional gasoline in 2004 (see Table 2.1-4 of Section 2.1.2.2) is a good indicator of this
trend.
In addition to the increasing availability of ethanol, consumer demand is also increasing
based on the growing number of ethanol-friendly vehicles on the road. Conventional vehicles
consume the majority of fuel ethanol and are limited to gasoline with 10 volume percent ethanol
(E10) or less. However, there are now about five million flexible fuel vehicles (FFVs) on the
road today with more being produced and sold each day. FFVs are specifically designed to be
able to handle a wide range of gasoline/ethanol blends up to 85 percent ethanol, or E85.
2.1.2 Development of the Base Case
In order to evaluate the impact of increased ethanol use on gasoline, we had to develop a
2012 reference case as a point of comparison for the two 2012 control cases (discussed further in
DRIA Section 2.1.4). In order to develop the reference case, we first needed to establish a base
case or a historical foundation representing pre-RFS gasoline conditions. A more in-depth
discussion of how the base case was established is presented below.
2.1.2.1 Strategy for Establishing the 2004 Base Case
For the purpose of this draft regulatory impact analysis, the 2004 calendar year was
selected to reflect current (base case) conditions. This period represented the most current year
for which gasoline and oxygenate data were available and also captured the recent California,
New York, and Connecticut MTBE bans (effective 1/1/04) while avoiding the 2005 calendar
year hurricane upsets.
The approach for assembling the 2004 base case consisted of obtaining gasoline, ethanol,
and MTBE usage for all 50 states as well as the District of Columbia. As mentioned earlier,
other low-volume oxygenate use (e.g. ETBE, TAME, etc.) was assumed to be negligible and
thus ignored. All ethanol-blended gasoline was assumed to be E10, with the exception of
California gasoline (both Federal RFG and California Phase 3 RFG (CaRFG3)).p All California
0 This RFS proposal is based on MSAT1 conditions. Impacts of the recent MS AT2 NPRM which proposes to
remove individual refinery toxic performance standards (baselines) in exchange for a nationwide benzene standard
will be reflected in the analysis for that rulemaking.
p The small volumes of E85 (85 percent ethanol) gasoline have been ignored for this analysis.
42
-------
"RFG" was assumed to contain approximately 5.7 percent ethanol (E5.7) based on discussions
with California Air Resource Board (CARB) officials. This includes all California "RFG"
supplied to the Phoenix metropolitan area in the summertime under Arizona's clean burning
gasoline (CBG) program.Q Finally, MTBE use in the base case was assumed to occur in 11
volume percent proportions.
Total gasoline consumption was obtained from the 2004 Petroleum Marketing Annual
(PMA) report published by the Energy Information Administration (EIA).23 The reported annual
average sales volume for each state was interpreted as total blended gasoline (including
additives, namely oxygenates). 2004 MTBE usage by state was obtained from EIA.R24 The data
received was exclusive to states with RFG programs (including Arizona's CBG program). Thus,
for the purpose of the 2004 base case analysis, MTBE use was assumed to be limited to RFG
areas. 2004 ethanol usage by state was derived from a compilation of data sources and
assumptions. As a starting point, total domestic ethanol consumption was acquired from EIA's
June 2006 Monthly Energy Review.25 State ethanol contributions originated from the 2004
Federal Highway Administration (FHWA) gasohol report.26 However, there was some
ambiguity with the 2004 FHWA data. First, the total ethanol consumption did not match up with
EIA's reported value (3.7 billion gallons compared to 3.5 billion gallons). Second, the gasohol
(and thus ethanol) volumes were derived from potentially imprecise motor vehicle fuel tax
reports.8 And third, not all states using ethanol reported their gasohol usage so FHWA had to
model-estimate 19 states' ethanol usage (accounting for 60% of the total ethanol volume). To
improve upon the FHWA data, a series of oxygenate verification tools were applied including
knowledge of state ethanol mandates, state MTBE bans, Arizona's CBG program, as well as fuel
T7 0£
survey results. The state-by-state FHWA data was adjusted accordingly and allocated by
fuel type (RFG, CG, and/or oxy-fuel). The summarized oxygenate results are presented
throughout this section.
2.1.2.2 2004 Gasoline/Oxygenate Consumption by PADD
In 2004, 3.5 billion gallons of ethanol and 1.9 billion gallons of MTBE were blended into
gasoline to supply the transportation sector with a total of 136 billion gallons of gasoline. A
breakdown of the 2004 gasoline and oxygenate consumption by PADD is found below in Tables
2.1-3.
Q For the purpose of this analysis, all Arizona CBG was classified as "RFG". In 2004, wintertime Arizona "RFG"
was assumed to contain 10% ethanol (governed by the Phoenix oxy-fuel program). Summertime "RFG" was
assumed to be comprised of 2/3 California "RFG" (containing 5.7 percent ethanol) and 1/3 PADD 3 RFG
(containing either 10 percent ethanol or 11 percent MTBE in 2004).
R EIA reported 2004 total MTBE usage (in RFG) as 2.0 billion gallons. The reported MTBE usage was reduced
from 2.0 to 1.9 billion gallons under the assumption that CA, NY, and CT implemented their state MTBE bans on
time by 1/1/04. (EIA showed small amounts of MTBE use in these states in 2004). EIA's allocation of MTBE by
state was also adjusted based on fuel survey results. Most noteworthy, EIA reported MTBE usage in Arizona "RFG"
as zero. However, the 2004 Phoenix fuel survey results suggest otherwise. As such, an appropriate amount of
MTBE was allocated to Arizona based on the assumption that 1/3 of all summertime Arizona "RFG" resembles
PADD 3 RFG (which contained some level of MTBE in 2004).
s The U.S. Department of Treasury requires a distinction between gasohol and gasoline on motor vehicle fuel tax
reports for states with gasohol sales tax exemptions. These financial records are the source of FHWA's
gasohol/ethanol data. However, since state gasohol tax exemptions have become virtually nonexistent over the past
several years, gasohol reporting (namely the distinction between gasoline and gasohol) has suffered.
43
-------
Table 2.1-3.
2004 Gasoline & Oxygenate Consumption by PADD
PADD
PADD1
PADD 2
PADD 3
PADD 4
PADD 5b
California
Total
Gasoline Ethanol
MMaal MM gal %
49,193
38,789
20,615
4,542
7,918
14,836
135,893
660
1,616
79
83
209
853
3,500
1 .34%
4.17%
0.38%
1 .83%
2.63%
5.75%
2.58%
MTBEa
MM gal %
1,360
1
498
0
19
0
1,878
2.76%
0.00%
2.42%
0.00%
0.23%
0.00%
1.38%
aMTBE blended into RFG
bPADD 5 excluding California
As shown above, nearly half (or about 45 percent) of the ethanol was consumed in PADD
2 gasoline in 2004, not surprisingly, where the majority of ethanol was produced. The next
highest region of use was the State of California which accounted for about 25 percent of
domestic ethanol consumption. This makes sense since California alone accounts for over 10
percent of the nation's total gasoline consumption and all fuel (both Federal RFG and CaRFGS)
was presumed to contain ethanol in 2004 (following their recent MTBE ban) at 5.7 volume
percent.
In 2004, total ethanol use exceeded MTBE use. Ethanol's lead oxygenate role is
relatively new, however the trend has been a work in progress over the past few years. From
2001 to 2004, ethanol consumption more than doubled (from 1.7 to 3.5 billion gallons), while
MTBE use (in RFG) was virtually cut in half (from 3.7 to 1.9 billion gallons). A plot of
oxygenate use over the past decade is provided below in Figure 2.1-1.
44
-------
4.5 -,
4.0
3.5
=- 3.0
co
0)
m.
a 2.5
g 2.0
01
O)
O 1.5
Figure 2.1-1. Oxygenate Consumption Over Time
X
0.5
0.0
1995 1996 1997 1998 1999 2000 2001 2002 2003 2004
Year
•Ethanol Consumption, Total
•MTBE Consumption, RFG
Source: Energy Information Administration
The nation's transition to ethanol is linked to states' responses to recent environmental
concerns surrounding MTBE groundwater contamination. Traces of MTBE have been found in
both surface and ground water in and around RFG areas. The MTBE is thought to have made its
way into the water from leaking underground storage tanks, gasoline spills, and engines.
Concerns over drinking water quality have prompted several states to significantly restrict or
completely ban MTBE use in gasoline. At the time of this analysis, 19 states had adopted MTBE
bans. Ten states had bans that impacted the entire 2004 calendar year, four states had bans that
impacted a portion of the year, and five states had bans that will become effective in 2005 and
beyond. A list of the states with MTBE bans (listed in order of phaseout date) is provided in
Table 2.1-4.
45
-------
Table 2.1-4. States with MTBE Bans Enacted as of June 2004
State3
Iowa
Minnesota
Nebraska
South Dakota
Colorado
Michigan
California
Connecticut
New York
Washington
Kansas
Illinois
Indiana
Wisconsin
Ohio
Missouri
Kentucky
Maine
New Hampshire
MTBE Phaseout Date
07/01/00
07/02/00; 07/02/05
07/13/00
07/01/01
04/30/02
06/01/03
12/31/03
01/01/04
01/01/04
01/01/04
07/01/04
07/24/04
07/24/04
08/01/04
07/01/05
07/31/05
01/01/06
01/01/07
01/01/07
Type of MTBE Banb
Partial
Partial; Complete
Partial
Partial
Complete
Complete
Complete
Complete
Complete
Partial
Partial
Partial
Partial
Partial
Partial
Partial
Partial
Partial
Partial
aArizona is not included because they do not have an official state MTBE ban.
Arizona adopted legislation on 4/28/00 calling for a complete phaseout of MTBE
as soon as feasible but no later than 6 months after California's phaseout. The
legislation expired on June 30, 2001, so it's not official policy although the state
still informally encourages the phaseout of MTBE.
bA partial ban refers to no more than 0.5 vol% MTBE except in the case of MN
(1/3%), NE (1%), and WA (0.6%)
Source: U.S. EPA, State Actions Banning MTBE (Statewide), June 2004
In 2004, all remaining MTBE consumption was assumed to occur in reformulated
gasoline (explained in 2.1.2.1). As shown above in Table 2.1-3, 99 percent of MTBE use
occurred in PADDs 1 and 3. This reflects the high concentration of RFG areas in the northeast
(PADD 1) and the local production of MTBE in the gulf coast (PADD 3). PADD 1 receives a
large portion of its gasoline from PADD 3 refineries who either produce the fossil-fuel based
oxygenate or are closely affiliated with MTBE-producing petrochemical facilities in the area.
2.1.2.3 2004 Gasoline/Oxygenate Consumption by Season
In 2004, according to EIA Petroleum Marketing Annual (PMA), approximately 40
percent of gasoline was consumed in the summertime and 60 percent was consumed in the
wintertime.131 Similarly, according to EIA Monthly Energy Review June 2006, 38 percent of
T Reported seasonal splits for gasoline and ethanol (presented throughout this section) were computed based on RFG
production seasons (Summer: May 1 through September 15th; Winter: January 1st through April 30th and September
16th through December 31st).
46
-------
the ethanol was consumed in the summertime and 62 percent was consumed in the wintertime in
200432.
Total gasoline use is higher in the wintertime because it's a longer season. The RFG
regulations define summertime fuel as gasoline produced from May 1st to September 15th (4.5
months total).u The remaining 7.5 months are considered wintertime gasoline. Even though on
an average per day basis summertime consumption is higher, more gasoline is still sold and
consumed in the wintertime based on the length of the season.
Seasonal ethanol use follows the same general trend as gasoline. However, besides the
associated correlation with seasonal gasoline consumption, there are additional reasons why
2004 ethanol use may have been higher in the wintertime. First, the oxy-fuel program requires
oxygenate to be used in certain areas in the wintertime only. These same areas, which do not
require oxygenate in the summer, are all believed to use ethanol as their oxygenate (as described
in 2.1.1.2). Thus, more areas use ethanol during the winter months than the summer. Secondly,
there is an economic penalty associated with blending ethanol into summertime RFG. Refiners
supplying summertime gasoline to RFG areas have to remove butanes and pentanes from the
gasoline they produce in order to add ethanol and still comply with the Reid vapor pressure
(RVP) requirements.
2.1.2.4 2004 Gasoline/Oxygenate Consumption by Fuel Type
Of the 3.5 billion gallons of ethanol blended into gasoline in 2004, approximately 2.2
billion gallons were used in reformulated gasoline and the remaining 1.3 billion gallons were
used in conventional gasoline (including wintertime oxygenated fuel).v A breakdown of the
2004 ethanol consumption by fuel type and PADD is found in Table 2.1-5.
u We acknowledge that the aforementioned seasonal split does not exactly match the new summer/winter seasons
defined in the Energy Act (Summer: April 1st through September 30th; Winter: January 1st through March 31st and
November 1st through December 31st).
v Ethanol allocation to reformulated gasoline based on U. S. EPA Office of Transportation & Air Quality, 2004 RFG
Fuel Survey Results (http://www.epa.gov/otaq/regs/fuels/rfg/properf/rfgperf.htmX Ethanol allocation to
conventional gasoline based on Alliance of Automobile Manufacturers (AAM) North American Fuel Survey 2004
(report can be purchased at http://autoalliance.org/fuel/fuel survevs.php). Ethanol allocation to oxyfuel based on
knowledge of 2004 oxyfuel areas (refer to Table 2.1.2) and assumption that all oxyfuel contained ethanol in 2004
(according to regional fuel contacts).
47
-------
Table 2.1-5. 2004 Ethanol Consumption by Fuel Type
PADD
PADD1
PADD 2
PADD 3
PADD 4
PADD 5C
California
Total
Ethanol
CG
0
1,072
31
0
45
0
1,149
Consumption (MM gal)
OXYa RFGb
0
0
21
83
89
0
193 2
660
544
26
0
75
853
,158
Total
660
1,616
79
83
209
853
3,500
aWinter oxy-fuel programs
"Federal RFG plus CA Pha
CPADD 5 excluding California
"Federal RFG plus CA Phase 3 RFG and Arizona CBG
As mentioned above in Section 2.1.2.2, 100 percent of the 1.9 billion gallons of MTBE
blended into gasoline in 2004, was assumed to be consumed in reformulated gasoline.
2.1.2.5 2004 Gasoline/Oxygenate Consumption by State
In 2004, ethanol was blended into gasoline in 34 of the 50 states. No ethanol use was
observed in the remaining 16 states: Maine, New Hampshire, Vermont, Pennsylvania, Delaware,
Georgia, North Carolina, South Carolina, West Virginia, Tennessee, Oklahoma, Mississippi,
Arkansas, Louisiana, Idaho, and West Virginia, nor was any ethanol used in Washington DC. A
summary of these results are provided in Table 2.1-6 and Figure 2.1-2. Note that a state ethanol
percentage less than 10 indicates that only a percentage of the gasoline pool was blended with
ethanol, not that ethanol itself was blended in less than 10 volume percent (E10) proportions,
except in the case of California gasoline (E5.7).
The states consuming the highest volumes of ethanol in 2004 were California, Illinois,
New York, Minnesota, and Ohio, respectively. With respect to gasoline use, the highest
percentage of ethanol use occurred in Minnesota, Hawaii, Connecticut, Illinois, and Iowa. Four
out of the five states are not surprising. The first two states have ethanol mandates and the last
two are located in the "corn belt" where ethanol is produced. Connecticut's high percentage of
ethanol use may come as a surprise at first glance. However, the entire state operates under the
RFG program (refer to Table 2.1-1), and since they also have a state MTBE ban, ethanol is found
in each gallon of gasoline.
48
-------
Table 2.1-6.
2004 Gasoline/Ethanol Consumption by State
State
California
Illinois
New York
Minnesota
Ohio
New Jersey
Connecticut
Indiana
Missouri
Iowa
Wisconsin
Arizona
Colorado
Michigan
Kentucky
Hawaii3
Kansas
Texas
Nebraska
Alabama
Oregon
South Dakota
Nevada
Massachusetts
Washington
North Dakota
New Mexico
Alaska
Utah
Montana
Rhode lslandb
Maryland13
Floridab
Virginia13
Total
Gasoline
MM gal
14,836
5,177
5,626
2,684
5,156
4,235
1,522
3,059
3,159
1,635
2,471
2,187
1,999
4,861
2,177
452
1,396
1 1 ,948
819
2,392
1,500
434
857
2,934
2,621
350
966
302
1,097
503
490
2,480
8,605
3,920
104,853
Ethanol
MM gal
853
422
301
268
192
188
152
148
122
117
109
88
80
77
50
45
41
39
37
31
31
24
23
18
18
11
8
3
2
1
0
0
0
0
3,500
Percent
Ethanol
5.75%
8.14%
5.35%
10.00%
3.72%
4.43%
10.00%
4.84%
3.86%
7.14%
4.39%
4.04%
4.01%
1 .58%
2.29%
10.00%
2.92%
0.33%
4.54%
1.31%
2.05%
5.51%
2.69%
0.63%
0.68%
3.00%
0.83%
1.11%
0.17%
0.22%
0.06%
0.01%
0.00%
0.00%
3.34%
aHawaii was assumed to have a state mandate in the 2004 base
case (Source: Renewable Fuels Associaion, Homegrown for the
Homeland: Ethanol Industry Outlook 2005)
bTrace amounts of ethanol consumption (<1 MMGal) in Rl, MD, FL,
andVA
49
-------
Figure 2.1-2. 2004 Ethanol Distribution, % ETOH by State
% ETOH by State
CH < 1 % (trace)
CH 1 to <5%
• 5 to < 10%
CH 10%
Not Pictured
AK: 1% ETOH
HI: 10% ETOH
DC: 0% ETOH
2.1.3 Development of the 2012 Reference Case
To conduct the subsequent economic and environmental analyses, we compared a base
year without RFS fuel changes (reference case) to a modeled year with renewable fuel changes.
Or more accurately, we compared a 2012 reference case to four potential 2012 renewable fuel
control cases (discussed further in DRIA Section 2.1.4).
To establish the 2012 reference case, we started with the 2004 base case (presented in
Table 2.1-3) and grew out gasoline/oxygenate use according to the EIA AEO 2006 motor
gasoline energy growth rate from 2004 to 2012.33 Accordingly, in the resulting 2012 reference
case, ethanol and MTBE use was proportional to 2004 use both by region and fuel type. A
summary of the 2012 reference case is found in Table 2.1-7.
50
-------
Table 2.1-7.
2012 Gasoline & Oxygenate Consumption by PADD
(Reference Case)
PADD
PADD1
PADD 2
PADD 3
PADD 4
PADD 5b
California
Total
Gasoline Ethanol
MMaal MM gal %
54,788
43,201
22,959
5,059
8,819
16,523
151,349
735
1,800
88
93
232
950
3,898
1.34%
4.17%
0.38%
1 .83%
2.63%
5.75%
2.58%
MTBEa
MM gal %
1,515
2
555
0
21
0
2,092
2.76%
0.00%
2.42%
0.00%
0.23%
0.00%
1.38%
aMTBE blended into RFG
br
PADD 5 excluding California
2.1.4 Development of the 2012 Control Cases
In Section 2.1.2 we described the development of the 2004 base case, which was used to
develop the 2012 reference case as described in Section 2.1.3. In this section we describe the
development of the two 2012 control cases representing increased use of renewable fuel. As
described in the preamble, the first control scenario represented the statutorily required minimum
volume of 7.5 billion gallons, while the second control scenario reflected the 9.9 billion gallon
volume estimated by EIA. Both control scenarios were used in comparison to the reference case
to evaluate the economic and environmental impacts of increased use of renewable fuels.
2.1.4.1 Strategy for Forecasting Ethanol Consumption
As mentioned earlier in Section 2.1.2.2, groundwater contamination concerns have
caused many states to ban the use of MTBE in gasoline. In response to the Energy Policy Act
(the Energy Act) of 2005, essentially all U.S. refiners are expected to eliminate the use of MTBE
in gasoline in 2006 or 2007, and certainly prior to 2012. Ethanol consumption, on the other hand
is expected to continue to grow at unprecedented rates in the future. Not only is ethanol
replacing MTBE, ethanol will fuel the growing number of ethanol-friendly vehicles being
produced, as well as satisfy the growing number of state ethanol mandates (Washington and
Montana recently joined Minnesota and Hawaii).W34 Additionally, the Energy Act requires a
minimum amount of renewable fuels to be used beginning in 2006. By 2012, the Act requires
7.5 billion gallons of renewable fuels to be consumed domestically, most of which is expected to
be ethanol.
w Montana state mandate requires all gasoline to contain 10% ETOH once plant production ramps up to 40
MMgal/yr and Washington state mandate requires 20% of all gasoline to contain 10% ethanol by 12/1/08. At the
time of our analysis, these were the only two new state ethanol mandates. However, EPA recognizes that as of
7/13/06, several states have enacted biofuel standards (Iowa, Louisiana, and Missouri) and several others have
biofuel standards pending (California, Colorado, Idaho, Illinois, Indiana, Kansas, New Mexico, Pennsylvania,
Virginia, and Wisconsin) which mandate a minimum amount of ethanol use.
51
-------
However, it's predicted that renewable fuel consumption will be much higher than 7.5
billion gallons in 2012. In Annual Energy Outlook (AEO) 2006, EIA forecasts that by 2012,
total ethanol use (corn, cellulosic, and imports) would be about 9.6 billion gallons and biodiesel
use would be about 300 million gallons.35 A comparison between the EIA AEO 2006 forecasted
renewable fuel consumption and the Energy Act renewable fuels standard is presented below in
Table 2.1-8.
Table 2.1-8. Renewable Fuel Consumption Forecast
Year
2006
2007
2008
2009
2010
2011
2012
EIA AEO
2006 Forecasted Renewable Fuel
Consumption (Bgal)
Ethanol3 Biodiesel
4.1
5.2
6.0
6.9
7.9
8.8
9.6
0.2
0.4
0.4
0.4
0.3
0.3
0.3
Total
4.3
5.6
6.4
7.3
8.2
9.1
9.9
EPAct
Renewable
Fuels Standard
(Bqal)
4.0
4.7
5.4
6.1
6.8
7.4
7.5
aSum of corn ethanol, cellulosic ethanol, and imports
As shown above in Table 2.1-8, EIA's renewable fuel projection in 2012 (9.9 billion
gallons total) greatly exceeds the 7.5 billion gallon RFS requirement. More specifically, EIA
predicts that ethanol production alone would exceed the RFS in 2006 through 2012. The
projected AEO 2006 fuel consumption levels were estimated using EIA's LP refinery model. In
2012, EIA's renewable fuel projection was based on a crude oil price of $47/bbl, which is
significantly lower than today's crude oil price (tracking above $70/bbl at the time of this
analysis).X36 Therefore, current market conditions indicate that ethanol and biodiesel production
could be even more favorable and/or prevalent in the future based on economics. However,
EIA's AEO 2006 analysis also considers the feasibility of building production facilities to
accommodate the growing renewable fuel demand. As such, we interpret EIA's 2012 ethanol
and biodiesel projections to be reasonable estimates considering both economics and the rate at
which new plants could feasibly come on-line.
To summarize, it is abundantly clear that renewable fuel use is growing rapidly, faster
than the RFS requires. However quantifying future renewable fuel consumption, namely ethanol
growth, is a difficult task. The gasoline refining industry and ethanol industry are currently
undergoing a variety of changes/expansions and there's no definite way to know exactly how
things are going to "fall out" in the future. Accordingly, EPA has chosen to model two different
2012 renewable fuel consumption scenarios to represent a reasonable range of ethanol use - 7.2
billion gallons (based on the Energy Act RFS requirement less EIA's biodiesel projection) and
9.6 billion gallons (based on EIA's AEO 2006 ethanol projection). The Agency is not
concluding that ethanol consumption could not possibly exceed 9.6 billion gallons by 2012, but
x West Texas Intermediate (WTI) crude oil pricing was $70.84/bbl in May, 2006; $70.95/bbl in June, 2006; and
$74.41/bbl in July 2006 according to EIA.
52
-------
rather that this volume is a reasonable "ceiling" for our analysis. The two future consumption
estimates are reasonably consistent with the total "under construction" (7.3 billion gallons) and
"planned" (9.8 billion gallons) ethanol production capacities discussed earlier in Section 1.2.2.1.
For each renewable fuel consumption scenario, EPA has considered cellulosic ethanol and
biodiesel consumption to be fixed at 250 million gallons (required by the Energy Act) and 300
million gallons (projected by EIA), respectively.
In addition to modeling two different 2012 ethanol consumption levels, two scenarios
were considered based on how refineries could potentially respond to the recent removal of the
RFG oxygenate mandate. In the maximum scenario ("max-RFG"), refineries could continue to
add oxygenate (ethanol) into all batches of reformulated gasoline. In this case, refineries
currently blending MTBE (at 11 volume percent) would be expected to replace it with ethanol (at
10 volume percent). In the minimum scenario ("min-RFG"), some refineries could respond by
using less (or even zero) ethanol in RFG based on the minimum amount needed to meet volume,
octane, and/or total toxics performance requirements. The rationale behind the max-RFG and
min-RFG assumptions for each area is explained in greater detail in Section 2.1.3.1. The max-
RFG and min-RFG criteria result in a total of four different 2012 ethanol consumption control
cases:
• 7.2 billion gallons of ethanol, maximum amount used in RFG areas;
• 7.2 billion gallons of ethanol, minimum amount used in RFG areas;
• 9.6 billion gallons of ethanol, maximum amount used in RFG areas; and
• 9.6 billion gallons of ethanol, minimum amount used in RFG areas.
Each of these control cases has been analyzed in more detail and the results are presented in
Section 2.1.4.6.
2.1.4.2 Forecast for RFG Ethanol Use
In the 2004 base case, there were 19 states with RFG programs covering a total of 181
counties (summarized previously in Table 2.1-1). For this analysis, we are assuming that in the
future the number of RFG areas would not change. As such, the RFG fuel contribution to the
gasoline pool of each state would remain the same, yet the amount of ethanol added to RFG
could change as discussed below.
In the past, all RFG areas were required to use a minimum amount of oxygenate in their
reformulated gasoline year-round, as discussed earlier in 2.1.1.1. However, effective May 5,
2006, EPA removed the RFG oxygenate requirement in response to the Energy Act.37 Although
the oxygenate requirement has already been eliminated, many refiners are still operating under
contracts with ethanol blenders. As such, refiners true response to the removal of the oxygenate
requirement is relatively unknown at this time. While it is difficult to predict exactly how each
refinery supplying an RFG area would behave, we have made an attempt to bracket the responses
as described below.
53
-------
Some refineries may continue to add oxygenate (ethanol) to all their reformulated
gasoline in 2012 based on octane, volume, and/or toxic performance requirements. Others,
particularly those located in close proximity to the ethanol production facilities (namely PADD
2), may continue to add ethanol as if the oxygenate requirement was still effective or may even
increase their ethanol use due to favorable economics. Still for others it may be more
economical to pare back or eliminate RFG ethanol use completely.
For the purpose of this analysis, future RFG area behavior (with respect to ethanol use)
was considered to be uniform within a PADD. Therefore RFG areas located in PADD 1 would
respond the same but perhaps differently from PADD 2 and PADD 3 RFG areas. Additionally,
California "RFG" (Federal RFG and CA Phase 3 RFG) would behave according to its own set of
RFG assumptions as would Arizona "RFG" (Arizona CBG in Phoenix Metropolitan Area).
For the max-RFG sensitivity, ethanol blending was assumed to be favorable year-round
throughout the country. Hence, in Table 2.1-10 (below), the resulting percent market share for
ethanol-blended gasoline is 100% in both summer and winter for all RFG areas.
For the min-RFG sensitivity, determining the percent market share for each area was
more involved. Since this proposal assumes that MSAT1 baselines are still in place, a minimum
level of ethanol blending (market share) could be estimated based on what would be required to
maintain required toxics performance accounting for the MTBE phase-out. This was carried out
at a PADD level for summer and winter gasoline using aggregated fuel parameters from 2001-02
batch data and MSAT1 baseline toxics performance figures.
In general, this analysis consisted of generating PADD-level estimated toxics baselines
for future years and comparing those to results of Complex Model runs using estimated future
fuel parameters. The amount of ethanol was reduced from 3.5 weight percent oxygen toward
zero until the Complex Model performance of the fuel parameters just met the estimated toxics
baselines, and that amount of ethanol was determined to be the minimum quantity to maintain
compliance. This estimation was made for calendar year 2012.
To estimate toxics baselines for the future years, new RFG volume was added at a fixed
annual growth rate of 1.7% based on historical production volume data, and this new volume
was assumed to come in at the minimum required toxics performance level of 21.5% total toxics
reduction. This resulted in a lower effective PADD-average MSAT1 baseline going into the
future.
Next, 2001-02 seasonal aggregate fuel parameters were modified using a balance
between 10 volume percent ethanol (3.5 weight percent oxygen) and 5 volume percent
aromatics. This adjustment to the aromatics values was determined from examining fuel quality
surveys, and corresponds to an adjustment a refiner could make to replace the octane value in 10
percent oxygenate in RFG. As the ethanol quantity was stepped down, aromatics were added
proportionally. This addition was done incrementally to find the point where annual average
MS ATI total toxics compliance would be just met. The results are presented below in Table 2.1-
9.
54
-------
This analysis did not make changes to other gasoline parameters. Discussion of changes
to other fuel parameters, and their relationship to the RFG VOC standard, can be found in
Section 2.2.4. We expect that for some refiners, their toxics standard would limit their ability to
remove oxygenate from their gasoline, while for others the VOC standard would be more
restrictive. More rigorous refinery modeling work is underway that will provide further details
for the final rulemaking; therefore the analysis presented in this section should be considered an
estimate.
Table 2.1-9.
Fuel Parameter Adjustments for "Min-RFG" Estimation
2001-02 Estimated 2012
wt% Oxygen" wt% Oxygen
Summer
PADD 1
PADD2
PADD 3
Winter
PADD 1
PADD 2
PADD 3
Annual
PADD 1
PADD 2
PADD 3
2.12
3.25
2.13
1.93
3.18
1.99
2.01
3.20
2.06
0.35
0.00
0.00
3.50
0.00
0.70
2.13
0.00
0.36
a PADD 1 & 3 oxygenate was primarily MTBE and TAME during this time
period, while PADD 2 used primarily ethanol.
b All future oxygenate is assumed to be ethanol.
With respect to PADD 1, Table 2.1-9 shows that next to zero oxygen (as ethanol) would
be required in summertime RFG and 3.5 weight percent (10 percent ethanol in every batch)
would be required in wintertime RFG. Accordingly, PADD 1 RFG has been assigned ethanol
blending market shares of zero percent in the summer and 100 percent in the winter (shown in
Table 2.1-10).
With respect to PADD 2, Table 2.1-9 shows that no oxygen would be required in RFG to
meet MSAT1 requirements. However, while this analysis suggests that PADD 2 RFG could go
without oxygenate in the future, ethanol blending is expected to occur due to proximity to
ethanol production and desire to support local economies. Ethanol blending is expected to be
lower in the summer compared to the winter due to economic penalties associated with
summertime ethanol blending (necessity to remove butanes and pentanes to meet RVP
requirements). Accordingly, PADD 2 RFG has been assigned ethanol blending market shares of
50 percent in the summer and 100 percent in the winter (shown in Table 2.1-10).
With respect to PADD 3, Table 2.1-9 shows that no oxygen would be required in
summertime RFG and 0.7 weight percent would be required in wintertime RFG. Accordingly,
55
-------
PADD 3 RFG has been assigned ethanol blending market shares of zero percent in the summer
and 25 percent in the winter (shown in Table 2.1-10).
A separate approach was used to determine the minimum ethanol blending market shares
for California "RFG" (Federal RFG and California Phase 3 RFG) and Arizona "RFG" (Arizona
CBG in Phoenix Metropolitan Area). In 2001, MathPro Inc. conducted a study to determine the
amount of ethanol blending expected to occur in California Phase 3 Reformulated Gasoline
(CaRFG3) under an oxygen waiver.38 MathPro concluded that ethanol blending in CaRFG3
would be in the range of 25 to 65 percent (E5.7). For the purpose of this analysis, we assumed
that the entire State of California would behave uniformly or more specifically like CaRFG3.
Thus, we applied the MathPro range to all California gasoline (both Federal RFG and CaRFG3).
We assumed that minimum California ethanol blending would be 25 percent (E5.7) in the
summertime (as suggested by the lower limit of the study). However, instead of selecting 65
percent in the wintertime (to match the upper limit of the study) we selected a higher value (100
percent) based on increased crude oil prices, increased ethanol availability (since 2001), and
favorability based on existing infrastructure. Accordingly, California "RFG" has been assigned
ethanol blending market shares of 25 percent (E5.7) in the summer and 100 percent (E5.7) in the
winter (shown in Table 2.1-10).
Finally, we assumed that Arizona "RFG" would be governed by winter oxy-fuel
requirements (Phoenix CBG is also covered by state oxy-fuel program). As such, we assumed
that all wintertime Arizona "RFG" would contain 10 percent ethanol. With respect to
summertime fuel, we assumed that Arizona "RFG" would be comprised of 2/3 CA "RFG" and
1/3 PADD 3 RFG. These seasonal assumptions are identical to the 2004 base case methodology
described in Section 2.1.2.1. However, in the future, the gasoline received from California is
assumed to be a single fuel to minimize the number of gasoline blends shipped via pipeline and
the predominant fuel available in California (75% of summer California fuel contains no ethanol
according to Table 2.1-10). As a result, no Arizona "RFG" would contain ethanol in the summer
(2/3 x 0 percent from CA and 1/3x0 percent from PADD 3). Accordingly, Arizona "RFG" has
been assigned ethanol blending market shares of 0 percent (E5.7) in the summer and 100 percent
(E10) in the winter (shown in Table 2.1-10).
Table 2.1-10. 2012 RFG Area Assumptions
ETOH-Blended Gasoline (% Market Share)3
Min-RFG Scenario Max-RFG Scenario
RFG Areas Summer Winter Summer Winter
PADD 1
PADD 2
PADD 3
California13
Arizona0
0%
50%
0%
25%
0%
100%
100%
25%
100%
100%
100%
100%
100%
100%
100%
100%
100%
100%
100%
100%
aPercent marketshare of E10, with the exception of California (E5.7 year-round)
and Arizona (E5.7 summer only)
bPertains to both Federal RFG and California Phase 3 RFG
cPertains to Arizona Clean Burning Gasoline (CBG)
56
-------
2.1.4.3 Forecast for CG Ethanol Use
Once we determined the range of potential ethanol use in RFG (by PADD), we needed a
systematic way to allocate the remaining ethanol into CG areas. Since the primary motivation to
blend or not blend ethanol is expected to be economic, we devised a way to rank CG areas, on a
state-by-state and urban/rural basis, as to the economic favorability of ethanol blending. This
was done by calculating an ethanol margin, which is equal to gasoline price minus ethanol
delivered price. Ethanol delivered price is equal to ethanol plant gate price plus transportation
costs minus any additional state plus other adjustments (explained below). The greater the
ethanol margin, the greater the economic incentive and the more likely ethanol is to be used in
that area.
At the time the analysis was carried out, ethanol plant gate price was taken from an older
EIA NEMS model. However, since this price was assumed to be the same for all ethanol, the
actual value is not important when trying to estimate relative allocation preferences between
areas. All ethanol blending was assumed to be done at 10 volume percent. The gasoline prices
for each state were the weighted average rack price of all conventional grades and all months,
taken from EIA Petroleum Marketing Annual 2004.39
Ethanol distribution costs were taken from figures given in the documentation for the
EIA NEMS model, and are based on a 2002 study by DAI, Inc.40 For the purpose of this
consumption analysis, all ethanol was assumed to be produced in the Midwest in census
divisions 3 and 4 (corresponding closely to PADD 2). Unfortunately, at the time of this analysis
we had not yet completed the production analysis to better understand where the future ethanol
plants would be located. However, while the results of the production analysis do not
completely coincide with this assumption (as shown in Table 1.2-14, only 92 percent of the total
anticipated plant capacity would actually come from PADD 2 and the rest would originate from
other areas throughout the country), this simplifying assumption is still very reasonable.
Ethanol consumed within census divisions 3 and 4 was assumed to be transported by
truck, while distribution outside of those areas was via rail, ship, and/or barge. A single average
distribution cost for each destination census division was generated by weighting together the
2012 freight costs given for each mode in both census divisions 3 and 4 according to their
volume share. These cent per gallon figures were first adjusted upward by 10 percent to reflect
higher energy prices, and then additional adjustments were applied to some individual states
based on their position within the census division. In the cases of Alaska and Hawaii,
differences in ethanol delivery prices from the mainland were inferred from gasoline prices.
Table 2.1-11 shows the gasoline price and ethanol distribution cost for each state as used in this
analysis.
57
-------
Table 2.1-11. Gasoline Price & Ethanol Distribution Costs
State
Alabama
Alaska
Arizona
Arkansas
California
Colorado
Connecticut
DC
Delaware
Florida
Georgia
Hawaii
Idaho
Illinois
Indiana
Iowa
Kansas
Kentucky
Louisiana
Maine
Maryland
Massachusetts
Michigan
Minnesota
Mississippi
Missouri
Montana
Nebraska
Nevada
New Hampshire
New Jersey
New Mexico
New York
North Carolina
North Dakota
Ohio
Oklahoma
Oregon
Pennsylvania
Rhode Island
South Carolina
South Dakota
Tennessee
Texas
Utah
Vermont
Virginia
Washington
West Virginia
Wisconsin
Wyoming
CD
6
9
8
7
9
8
1
5
5
5
5
9
8
3
3
4
4
6
7
1
5
1
3
4
6
4
8
4
8
1
2
8
2
5
4
3
7
9
2
1
5
4
6
7
8
1
5
9
5
3
8
Gasoline Rack
Price (c/gal)
123.2
157.0
138.0
123.3
142.1
129.5
No CG sold
No CG sold
No CG sold
124.9
125.8
151.7
134.2
125.7
125.6
127.5
124.3
125.9
123.1
125.5
124.8
No CG sold
126.5
127.4
123.0
126.0
130.5
126.0
141.6
125.3
No CG sold
128.4
126.0
124.4
127.7
126.2
123.4
133.8
126.1
No CG sold
124.9
127.8
124.5
122.5
132.3
127.3
123.4
132.1
125.8
125.2
130.4
Ethanol Distribution
Cost (c/gal)
7.2
41.5
15.4
7.3
16.5
10.4
11.4
11.4
11.4
8.4
11.4
36.5
15.4
4.4
5.4
3.4
4.4
6.2
7.3
13.4
11.4
11.4
6.4
4.4
6.2
4.4
13.4
4.4
16.4
12.4
11.4
12.4
11.4
11.4
5.4
5.4
8.3
16.5
8.4
11.4
11.4
4.4
6.2
10.3
13.4
12.4
11.4
16.5
11.4
4.4
12.4
58
-------
As the final step in the calculation, subsidies and other adjustments were applied. The
federal blending credit of 51 cents per gallon was given to all areas, and five state retail
incentives were included as follows (all cents per gallon of ethanol): Iowa, 29.5; Illinois, 20.1,
South Dakota, 20; Maine: 7.5; Oklahoma, 1.6.Y41
In addition to state subsidies, small penalty adjustments were made for distributing
ethanol into rural areas in several states. These are given in Table 2.1-12. The reasoning behind
this is that when large shipments of ethanol come from the Midwest by barge, ship, or rail, they
will be unloaded initially at large terminals near metropolitan areas. Further storage and
handling will be required to allow smaller quantities to be distributed via truck into rural areas.
Several states have gasoline pipelines that traverse them with connections at various points,
helping to reduce distribution burdens, but ethanol is not expected to be shipped via pipeline.
Based on these considerations, the largest adjustment was applied to the Rocky Mountain states
since they are generally large in area and additional expense is required to transport freight
through higher elevations and rugged terrain. Smaller adjustments were applied to states that are
smaller, flatter, or have navigable water access on one or more sides. The states that do not
appear on this list of adjustments were generally in the Midwest where ethanol is produced and
were not believed to have significant differences in rural and urban distribution costs.
Table 2.1-12.
Adjustment for Ethanol Distribution into Rural Areas
Rural Area
AJ- 4. ±( I l\
Adjustment (c/gal)
OH 1
AK, AL, AR, FL, GA, KY, LA, MD,
ME, MS, NC, NH, NY, OK, OR PA, 2
SC, TN, TX, VA, VT, WA, WV
AZ, CO, ID, NM, NV, UT, WY 3
The resulting ranking system for distributing ethanol into conventional gasoline by state
and region is summarized below in Table 2.1-13. The amount of ethanol leftover after filling
both RFG (according to RFG assumption Table 2.1-10) and winter oxy-fuel (discussed below in
Section 2.1.4.4), determined the cut off point or last state to receive ethanol in conventional
gasoline.
Y EPA acknowledges that other states are considering (or may have even approved) retail pump incentives for
gasohol. However, at the time this consumption analysis was completed, these were the only five states offering
retail pump incentives that seemed likely to be applicable in 2012.
59
-------
Table 2.1-13. 2012 Precedence for Adding ETOH to Conventional Gasoline
Rank
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
State3
IA
SD
IL
NV
AZ
ND
NV
MO
NE
OH
Wl
IN
Ml
KS
OH
KY
AZ
ME
CO
UT
ID
TN
WY
KY
PA
ME
OR
MS
OK
FL
TN
CO
NM
AR
AL
UT
Region15
All
All
All
Urban (s)
Urban (s)
All
Rural
All
All
Urban
All
All
All
All
Rural
Urban
Rural
Urban
Urban
Urban
Urban
Urban
Urban
Rural
Urban
Rural
Urban (s)
Urban
Urban
Urban
Rural
Rural
Urban (s)
Urban
Urban
Rural
Rank
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
State3
ID
LA
PA
WA
AK
OR
WY
VT
MS
OK
NY
FL
GA
WV
AR
AL
LA
WA
AK
SC
MD
NM
NC
NH
VT
NY
GA
WV
TX
VA
SC
MD
NC
NH
TX
VA
Region15
Rural
Urban
Rural
Urban
Urban
Rural
Rural
Urban
Rural
Rural
Urban
Rural
Urban
Urban
Rural
Rural
Rural
Rural
Rural
Urban
Urban
Rural
Urban
Urban
Rural
Rural
Rural
Rural
Urban (s)
Urban
Rural
Rural
Rural
Rural
Rural
Rural
aMN, HI, and MT are not included on the CG order of precedence table because
they have state mandates requiring ETOH in all gasoline. WA is included
because their state mandate only accounts for 20% of their fuel.
bWith respect to state ethanol distribution, "all" means the entire state
fills with ethanol at the same precedence level, whereas "urban" and
"rural" imply that these regions fill separately. An (s) next to urban refers
to summer gasoline only (winter is covered by respective state oxy-fuel
programs).
60
-------
2.1.4.4
Forecast for Winter Oxy-fuel Ethanol Use
In the 2004 base case, there were 14 state-implemented winter oxy-fuel programs in 11
states (summarized previously in Table 2.1-2). Of these programs, 9 were required in response
to non-attainment with the CO National Ambient Air Quality Standards (NAAQS) and 4 were
implemented to maintain CO attainment status. However, in the future 4 of the 9 required oxy-
fuel areas are expected to be reclassified from non-attainment to attainment and discontinue
using oxy-fuel in the wintertime2. These areas are: Anchorage, AK; Las Vegas, NV;
Provo/Orem, UT; and Spokane, WA. In addition, Colorado is expected to discontinue using
winter oxy-fuel in Denver/Boulder and Longmont to maintain CO attainment status. The use of
oxy-fuel in the above-mentioned areas is expected to discontinue by 2012 or sooner. With the
removal of these 6 state-implemented programs, that leaves oxyfuel areas in Tuscon and
Phoenix, AZ; Los Angeles, CA; Missoula, MT; Reno, NV; Albuquerque, NM; Portland, OR; and
El Paso, TX. We assumed that these remaining areas would continue to use 10 percent ethanol
for their entire winter oxy-fuel period (duration varies by area, six month maximum) in the 2012
control cases.
2.1.4.5
2012 Forecasted Ethanol Consumption by Season
In 2012, for the purpose of this analysis, we have assumed that 45 percent of the gasoline
would be consumed in the summertime and 55 percent would be consumed in the wintertimeAA.
Additionally, we made the assumption that 100 percent of the winter oxy-fuel would be
consumed in the wintertime. Applying the RFG assumptions along with the CG order of
precedence, the resulting seasonal ethanol use for the four 2012 control cases is shown below in
Table 2.1-14.
Table 2.1-14. 2012 Forecasted Ethanol Consumption by Season
201 2 Control Case
7.2
7.2
9.6
9.6
Bgal/
Bgal/
Bgal/
Bgal/
Max-RFG
Min-RFG
Max-RFG
Min-RFG
CG
Summer
1
2
2
3
,269
,144
,356
,223
Winter
1,537
2,571
2,830
3,881
Ethanol
OXYa
Winter
72
72
73
73
Consumption
RFG
Summer
1,932
244
1,941
246
(MM gal)
b
Winter
2,389
2,168
2,400
2,178
Total
Summer
3
2
4
3
,201
,388
,297
,468
Winter
3,999
4,812
5,303
6,132
aWinter oxy-fuel programs
bFederal RFG plus CA Phase 3 RFG and Arizona CBG
Based on conversations with state officials and regional EPA officials.
^ We acknowledge that the volumetric seasonal split used in this analysis may or may not correspond with the new
summer/winter seasons defined in the Energy Act (Summer: April 1st through September 30th; Winter: January 1st
through March 31st and November 1st through December 31st). However, we believe it is a reasonable assumption
for this analysis.
61
-------
2.1.4.6 2012 Gasoline/Oxygenate Consumption by Fuel Type
7.2 Bgal / Max-RFG Control Case
In 2012, when modeling the minimum or "floor" amount of ethanol (7.2 billion gallons)
coupled with the "Max-RFG" assumption, there was only 2.9 billion gallons leftover to fill
conventional gasoline. After satisfying state mandates and winter oxy-fuel requirements, this left
1.8 billion gallons to be used in the CG pool. This leftover ethanol filled about two thirds (by
volume) of PADD 2 conventional gasoline and made its way to the State of Kansas (Rank #14
on the CG order of precedence table), filling 29 percent of the state's CG before reaching the 1.8
billion gallon amount (7.2 billion gallons total). A summary of the ethanol consumption by fuel
type and PADD is found in Table 2.1-15. Additionally, a summary of ethanol consumption by
state is found in Table 2.1-19 and a graphical representation (by season) is found in Figures 2.1-3
and 2.1-4.
Table 2.1-15. 2012 Ethanol Consumption by Fuel Type (MMgal)
7.2 Bgal / Max-RFG Control Case
PADD
PADD1
PADD 2
PADD 3
PADD 4
PADD 5C
California
Total
CG
Summer
0
1,144
0
25
100
0
1,269
Winter
0
1,398
0
30
109
0
1,537
OXYa
Winter
0
0
24
1
47
0
72
RFG
Summer
956
274
241
0
31
430
1,932
b
Winter
1,168
335
295
0
66
525
2,389
Total
2,124
3,151
560
56
353
955
7,200
aWinter oxy-fuel programs
"Federal RFG plus CA Pha
CPADD 5 excluding California
bFederal RFG plus CA Phase 3 RFG and Arizona CBG
7.2 Bgal / Min-RFG Control Case
In 2012, when modeling the 7.2 billion gallon case coupled with the "Min-RFG"
assumption, there was 4.8 billion gallons leftover to fill conventional gasoline. After satisfying
state mandates and winter oxy-fuel requirements, this left 3.7 billion gallons to be used in the CG
pool. This leftover ethanol filled an even larger portion of PADD 2 conventional gasoline than
in the 7.2 Bgal / Max-RFG control case (91 percent by volume compared to 67 percent).
Further, the ethanol made its way to the urban portion of Florida (Rank #30 on the CG order of
precedence table), filling 24 percent of the state's CG before reaching the 3.7 billion gallon
amount (7.2 billion gallons total). A summary of the ethanol consumption by fuel type and
PADD is found in Table 2.1-16. Additionally, a summary of ethanol consumption by state is
found in Table 2.1-19 and a graphical representation (by season) is found in Figures 2.1-5 and
2.1-6.
62
-------
Table 2.1-16. 2012 Ethanol Consumption by Fuel Type/PADD (MMgal)
7.2 Bgal / Min-RFG Control Case
PADD
PADD1
PADD 2
PADD 3
PADD 4
PADD 5C
California
Total
CG
Summer
231
1,544
41
150
177
0
2,144
Winter
283
1,888
50
182
169
0
2,571
OXYa
Winter
0
0
24
1
47
0
72
RFG
Summer
0
137
0
0
0
107
244
b
Winter
1,168
335
74
0
66
525
2,168
Total
1,682
3,904
188
334
459
633
7,200
aWinter oxy-fuel programs
bFederal RFG plus CA Phase 3 RFG and Arizona CBG
CPADD 5 excluding California
9.6 Bgal / Max-RFG Control Case
In 2012, when modeling the maximum or "ceiling" amount of ethanol (9.6 billion
gallons) coupled with the "Max-RFG" assumption, there was a significant amount of ethanol
leftover (about 5.3 billion gallons) to fill conventional gasoline. After satisfying state mandates
and winter oxy-fuel requirements, this left 4.2 billion gallons to be used in the CG pool. This
leftover ethanol filled an even larger portion of PADD 2 conventional gasoline than the 7.2 Bgal
/ Min-RFG control case (97 percent by volume compared to 91 percent). Further, the ethanol
made its way to the rural portion of Colorado (Rank #32 on the CG order of precedence table),
filling 80 percent of the state's CG reaching the 4.2 billion gallon amount (9.6 billion gallons
total). A summary of the ethanol consumption by fuel type and PADD is found in Table 2.1-17.
Additionally, a summary of ethanol consumption by state is found in Table 2.1-19 and a
graphical representation (by season) is found in Figures 2.1-7 and 2.1-8.
Table 2.1-17. 2012 Ethanol Consumption by Fuel Type/PADD (MMgal)
9.6 Bgal / Max-RFG Control Case
PADD
PADD1
PADD 2
PADD 3
PADD 4
PADD 5C
California
Total
CG
Summer
345
1,634
41
158
178
0
2,356
Winter
421
1,997
50
192
170
0
2,830
OXYa
Winter
0
0
24
1
48
0
73
RFG
Summer
960
275
243
0
31
432
1,941
b
Winter
1,173
336
296
0
66
528
2,400
Total
2,900
4,243
654
352
492
960
9,600
aWinter oxy-fuel programs
Federal RFG plus CA Phase 3 RFG and Arizona CBG
CPADD 5 excluding California
b
63
-------
9.6 Bgal / Min-RFG Control Case
In 2012, when modeling the 9.6 billion gallon case coupled with the "Min-RFG"
assumption, there was a maximum amount of ethanol leftover (about 7.2 billion gallons) to fill
conventional gasoline. After satisfying state mandates and winter oxy-fuel requirements, this left
6.1 billion gallons to be used in the CG pool. The leftover ethanol filled PADD 2 conventional
gasoline entirely and made its way to the urban portion of Georgia (Rank #49 on the CG order of
precedence table), filling 26 percent of the state's CG before reaching the 6.1 billion gallon
amount (9.6 billion gallons total). A summary of the ethanol consumption by fuel type and
PADD is found in Table 2.1-18. Additionally, a summary of ethanol consumption by state is
found in Table 2.1-19 and a graphical representation (by season) is found in Figures 2.1-9 and
2.1-10.
Table 2.1-18. 2012 Ethanol Consumption by Fuel Type/PADD (MMgal)
9.6 Bgal / Min-RFG Control Case
PADD
PADD1
PADD 2
PADD 3
PADD 4
PADD 5C
California
Total
CG
Summer
788
1,689
243
230
273
0
3,223
Winter
963
2,064
288
280
286
0
3,881
OXYa
Winter
0
0
24
1
48
0
73
RFG
Summer
0
138
0
0
0
108
246
b
Winter
1,173
336
74
0
66
528
2,178
Total
2,925
4,226
629
511
672
636
9,600
b
aWinter oxy-fuel programs
Federal RFG plus CA Phase 3 RFG and Arizona CBG
CPADD 5 excluding California
2.1.4.7 2012 Gasoline/Oxygenate Consumption by State
A summary of each state's total ethanol consumption for each of the four 2012 control
cases is found below in Table 2.1-19. Additionally Figures 2.1-3 through 2.1-10 graphically
show the percent ethanol use by state for each of the control cases broken down by season
(summer versus winter).
64
-------
Table 2.1-19. 2012 Ethanol Consumption by State
State
Alabama
Alaska
Arizona
Arkansas
California
Colorado
Connecticut
Delaware
District of Columbia
Florida
Georgia
Hawaii
Idaho
Illinois
Indiana
Iowa
Kansas
Kentucky
Louisiana
Maine
Maryland
Massachusetts
Michigan
Minnesota
Mississippi
Missouri
Montana
Nebraska
Nevada
New Hampshire
New Jersey
New Mexico
New York
North Carolina
North Dakota
Ohio
Oklahoma
Oregon
Pennsylvania
Rhode Island
South Carolina
South Dakota
Tennessee
Texas
Utah
Vermont
Virginia
Washington
West Virginia
Wisconsin
Wyoming
Total
7.2 Bgal /
MMgal
0
0
113
0
955
0
171
50
13
0
0
51
0
580
343
183
46
56
0
0
247
329
544
301
0
354
56
92
96
58
474
9
337
0
39
289
0
34
144
55
0
49
0
551
0
0
246
59
0
277
0
7,200
Max-RFG
%
0.0%
0.0%
4.6%
0.0%
5.8%
0.0%
10.0%
10.0%
10.0%
0.0%
0.0%
10.0%
0.0%
10.0%
10.0%
10.0%
2.9%
2.3%
0.0%
0.0%
8.9%
10.0%
10.0%
10.0%
0.0%
10.0%
10.0%
10.0%
10.0%
7.4%
10.0%
0.8%
5.4%
0.0%
10.0%
5.0%
0.0%
2.1%
2.7%
10.0%
0.0%
10.0%
0.0%
4.1%
0.0%
0.0%
5.6%
2.0%
0.0%
10.0%
0.0%
4.7%
7.2 Bgal
MMgal
0
0
191
0
633
163
94
28
7
233
0
51
35
502
332
183
156
231
0
85
136
181
544
301
91
334
56
92
96
32
261
9
186
0
39
577
121
63
275
30
0
49
182
88
61
0
135
59
0
261
17
7,200
/ Min-RFG
%
0.0%
0.0%
7.8%
0.0%
3.8%
7.3%
5.5%
5.5%
5.5%
2.4%
0.0%
10.0%
5.0%
8.6%
9.7%
10.0%
10.0%
9.5%
0.0%
10.0%
4.9%
5.5%
10.0%
10.0%
5.0%
9.4%
10.0%
10.0%
10.0%
4.1%
5.5%
0.8%
2.9%
0.0%
10.0%
10.0%
5.0%
3.7%
5.1%
5.5%
0.0%
10.0%
5.0%
0.7%
5.0%
0.0%
3.1%
2.0%
0.0%
9.4%
5.0%
4.7%
9.6 Bgal
MMgal
0
0
223
0
960
180
171
50
13
484
0
51
36
582
344
184
157
245
0
85
248
330
547
302
91
355
57
92
96
58
477
9
339
0
39
580
121
63
341
55
0
49
366
554
62
0
247
59
0
278
17
9,600
/ Max-RFG
%
0.0%
0.0%
9.1%
0.0%
5.8%
8.0%
10.0%
10.0%
10.0%
5.0%
0.0%
10.0%
5.0%
10.0%
10.0%
10.0%
10.0%
10.0%
0.0%
10.0%
8.9%
10.0%
10.0%
10.0%
5.0%
10.0%
10.0%
10.0%
10.0%
7.4%
10.0%
0.8%
5.4%
0.0%
10.0%
10.0%
5.0%
3.7%
6.3%
10.0%
0.0%
10.0%
10.0%
4.1%
5.0%
0.0%
5.6%
2.0%
0.0%
10.0%
5.0%
6.3%
9.6 Bgal
MMgal
135
17
192
79
636
225
94
28
7
968
138
51
71
504
333
184
157
232
129
85
136
182
547
302
182
335
57
92
96
32
262
16
333
0
39
580
243
169
473
30
0
49
366
89
123
19
136
147
0
262
35
9,600
/ Min-RFG
%
5.0%
5.0%
7.8%
5.0%
3.8%
10.0%
5.5%
5.5%
5.5%
10.0%
2.6%
10.0%
10.0%
8.6%
9.7%
10.0%
10.0%
9.5%
5.0%
10.0%
4.9%
5.5%
10.0%
10.0%
10.0%
9.4%
10.0%
10.0%
10.0%
4.1%
5.5%
1.5%
5.3%
0.0%
10.0%
10.0%
10.0%
10.0%
8.8%
5.5%
0.0%
10.0%
10.0%
0.7%
10.0%
5.0%
3.1%
5.0%
0.0%
9.4%
10.0%
6.3%
65
-------
Figure 2.1-3. 2012 7.2 Bgal / Max-RFG Summer Ethanol Distribution
(% ETOH by State)
2012%ETOHbvState
CH 1 to <5%
CH 5 to < 10%
10%
Not Pictured
AK: 0% ETOH
HI: 10% ETOH
DC: 10% ETOH
Figure 2.1-4. 2012 7.2 Bgal /Max-RFG Winter Ethanol Distribution
(% ETOH by State)
2012% ETOH by State
CH 1 to <5%
d 5 to < 10%
I—I 10%
Not Pictured
AK: 0% ETOH
HI: 10% ETOH
DC: 10% ETOH
66
-------
Figure 2.1-5. 2012 7.2 Bgal / Min-RFG Summer Ethanol Distribution
(% ETOH by State)
2012%ETOHbvState
CH 1 to <5%
n 5 to < 10%
I—I 10%
Not Pictured
AK: 0% ETOH
HI: 10% ETOH
DC: 0% ETOH
Figure 2.1-6. 2012 7.2 Bgal / Min-RFG Winter Ethanol Distribution
(% ETOH by State)
2012% ETOH by State
CH 1 to <5%
CH 5 to < 10%
• 10%
Not Pictured
AK: 0% ETOH
HI: 10% ETOH
DC: 10% ETOH
67
-------
Figure 2.1-7. 2012 9.6 Bgal / Max-RFG Summer Ethanol Distribution
(% ETOH by State)
2012%ETOHbvState
CH 1 to <5%
n 5 to < 10%
I—I 10%
Not Pictured
AK: 0% ETOH
HI: 10% ETOH
DC: 10% ETOH
Figure 2.1-8. 2012 9.6 Bgal / Max-RFG Winter Ethanol Distribution
(% ETOH by State)
2012% ETOH by State
CH 1 to <5%
CH 5 to < 10%
• 10%
Not Pictured
AK: 0% ETOH
HI: 10% ETOH
DC: 10% ETOH
68
-------
Figure 2.1-9. 2012 9.6 Bgal / Min-RFG Summer Ethanol Distribution
(% ETOH by State)
2012%ETOHbvState
CH 1 to <5%
n 5 to < 10%
I—I 10%
Not Pictured
AK: 50% ETOH
HI: 10% ETOH
DC: 10% ETOH
Figure 2.1-10. 2012 9.6 Bgal / Min-RFG Winter Ethanol Distribution
(% ETOH by State)
2012% ETOH by State
CH 1 to <5%
CH 5 to < 10%
• 10%
Not Pictured
AK: 50% ETOH
HI: 10% ETOH
DC: 10% ETOH
69
-------
2.2 Effects of Ethanol and MTBE on Gasoline Fuel Properties
2.2.1 Effect of Ethanol on Conventional Gasoline Fuel Properties
Gasoline fuel properties include parameters such as aromatics and olefins levels, and
vapor pressure. When ethanol is added to gasoline, it modifies these properties. The changes in
these properties are not simply a factor of how much ethanol is added to gasoline, but can also
depend on changes made to the hydrocarbon portion of the blend which the refiner may have
made in anticipation of ethanol blending. Two methods by which ethanol is added to gasoline
include splash-blending and match-blending. In splash-blending, ethanol is typically added to
gasoline in a fuel delivery truck containing gasoline that otherwise meets all applicable
specifications. The finished blend is a product of the controlled volumes of ethanol and gasoline,
but properties of the gasoline portion of the finished blend were not specifically designed for
ethanol blending. Splash-blending is a common method by which ethanol is added to
conventional gasoline (CG), since EPA regulations allow it. Only a few states require that
conventional gasoline with ethanol meet the same RVP standards as gasoline. This effectively
prohibits splash-blending, since splash-blending increases RVP by roughly 1.0 psi.
The downside to splash-blending is that the ethanol blend contains more octane than the
original gasoline. While some of the value of this octane increase can be recovered by
increasing the grade of the ethanol blend from regular to midgrade or midgrade to premium,
practically, this can only be done for a fraction of the gasoline. Thus, splash-blending tends to
give away octane value. The alternative is to match-blend the ethanol. With match-blending, the
refiner produces a hydrocarbon component which is designed to meet applicable gasoline
specifications after 10 vol% ethanol has been added. Thus, this hydrocarbon component can
have a lower octane value than required for finished gasoline. The downside to match-blending
is that the low octane hydrocarbon component must be distributed separately from finished
gasoline and it must be blended with ethanol prior to sale.
Historically, most ethanol has been splash-blended into conventional gasoline. However,
whenever the market share of ethanol blending reaches a sufficient level, refiners serving that
market tend to supply a sub-octane gasoline for match-blending with ethanol. With the dramatic
increase in ethanol blending already occurring, plus that which is anticipated over the next
several years, we believe that most ethanol will be match-blended into gasoline to allow refiners
to reduce their octane requirements. Due to the way in which gasoline is refined, this has the
beneficial side effect of increasing the total supply of hydrocarbon gasoline.
Reformulated gasoline (RFG) requires more precise control of fuel properties, such as
vapor pressure. This control can only be, and has historically been, achieved through match-
blending.
Our purpose in this section is to estimate the impact of blending ethanol on the properties
of both CG and RFG. Typically, EPA has estimated such impacts using refinery linear
programming models. These models simulate the feedstocks and chemical processes used in
refineries and determine the types of processes needed to produce specific quantities of finished
products and their properties. As discussed in Chapter 7, EPA is currently conducting such
70
-------
modeling via contract. The results of this modeling will be completed in time for use in the
analyses supporting the final RFS rule. In addition, past refinery modeling does not sufficiently
reflect the conditions now existing (such as high crude oil prices), nor the specific volumes of
ethanol expected to occur in the future. Thus, in the absence of such refinery modeling, we
opted to analyze empirical gasoline property data available through annual fuel survey data
conducted by the Alliance of Automobile Manufacturers (AAM).42 The AAM data reflect the
properties of gasoline from many refineries and in many geographic locations. By investigating
the relationship between ethanol content and other fuel properties used in emission inventory
models, we can predict the changes in gasoline quality which will occur with increased use of
ethanol and the resultant changes in in-use emissions. For the final rule analysis, we plan to
update these estimates using the refinery modeling which will then be complete.
The first step in assessing the effect of ethanol content on gasoline properties was to
determine which of the AAM data to consider. The AAM reports include fuel sample data from
across North America. Given the focus of our analysis is ethanol blending in the U.S., we
decided to only use the data for the 26 U.S. cities represented in the survey, thereby excluding
data from Canada and Mexico. We then examined the data in order to identify those cities which
had data for both ethanol and non-ethanol blends. We could have simply averaged all of the data
for ethanol and non-ethanol blends and compared the two results. However, this comparison is
likely to include factors which affect fuel quality other than simply the addition of ethanol (e.g.,
regional differences in crude oil quality and refinery configuration). Even restricting the
comparison to ethanol and non-ethanol blends likely includes differences between specific
refineries serving the same city. However, this potentially confounding factor cannot be avoided
in this type of analysis.
Specifically, we counted the number of winter and summer samples in each city, as well
as the number of samples in each season near 10 vol% ethanol content (E10) and the number of
samples at or near 0 vol% ethanol content (EO). (We considered any gasoline that contained less
than 5 vol% ethanol as being representative of EO, and gasoline that contained 5 vol% or more
ethanol as being representative of E10.) The number of samples according to this breakdown is
shown in Table 2.2-1.
71
-------
Table 2.2-1. Number of Fuel Samples Collected in AAM Fuel Surveys,
U.S. Cities, 2001-2005.
City
Albuquerque, NM
Atlanta, GA
Billings, MT
Boston, MA
Chicago, IL
Cleveland, OH
Dallas, TX
Denver, CO
Detroit, MI
Fairbanks, AK
Houston, TX
Kansas City, MO
Las Vegas, NV
Los Angeles, CA
Miami, FL
Minneapolis/St. Paul, MN
New Orleans, LA
New York City, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
San Antonio, TX
San Francisco, CA
Seattle, WA
St. Louis, MO
Washington, DC
RFC
Area
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Total # of
Samples
112
202
138
183
154
174
154
180
191
80
66
154
162
151
222
183
168
156
166
221
69
129
166
151
156
183
Number of Winter Samples
Winter
Total
54
104
68
92
78
87
77
92
93
40
0
74
80
76
110
87
85
73
85
110
0
64
83
77
78
92
E10a
53
0
0
6
78
50
0
92
23
0
0
0
80
41
0
84
0
25
2
110
0
0
37
16
61
0
E0b
1
104
68
86
0
37
77
0
70
40
0
74
0
35
110
3
85
48
83
0
0
64
46
61
17
92
Number of Summer Samples
Summer
Total
58
98
70
91
76
87
77
88
98
40
66
80
82
75
112
96
83
83
81
111
69
65
83
74
78
91
E10a
12
0
0
0
76
57
0
47
26
0
0
1
5
46
0
91
0
30
0
0
11
0
43
14
62
0
E0b
46
98
70
91
0
30
77
41
72
40
66
79
77
29
112
5
83
53
81
111
58
65
40
60
16
91
a "E10," or 10 vol% ethanol, represents gasoline that contains 5 vol% or more ethanol.
b "EO," or 0 vol% ethanol, represents gasoline that contains less than 5 vol% ethanol.
We identified four cities that contained a reasonable number of EO and E10 samples for
each season. Cleveland (PADD 2), Detroit (PADD 2), Denver (PADD 4), and Seattle (PADD 5)
met these criteria while representing various geographic and fuel-processing regions. Denver
was an exception in that only the summer survey data showed a mix of fuels, since Denver has
an oxygenated fuel mandate in the winter (i.e., there were no EO samples in the winter months).
Overall, very few of the data from any of the four cities deviated more than a few tenths of a
percent from EO and E10. The exception to this was Seattle, where 4 of the 16 winter data points
were between 5 and 6 vol% ethanol. Table 2.2-2 shows the fuel properties for EO and E10 for
each city, by season.
72
-------
Table 2.2-2. Effect of Ethanol Content on Gasoline Properties in Individual Cities
City
Cleveland,
OH
Denver,
CO
Detroit,
MI
Seattle,
WA
Sea-
son
Qnm
Win
Qnm
Win
Qnm
Win
Qnm
Win
Fuel
Type
EO
E10
EO
E10
EO
E10
EO
E10
EO
E10
EO
E10
EO
E10
EO
E10
#
Samp
les
30
57
37
50
41
47
0
0
72
26
70
23
60
14
61
16
T50
(°F)
223
190
208
163
212
178
—
—
220
202
202
161
218
195
208
179
T90
(°F)
332
327
326
320
330
319
—
—
338
332
335
327
326
324
316
310
E200
(%)
39.0
51.4
46.6
57.4
44.5
54.8
—
—
41.0
48.8
49.1
57.4
40.4
51.0
46.0
54.4
E300
(%)
81.4
83.8
83.5
85.8
82.6
85.7
—
—
80.3
82.7
81.3
84.3
83.1
84.6
85.7
87.3
Aromatics
(Vol%)
34.2
26.1
24.0
19.9
29.1
23.4
—
—
32.2
27.1
22.6
19.8
32.5
29.9
26.7
22.2
Olefins
(Vol%)
7.6
7.1
22.3
16.0
9.5
9.2
—
—
6.5
7.5
19.9
17.4
8.0
5.4
17.2
18.3
RVP
(psi)
8.9
9.9
—
—
8.3
9.3
—
—
7.5
8.7
—
—
7.6
8.7
—
-
Octane
(R+M)/2
90.0
90.0
89.4
90.6
86.9
86.7
—
—
89.8
90.4
89.6
90.7
89.5
89.9
89.7
90.3
Benzene
(Vol%)
0.9
1.1
0.9
1.0
1.4
1.5
—
—
.2
.0
.3
0.9
.5
.6
.6
.6
Conceivably, the effect of ethanol blending on gasoline properties could vary regionally.
However, given the availability of comparable data in only four cities, none of which is located
in the southern or northeastern U.S., we decided to combine the results for the four cities to
develop a single set of fuel quality changes for the entire U.S. Table 2.2-3 shows the average
fuel properties across the four cities, where the averages have been weighted by the number of
samples from each city.
Table 2.2-3. Fuel Properties for EO and E10 (Four-City Average)
Season
Summer
Winter
Fuel
Type
EO
E10
EO
E10
T50 (°F)
218
189
205
165
T90 (°F)
332
325
326
320
E200 (%)
41.2
52.0
47.4
56.9
E300 (%)
81.8
84.3
83.3
85.7
Aromatics
(Vol%)
32.0
25.8
24.4
20.3
Olefins
(Vol%)
7.7
7.7
19.5
16.8
RVP (psi)
7.9
9.4
—
~
Octane
(R+M)/2
89.2
89.0
89.6
90.6
Benzene
(Vol%)
1.3
1.3
1.3
1.1
We then calculated the differences between the properties of EO and E10 conventional
gasoline. Table 2.2-4 shows how fuel properties change when adding ethanol to create a 10
vol% ethanol blend from gasoline with no ethanol.
Table 2.2-4. Change in Fuel Properties Due to Addition of Ethanol (EO to E10)
Season
Summer
Winter
T50 (°F)
-29
-40
T90 (°F)
-7
-6
E200 (%)
10.8
9.5
E300 (%)
2.6
2.4
Aromatics
(Vol%)
-6.2
-4.1
Olefins
(Vol%)
0.0
-2.7
RVP (psi)
1.5
~
Octane
(R+M)/2
-0.2
1.0
Benzene
(Vol%)
0.0
-0.2
Finally, Table 2.2-5 averages the effects of Table 2.2-4, weighting the average by season
(assuming five summer months and seven winter months). Where the final values in Table 2.2-5
indicate the changes in fuel properties as the level of ethanol in gasoline increases from 0 vol%
to 10 vol%, we assumed adding smaller amounts of ethanol would simply change properties
73
-------
proportionally. Table 2.2-5 also includes the change in fuel property on a per vol% ethanol
basis, which provides a useful factor for later adjustments to fuel properties based on changes in
ethanol content. As will be discussed below, we assume that ethanol is always blended into
conventional gasoline at 10 vol% (E10). Thus, the ethanol content of an area's gasoline is only
less than 10 vol% when the E10 market share is less than 100%. In this case, simple linear
interpolation of properties is reasonable.
Table 2.2-5.
Change in Properties of Conventional Gasoline Due to Addition of Ethanol
Change between
EO and E10
Change per 1 vol%
increase in ethanol
T50
(°F)
-36
-3.6
T90
(°F)
-7
-0.7
E200
(%)
10.0
1.0
E300
(%)
2.4
0.24
Aromatics
(Vol%)
-5.0
-0.5
Olefins
(Vol%)
-1.6
-0.16
RVP
(psi)
1.0a
~
Octane
(R+M)/2
0.5
0.05
Benzene
(Vol%)
-0.1
-0.01
Oxygen
(Wt%)
3.5
0.35
a Summer only. Based on average of city-specific differences shown in Table 2.2-2.
The first item to note about the differences shown in Table 2.2-5 is that the difference in
octane ((R+M)/2) is 0.5. Splash-blending ethanol typically increases octane by 2 to 2.5 octane
numbers. Thus, it appears that most of the ethanol blending being performed in these cities is
match-blending. Our projection of match-blending for the future appears very reasonable in light
of this. The presence of match-blending is also confirmed by the 5.0 vol% decrease in aromatic
content. As indicated in Table 2.2-3, the aromatic content of non-oxygenated gasoline tends to
average just under 30 vol%. Splashblending 10 vol% ethanol should reduce this value by 3
vol%. Reforming tends to be the refinery process which increases octane on the margin and does
so by increasing the aromatic content. Thus, the fact that aromatics decreased by well above 3
vol% indicates a reduction in the severity of reforming when the fuel is being blended with
ethanol.
The other significant item to note is the difference in RVP. We do not show a seasonally
weighted value for RVP, as RVP is usually only relevant for summertime emission projections.
In Table 2.2-4, the difference in summer RVP is 1.5 psi. This is well above the 1.0 psi value
typically found for ethanol blending. The 1.5 psi difference is due to the sample weighting
scheme used to develop the figures in Tables 2.2-4 and 2.2-5. The number of ethanol and non-
ethanol samples is not evenly weighted across the four cities and the applicable RVP standards in
each city differ. As can be seen in Table 2.2-2, the difference in RVP in each city is 1.0-1.2 psi.
Thus, we will assume the typical RVP increase associated with ethanol blending of 1.0 psi
applies here. We do not believe that the sample weighting scheme affects any of the other fuel
properties in this manner.
In our above approach, aggregating data within urban areas loses any refinery-specific
effects. Also, we lose the ability to apply region-specific effects since we only include four cities
and do not have any cities with both ethanol and non-ethanol fuels from the Gulf area or east
coast. However, the results use available data to provide an acceptable national assessment of
ethanol and fuel properties. We checked our results against the AAM data from all U.S. cites,
comparing all conventional gasoline non-ethanol blends (with an RVP greater than or equal to
8.2 psi) to all conventional gasoline ethanol blends (with an RVP greater than or equal to 8.7
74
-------
psi). The results were very similar to those from the four cities, supporting the validity of our
approach. These results are shown in Table 2.2-6 below.
Table 2.2-6.
Change in Properties of Conventional Gasoline Due to Addition of Ethanol,
Using All U.S. Conventional Gasoline Data from AAM Survey
Season
Summer
Winter
Fuel Type
EO
E10
EO
E10
Average change
between EO and E10a
T50 (°F)
216
179
203
171
-34
T90 (°F)
328
325
329
322
-5
E200
(%)
42.3
54.0
48.5
56.4
9.5
E300
(%)
82.4
84.5
82.8
85.3
2.3
Aromatics
(Vol%)
29.7
24.3
22.7
18.8
-4.5
Olefins
(Vol%)
7.4
7.8
20.2
18.5
-0.9
RVP
(psi)
8.6
9.6
~
~
1.0b
Octane
(R+M)/2
88.5
88.8
89.4
89.3
0.1
Benzene
(Vol%)
1.1
1.2
1.0
1.1
0.1
a Weighted by seasons of five summer months and seven winter months.
b Summer only.
2.2.2 Effects of Ethanol on Reformulated Gasoline Fuel Properties
RFG must meet tight specifications for VOC, NOx and toxic emission performance.
These emission performance standards result in particularly tight control of RVP, benzene and
aromatics. This means that the RVP increase shown above in Table 2.2-5 cannot occur and must
be compensated for through the removal of low molecular weight, high RVP hydrocarbon
components.
Until recently, all RFG was required to contain 2.0 wt% oxygen on average. Thus, all
RFG contained either MTBE or ethanol. Any additional ethanol use in RFG relative to our 2004
base case will thus replace MTBE. Both MTBE and ethanol are high octane components and
have relatively low vapor pressures (i.e., they both tend to decrease T50 substantially). RFG has
typically contained 11 vol% MTBE or 10 vol% ethanol. Given their similar usage levels and
generally similar properties other than RVP, plus the restrictions imposed by the applicable
VOC, NOx and toxic emission performance standards, we assume here that the replacement of
MTBE by ethanol in RFG will not change any fuel properties other than the type of oxygenate
and oxygen content.
2.2.3 Effects of MTBE on Conventional Gasoline Fuel Properties
The purpose of this section is to estimate the impact of removing MTBE from
conventional gasoline. Unlike the situation with respect to ethanol blending, we do have refinery
modeling available which indicates the impact of MTBE blending. This modeling is somewhat
dated (circa 1993). However, since removing MTBE does not involve any predictions of its total
usage level, nor the location of its use, economics, such as crude oil price, are not a factor. It is
primarily an issue of chemical properties and general refinery operation, such as octane
management. Also, MTBE is always match-blended, since gasoline can be shipped with MTBE
75
-------
through pipelines. Thus, MTBE is always added at the refinery, allowing the refiner to take full
advantage of its properties.
In support of the final rule implementing the RFG program in 1993, refinery modeling
was performed which estimated the impact of MTBE blending on the various gasoline
properties.43 While this modeling was performed in the context of projecting the cost of
producing RFG, it is applicable to the use of MTBE in CG, as well. The refinery modeling
examined a number of incremental steps involved in the production of RFG. Because RFG was
mandated to contain oxygen and MTBE was expected to be the oxygenate of choice, MTBE was
added in the first step of the analysis, before the fuel met the rest of the RFG requirements.
Table 2.2-7 shows the results of adding MTBE based on this refinery modeling.
Table 2.2-7. Effect of MTBE on Gasoline Properties: RFG Final Rule
Fuel Parameter
RVP (psi)
T50a
T90a
E200 (vol%)
E300 (vol%)
Aromatics (vol%)
Olefms (vol%)
Oxygen (wt%)
Sulfur (ppm)
Benzene (vol%)
Base 9 RVP Gasoline
8.7
218
329
41
83
32.0
13.1
0
339
1.53
MTBE Blend
8.7
207
321
46.7
84.9
25.5
13.1
2.1
309
0.95
Difference
0
-11
-8
5.7
1.9
-6.5
0
2.1
-30
-0.58
a Estimated using correlations developed in support of EPA RFG final rule, Docket A-92-12, February 1994.
T50 = 302 - E200 / 0.49 and T90 = 707 - E300 / 0.22
As with ethanol blending, MTBE blending reduces aromatic content significantly as
refiners take advantage of MTBE's high octane level. Like ethanol, MTBE also tends to
increase E200 and E200 and decrease T50 and T90. Unlike ethanol, MTBE does not increase
RVP.
MTBE blending is shown to modestly reduce sulfur and benzene levels, as well. This
refinery modeling was performed prior to the development of the Tier 2 sulfur standards for
gasoline. With these standards, gasoline must meet a 30 ppm sulfur standard on average with or
without MTBE blending. As refiners can adjust the severity of their hydrotreating processes to
account for various changes in feedstocks and oxygenate use, we do not expect that the removal
of MTBE will result in any increase in sulfur content. Otherwise, the reversal of the differences
shown in Table 2.2-7 are expected to occur when MTBE is removed from gasoline (when the
MTBE content was 11 vol%). Table 2.2-8 shows these changes (in terms of the addition of
MTBE) for both a fuel containing 11 vol% MTBE and on the basis of 1 vol% MTBE.
76
-------
Table 2.2-8.
Change in Properties of Conventional Gasoline Due to Addition of MTBE
Change between 0 vol%
and 11 vol%MTBE
Change per 1 vol%
increase in MTBE
T50
(°F)
-11
-1.0
T90
(°F)
-8
-0.7
E200
(%)
5.7
0.52
E300
(%)
1.9
0.17
Aromatics
(Vol%)
-6.5
-0.6
Olefins
(Vol%)
0
0
RVP (psi)
0
0
Oxygen
(Wt%)
2.1
0.2
2.2.4 Effects of MTBE on Reformulated Gasoline Fuel Properties
Reformulated Gasoline (RFG) has historically contained oxygenate due to the applicable
2.0 weight percent oxygen content requirement. RFG has contained 11 vol% MTBE or ten vol%
ethanol, except in California, where 6 vol% ethanol blends have been common. As discussed in
Section 2.1, we expect that the use of MTBE in RFG will cease soon. It will be replaced by
either 10 vol% or ethanol or high octane hydrocarbon blending components, such as alkylate or
reformate. In either case, as discussed in Section 2.2.2, RFG will continue to have to meet
stringent VOC, NOx, and toxic emission performance standards.
Compliance with the NOx standard is essentially assured with compliance with the Tier 2
sulfur standards applicable to all gasoline. Compliance with the MS ATI toxics standards was
discussed in Section 2.1.4.2 above. There, we concluded that if refiners used reformate to
compensate for MTBE's octane, then aromatic content would increase, limiting the volume of
non-oxygenated RFG which could be produced and still comply with the MS ATI toxics
standards. This was the basis for our projections of the use of ethanol in RFG under the
"minimum ethanol use in RFG" scenarios.
The VOC emission performance standard could also limit the production of non-
oxygenated RFG if reformate was used to compensate for MTBE's octane. Assuming an RVP
level of 6.8 psi and a sulfur content of 30 ppm, per the Complex Model, refiners still have to
increase E200 and E300 and reduce aromatic content relative to typical conventional gasoline in
order to meet the RFG VOC standard. Refiners have some flexibility in which of these
parameters to adjust and to what degree. They could also reduce RVP below 6.8 psi, as this level
is well above that needed for the hydrocarbon portion of ethanol RFG, if the latter is to be at 6.8
RVP after ethanol blending.
The refinery modeling currently underway will provide considerable insight into both the
potential market share of non-oxygenated RFG and its likely properties. For the purpose of this
analysis, we assume that most of the properties of non-oxygenated RFG will be very similar to
those of ethanol RFG. We decreased the levels of E200 and E300 of non-oxygenated RFG as
much as possible while still complying with the VOC performance standard for southern RFG of
29%. We plan to update this estimate of the quality of non-oxygenated RFG for the final RFS
rule analysis based on the refinery modeling to be completed soon. For comparison purposes,
Table 2.2-9 also shows the specifications of a comparable ethanol RFG.
77
-------
Table 2.2-9. RFG Quality With and Without Oxygenate
Fuel Parameter
RVP (psi)
T50
T90
E200 (vol%)
E300 (vol%)
Aromatics (vol%)
Olefms (vol%)
Oxygen (wt%)
Sulfur (ppm)
Benzene (vol%)
Non-Oxygenated RFG
6.8
210
320
45
85
25.5
13.1
0
30
0.65
MTBE RFG
6.8
212
321
44
85
25.5
13.1
2.1
30
0.65
Ethanol RFG
6.8
194
322
53
85
25.5
13.1
3.5
30
0.65
2.2.5 Estimation of County-Specific Gasoline Properties
In order to estimate the impact of increased ethanol use and reduced MTBE use on
national emissions and air quality (described in Chapters 4 and 5), we need to estimate gasoline
properties on a county-specific basis throughout the U.S. In support of previous analyses of
national impacts of various rules, EPA has developed a set of gasoline specifications for each
county in the U.S. for various months and calendar years.44 We based our analysis on the fuel
quality specifications for January and July of 2008, since 2008 is the first year of full
implementation of the Tier 2 sulfur standard of 30 ppm. Some of the EPA county-level gasoline
specifications were based on old data, so we reviewed the estimates and made several
modifications before applying the changes expected due to ethanol addition and MTBE removal.
First, we adjusted RVP values using more recent information on local RVP programs and
to reflect commingling. Second, we revised the oxygenate content and type in each county to
match the levels estimated in Section 2.1 to be sold there under each of the five ethanol use
scenarios. Third, we adjusted the other properties of gasoline which are affected by the
oxygenate use determined in step three. These modifications are described in more detail below.
2.2.5.1
Adjustments to RVP Levels Prior to Oxygenate Use
Our review of the NMIM database of county-specific RVP levels for July indicated that
the same RVP level was often applied to all the counties of a specific state. In many cases, this
appeared reasonable, since the same RVP standard applied throughout the entire state. However,
in other cases, for example, Florida, most counties have a 9.0 RVP standard, while those
comprising several large urban areas have a 7.8 RVP standard. The RVP levels in the NMIM
database were consistent with the 7.8 RVP control programs, implying that the 7.8 RVP fuel was
sold throughout the entire state. This was true for much of the south.
As mentioned above, the NMIM fuel quality database was based primarily on fuel survey
data from 1999. Fuel surveys tend to focus on large urban areas, as opposed to smaller urban or
rural areas. Thus, the only available fuel survey data was likely from the areas with the tighter
local RVP controls. RVP control reduces gasoline supply, since lighter hydrocarbons must be
78
-------
removed in order to reduce RVP. Some, but not all of these hydrocarbon components can be
moved to higher RVP fuel sold elsewhere. Obviously gasoline prices are now much higher than
they were in 1999. So the incentive to increase supply is greater now than in 1999. As discussed
in Chapter 7, high gasoline prices are projected for the foreseeable future, at least relative to
those existing in 1999. Thus, we believe that it is reasonable to project that refiners will market
gasoline blends with as high a level of RVP as practical given the applicable standards. For
example, in Florida, two fuels will be marketed: one to meet the 7.8 RVP standard in several
urban areas and another to meet the 9.0 RVP standard applicable elsewhere. There certainly
could be some spillover of the 7.8 RVP fuel into adjacent 9.0 RVP counties. However, we lack
data indicating the degree to which this is occurring and might occur in the future. Lacking this
data, it seems more reasonable to project only that level of RVP control which is guaranteed by
the applicable standards than to assume that refiners will over-comply with RVP standards and
reduce the volume of gasoline which they can produce.
Past studies have shown that a typical compliance margin for RVP is about 0.3 psi. Thus,
for those counties where the standard 9.0 RVP standard applies, we set the July RVP level to 8.7
psi.
EPA maintains a list of counties where its 7.8 RVP standard applies, as well as any local
standards more stringent than 9.0 RVP.45 Using this list, we assigned RVP values in each
county equal to 0.3 psi less than the standard applicable in July. We also reduced the RVP levels
of two sets of counties which had voluntary local RVP control programs (and therefore not listed
the above Guide). These two areas were Seattle and Tulsa. Based on a review of AAM fuel
survey data in these two areas, the fuel being sold in these areas was very similar to that for an
area with a 7.8 RVP standard. Thus, we assigned a value of 7.5 psi RVP to Tulsa County,
Oklahoma, and to King, Pierce, and Snohomish Counties, Washington.
We then assigned an RVP value of 6.8 psi to counties subject to the Federal RFG
program, again based on an EPA list of the counties subject to this program.46 The EPA list of
RFG counties includes the Baton Rouge, Louisiana, area. However, litigation has held up
implementation of this program, so these counties were assigned RVP values consistent with the
currently applicable 7.8 RVP standard instead. The RVP value of 6.8 psi was typical for the
RFG areas included in the AAM fuel surveys.
For the purposes of our analysis, we also assigned the entire State of California an RVP
of 6.8 psi, since California fuel must meet a similar VOC performance standard to RFG.
Likewise, RVP in Maricopa and Final counties in Arizona were assigned a level of 6.8 psi.
These two counties are subject to Arizona's unique reformulated gasoline program. This
program basically requires that gasoline sold in these two counties meet either the California
RFG or Federal RFG standards. Thus, RVP in these two counties will be the same as in those
other two areas, similar to national RFG fuel.
These RVP levels for 9.0 RVP and low RVP areas are appropriate when no ethanol is
being blended into gasoline. However, most of these areas increase the applicable standard by
1.0 psi for ethanol blends, which is the typical impact of ethanol blending. Therefore, these
levels need to be adjusted for the expected level of ethanol use, which is discussed below.
79
-------
2.2.5.2 County-Specific Oxygenate Type and Content
The five ethanol use scenarios developed in Section 2.1 assign ethanol and MTBE use by
state and fuel type (i.e., conventional gasoline, RFG, oxyfuel). In order to develop county level
estimates of ethanol and MTBE use, we simply assume that ethanol and MTBE use within a state
and fuel type is uniform. For example, if the E10 market share in conventional gasoline Iowa is
34%, then ethanol use in every county receiving conventional gasoline in Iowa was assigned an
E10 market share of 34%.
As described above, we nearly always assume that ethanol use is in the form of a 10 vol%
blend with gasoline. The two exceptions are California fuel and Arizona RFG. California fuel
containing ethanol is assumed to contain 5.7 vol% ethanol. Arizona RFG is assumed to be a mix
of 67% California fuel and 33% Federal RFG produced in PADD 3. Therefore, its ethanol
content is a 2/1 mix of the ethanol contents of California RFG and PADD 3 Federal RFG.
Similarly, we assume that MTBE is used at an 11 vol% level in RFG, since this meets the
previously mandated oxygen content of 2.1 wt%. MTBE in conventional gasoline was assumed
to be used at a 3 vol% level. This was somewhat arbitrary, but does not affect the outcome of the
analysis. The effect of MTBE blending on emissions is very linear. Therefore, whether the fuel
pool in a particular area consists of 10% of a 10 vol% MTBE blend or 33% of a 3 vol% MTBE
blend is immaterial.
EPA's NMIM model (described in more detail in Chapter 4) will only accept a single
composite fuel for each county. Therefore, we could not use the mix of fuels often projected to
be supplied to counties developed in Section 2.1. In order to produce a single, composite fuel,
we simply multiplied the ethanol and MTBE contents of each blend by their market share in that
county in order to determine the average ethanol and MTBE contents of each county's fuel pool,
respectively. For example, if the E10 market share in a specific county was 50%, the ethanol
content for that fuel was set to 5 vol%. We then adjusted the other fuel properties to account for
these oxygenates, which is discussed below.
2.2.5.3 Adjustments to Other Gasoline Properties for Oxygenate Use
We next adjusted other gasoline properties to account for the level of county-specific
oxygenate use projected to occur under the five ethanol use scenarios. Our review of the NMEVI
fuel database indicated that properties, such as aromatics, reflected the level of oxygenate use
existing in 1999. Therefore, we used the oxygenate levels in the NMIM database, which differ
from those developed in Section 2.1. for 2004, as the basis for our adjustments of the other fuel
properties. For example, if the NMEVI database indicated an ethanol content of 3 vol% for fuel
sold in Wayne County, Michigan, and the 2004 projection for this county was 5 vol%, we
adjusted the NMEVI fuel properties for this county to reflect the addition of 2 vol% ethanol.
The basis for these adjustments were those developed in Sections 2.2.1 through 2.2.4.
above. As described there these adjustments apply primarily to conventional gasoline. These
adjustments are summarized in Table 2.2-10 below.
80
-------
Table 2.2-10.
Change in Property per 1 Vol% Increase in Ethanol and MTBE Content
E200 (%)
E300 (%)
Aromatics (Vol%)
Olefins (Vol%)
RVP (psi)
Conventional Gasoline
Ethanol
MTBE
+1.0
+0.52
+0.24
+0.17
-0.5
-0.59
-0.16
0
+0.1
0
Reformulated Gasoline
Ethanol
MTBE
0
0.1
0
0.1
0
0
0
0
0
0
To calculate new fuel properties for each county, we applied the ethanol and MTBE
factors to the change in county-level ethanol and MTBE content. The overall adjustment to the
fuel property was the addition of the ethanol effect and the MTBE effect to the baseline fuel
property, or in equation form:
New Fuel
Property
Level
NMIM Database Fuel , Ethanol
Property Level Effect
MTBE Effect
For example, the equation for the ethanol effect is as follows:
Ethanol
Effect
URFS x s RFS x s NMIM x s NMIM x -v s
Ethanol II Market \ — I Ethanol If Market \ I X I
Content J\ Share / \ Content J\ Share / ( \
Fuel Property Change
per 1 vol% Ethanol
Increase
For the impact of ethanol blending on aromatic and olefin contents, we followed a
slightly different approach. We assumed that the ethanol present in 1999 had been splash-
blended, while that being used in the future will be match-blended. This difference doesn't
affect the adjustment of RVP, E200, or E300, since we assume that these parameters are affected
in the same way regardless of whether the ethanol is splash- or match-blended. However, as
discussed above, the change in aromatics does depend on which blending approach is used. The
situation is similar for olefins, though to a lesser extent. Thus, we employed what can be thought
of as a two step process in adjusting aromatic and olefin contents for the change in ethanol
content between the NMIM estimate and those for the five ethanol use scenarios developed in
Section 2.1.
The first step is to account for any splash-blended ethanol in the NMIM database. With
splash-blending, aromatic and olefin contents are reduced simply by dilution, since ethanol
contains is neither an aromatic nor an olefin. The following equation shows how the NMIM
level of aromatics was adjusted:
81
-------
Intermediate
Aromatic =
Content
x NMIM \ / / NMIM x s NMIM x x
I Aromatic 1 4- I 1 — I Ethanol II Ethanol 1 4- 100 1
\ Content / \ \ Content / \ Market Share / /
Then, the effect of any ethanol projected to be sold in that county in the five ethanol use
scenarios developed in Section 2.1 was applied using the approach described above for RVP,
E200 and E300 (and for the effect of MTBE on aromatics and olefms). In this case, the NMIM
ethanol content and market share is zero, since we already adjusted the NMEVI aromatic and
olefin contents to represent those existing for a zero ethanol content. For example, the equation
for the ethanol effect is as follows:
New Fuel
Property Level
Intermediate
Fuel Property +
Level
,
Ethanol
Content
S Market
,
Share
)"(
Fuel Property Change per 1
vol% Ethanol Increase
We make one final adjustment to RVP to add a commingling effect to account for areas
where vehicles may be fueled by a mix of ethanol -blend gasoline. Commingling of ethanol and
non-ethanol blends can increase the average RVP of gasoline in vehicle fuel tanks by 0.1-0.3 psi.
EPA's MOBILE6.2 model normally accounts for this effect automatically. However, when
NMEVI is used to run MOBILE6.2, the commingling effect in MOBILE6.2 is by-passed.
Therefore, any effect of commingling needs to be accounted for in the average fuel specified to
be sold in each county. To roughly account for this effect, we increased RVP by 0.1 psi in all
states where the E10 market share was significant (i.e., more than 10%) but less than 100%. In
the four, future ethanol use scenarios, we tend to project that ethanol use will be either zero or
100% in any particular state, due to the difficulty in projecting different ethanol use levels within
a state. Theoretically, commingling would not exist under these situations. However, in reality,
ethanol blending will not often stop at a state line between two states, one with a projection of
zero E10 market share and the other with 100% market share. The former will likely receive
some ethanol, while the latter will be less than 100%. Therefore, we added a commingling effect
of 0.1 psi RVP to counties in those states where the projected level of ethanol blending changed
from 100% to zero. These states are shown in Table 2.2-1 1.
Table 2.2-11. States Where RVP was Increased Due to Commingling
7.2 Min
7.2 Max
9. 6 Min
9.6 Max
Arizona, Arkansas, Kansas, Kentucky, Maine, Missouri, Montana, Nebraska,
Nevada, Ohio, West Virginia
Arkansas, Colorado, Idaho, Indiana, Kentucky, Missouri, Montana, Nebraska,
Nevada, Pennsylvania, Utah, West Virginia, Wyoming
Arizona, Colorado, Florida, Idaho, Kentucky, Maine, Maryland, Mississippi,
Missouri, Nevada, New Hampshire, Ohio, Oklahoma, Oregon, Pennsylvania,
Tennessee, West Virginia
Alabama, Arizona, Arkansas, Kansas, Kentucky, Maine, Maryland, Missouri,
Montana, Nevada, Ohio, Tennessee, West Virginia
82
-------
2.3 Effects of Biodiesel on Diesel Fuel Properties
Our assessment of the effects of biodiesel on diesel fuel properties is found in the 2002
EPA report "A Comprehensive Analysis of Biodiesel Impacts on Exhaust Emissions"47. Table
2.3-1 below displays the difference in fuel properties between biodiesel (B100) and conventional
diesel. Note that by 2010, all highway and nonroad diesel fuel will meet a 15 ppm cap on sulfur.
The data in the table below were derived from a wide-range of biodiesels, primarily
plant- and animal-based. The 2002 EPA report did not provide properties for soy-only based
biodiesel, which is the type of diesel considered in the biodiesel emissions effects in Chapter
3.1.3.
Table 2.3-1. Comparison Between Biodiesel and Conventional Diesel Fuel"
Natural cetane number
Sulfur, ppm
Nitrogen, ppm
Aromatics, vol%
T10, degF
T50, deg F
T90, deg F
Specific gravity
Viscosity, cSt at 40 deg F
Average Biodiesel
55
54
18
0
628
649
666
0.88
6.0
Average Diesel
44
333
114
34
422
505
603
0.85
206
' Conventional diesel fuel sold outside of California.
83
-------
Chapter 3: Impacts on Emissions from Vehicles, Nonroad
Equipment, and Fuel Production Facilities
As described in Chapter 2, there are a large number of potential fuels that qualify as
renewable. However, only two are expected to be used in significant volumes by 2012: ethanol
and biodiesel. Of these, ethanol use is expected to predominate. In particular, ethanol is
expected to dominate the "growth" in renewable fuel use between now and 2012. Thus, our
primary focus here will be on the impact of the use of ethanol on emissions. We will more
briefly touch on the impact of biodiesel fuel use on emissions.
Similarly, we expect that the bulk of the impact of ethanol use on emissions and air
quality will be associated with emissions from vehicles and equipment using low level ethanol-
gasoline blends. We expect the use of high level ethanol-gasoline blends, like E85 to be
relatively small in comparison. Thus, the discussion here will focus on emissions from the use of
low level ethanol blends. We will more briefly discuss the per vehicle impacts of use of high
level ethanol-gasoline blends relative to gasoline.
Finally, we present estimates of the emissions related to the production and distribution
of both ethanol for use in gasoline and biodiesel.
3.1 Effect of Fuel Quality on Vehicle and Equipment Emissions
Ethanol belongs to a group of gasoline additives commonly referred to as oxygenates.
The two most commonly used oxygenates are ethanol and MTBE, though TAME has been used
in significant volumes, as well. All oxygenates have relatively high levels of octane (i.e., greater
than 100 R+M/2). Both ethanol and MTBE have been used historically to meet the gasoline
oxygen requirements for oxyfuel and RFG. Historically, MTBE was the predominant oxygenate
used in gasoline in the U.S. Over time, MTBE use has decreased in the U.S, while ethanol use
has increased, to the point where ethanol use now predominates. This trend appears to be
accelerating, to the point where it appears that essentially all MTBE use will cease in the U.S
sometime in 2007.
The impact of oxygenate use on emissions from motor vehicles and nonroad equipment
has been evaluated since the mid-1980's. Several models of the impact of gasoline quality on
motor vehicle emissions were developed in the early 1990's and updated periodically since that
time. We use the most up-to-date versions of these models here to estimate the impact of
changes in oxygenate use on emissions. Still, as will be described below, significant uncertainty
exists as to the effect of these gasoline components on emissions from both motor vehicle and
nonroad equipment, particularly from the latest models equipped with the most advanced
emission controls. Assuming adequate funding, we plan to conduct significant vehicle and
equipment testing over the next several years to improve our estimates of the impact of these
additives and other gasoline properties on emissions. The results of this testing will not be
available for inclusion in the analyses supporting the final Renewable Fuel Standard (RFS) rule.
84
-------
We hope that the results from these test programs will be available for reference in the
comprehensive evaluation of the emission and air quality impacts of all the fuel-related
requirements of the Energy Act required by Section 1506. This draft study is required to be
completed in 2009. As we discuss the emission impacts of increased ethanol use below, we
identify the areas where current estimates appear to be the most uncertain and where we hope to
obtain additional data prior to the 2009 study.
3.1.1 Low Level Ethanol and MTBE-Gasoline Blends
This section is divided into two parts. The first evaluates the impact of ethanol and
MTBE use on motor vehicle emissions. The second evaluates the impact of ethanol and MTBE
use on emissions from nonroad equipment.
3.1.1.1 Motor Vehicles
EPA has developed a number of emission models relating the impact of gasoline quality
on emissions from motor vehicles. In 1993, EPA published the Complex Model, which predicts
the effect of gasoline quality on VOC, NOx and air toxic emissions from 1990 model year light-
duty motor vehicles (i.e., Tier 0 vehicles). This model is used to determine refiners' compliance
with RFG and anti-dumping standards. The Complex Model also contains estimates of the
impact of gasoline RVP on non-exhaust VOC emissions. These estimates were taken from the
then-current version of the MOBILE emissions model, MOBILES.
In 2000, in responding to California's request for a waiver of the RFG oxygen mandate,
EPA performed a new analysis of the impact of gasoline quality on exhaust VOC and NOx
emissions from Tier 0 vehicles. This analysis included essentially all of the data used to develop
the Complex Model, as well as some additional data developed since 1993. It also used more
advanced statistical tools, such as a mixed model, which were not available in 1993. These VOC
and NOx models are referred to here as the EPA Predictive Models. Thus, in terms of both
supporting data and modeling tools, the EPA Predictive Models represent an improvement over
the Complex Model, at least for exhaust VOC and NOx emissions. Because the criteria for
granting California a waiver of the oxygen requirement focused on ozone and PM impacts, EPA
did not develop a similar model for toxics or CO emissions.
In roughly the same timeframe, EPA developed its latest motor vehicle emission
inventory model, MOBILE6. Some of the fuel-emission relationships from the Complex Model
were incorporated into MOBILE6. These included the effect of selected gasoline properties on
exhaust VOC and NOx emissions and the fraction of VOC emissions represented by several air
toxics (benzene, formaldehyde, acetaldehyde, and 1,3-butadiene). The EPA Predictive Models
were not available in time for their incorporation into MOBILE6. MOBILE6 also contains
estimates of the effect of certain gasoline parameters on CO emissions, namely RVP and oxygen
content. The effect of RVP on non-exhaust VOC emissions contained in MOBILE6.2 represents
an update of the MOBILES and Complex Model estimates.
We desire in this RFS analysis to utilize the most up to date estimates of the impact of
gasoline quality on emissions currently available. No one model contains the most up to date
85
-------
estimates for all the pollutants of interest. Therefore, we have broken up the remainder of this
sub-section into six parts. The first discusses emissions of VOC and NOx, as the EPA Predictive
Models address these pollutants. The second discusses CO emissions, as neither the Complex
Model nor the EPA Predictive Models address this pollutant. The third section addresses
emissions of air toxics, as a combination of models represents the best estimate of the impact of
fuel quality on these emissions. The fourth section addresses non-exhaust VOC emissions. The
fifth section addresses PM emissions. The sixth section addresses emissions of aromatic
hydrocarbons. The seventh and final section presents the impact of ethanol and MTBE blending
on per mile emissions from gasoline-fueled motor vehicles.
3.1.1.1.1 Exhaust VOC and NOx Emissions
3.1.1.1.1.1 Complex Model and EPA Predictive Models
In 1993, EPA published the Complex model to investigate the effects of changing
gasoline fuel parameters on the exhaust emissions of Tier 0 and older vehicles. This model is
used to determine compliance with the emissions performance requirements for federal RFG by
comparing the predicted emissions of a candidate fuel to that of a baseline fuel for common
baseline vehicle technology. The baseline fuel and the baseline vehicle technology represent
those fuels and vehicles included in the 1990 US light duty vehicle fleet (Tier 0 technology).
In 1999, the state of California petitioned EPA for a waiver of the oxygen requirement in
RFG. The reasoning behind the waiver request centered on the California Air Resources Board's
(CARB) analysis which showed that reducing the amount of oxygen in RFG would lead to
reduced NOx exhaust emissions. The model that CARB developed to support their claim was
called the Phase 3 predictive model48. This model differed from the previous version of CARB's
predictive model (the Phase 2 model) in a number of ways. The most significant difference
included a substantially expanded database, mainly for model year 1986 and newer vehicles, as
well as an improved version of the statistical analysis software package used to develop the
model (SAS® PROC MIXED). According to CARB, the Phase 3 predictive model displays a
steeper NOx/Oxygen response than the Phase 2 Predictive model as a result of eliminating the
RVP by Oxygen term which the previous model had erroneously included. This caused an
increase in the NOx exhaust emissions predicted, and for many areas this increase would lead to
NOx levels exceeding those set by National Ambient Air Quality Standards (NAAQS).
According to the 1990 Clean Air Act, EPA can waive the RFG oxygen requirement if it
prevents compliance with a NAAQS. In order to properly perform an environmental impact
analysis in response to this waiver request, EPA considered using both its Complex Model, as
well as CARB's Phase 3 Predictive Model to estimate the impact of gasoline quality on
emissions. The EPA Complex model, while considered statistically robust due to the large
number of vehicles comprising the dataset, was not considered to be adequate for a number of
reasons. First, the Complex Model was based on data which did not include several studies
which has since been published. Second, the EPA Complex Model was developed using a fixed
effects statistical modeling approachBB. In contrast, both the CARB Phase 2 and 3 models were
BB A "fixed effects" model of this kind makes no attempt to estimate the error introduced by sampling from some
larger population of vehicles or fuels. The model just describes quantitatively the relationships among variables that
86
-------
mixed models, employing a more sophisticated statistical approach than was available at the time
of development of the Complex model.
EPA also rejected using CARS's Phase 3 Predictive Model in its analysis of the waiver
request. While CARB had developed a very detailed protocol for developing the Phase 3 model,
it rejected the results of this protocol because they differed too substantially from the Phase 2
model. Thus, EPA decided to create its own "predictive models" for exhaust VOC and NOx
emissions which combined the protocols used to develop the Complex Model with the expanded
database and improved statistical tools which were now available. EPA relied on existing EPA
models for evaporative VOC emission effects. However, these latter estimates were augmented
with recent data indicating that ethanol increased permeation emissions, as well as the
consideration of several commingling models and associated assumptions about driver's
refueling behavior.
One main conclusion drawn by EPA in the California Oxygen waiver analysis was that
insufficient data existed at that time to conclusively determine the response of Tier 1 and newer
vehicles to fuel parameters other than sulfur.cc Some data indicated that oxygen increased NOx
emissions from Tier 1 and later vehicles, while other data contradicted this. Due to this
inconsistency, EPA assumed that oxygen did not affect exhaust VOC, NOx or CO emissions
from Tier 1 and later vehicles in its analysis of CARS's request for a waiver of the RFG oxygen
mandate.
Recently, the Coordinating Research Council (CRC) completed an emission testing and
modeling effort (the E-67 study) involving low emission vehicles (LEVs), ultra low emission
vehicles (ULEVs), and one super ultra low emission vehicles (SULEV). This new data provides
the opportunity to confirm the assumption made in EPA's analysis of the California waiver
request. The data from this study is evaluated in the next section, below.
3.1.1.1.1.2 CRC E-67 Study
In early 2006, CRC published the results of their E-67 study investigating the effects of
three fuel parameters, ethanol, T50 and T90, on exhaust emissions from recent model year
vehicles.2 The twelve vehicles tested included both cars and light trucks, certified to California
LEV, ULEV and SULEV standards, with model years ranging from 2001 to 2003. A matrix of
twelve (12) fuels was tested in this program, with varying levels of ethanol, T50, and T90. Each
fuel parameter (ethanol, T50, and T90) was tested at each of three levels. However, a full
are present in the dataset that was analyzed. A "mixed" model, as was used by CARB in both the Phase 2 and Phase
3 predictive models' construction, attempts to go beyond description of the available data to make statistical
inference to some larger population from which the available data were sampled. In this case CARB treated the
vehicle effects as random (assuming that the test vehicles were sampled from some larger fleet) while fuel effects
were treated as "fixed" (assuming that all fuels of interest were represented in the data). Such a modeling approach
makes it possible to estimate the probable error in modeled effects in a way that is not possible with a fixed effects
model. The approach, moreover, improves the accuracy of the significance measures used to decide which terms to
include in the model.
CGAt the time of that 1999 analysis, sufficient data existed on the emissions effects of Sulfur on Tier 1 vehicles to be
modeled. However, sulfur levels were not expected to change as a result of the removal of oxygen from RFG and so
the effect of sulfur is moot.
87
-------
factorial matrix of 27 test fuels was deemed unnecessarily large due to subtle differences
between fuels that may not have yielded statistically significant results, or due to practical
considerations regarding the fuels that could be blended using existing refinery streams.
The E-67 report presents the results of emission testing for each fuel, as well as a mixed
statistical model created from the emission data. The model indicates that each of the three fuel
parameters always has a statistically significant effect on both NMHCDD and NOx emissions. In
addition, significant interactions between the three fuel parameters are also often present.
The first step in our analysis of the CRC E-67 model was to compare the emissions
changes predicted by the CRC mixed model to the actual emissions changes observed for each
fuel in the test program. We calculated average NMHC and NOx emissions over the Federal
Test Procedure for all twelve vehicles on each fuel. The CRC mixed model predicts the
percentage change in emissions for each fuel relative to another fuel. These predicted changes in
measured exhaust emissions of each fuel were applied to the measured emissions for fuel "H" in
order to create a set of absolute emission levels for each fuel. We then compared the emissions
predicted by the CRC E-67 model to the measured emission levels to observe how well the
model predicted the effects of each fuel. The fuel properties of the CRC E-67 test fuels are listed
in Table 3.1-149, below, and in greater detail in Table 3A-1 of Appendix 3 A. We selected CRC
fuel "H" as the "base" fuel since its properties are the closest to a national average non-
oxygenated conventional gasoline (0% Ethanol, 215°F T50, 330 °F T90). (See Table 2.2-6 in
Chapter 2 for the results of gasoline survey results across the U.S.)
Table 3.1-1. CRC E-67 Test Program Fuels Properties'1
Fuel
A
B
C
D
E
F
G
H
1
J
K
L
Target Pro
T50 (°F)
195
195
195
195
195
215
215
215
215
235
235
235
parties for Design Variables
T90 (°F)
295
295
330
355
355
295
295
330
355
330
355
355
Ethanol (%)
0
5.7
10
0
10
0
10
0
5.7
5.7
0
10
Actual Values
T50 (°F)
195
191
193
199
198
217
212
216
216
237
236
233
T90 (°F)
294
290
329
355
352
295
291
327
354
329
355
349
Ethanol
0
5.6
10.4
0
10.3
0
10.1
0.1
5.9
5.9
0
10.5
See Table 3A-1 in Appendix 3 A for detailed properties of all E-67 test fuels
The comparison of predicted to measured NOx emissions are shown in Figure 3.1-1. The
fuels are shown to indicate a trend in ethanol content, from the lowest levels on the left to the
highest levels on the right. Within a constant level of ethanol content, the fuels are then shown
in order of their level of T90 (lowest again on the left and highest on the right). The y-axis scale
DD
NMHC is essentially equivalent to VOC for our purposes in this study.
88
-------
in this figure is set to match that for NMHC emissions, which will be presented and discussed
next.
Figure 3.1-1. CRC E-67 Predicted vs. Actual Emissions
Relative NOx Emissions Change from Base (CRC H: 215 T50, 330 T90, 0% EtOH)
0% Ethanol
5.7% Ethanol
10% Ethanol
50.0%
40.0%
30.0%
X
O
01
0)
c
ro
O
5?
20.0%
10.0%
0.0%
-10.0%
-CRC E-67 NOx Predicted
-Actual CRC Emissions
D K B J I G
CRC Fuels (order of increasing O2, T50, T90)
As shown in Figure 3.1-1, the CRC model for NOx emissions predicts the general trend
in the emission data, which roughly indicates an increase in NOx emissions with increasing
ethanol content. However, the model clearly does not reflect many of the fuel to fuel differences
indicated by the actual emissions data. One example of this is the change between fuels G and C
- twolO% ethanol blends with relatively low distillation temperatures. In changing from fuel G
to fuel C, the CRC E-67 model predicts a 4.3% increase in NOx emissions whereas the actual
test data clearly shows a 9.2% decrease. This likely indicates the existence of interactions
between the fuel parameters which are more complex than those which could be included in the
model. While fuel parameters other than ethanol, T50 and T90 were held as constant as possible
among all the test fuels, the level of specific compounds, such as toluene or the various xylenes,
could not be held constant. It is possible that some of these compounds are affecting NOx
emissions and confounding the ability of the model based just on ethanol, T50 and T90 to predict
the observed changes.
Figure 3.1-2 repeats this comparison for NMHC.
89
-------
Figure 3.1-2. CRC E-67 Predicted vs. Actual Emissions
Relative NMHC Emissions Change from Base (CRC H: 215 T50, 330 T90, 0% EtOH)
0% Ethanol
10% Ethanol
50.0%
-10.0%
CRC Fuels (order of increasing O2, T50, T90)
First, it is very apparent that NMHC emissions are much more sensitive to fuel quality
than NOx emissions. The largest increase in NMHC emissions relative to Fuel H is three times
that for NOx emissions. Except for Fuels A and F (and of course Fuel H), the CRC model
generally under-predicts the measured NMHC data. However, directionally the predicted
emissions changes are very consistent with those observed in the test results. For this dataset at
least, the effect of fuel quality on NMHC emissions are much more predictable than NOx
emissions.
The fuels studied in this test program were varied independently at low, medium, and
high levels of T50, T90, and Ethanol. If you include all the possible linear, quadratic, and
interactive terms, there are a total of possible 10 combinations. The CRC E-67 models included
8 out of thelO possible fixed effects for the NOx, NMHC, and CO models. These terms were:
T50, T90, ethanol (EtOH), T50 squared, T90 squared, EtOH squared, T50 by EtOH, and T90 by
EtOH. The excluded terms were T50 by T90, and T50 by T90 by EtOH, which CRC excluded
from consideration since previous studies had indicated that these terms had little effect on
emissions.
Also, several of the terms that were included in the CRC model had p-values greater than
0.1, indicating that those terms are less than marginally significant.EE Specifically, the EtOH by
EtOH term in the NMHC model and the T90, T90 by T90 and T90 by EtOH terms in the NOx
model all had p-values above 0.10. In developing both the Complex Model and the EPA
Predictive Models, our procedures would normally exclude the least significant term. A new
EE In general, p < 0.05 is considered to be statistically significant, 0.05 < p < 0.10 is marginally significant, and p >
0.10 is not considered statistically significant.
90
-------
regression would then be performed without this term being included in the model. This process
would be repeated until all the remaining terms were statistically significant.
FF
We desired to determine how these statistically insignificant terms might be affecting the
predicted emission changes. This, plus the discrepancies between the CRC E-67 model and the
actual emissions data, especially for the NOx model, prompted us to create our own NOx and
NMHC models using the CRC E-67 dataset. Conducting our own modeling also provides us
with the opportunity to apply a wide range of statistical tests in order to better understand the
role of various fuel parameters in affecting emissions from these vehicles. The following
sections provide details pertaining to the verification of the CRC model and the motivation for
constructing a new model from this data.
3.1.1.1.1.3 Development of a New Mixed Model: The EPA E-67 Model
Using the E-67 dataset provided by the CRC, EPA first verified the coefficients and p-
values of the CRC E-67 model using the full E-67 dataset (no outliers were removed) with the
same 8 fixed fuel effects that were included by CRC. This was successful and the coefficient
and p-values resulting from this modeling are shown in Table 3.1-2.
Table 3.1-2. CRC E-67 Model P-Values and Coefficients
CRC E-67 NMHC CO NOx
Effect
Intercept
T50
T50*T50
T90
T90*T90
EtOH
EtOH*EtOH
T50*EtOH
T90*EtOH
P-Value
<0001
<0001
<0001
0.0541
0.0035
0.1124
0.2816
0.084
0.0004
Coefficient
-3.2942
0.0063
0.000176
0.001685
0.000058
0.005679
0.000722
0.000195
0.000244
P-Value
0.0001
0.3099
0.0428
0.0051
0.0815
0.0174
0.0005
0.0182
0.0534
Coefficient
-0.7966
0.001227
0.000099
-0.0045
0.000045
-0.01581
0.003118
0.000355
0.000174
P-Value
<.0001
0.8939
0.2182
0.762
0.1163
0.0504
0.0861
0.0414
0.99
Coefficient
-2.6183
-0.00013
-0.00006
0.00024
0.000043
0.00571
0.001622
-0.00032
-1.19E-06
EPA then created a new model starting with all combinations of T50, T90, and EtOH
along with their squares, cross products, and the interactive terms T50 by T90 by EtOH for a
total of 10 fixed effects. From this "full model", variables were eliminated in order to improve
the fit statistics between the model and the test data until a "final model" was created that
contained 7 fixed fuel effects for NMHC and CO, and 6 fixed fuel effects for NOx. Table 3.1-3,
on the following page, shows the p-values and coefficients for the fixed effect terms of each
model.
FF One exception to this process is that the linear form of a variable, such as ethanol, would always be retained in
the model if a second order term included ethanol (e.g., the ethanol by T90 term).
91
-------
Table 3.1-3. EPA E-67 Model P-Values and Coefficients
EPA E-67 NMHC CO NOx
Effect
Intercept
T50
T50*T50
T90
T90*T90
EtOH
T50*EtOH
T90*EtOH
T50*T90
T50*T90*EtOH
P-Value
<.0001
<0001
<.0001
0.0498
0.0039
0.101
0.0987
0.0002
Coefficient
-3.2773
0.006272
0.000168
0.00172
0.000057
0.005892
0.000186
0.00025
P-Value
0.0002
0.5815
0.0059
0.0111
<.0001
0.0299
<.0001
0.0003
Coefficient
-0.7684
0.00066
-0.00437
-0.01726
0.003843
0.000178
0.000126
0.000023
P-Value
<.0001
0.687
0.7761
0.0735
0.062
0.0446
0.0426
Coefficient
-2.6418
-0.00037
0.000224
0.000047
0.005393
0.001854
-0.00031
As shown in Table 3.1-3, the EPA E-67 Model does not include terms with p-values
greater than 0.10 (except for linear terms included in statistically significant higher order terms).
Statistical tests show that these two models are not significantly different from one another. The
null hypothesis in this case is that EPA E-67 fits the data just as well as the original CRC E-67
model based on a chi-squared test. However, based on several fit statistics (AIC, AICC, and
BIC) the EPA E-67 model provides a slightly better fit to the test data than either the original
CRC E-67 model or the full model with all 10 terms included. The next step is to compare the
EPA E-67 model predictions to both the E-67 data and the predictions of the EPA Predictive
Models, which reflect the emission effects for older vehicles.
(a) Model Comparison: EPA E-67 vs. EPA Predictive Models
Both the EPA E-67 and EPA Predictive models are mixed models that predict the relative
changes in exhaust emissions due to carefully controlled changes in gasoline quality, including
the addition of an oxygenate such as ethanol. The models are not intended to be accurate at
predicting absolute emission levels, but rather the difference in emissions when fuel properties
are varied. The goal of this analysis is to determine if the EPA Predictive models, which were
developed using data from Tier 0 and earlier vehicles, predict the same relative changes in
emissions as the Tier 1 vehicles used for the EPA E-67 model.
A key difference between the models is that there are only three fuel parameters used as
inputs for the EPA E-67 model: T50, T90, and ethanol content. The EPA Predictive Models use
these three properties along with RVP, aromatic content, olefm content, and sulfur content as
fuel parameter inputs to the model.
We ran the EPA E-67 and EPA predictive models with the 12 fuels used in the CRC test
program, inputting the applicable fuel properties used in each model. Following the same
procedure as outlined above, CRC test fuel H was selected as a "base" fuel in order to compare
relative changes between this fuel and others with varying amounts of ethanol, T50, and T90.
The NOx emissions predicted by the EPA E-67 and EPA Predictive models, together with the
actual E-67 study data, are shown graphically in Figure 3.1-3, below.
92
-------
Figure 3.1-3. EPA E-67 and EPA Predictive vs. Actual Emissions
Relative NOx Emissions Change from Base (CRC H: 215 T50, 330 T90, 0% EtOH)
50.0%
40.0%
30.0%
O 20.0%
10.0% -
0.0% -
-10.0%
-20.0%
CRC Test Fuels (in order of increasing O2, T50, T90)
As shown in Figure 3.1-3, neither model predicts the actual test data with complete
accuracy. The EPA E-67 shows the same general relationship to the emission data as did the
CRC E-67 NOx model. Thus, removing the statistically insignificant terms had little impact on
the relative fit of the model to the data. The EPA Predictive NOx models, on the other hand,
appear to be primarily sensitive to ethanol, with T50 and T90 playing very limited roles in
affecting NOx emissions. In contrast, the E-67 model shows sensitivities to all three parameters.
Overall, the E-67 study indicates that NOx emissions from recent model year vehicles
(LEVs, ULEVs and SULEVs) are still sensitive to at least several fuel parameters. As indicated
by the inability of either the EPA E-67 model or the EPA Predictive Models to accurately predict
all of the changes seen in the E-67 data, this study is very valuable in identifying the continued
sensitivity of LEV and cleaner vehicles to changes in fuel quality.
Figure 3.1-4, below, shows the comparison of NMHC emissions predicted by the EPA E-
67 and Predictive models together with the E-67 study data.
93
-------
Figure 3.1-4. EPA E-67 and EPA Predictive vs. Actual Emissions
Relative NMHC Emissions Change from Base (CRC H: 215 T50, 330 T90, 0% EtOH)
50.0%
40.0%
-20.0%
CRC Test Fuels (in order of increasing O2, T50, T90)
From Figure 3.1-4, it is apparent that both models do a better job at predicting changes in
NMHC emissions than was the case for NOx emissions. The EPA E-67 model is clearly the
more accurate of the two models. However, this is to be expected given it was based on the data
being depicted. The ability of the EPA Predictive Model to predict the general trend of nearly all
the CRC E-67 test fuels is impressive, given it is based on data from Tier 0 vehicles with 5-10
times the NMHC emission levels of the vehicles in the E-67 test program. Overall, it appears
that NMHC emissions from LEVs and cleaner vehicles are even more sensitive to changes in
fuel quality than NMHC emissions from Tier 0 vehicles.
The preceding figures illustrate the differences between the models for all 12 fuels
included in the E-67 test program. Some of these fuels are more practical, or likely to be
commercially produced, than others. Based on the results of AAM fuel surveys presented in
Chapter 2, summertime E10 blends will generally have levels of T50 and T90 that are about 29
°F and 7 °F lower than non-ethanol blends. Thus, it could be useful to focus on sets of fuels in
the CRC E-67 study which reflect these differences.
The fuel pair which most closely reflects these differences are CRC fuel "C", a 10 vol%
ethanol blend, and CRC fuel "H", a non-oxygenated fuel. Both fuels have a mid-range level of
T90. A second, more complex set of fuels involve those with higher levels of T90. The CRC
"E" fuel contains 10 vol% ethanol and has the high level of T90. However, there is not a good
match to this fuel which is non-oxygenated. Yet two non-oxygenated fuels ("D" and "K"), when
considered together, represent a reasonable match to fuel "E. Fuel D reflects no change in T50
relative to fuel E, while fuel K reflects slightly more than a typical drop in T50. Thus, by
averaging the emissions for fuels K and D and then comparing this to the emissions with fuel E,
94
-------
we are able to generate a second direct indication of the impact of ethanol blending on emissions
from these low emitting vehicles. The general properties of these five fuels and the emissions
changes predicted by the two models are shown in Table 3.1-4 below.
Table 3.1-4. Predicted NOx and NMHC Emissions Changes
for EPA E-67and Predictive Models
GG
Fuel Changes
T50 (°F)
T90 (°F)
Oxygen (vol%)
Change in Emissions
EPA Predictive Model NOx
EPA
EPA E-67 NOx
Actual E-67 Data
Predictive Model NMHC
EPA E-67 NMHC
Actual E-67 Data
HtoC
Mid - Low
Mid - Mid
0-10
9.5%
1 1 .0%
3.3%
-3.7%
-3.8%
-6.3%
KtoE
High - Low
High - High
0-10
8.4%
5.8%
1 .6%
-11.1%
-17.3%
-21 .0%
DtoE
Low - Low
High - High
0-10
9.4%
10.4%
-1 .8%
7.3%
8.8%
9.2%
As shown in Table 3.1-4, the two models agree quite closely on the effect of fuel C
relative to fuel H on both NMHC and NOx emissions. However, that said, both models tend to
overestimate the impact of fuels C and E on NOx emissions and underestimate the impact of
these fuels on NMHC emissions.
Regarding the comparison of fuel E to fuels K and D, the two models tend to agree on the
effect of fuel E to fuel D, but differ more with respect to the effect of fuel E to fuel K,
particularly for NMHC emissions. One reason for the difference in the latter comparison is that
the EPA E-67 NMHC model is more sensitive to very high levels of T50 than the EPA
Predictive Model for NMHC.
Overall, the results of the E-67 study suggest that our assumption that Tier 1 and later
vehicles would not be sensitive to fuel parameters such as ethanol, T50 ad T90 (made in our
consideration of California's request for a waiver of the RFG oxygen requirement) may not be
valid. The observation that NMHC emissions from LEVs, et. al. could actually be more
sensitive than Tier 0 vehicles (on a percentage basis), particularly challenges our assumption.
While the effect of fuel quality on NOx emissions from low emitting vehicles is still not clear
from the recent test data, these emissions do appear to be sensitive to fuel quality.
The data from the CRC E-67 study are too limited, both in terms of the number of
vehicles tested and the fuel parameters evaluated, to be used here to predict the effect of
increased ethanol use on exhaust emissions. However, these data clearly indicate that assuming
no effect of fuel quality on emissions from these vehicles could very well be incorrect.
Therefore, we believe that it is appropriate to estimate the potential impact of the sensitivity of
these vehicles to fuel quality via a sensitivity analysis. In this sensitivity analysis, we will extend
GG For an additional comparison between the models with an expanded set of fuels to be used later in this analysis,
refer to Table 3 A-2 in Appendix 3 A.
95
-------
the effect of the EPA Predictive Models to Tier 1 and later vehicles. Therefore, in this case, all
gasoline vehicles are assumed to be sensitive to fuel quality to the degree predicted by the EPA
Predictive Models for NMHC and NOx. At the same time, our primary analysis here will
continue to use the EPA Predictive models to predict the fuel-emission effects for Tier 0 vehicles
and assume that exhaust NMHC and NOx emissions from Tier 1 and later vehicles are not
affected by fuel quality.
As mentioned previously, the difference in sensitivity between the models, as well as the
very limited dataset used to develop the E-67 model, further illustrates the need to conduct
additional testing using newer vehicle technology.
3.1.1.1.1.4 MOBILE6.2
The exhaust emission effects contained in MOBILE6.2 often differ for normal and high
emitting vehicles. They can also vary by model year. As it is difficult to determine the fraction
of emissions coming from each model year's vehicles in MOBILE6.2, as well as normal and
high emitters, it is not feasible to predict outside of the model how a specific fuel is going to
affect in-use emissions. In addition, the split between normal and high emitters varies depending
on the presence and type of inspection and maintenance (I/M) program applicable in a particular
local area. Thus, the effect of a specific fuel on emissions can vary to some degree from one
county to another.
In order to quantify the effect of various fuel parameters on exhaust emissions in
MOBLIE6.2 under the conditions existing in the 2012-2020 timeframe, we compared the
changes in emissions predicted by the NMEVI modeling described in Chapter 4 with the changes
in fuel quality occurring in the ethanol use scenarios. Specifically, we first determined the
percentage change in exhaust VOC, CO and NOx emissions by county for the base and 7.2
Minimum RFG Use scenarios. We then performed a series of linear regressions of these ratios
against the change in fuel RVP, ethanol content and MTBE content. We did this for the 2012,
2015 and 2020 emission projections separately. For each combination of county and calendar
year, the only property that changed was fuel quality. All other parameters relevant to emissions
(e.g., the distribution of vehicles by age and class, VMT, ambient temperature, etc.) were
otherwise identical. The results are summarized in Table 3.1-5. We show the effect of fuel
quality on CO emissions here for convenience. These effects will be discussed further in the
next section.
96
-------
Table 3.1-5. Fuel-Exhaust Emission Effects in MOBILE6.2
RVP (% / psi)
Ethanol (% / Vol %)
MTBE (% / Vol%)
Adjusted r-Square
20 12 (fleet average)
voc
NOx
CO
7.1%
0.6%
12.7%
-1.1%
0.0%
-0.7%
-0.7%
0.0%
-0.4%
0.83
0.95
0.36
2015 (fleet average)
VOC
NOx
CO
7.0%
0.6%
12.7%
-1.2%
0.0%
-0.7%
-0.7%
0.0%
-0.4%
0.85
0.95
0.36
2020 (fleet average)
VOC
NOx
CO
6.7%
0.6%
12.6%
-1.2%
0.0%
-0.7%
-0.7%
0.0%
-0.4%
0.87
0.95
0.39
For comparative purposes, the effect of RVP, ethanol and MTBE on exhaust VOC and
NOx emissions from the EPA Predictive Models are shown in Table 3.1-6. The base fuel is a
typical non-oxygenated, summertime, conventional gasoline, with 8.7 RVP, 30 ppm sulfur, 32
vol% aromatics, 13 vol% olefins, T50 of 218 F, T90 of 329, and no oxygen.
Table 3.1-6.
Fuel-Exhaust Emission Effects per the EPA Predictive Models
VOC
NOx
RVP (% / psi)
1.1%
1.1%
Ethanol (% / Vol %)
-0.16%
0.75%
MTBE (% / Vol%)
-0.17%
0.36%
As can be seen, the exhaust emission effects contained in the EPA Predictive Models
differ quite dramatically from those in MOBILE6.2. Regarding the effect of RVP, both models
predict that an increase in RVP will increase both exhaust VOC and NOx emissions. However,
MOBILE6.2 predicts that an increase of 1.0 psi will increase exhaust VOC by roughly 7%, while
the EPA Predictive Models predict only a 1% increase. Regarding NOx emissions, the EPA
Predictive Models predict the larger effect (1%), while the effect in MOBILE6.2 is smaller
(0.6%). While the ratio of these two effects is significant, the absolute difference (0.4%) is very
small.
Regarding the addition of ethanol, the two models again predict very different results.
MOBILE6.2 predicts roughly 7 times the exhaust VOC reduction per volume percent of ethanol
added, with no increase in NOx. The EPA Predictive Models project a significant increase in
NOx emissions. The relative differences are similar for the addition of MTBE to gasoline,
though the difference between the two estimates of exhaust VOC reduction is smaller.
3.1.1.1.1.5
Selection of Models for Each Pollutant
For Tier 0 vehicles, the EPA Predictive Models are based on more data and the most
advanced statistical tools. Therefore, we will use these models here to project the effect of
increased ethanol use and decreased MTBE use on exhaust VOC and NOx emissions.
97
-------
For Tier 1 and later vehicles, the choice is much less clear. In our analysis of California's
request for an RFG oxygen waiver, we assumed that there was too little data upon which to
project the effect of fuel quality on the emissions of these vehicles. The CRC E-67 study
indicates that these vehicles' emissions are at least as sensitive to changes in ethanol, T50 and
T90 as Tier 0 vehicles. However, the study only tested 12 vehicles on 12 fuels. It also did not
investigate the effect of other fuel parameters, such as aromatics, olefins and RVP. As discussed
above, there are also problems with trying to substitute the CRC E-67 effects for the three fuel
parameters tested with the other fuel effects in the EPA Predictive Models.
Overall, we believe that we still lack reasonable estimates of the effect of fuel quality on
exhaust VOC and NOx emissions from Tier 1 and later vehicles. Given this, we believe that it is
valuable to maintain consistency with our analysis conducted in response to California's request
for an RFG oxygen waiver. Thus, we will continue to assume here in our primary analysis that
exhaust VOC and NOx emissions from Tier 1 and later vehicles are unaffected by fuel quality.
However, in recognition of the strong evidence presented by the CRC E-67 study, we believe
that it is important to evaluate the possibility that these vehicles respond to changes in fuel
quality. Therefore, in a sensitivity analysis, we will assume that Tier 1 and later vehicles
respond to fuel quality in the same way as Tier 0 vehicles.
3.1.1.1.2 CO Emissions
Fewer options are available to project the impact of fuel quality on CO emissions. The
Complex and EPA Predictive Models do not address CO emissions. Historically, this is because
the RFG program did not mandate a specific reduction in CO emissions and the lesser role of CO
emissions in forming ambient ozone. The only EPA model which predicts the impact of fuel
quality on CO emissions is MOBILE6.2. The effect of RVP, ethanol and MTBE on CO
emissions were shown in Table 3.1-5 above. MOBILE6.2 does not project any impact of the
other relevant fuel parameters (aromatics, olefins, T50, and T90) on CO emissions.
It is interesting to compare the effect of ethanol contained in MOBILE6.2 to that from the
EPA E-67 model discussed above. Changing just ethanol content in the EPA E-67 model
produces a 1.15% reduction in CO emissions per 1 vol% ethanol. This is larger than that the
MOBILE6.2 effect of 0.7% shown in Table 3.1-5. As mentioned above, MOBILE6.2 does not
project the effect of changes in most fuel parameters on CO emissions, like aromatics, olefins,
T50 and T90. The effect of increasing ethanol content on CO emissions in MOBILE6.2 is based
on the testing both splash-blended and match-blended ethanol fuels. Therefore, the fuel-
emission effect includes the typical effect of ethanol blending on these other fuel parameters.
Adding ethanol and decreasing T50 and T90 per the relationships described in Section 2.2.3
above increases the CO emission reduction per the EPA E-67 model to 1.25% per vol% ethanol.
Thus, considering these associated effects of ethanol on T50 and T90, the EPA E-67 model
suggests a larger impact than that in MOBILE6.2.
As discussed above, the models representing the CRC E-67 study are not sufficient for
use in quantitatively projecting the impact of fuel quality on emissions. Additional data must
98
-------
still be collected over a broader set of vehicles, fuel changes, and conditions. Therefore, we use
MOBILE6.2 here to project the impacts of ethanol use on CO emissions.
3.1.1.1.3 Exhaust Toxic Emissions
Two EPA models project the impact of fuel quality on exhaust toxic emissions: the
Complex Model and MOBILE6.2. The Complex Model projects the impact of fuel quality on
toxic emissions directly. That is, any impact of fuel quality on total exhaust VOC emissions
(which includes the air toxics) is implicitly included in the model's predictions. MOBILE6.2
separates the process into two steps. Total exhaust VOC emissions are projected first, in part
based on fuel quality. Then, the fraction of VOC represented by each air toxic is estimated, in
part based on fuel quality.
The effect of fuel quality on exhaust VOC emissions in MOBILE6.2 was already
discussed above. The effect of fuel quality on the fraction of exhaust VOC emissions
represented by each air toxic in MOBILE6.2 is based on the projections contained in the
Complex Model. These Complex Model's effects of fuel quality on exhaust toxic emissions
were used with the effect of fuel quality on exhaust VOC emissions backed out. Thus, with
respect to the effect of fuel quality on the fraction of exhaust VOC emissions represented by each
air toxic, the Complex Model is the basis of both the Complex Model and MOBILE6.2
predictions.
With respect to exhaust VOC emissions, we already decided above that the EPA
Predictive Models represent the best estimate for Tier 0 vehicles. For Tier 1 and later vehicles,
we assume in our primary analysis that these vehicles' exhaust VOC emissions are unaffected by
fuel quality. As a sensitivity analysis, we decided to extend the impacts indicated by the EPA
Predictive Models to all vehicles.
We follow the two-step process taken in MOBILE6.2 here in modeling the impact of fuel
quality on exhaust toxic emissions. We will use the EPA Predictive Models to project the impact
of fuel quality on exhaust VOC emissions, as discussed above. We will use the effects of fuel
quality on the toxic fractions of exhaust VOC emissions contained in MOBILE6.2.
3.1.1.1.4 Non-Exhaust Emissions
Both the Complex Model and MOBILE6.2 evaluate the effect of gasoline quality on non-
exhaust VOC emissions. However, the effects in the Complex Model were taken from an older
version of MOBILE, as was mentioned above. Therefore, MOBILE6.2 represents the better of
the two estimates of the effect of gasoline quality on non-exhaust VOC emissions. The EPA
Predictive Models do not address non-exhaust emissions, so they are not applicable here.
In EPA's second analysis of California's request for a waiver of the RFG oxygen
requirement, we enhanced the estimate of non-exhaust emissions in MOBILE6.2 by adding
additional permeation emissions related to the use of ethanol. Recent testing at that time
indicated that ethanol increases the rate of permeation of hydrocarbons through plastic fuel tanks
and elastomers used in fuel line connections, as well as permeating itself. Subsequent testing as
99
-------
confirmed this effect. Therefore, we have added the effect of ethanol on permeation emissions to
MOBILE6.2's estimate of non-exhaust VOC emissions in assessing the impact of gasoline
quality on emissions here.
Air Improvement Resource, Inc. for the American Petroleum Institute, recently
summarized the available test data on the effect of ethanol on permeation emissions and
developed a methodology for estimating in-use permeation emissions in several U.S. cities.50
This study provides a useful starting point for incorporating these emissions into this RFS
analysis.
As a starting point, it is useful to point out that the non-exhaust emission estimates in
MOBILE6.2 include permeation emissions for non-oxygenated gasoline. Typical extended
diurnal emission tests (e.g., those lasting 2-3 days) automatically include any emissions
permeating through plastic and elastomeric fuel system components. However, since the
emission tests used as the basis of the MOBILE6.2 estimates of non-exhaust emissions primarily
were performed with non-oxygenated gasoline. Those tests that did include ethanol blends only
exposed the vehicle to this fuel for a few days. The CRC study of ethanol-related permeation
indicates that it takes at least a week or two for the effect of ethanol to fully develop. Therefore,
it is very unlikely that the tests performed by EPA and others to assess the impact of ethanol and
other fuel components on non-exhaust emissions included the effect of ethanol on permeation
emissions. In those cases where a vehicle may have been exposed to an ethanol blend for some
time prior to testing, the increased permeation emissions likely were still present when the
vehicle was tested on a non-oxygenated gasoline, still masking the effect. Therefore, our task
here is to develop an estimate of the incremental impact of ethanol use on permeation emissions,
and not an estimate of total permeation emissions with and without ethanol.
The primary source of ethanol permeation emission data is the CRC E-65 study.51 This
study tested 10 vehicles, 6 cars and 4 light trucks, ranging in model year from 1989 to 2001.
AIR placed these vehicles into three groups, based on a combination of model year and
applicable evaporative emission standards. The vehicles in the test program were certified to
two distinct evaporative emission requirements. The older vehicles were certified to EPA's or
California's 2 gram hot soak plus diurnal emission standard based on an accelerated one-hour
diurnal test. The three newest vehicles were certified to the enhanced evaporative emission
requirements first implemented in the 1996 model year, which included an extended two or three
day diurnal test. In addition, the data indicated that the three pre-1990 model year vehicles had
much larger incremental ethanol permeation emissions than the later pre-enhanced evaporative
emission vehicles. Therefore, AIR split the pre-enhanced evaporative emission category into
two groups, pre-1990 model year vehicles and 1990 and later model year vehicles. We believe
that this is appropriate and apply this split here, as well.
Since the earliest calendar year during which emissions are assessed in the RFS analysis
is 2012 and MOBILE6.2 only considers vehicles which are 24 years old or newer, at most only
two model years of pre-1990 vehicles are present in our analysis. Due to accumulated
scrappage, these vehicles comprise a very small percentage of the on-road fleet in 2012 and
disappear from our analysis by 2015. Therefore, we decided to ignore the pre-1990 model year
data here.
100
-------
AIR estimated the average incremental ethanol permeation emission rates for the 1990
and later model year pre-enhanced evaporative emission vehicles to be 0.86 gram per day
(g/day), while that enhanced evaporative emission vehicles was 0.80 g/day. Given the small
number of vehicles tested and the variability in the rates measured for individual vehicles, for the
purposes of this analysis, we consider these two levels to be generally equivalent. Therefore, we
use an average incremental ethanol permeation emission rate of 0.8 g/day for all vehicles.
Beginning with the 2004 model year, EPA and California implemented further
enhancements to their evaporative emission standards. The EPA "Tier 2" requirements include
accumulating mileage on durability data vehicles with an ethanol blend. However, actual
emission testing is still performed using non-oxygenated gasoline. We believed that this
combination of requirements would incorporate any effects of ethanol on emissions, including
potentially permeation effects. Because of these and other aspects of the 2004 and later
standards, AIR estimated that the permeation emissions due to ethanol would be reduced to 0.43
g/day for these vehicles.
We believe that it is likely that permeation emissions for non-oxygenated blends will be
lower for these vehicles, due to the fact that the diurnal emission standard was reduced from 2
g/day to 0.95 g/day and lower in some cases. However, as mentioned above, the effect of
ethanol on permeation emissions takes about 2 weeks to fully develop and to fully disappear.
Therefore, it is possible to accumulate mileage on a certification vehicle using an ethanol blend,
change the fuel to the emission test fuel, wait two weeks and then test the vehicle. In this case,
the effect of ethanol on permeation will have disappeared during the certification testing. Thus,
until EPA requires certification emission testing with ethanol blends, we have no assurance that
manufacturers will modify their vehicle designs to address the effect of ethanol on permeation.
Thus, for the purposes of this analysis, we maintain the estimate of 0.8 g/day for Tier 2, as well
as earlier vehicles.
Permeation emissions vary significantly with ambient temperature, with emission
increasing with increases in temperature. The 0.8 g/day emission estimate applies at an average
temperature of 95 F. The literature indicates that permeation varies exponentially with
temperature. The CRC testing indicates that permeation emissions double with every increase in
temperature of 18 F. Vice versa, permeation emissions drop 50% with every decrease in
temperature of 18 F. We apply this relationship in Chapter 4 in developing incremental ethanol
permeation emissions for each hour of the day in each county in the U.S.
We plan to update our projections of the effect of gasoline quality on non-exhaust VOC
emissions from Tier 1 and later model year vehicles based on additional testing which is
expected to begin next year. Additional testing of permeation emissions is already underway
with the CRC E-77 test program. These updated projections will be used in the comprehensive
assessment of the impact of the fuel-related provisions of the Energy Act which due in 2009.
Non-exhaust emissions are a function of ambient temperature and temperatures vary
across the nation. Therefore, it is not as simple to determine the effect of RVP and other fuel
qualities on non-exhaust emissions on a per vehicle basis as it is for exhaust emissions.
101
-------
Therefore, we performed a regression of the non-exhaust VOC and benzene emissions developed
in Chapter 4 as a function of fuel properties in order to estimate these effects on a per vehicle
basis. Specifically, we regressed the ratio of non-exhaust VOC and benzene emissions in each
county in July between two fuel scenarios (the 7.2 billion gallon ethanol, minimum ethanol use
in RFG scenario and the base case scenario) against the change in RVP, ethanol content and
MTBE content. The results are summarized in Table 3.1-7.
Table 3.1-7. Fuel-Non-Exhaust Emission Effects in MOBILE6.2: 2012
VOC
Benzene
RVP (%/psi)
15.6%
14.8%
Ethanol (%/Vol %)
-0.1%
-1.3%
MTBE (%/Vol%)
0.0%
-0.7%
Adjusted r-Square
0.98
0.08
3.1.1.1.5 PM Emissions
The amount of data evaluating the impact of ethanol and MTBE blending on direct
emissions of PM from gasoline-fueled vehicles is extremely limited. Three fairly limited studies
have evaluated the impact of ethanol blending on PM emissions from gasoline vehicles. These
studies are summarized below.
The Colorado Department of Public Health and Environment tested 24 vehicles on two
^0
winter grade commercial fuels at 35°F in Denver (i.e., at high altitude). Both fuels were
obtained from a local refinery. One fuel was non-oxygenated and represented fuel sold outside
of the Denver area. The other contained 10 vol% ethanol and represented fuel sold in the Denver
area, which has an oxygenated fuel mandate. As would be expected, the fuels differed in other
qualities besides ethanol content. The ethanol blend had a 2 vol% lower aromatic content, which
is somewhat less than expected. However, it also had a 53 F lower T50 level, which is a much
greater difference than is typical. The two fuels used during this testing appear to have been
used in random order (i.e., sometimes the non-oxygenated fuel was tested first, other times the
E10 fuel was tested first).
Half of the 24 vehicles were certified to Tier 0 emission standards, while the other half
were certified to Tier 1 standards. Each group of 12 vehicles included 8 cars and 4 light trucks.
The study found that PM emissions for the 24 vehicles over the FTP decreased from
about 9 mg/mi to about 6 mg/mi with the ethanol blend, for a reduction of 36%. In addition, the
vehicles with the highest base PM emission rates showed by far the largest reductions, both in
absolute terms and in terms of percentage. PM emissions from Tier 1 vehicles decreased from
roughly 5.5 mg/mi to 4 mg/mi with the ethanol blend, for a reduction of 27%. Essentially all of
the emission reduction occurred during Bag 1 of the test (i.e., related to the cold start).
PM emissions were also measured over a warmed up California Unified Cycle (i.e., no
cold start). PM emissions for the 24 vehicles over this cycle for the two fuels were not
statistically different. The ethanol blend increased PM emissions from the Tier 0 vehicles
slightly and decreased those from the Tier 1 vehicles slightly.
102
-------
Finally, PM emissions were also measured over an EPA REP05 Cycle, again with no
cold start. PM emissions over this cycle were 4-5 times those over the California Unified Cycle,
indicating the impact of high speed, aggressive driving on PM emissions. However, despite this
general increase in PM emissions, for the 24 vehicles PM emissions over the REP05 cycle were
again very similar for the two fuels. This time, however, the ethanol blend decreased PM
emissions from the Tier 0 vehicles slightly and increased those from the Tier 1 vehicles slightly.
Overall, this testing indicates that the effect of ethanol (together with lower aromatics and
T50 levels) may reduce PM emissions due to cold starting at 35 F under high altitude conditions.
However, PM emissions during warmed up driving are very low and an effect of fuel quality was
indiscernible.
The State of Alaska, in conjunction with General Motors Corp. and EPA, measured PM
emissions from ten vehicles ranging in model year from 1977 to 1994 using two fuels.53 The
non-oxygenated fuel was a commercial wintertime fuel from the Fairbanks areas. The ethanol
blend in this study was created from the non-oxygenated fuel via splash blending. Testing was
performed both in Alaska using a portable dynamometer. Three of the vehicles were also tested
at EPA's laboratory in Research Triangle Park, N.C, ranging in model year from 1987 to 1994.
The testing in Alaska was performed at -20°F, 0°F, and 20°F. The EPA testing was performed at
these same temperatures plus 75 F. Both sets of testing began with testing with the non-
oxygenated fuel, followed by testing with the E10 fuel. This could introduce a bias into the
results, but the degree of this is unknown.
The cold conditions led to difficulties in measuring PM emissions in Alaska. Therefore,
few acceptable measurements of PM were made and the results were not presented in the paper.
The fact that the EPA testing was conducted in a laboratory made vehicle conditioning and
operation and particulate collection more feasible. The PM emissions from the three vehicles
tested by EPA on the two fuels are presented in the paper.
Only one measurement of PM emissions was made for each combination of vehicle, fuel
and temperature. Thus, no direct measurement of test to test variability is available. We
calculated the percentage difference in PM emissions between the E10 and EO fuel for each of
the eleven combinations tested. PM emissions with the ethanol blend ranged from 81% lower to
84% higher than those with the EO fuel. Thus, there appears to be considerable variability in the
test results. Taken together, the average of the percentage changes for each condition showed
the ethanol blend reducing PM emissions by 21%. However, this decrease was not statistically
significant at the 90% confidence level. The ethanol blend more consistently decreased PM
emissions at -20 F and OF, but not at 20 F or 75 F. The paper states that PM emissions at the
higher two temperatures were very low and the differences tended to be within measurement
accuracy. It is important to note, however, that the lower end of this range is 20 F. Only a few
percent of driving in the U.S. occurs below this temperature.
The third and final study was performed by EPA's laboratory in Research Triangle Park,
N.C.54 This study was conducted in three phases; the last two of which are relevant here. In
Phase II, PM emissions from two 1993-1995 model year vehicles were tested at at -20 F, 0 F,
and 20 F. In Phase III, PM emissions from an additional five 1987-2001 model year vehicles
103
-------
were tested at -20 F, 0 F, 20 F and 40 F. Both phases utilized two fuels, one a wintertime non-
oxygenated fuel and the other a 10 vol% ethanol fuel created from the non-oxygenated fuel via
splash blending. Both phases measured PM emissions over the FTP and over a series of four
back-to-back EVI240 tests.HH It is not clear whether the fuels were always tested in the same
order or tested randomly. Some testing was performed with various malfunctions induced on the
vehicles, like disconnecting the oxygen sensor. We focus on the emissions from the properly
operating vehicles here.
Of the 26 combinations of vehicle and temperature tested, valid PM measurements over
the FTP were successfully obtained for both fuels in 21 of them. The average percentage change
in PM emissions due to ethanol blending was +1%, in other words a very slightly increase. In
contrast to the results of the two test programs discussed previously, the ethanol blend did not
show a benefit at -20 F, and showed only a very slight 1% reduction in PM emissions at 0 F.
The data show some tendency for the ethanol blend to produce a greater PM emission reduction
for the highest emitting vehicles. However, this trend is not as clear as in the Colorado study.
Thus, this study indicates no clear effect of ethanol on PM emissions.
The IM240 testing showed much lower PM emission levels due to the warmed up nature
of the test. There was also no clear trend in the effect of ethanol on PM emissions in this testing.
The available data indicate that ethanol blending might reduce exhaust PM emissions
under very cold weather conditions (i.e., 0 F or less), particularly at high altitude. There is no
indication of PM emission reductions at higher temperatures and under warmed up conditions.
The data are certainly too limited to support a quantitative estimate of the effect of ethanol on
PM emissions.
Fine particles can also be formed through a series of chemical reactions in the atmosphere
from gasses such as sulfate (SC>42"), nitrate (N(V), ammonia (NH3), and volatile organic
compounds (VOC) emitted from motor vehicles. This aerosol formed secondarily in the
atmosphere through these gas to particle conversions will be discussed in greater detail in
Chapter 5 of this document. Emerging science is indicating that gaseous aromatic compounds
are likely among the most important VOCs which are precursors of carbonaceous PM which is
formed in the atmosphere. Therefore, we discuss the effect of fuel quality on aromatic
hydrocarbon emissions in the next section.
3.1.1.1.6 Aromatic Emissions
The Auto/Oil Air Quality Improvement Research Program tested over 100 vehicles from
model years 1983 - 1989 on a fuel matrix of over 80 fuel blends to determine the exhaust
emission effects of varying fuel parameters - including ethanol and aromatics.55 Phase 1 of this
study tested two fleets of vehicles: twenty (20) 1989 model year vehicles, and fourteen (14)
1983-1985 model year vehicles. A matrix of 16 fuels (Matrix A) was developed in the first
portion of the study with half the fuels containing 20% aromatics by volume and half with 45%
aromatics by volume. This data was used to investigate the impact of changing aromatic levels
HH The IM240 test is a warmed up test consisting of a portion of the FTP driving cycle. It was designed as a short
transient test cycle for use in vehicle inspection and maintenance programs.
104
-------
in the fuel on the aromatics emitted in the exhaust as a function of total hydrocarbon emissions.
Linear regression of the test data indicates that there is a linear relationship between the level of
aromatics by volume in the fuel and the mass of aromatics emitted in the exhaust. Based on the
results of this regression, aromatics have a tendency to be emitted less than proportionally to
their percent volume in the fuel, as shown in the following equation:
Aromatics Exhaust(wt%) = 0.64 x Aromatics Fuel (vo/%)
The coefficient in the above equation was statistically significant at the 90% confidence level
(0.64 ± 0.02). However, when we considered the presence of an intercept, it was not statistically
significant (-0.35 ± 2.26). Therefore, we forced the regression line through zero and repeated the
regression.
The Auto/Oil program also produced data which allows the effect of ethanol on aromatic
hydrocarbon emissions to be assessed. As discussed in Section 2.2, ethanol blending tends to
reduce the aromatic content of gasoline. Of interest here is whether ethanol has any other effect
on aromatic hydrocarbon emissions beyond that associated with reducing the aromatic content of
gasoline.
The Auto/Oil data contained a subset of fuels designed specifically for this analysis. A
total of 4 ethanol blends were produced by splash blending ethanol into four non-oxygenated
fuels. Two of the non-oxygenated fuels came from the original group of sixteen tested during
Phase 1 of the research study. Base fuel A (industry average fuel) was a 9 RVP fuel with 32%
aromatics. Base fuel F was also a 9 RVP gasoline with 20% aromatics. Two new non-
oxygenated fuels were created from base fuels A and F. In both cases, and butane was removed
to lower the RVP level by 1 psi resulting in fuels V and S, respectively. To each of these 4 non-
oxygenated fuels, 10% ethanol was splash blended resulting in the final 8 fuel test matrix. A
summary of the differences between the expected and actual aromatic content in the fuel and in
the exhaust as a function of THC is shown in Table 3.1-8, on the following page.56
Table 3.1-8. Expected vs. Predicted Non-Oxy and E10 Fuel Properties and Exhaust
Aromatics Reductions for Auto/Oil AQIRP "Fuel Matrix B'
Non-
Oxy
Fuel
A
V
F
S
Fuel
Aromatics
32
33.5
20
21.2
E10
Fuel
X
w
u
T
Fuel
Aromatics
Expected3
28.8
30.15
18
19.08
Measured
Fuel
Aromatics
27.2
29.0
19.1
18.1
% Reduction in
Exhaust
Aromatics
Expected
2.05
2.14
1.28
1.36
% Reduction
in Exhaust
Aromatics
Actual
2.39
1.54
0.93
0.63
aBased on dilution as a result of splash blending 10% ethanol
The aromatic contents of the ethanol blends, as listed in Table 3.1-8, do not reflect the
10% reduction from the aromatic contents of their non-oxygenated base fuels which would be
expected from splash blending with 10 vol% ethanol. The discrepancies between the measured
and estimated fuel aromatic contents are small, on the order of 1-2 vol% aromatics. However,
105
-------
when the total difference is on the order of 3 vol% ethanol, these discrepancies are significant.
The discrepancies are likely the result of measurement uncertainty of both base and ethanol
fuels.
Using the relationship between fuel aromatic content and aromatic hydrocarbon
emissions developed above, we can predict the reduction in exhaust aromatic emissions
associated with the differences in the aromatic contents of the non-oxygenated fuels and their
ethanol containing counterparts. We believe that it is most accurate to use the expected aromatic
contents for the ethanol blends rather than the measured levels, since it is likely that the volume
of ethanol added was very close to 10 vol%.
The expected reduction in the percentage of VOC emissions represented by aromatic
hydrocarbons based on the expected reduction in fuel aromatic content is shown in the second to
the last column in Table 3.1-8. The measured reduction in the percentage of VOC emissions
represented by aromatic hydrocarbons is shown in the last column. In three out of four cases (all
but fuels A and X), the actual reduction in aromatic emissions is less than the predicted reduction
based on dilution. Had we used the measured aromatic contents for the ethanol blends, the
outcome would have been the same: In three out of four cases (in this case all but fuels F and U),
the actual reduction in aromatic emissions is less than the predicted reduction based on dilution.
Qualitatively, this indicates that there does not appear to be any additional benefit in
reducing aromatic hydrocarbon emissions associated with the use of ethanol beyond that
expected from the reduction in the aromatic content of gasoline portion associated with ethanol
blending.
Based on our analysis in Section 2.2, increased ethanol blending will significantly reduce
gasoline aromatic content. This could cause a corresponding reduction in the aromatic fraction
of exhaust VOC emissions relative to non-oxygenated conventional gasoline. In addition,
ethanol also reduces total exhaust VOC emissions from older vehicles and may do so from newer
vehicles, based on the CRC E-67 study. This would further reduce emissions of aromatic
hydrocarbons. As will be discussed further in Chapter 5, this reduction in aromatic hydrocarbon
emissions could reduce ambient levels of secondary organic PM.
3.1.1.1.7 Emission Effects Associated with Specific Fuel Blends
3.1.1.1.7.1 Conventional Gasoline Analysis
In Section 2.2 of Chapter 2, we estimated the effect of blending ethanol and MTBE on
gasoline quality. Here, we present the effect of these changes in fuel quality on emissions from
motor vehicles in percentage terms, relative to those of a typical non-oxygenated U.S. gasoline
blend. Table 3.1-9 presents the gasoline qualities of a typical 9 RVP CG, as well as MTBE and
ethanol blends which reflect the effect of adding these two oxygenates to gasoline.
106
-------
Table 3.1-9. CG Fuel Quality With and Without Oxygenates"
Fuel Parameter
RVP (psi)
T50
T90
Aromatics (vol%)
Olefins (vol%)
Oxygen (wt%)
Sulfur (ppm)
Benzene (vol%)
Typical 9 RVP CG
8.7
218
332
32
7.7
0
30
1.0
MTBECG Blend
8.7
206
324
25.5
7.7
2
30
1.0
Ethanol CG Blend
9.7
186
325
27
6.1
3.5
30
1.0
Assumes summer (July) conditions
Table 3.1-10 presents the differences in emissions of the MTBE and ethanol blends
relative to that of non-oxygenated conventional gasoline.
Table 3.1-10. Effect of Oxygenates on Conventional Gasoline Emissions"
Pollutant
Exhaust VOC
NOx
cob
Exhaust Benzene
Formaldehyde
Acetaldehyde
1,3-Butadiene
Non-Exhaust VOC
Non-Exhaust Benzene
Source
EPA Predictive Models
MOBILE6.2
EPA Predictive and
Complex Models
MOBILE6.2
MOBILE6.2 & Complex
Models
1 1 Volume
Percent MTBE
-9.2%
2.6%
-6% 7-11%
-22%
+ 10%
-8%
-12%
Zero
-10%
10 Volume
Percent Ethanol
-7.4%
7.7%
-11% 7-19%
-27%
+3%
+141%
-27%
+30%
+13%
Assumes summer (July) conditions
The first figure shown applies to normal emitters; the second applies to high emitters.
The two oxygenated blends both reduce exhaust VOC and CO emissions, but increase
NOx emissions. The MTBE blend does not increase non-exhaust VOC emissions due to the fact
that non-oxygenated and MTBE blends have to meet the RVP standard. Ethanol blending
increases non-exhaust VOC emissions in two ways. First, ethanol blends are allowed 1.0 psi
higher RVP levels in most areas with CG. Second, ethanol increases permeation emissions. The
most notable effect on toxic emissions in percentage terms is the increase in acetaldehyde with
the use of ethanol. Acetaldehyde emissions more than double. However, as will be seen below,
base acetaldehyde emissions are low relative to the other toxics. Thus, the absolute increase in
emissions is relatively low.
3.1.1.1.7.2 Reformulated Gasoline (RFG) Analysis
The previous section discussed the relative emission changes to expect when adding
ethanol to the conventional non-oxygenated gasoline pool. A second scenario to consider is the
107
-------
case where RFG areas change from MTBE, a commonly used oxygenate in RFG areas, to either
ethanol RFG or a non-oxygenated RFG.
Whether MTBE is removed from RFG and replaced by ethanol, or is removed and simply
left without an oxygenate, it is assumed for the purposes of this analysis that sulfur
concentrations in the fuel will remain at 30 ppm, and that olefm content and benzene would not
change. Since RFG has tighter aromatics control than conventional gasoline, we will assume
that aromatics will remain constant for toxics control when oxygen is either removed or added.
Therefore the only fuel properties that change in this analysis are oxygen, T50, and T90.
Table 3.1-11 presents the gasoline qualities of three types of RFG: non-oxygenated, a
typical MTBE RFG as has been marketed in the Gulf Coast and a typical ethanol RFG which has
been marketed in the Midwest.
Table 3.1-11. RFG Fuel Quality With and Without Oxygenates"
Fuel Parameter
RVP (psi)
T50
T90
Aromatics (vol%)
Olefms (vol%)
Oxygen (wt%)
Sulfur (ppm)
Benzene (vol%)
Non-Oxygenated RFG
6.7
214
325
25.5
13.1
0
30
0.65
MTBE RFG
6.7
212
321
25.5
13.1
2.1
30
0.65
Ethanol RFG
6.7
194
322
25.5
13.1
3.5
30
0.65
Assumes summer (July) conditions
Table 3.1-12 presents the emission impacts of these three types of RFG relative to the 9
RVP CG described in Table 3.1-9.
108
-------
Table 3.1-12.
Effect of RFG on Per Mile Emissions from Tier 0 Vehicles
Relative to a Typical Conventional Gasoline"
Pollutant
Source
Non-Oxy
RFG
1 1 Volume
Percent MTBE
10 Volume
Percent Ethanol
Exhaust Emissions
voc
NOx
CO
Exhaust
Benzene
Formaldehyde
Acetaldehyde
1,3-Butadiene
EPA Predictive
Models
MOBILE6.2
EPA Predictive
and Complex
Models
-7.7%
-1.7%
-24%
-18%
7%
7%
22%
-11.1%
2.4%
-28%
-30%
11%
-8%
2%
-12.9%
6.3%
-32%
-35%
2%
143%
-7%
Non-Exhaust Emissions
VOC
Benzene
MOBILE6.2 &
CRC E-65
MOBILE6.2 &
Complex Models
-30%
-5%
-30%
-15%
-18%
-7%
Assumes summer (July) conditions
As can be seen, the oxygenated RFG blends are predicted to produce a greater reduction
in exhaust VOC and CO emissions, but a larger increase in NOx emissions. Non-exhaust VOC
emissions with the exception of permeation are roughly the same due to the fact that the RVP
level of the three blends is the same. However, the increased permeation emissions associated
with ethanol reduces the overall effectiveness of ethanol RFG. The most notable effect on toxic
emissions in percentage terms is the increase in acetaldehyde with the use of ethanol.
Acetaldehyde emissions more than double. However, as will be seen below, base acetaldehyde
emissions are low relative to the other toxics. Thus, the absolute increase in emissions is
relatively low.
The exhaust emission effects shown for VOC and NOx emissions only apply to Tier 0
vehicles in our primary analysis. In the sensitivity analysis, these effects are extended to Tier 1
and later vehicles. The effect of RVP on non-exhaust VOC emissions is temperature dependent.
The figures shown above represent the distribution of temperatures occurring across the U.S.
under summer conditions (average July fuel specifications).
3.1.1.2
Nonroad Equipment
We use EPA's NONROAD emission model to estimate the effect of gasoline quality on
emissions from nonroad equipment. We use the 2005 version of this model, NONROAD2005,
which includes the effect of ethanol on permeation emissions from several types of equipment:
all small spark-ignition equipment (including handheld and non-handheld equipment less than or
equal to 25 hp), all spark-ignition recreational marine watercraft (includes all outboard, stern-
drive inboard, and personal watercraft). Note that these categories do not include recreational
vehicles (motorcycles, ATVs, and snowmobiles) or large spark-ignition equipment.
109
-------
Only a limited number of fuel parameters affect emissions in NONROAD. Exhaust
VOC, CO and NOx emissions are a function of sulfur and oxygen. Here, only the latter fuel
parameter is of interest. Emissions of all three pollutants are assumed to change proportionally
with fuel oxygen content. Table 3.1-13 shows the effect of moving to a 10 volume percent
ethanol blend on these emissions, either from a non-oxygenated fuel or from an 11 volume
percent MTBE blend.
Table 3.1-13.
Effect of a 10 Volume Percent Ethanol Fuel on Nonroad Exhaust Emissions
Base Fuel
VOC
CO
NOx
4-Stroke Engines
Non-
Oxygenated
-15%
-21%
+37%
1 1 Volume
Percent MTBE
-6%
-8%
+13%
2-Stroke Engines
Non-
Oxygenated
-1%
-12%
+18%
1 1 Volume
Percent MTBE
~ zero
-4%
+6%
As can be seen, the higher oxygen content of ethanol blends reduces exhaust VOC and
CO emissions. However, it also increases NOx emissions quite substantially, especially from 4-
stroke engines. However, it should be noted that NOx emissions from these engines tend to be
fairly low to start with, given the fact that these engines run richer than stoichiometric. Thus, a
large percentage increase of a relative low base value can be a relatively small increase in
absolute terms. This will be seen below in Chapter 4, when we evaluate the impact of increased
ethanol use on the local and national emission inventories.
Non-exhaust VOC emissions (other than permeation) are a function of gasoline RVP and
ethanol content in NONROAD2005. Ethanol content only affects permeation emissions. Both
of these emissions are temperature dependent, so the effect of ethanol and RVP is also
temperature dependent. Based on the results of modeling national emissions in July, a 10
volume percent ethanol blend increases non-exhaust VOC emissions by 15 percent. This
assumes a 1.0 psi increase in RVP.
11 Permeation emissions from nonroad equipment are not regulated. Thus, the elastomers used in the fuel systems of
nonroad equipment are likely to be more similar to those of older onroad vehicles than those of later onroad vehicles
which were subject to stringent non-exhaust VOC emission standards.
110
-------
1) Permeation emissions in the public version of NONROAD2005 were 450 grams per
meter-squared per day (g/m2/day) for both small spark-ignition engines and for the supply
hoses on portable fuel tanks in recreational marine watercraft, applicable to all fuels. For
both types of equipment, the permeation emission rates were changed to 122 g/m2/day for
gasoline and 222 g/m2/day for El0.
2) Permeation emissions in the public version of NONROAD2005 were 100 g/m2/day
for supply hoses on outboard recreational marine watercraft (> 25 hp), 300 g/m2/day for
supply hoses on personal watercraft (PWC), and 110 g/m2/day for fill neck hoses on both
outboards and PWC. These three permeation emission rates were changed to 42
g/m2/day for gasoline and 125 g/m2/day for E10.
3) Permeation emissions from sterndrive/inboard recreational marine watercraft in the
public version of NONROAD2005 were 100 g/m2/day for supply hoses and 110 g/m2/day
for fill neck hoses. Both of these permeation emission rates were changed to 22 g/m2/day
for gasoline and 40 g/m2/day for El0.
4) Permeation emissions in the public version of NONROAD2005 were 0 g/m2/day for
vent hoses on all recreation marine watercraft. This permeation emission rate was
changed to 2.5 g/m2/day for gasoline and 4.9 g/m2/day for E10.
5) One final adjustment was to double the vent hose length for all gasoline-fueled
outboards, personal watercraft, and sterndrive/inboard watercraft.
The NONROAD emissions model does not estimate emissions of toxic air pollutants
from nonroad equipment. However, the National Mobile Inventory Model (NMIM) does make
such estimates. NMIM utilizes the MOBILE and NONROAD models to develop national
emission estimates for motor vehicles and nonroad equipment. For the most part, NMIM
provides the relevant inputs to MOBILE6.2 and NONROAD and processes the results.
However, with respect to nonroad toxic emissions, NMEVI takes exhaust and non-exhaust VOC
emission estimates from NONROAD and applies a set of toxic fractions of VOC emissions
based on fuel quality.58 NMIM contains estimates of the toxic fractions of VOC emissions for
three fuels: a non-oxygenated gasoline, an MTBE blend and an ethanol blend. NMIM applies
the fraction of VOC emissions represented by each of the air toxics to either the exhaust or non-
exhaust VOC emissions estimated by NONROAD. The toxic fractions of VOC were derived
from motor vehicle testing. Thus, they are considered approximate.
We hope to update our projections of the effect of gasoline quality on exhaust and non-
exhaust emissions from nonroad equipment based on additional testing to be conducted over the
next several years if funding allows. These updated projections could be used in the
comprehensive assessment of the impact of the fuel-related provisions of the Energy Act which
is due in 2009.
Ill
-------
3.1.2 High-Level Ethanol Blends
The vast majority of ethanol blended into gasoline as a result of the RFS is expected to be
used in a 10 vol% ethanol blend (E10) rather than an 85 vol% ethanol blend (E85), as discussed
in Chapter 1. At the same time, some ethanol is likely to be used as E85, and its use is growing.
Therefore, it is useful to examine the available data regarding the emission impacts of E85.
3.1.2.1 Exhaust emissions
3.1.2.1.1 Regulated Pollutants
Relatively little data is available for investigating the effects of high level ethanol blends
on exhaust emissions. Part of the 1993 Auto/Oil Air Quality Improvement Research Program
(AQIRP) investigated the emissions associated with the use of E85 blends. Emissions over the
Federal Test Procedure (FTP) were measured from three Tier 0 and Tier 1 certified flexible-fuel
vehicles with three test fuels. Another source of emission data is EPA's Certification and Fuel
Economy Information System (CFEIS) database, which contains certification data for all FFVs
sold in the US.59
The Auto/Oil Study found that E85 reduced FTP composite NOx emissions by 49%
compared to conventional gasoline with 1988 industry average fuel properties. This is likely the
result of improved catalyst efficiency due to the low sulfur concentration in E85. The 2005 and
2006 CFEIS data from new FFVs, on the other hand, show a 4% increase is NOx emissions with
E85, with a large degree of vehicle to vehicle variability (standard deviation of 35%). Neither
the Auto/Oil study nor the certification data found statistically significant changes in CO.
Emissions of Non-Methane Organic Gases (NMOG) increased 33% and 56% in the CFEIS and
Auto/Oil data, respectively. The measurement used to determine NMOG in both CFEIS and
Auto/Oil data include the mass of oxygen in all measured organic species except methane.
While the emissions of NMOG are expected to increase with E85 compared to EO for
Tier 2 certified vehicles, the majority (-55%) of E85 NMOG emissions are direct emissions of
ethanol, which has a relatively low reactivity compared to other NMOG species. Thus there may
still be a slight NMOG benefit based on ozone reactivity despite a potential net increase in total
NMOG emissions. NMOG emissions with E85 at colder temperatures could be much greater (2
to 3 times higher than with EO). Much of this increase occurs during cold start, before the
combustion chamber has reached high enough temperatures to allow ethanol vaporization. As
with NMOG emissions at normal engine operating temperatures, a high percentage of the
increased NMOG emissions are ethanol and acetaldehyde.
3.1.2.1.2 Air Toxics
With increasing use of E85, some air toxics may increase while others decrease relative
to EO. Emissions of Benzene and 1,3-Butadiene decrease while acetaldehyde, formaldehyde, and
emissions of ethanol increase. The net result is an increase in total air toxics, but this is largely
driven by increase ethanol and acetaldehyde emissions. Table 3.1-14, below, shows the percent
change in FTP composite g/mile emissions of several air toxics for the three FFVs tested on three
112
-------
fuels as part of the 1993 Auto/Oil study. The fuels tested were AQIRP gasoline with!988
Industry average qualities (CG), a 1996 California phase 2 reformulated gasoline (RFG blended
with MTBE), and an E85 blend with identical gasoline specs as RFG.60
Table 3.1-14.
Percent Difference in Toxic Emissions Between EO and E85
% Difference Between Fuels
Formaldehyde
Benzene
1,3-Butadiene
Acetaldehyde
Total Toxics
RFG vs CG
-2
-55
-31
-18
-42
E85 vs CG
93
-87
-85
2620
108
E85 vs RFG
97
-72
-79
3220
255
The increase in acetaldehyde emissions is substantial, on the order of 20 to 30 times that
of EO. This is substantially higher than the 15 to 20 fold increase shown with Tier 2 FFVs in the
CFEIS data. Emissions of benzene, 1,3-butadiene, ethylbenzene, hexane, styrene, toluene, m-
xylene, p-xylene, o-xylene, and naphthalene are all expected to decrease significantly (50-80%)
with the use of E85 vs. EO according to CFEIS, which is consistent with the Auto/Oil results
presented in Table 3.1-14. Regardless of vehicle technology, the increased emissions of
acetaldehyde could be a potential concern due to its strong odor, as well as its respiratory system
irritating and potentially carcinogenic properties.
3.1.2.1.3 Parti culate Matter
Even less data exists to draw firm conclusions on direct particulate matter emissions due
to increased E85 use. Theoretically, E85 use has the potential to increase direct emissions of PM
under modes of rich engine operation. This is especially important at cold start, before the
catalyst has reached its operating temperature and when an E85 fueled vehicle runs substantially
richer than if it were fueled with EO. In this situation, the low temperatures in the combustion
chamber, compounded by the evaporative cooling effect of ethanol, makes fuel vaporization
difficult and may increases exhaust emissions of raw fuel and PM at cold start. Sustained
periods of high load may also have increased emissions of PM with E85 than with EO due to
richer operation with E85. Results from a 2003 SAE paper showed a negligible increase in
direct PM emissions from E85 vs. EO fueled vehicles over the European Test Cycle (Directive
70/220/EEC and its amendments).61 Tests conducted at 23°C and 16°C showed an increase in
PM emissions with decreasing temperature for both EO and E85, with slightly higher PM
emissions at cold temperature with E85. This study only used one E85 blend and one model year
2002 FFV, however, so these results cannot be considered entirely representative of the on-road
FFV vehicle fleet. Again, since the projected use of ethanol as E85 is very small compared to
its use as E10, the emissions impacts associated with E85 will be also be quite small on an
absolute scale.
As discussed above, emerging science is beginning to identify gaseous aromatics as an
important precursor to secondary organic aerosol. Exhaust aromatic emissions should be
reduced with E85 since the fuel aromatics content of E85 is much lower than that for EO or E10
113
-------
blends. This reduction in exhaust aromatics should reduce the formation of secondary organic
aerosol. However, as mentioned earlier, no specifications currently exist for the 15% gasoline
portion of E85. Thus, the degree that the aromatic content of E85 will be lower than gasoline is
not known with any confidence. Lack of data regarding on the speciation of VOC emissions also
prevents any quantitative estimate of any benefit in this area.
3.1.2.2 Non-Exhaust Emissions
We currently have very little data regarding non-exhaust emissions from E85 vehicles.
Theoretically, evaporative emissions of E85 fueled vehicles have the potential to be lower than
with EO or E10. This is because ethanol blended with a given gasoline at the 85% level is likely
to be less volatile than EO or E10 (with the same gasoline fuel quality). This is not entirely
certain, however, since there is no fuel specification for the hydrocarbon composition of the 15%
of E85 that is gasoline. Thus, the RVP of the final E85 blend could be closer to that of EO or
E10 fuels than commonly thought is the case. Moreover, since the volatility of ethanol blends
peaks between 6 and 30 vol% ethanol, the fuel in the tank of drivers of flex-fuel vehicles who
alternate between E85 and gasoline will experience a wide range of ethanol concentrations in the
fuel at any given time, and therefore a wide variation in the corresponding evaporative
emissions.
Similarly, we have very little data with which quantitative predictions of the impact of
E85 use on non-exhaust emissions of air toxics (e.g., benzene) can be drawn. The Auto/Oil
study mentioned in the previous section tested the same three Tier 0 and Tier 1 vehicles for hot-
soak evaporative emissions. They found no statistically significant change in NMOG or
OMHCEJJ evaporative emissions, yet found a statistically significant 60% reduction in benzene
emissions. Directionally, you would expect both hydrocarbon and air toxic evaporative
emissions to decrease due to the dilution of the hydrocarbon portion of the fuel with ethanol.
However, again, it is highly dependent on the volatility of the gasoline component of the specific
E85 used and its benzene content, neither of which is regulated.
3.1.3 Biodiesel
Biodiesel is expected to be one of two renewable fuels to be used in significant volumes
through 2020. While ethanol will dominate the market, biodiesel use is likely to grow
considerably reaching 300 million gallons by 2012, according to EIA estimates. It is produced
domestically from vegetable oils, animal fats and recycled cooking oils, with the majority of this
product coming from soybean oil. It is typically used in 2%, 5% and 20% blends with diesel fuel
which have been assigned B2, B5 and B20 designations, respectively.
In 2002, EPA issued a report entitled "A Comprehensive Analysis of Biodiesel Impacts
on Exhaust Emissions". This report included a technical analysis of biodiesel effects on
regulated and unregulated pollutants from diesel powered vehicles. It gathered existing data
from various test programs to investigate these effects. About 80% of engines tested in those
programs represented model years 1991 through 1997. The remaining engines spanned model
years 1983 to 1990. None of them were equipped with exhaust gas recirculation (EGR) or
11 Organic Material Hydrocarbon Equivalent
114
-------
exhaust aftertreatment devices. Since the majority of then-available data was collected on
production highway engines, it formed the basis of this study. Only test results generated using
the heavy-duty transient Federal Test Procedure (FTP) or multiple-mode steady-state cycles were
included in the quantitative analysis.
All base fuels met boiling range requirements of the ASTM D975 diesel fuel
specification and were either high sulfur (< 5000 ppm S) or low sulfur (< 500 ppm S) No. 1 or
No.2 grades. Fuels made from pure chemicals rather than refinery streams were excluded from
analysis. Base fuels included in this study were divided into "clean" and "average" depending
on their cetane number, aromatic content, density and conformance with CA diesel fuel
requirements.
Emission impacts of B20 biodiesel fuel blended using an "average" diesel base fuel and
soybean-based biodiesel are characterized in Table 3.1-15. The B20 blend is shown to
moderately reduce HC, CO and particulate emissions while slightly increasing NOx. This fuel
was also found to reduce fuel economy by 1.6%. Aggregate toxics were predicted to be reduced,
but results differed considerably from one species to another and should be treated as preliminary
due to limited sample size. It is important to note that the conclusions of this study should be
considered to only apply to heavy-duty highway engines as insufficient data on the effects of
biodiesel on exhaust emissions of light-duty and nonroad engines were available for analysis.
Table 3.1-15.
Effect of Soybean-Based B20 Biodiesel Fuel on Exhaust Emissions
from Diesel Engines - 2002 EPA Study.
Pollutant
NOx
PM
HC
CO
Change in Emissions
+ 2%
-10.1 %
-21.1 %
-11.0%
For this rulemaking, EPA again reviewed the technical literature related to biodiesel
effects on exhaust emissions from diesel engines and diesel powered vehicles. This review
covered technical papers and reports published between 2002 and 2006, as well as one 2001
report which was not included in the 2002 EPA study. The same data selection criteria were
used, but the review was focused exclusively on soybean-based B20 blends. Due to the much
shorter time period involved and the scope of analysis limited to B20 blends, the volume of
engine and vehicle test data available for analysis was considerably smaller than in the 2002
EPA study. This was true, in particular, of experimental data generated using newer technology
engines and vehicles. We do not perform any detailed statistical analysis of the new
experimental data here, due to the complexities involved in combining data obtained using
different test cycles. Also, the reasons why the engine and vehicle testing sometimes yield
different results are not yet clear.
Engine test data selected for analysis in this follow-up review corresponded mainly to
model years 1998 - 2003. Two engines included in the data set were equipped with oxidation
catalysts and two with EGR systems. This is a marked improvement over the kind of engines
which had been tested at the time of the 2002 study, where the vast majority of the data was
115
-------
obtained on 1997 and earlier engines. Only test data based on the heavy-duty transient FTP test
or the AVL 8-mode steady-state cycle was considered.
Vehicle emission test data utilized in this review was split between model years 1983 to
1993 and 2002 to 2004. The vehicles included a passenger van, six pick-up trucks and two non-
combat military vehicles. Three different driving cycles were represented in the data set,
including the light-duty FTP, US06 and one test schedule representative of a specific vehicle
application. It is important to note that a substantial volume of chassis test data related to B20
effects in diesel powered vehicles has recently been generated in the course of several ongoing
test programs, but has not yet been published. We expect to be able to incorporate this data into
our analysis for the FRM.
The base fuels included in the analysis were refinery products conforming to the ASTM
D975 diesel fuel specification and were either low sulfur (< 500 ppm S) or ultra low sulfur (< 15
ppm S) No.2 grades. The B20 blends met boiling range requirements of the ASTM D975
specification.
Data indicating the emission impacts of B20 biodiesel fuel on exhaust emissions from
diesel engines which has become available since the 2002 study are shown in Tables 3.1-16.
This table shows the effect of B20 on the emissions of various pollutants for individual engines
and vehicles, as well as minimum, maximum and average values for each pollutant across the
entire group of engines tested.
116
-------
Table 3.1-16.
Recent Studies of the Effect of B20 on Emissions from Diesel Engines
(% change relative to base fuel)
Author
McCormick R.L.
et. al. (2001) 62
Souligny M. et.
al. 63
Alam M. et. al.
64
McCormick R.L.
et. al. (2005) 65
Vehicles Tested
1991 DDC Series 60
1998 Cummins C
Series (w/oxycat)
2000 Cummins C
Series (w/oxycat)
2000 Cummins B
Series
2002 Cummins ISB
(w/EGR)
2003 DDC Series 60
(w/EGR)
Test Cycle
Hot FTP
Hot FTP
Hot FTP
AVL8-
mode, base
fuel #1
AVL8-
mode, base
fuel #2
Hot FTP
Hot FTP
Minimum
Maximum
Average
NOx
2.9
0
0
0
-3.0
3.6
6.0
-3.0
6.0
1.4
HC
-35.0
-12.2
-21.0
-
-
-4.2
13.5
-35
13.5
-11.8
CO
-7.7
-24.6
-28.1
-
-
-10.5
1.0
-28.1
1.0
-14.0
PM
-23.6
-30.8
-16.6
-27.0
6.0
-24.9
-24.1
-30.8
6.0
-20.1
The average B20 effects on diesel engine emissions shown in Table 3.1-16 reinforce the
general conclusions of the 2002 EPA study. More specifically, they indicate that the B20 blend
causes moderate reductions in HC, CO and particulate emissions and a small - if any - increase in
NOx emissions.
Data indicating the emission impacts of B20 biodiesel fuel on exhaust emissions from
diesel vehicles which has become available since the 2002 study are shown in Tables 3.1-17.
117
-------
Table 3.1-17.
Recent Studies of the Effect of B20 on Emissions from Diesel Powered Vehicles
(% change relative to base fuel)
Author
Durbin,
T.D., et al 66
Hoi den
Bruce et. al.
67
Vehicles Tested
1993 Ford 350
1990 Ford E3 50 van
1990 Chevy 2500
1989 Chevy 2500
1987 Chevy C-30
1985 Chevy C-30
1983 Ford F-250
2004 Humvee
2002 Thomas bus -
Lab 2
2002 Thomas bus -
Lab 1
Test Cycle
FTP chassis
FTP chassis
FTP chassis
FTP chassis
FTP chassis
FTP chassis
FTP chassis
FTP chassis
US06
Cheyenne
Mountain
Cycle
Minimum
Maximum
Average
NOx
0.7
2.8
3.3
3.2
1.7
8.7
3.5
-1
-1
3
0.2
-1
8.7
2.3
HC
-21.3
39.4
-33
40.9
15.7
-9.7
-0.7
113
3
-22
-11.2
-33
113
10.4
CO
-6.1
-0.4
-10.6
12
-1.3
-1.8
-5.2
26
44
-17
-1.3
-10.6
44
3.5
PM
17.2
-8.1
-2.5
22.1
15.1
-11.3
39.8
-9
-57
-29
-8.4
-57.0
39.8
-2.8
As expected, the vehicle test data provided in Table 3.1-17 are considerably more
variable reflecting the strong effect of the different driving cycles used in their generation. The
average B20 effects on vehicle HC, CO and particulate emissions differ markedly from the
effects observed during engine testing. However, the response of NOx emissions to B20 was
similar to that seen in engine tests. These results should be treated as preliminary due to the
limited size of the vehicle population included in this analysis and the variability in the observed
emission effects.
The results of two additional test programs are shown in Table 3.1-18. The first is the
ongoing National Renewable Energy Laboratory (NREL) study of Denver transit buses. The
second study is a North Carolina State University (NCSU) study of twelve dump trucks operated
on B20 fuels. These programs are considered separately as they did not meet the data selection
criteria established for this review. The results of the NREL study are still considered to be
preliminary. We currently only have the averages of the emission data for several buses. The
NCSU study used an on-board emissions measurement system to monitor exhaust emissions of
the trucks during normal duty cycles. This system measured HC, CO, NO and particulate
emissions, but was not capable of measuring NO2; NO2 emissions have been shown to be
affected by biodiesel use in laboratory testing. Therefore, the NOx results shown here are not
118
-------
reliable. Both sets of data show moderate improvements of HC, CO and particulate emissions
associated with B20 use, similar to those seen in the 2002 EPA study.
Table 3.1-18.
Recent Studies of the Effect of B20 on Emissions from Diesel Powered Vehicles:
Supplementary data (% change relative to base fuel)
Author
Proc K., et.
al.68
Frey
Christopher,
et. al. 69
Vehicles Tested
5 Denver
Regional Transit
District Buses
12 1998-2004
dump trucks
Test Cycle
City Suburban Heavy
Vehicle Cycle
On-road tests using a
portable emissions
measurement system
NO
-4
(ave)
-10
(ave)
HC
-29
(ave)
-22
(ave)
CO
-24
(ave)
-11
(ave)
PM
-18
(ave)
-10
(ave)
In summary, the additional data which has become available since the time of the 2002
EPA study generally supports the results of the 2002 EPA study. In addition, the more recent
data indicates that the impacts of B20 on emissions from new engines may not be that different
from those of older engines (on a percentage basis). However, there is still a need for additional
test data, particular for newer technology engines and across the board for nonroad engines.
3.2 Emissions from Fuel Production Facilities
3.2.1 Ethanol
The primary impact of renewable fuel production and distribution regards ethanol, since
it is expected to be the predominant renewable fuel used in the foreseeable future. We
approximate the impact of increased ethanol production, including corn farming, on emissions
based on DOE's GREET model, version 1.6. We also include emissions related to distributing
the ethanol and take credit for reduced emissions related to distributing displaced gasoline.
These emissions are summarized in Table 3.2-1.
119
-------
Table 3.2-1.
Well-to-Pump Emissions for Producing and Distributing Ethanol from Corn
(grams per gallon ethanol)
Pollutant
voc
CO
NOx
PM10
SOx
Corn Farming
and
Transportation
0.8
4.3
11.3
8.1
1.2
Ethanol
Production
6.8
3.0
4.9
0.4
6.8
Co-
Product
Credits
-3.7
-3.0
-6.4
-2.5
-0.8
Ethanol
Trans-
portation
0.5
0.2
1.5
0.0
0.2
Gasoline
Transportation
Credit
-0.9
-0.1
-0.4
0.0
-0.1
Total
Emissions
3.6
4.4
10.8
6.1
7.2
At the same time, areas with refineries might experience reduced emissions, not
necessarily relative to current emission levels, but relative to those which would have occurred in
the future had ethanol use not risen. However, to the degree that increased ethanol use reduces
imports of gasoline, as opposed to the domestic production of gasoline, these reduced refinery
emissions will occur overseas and not in the U.S.
Similarly, areas with MTBE production facilities might experience reduced emissions
from these plants as they cease producing MTBE. However, some of these plants are likely to be
converted to produce other gasoline blendstocks, such as iso-octane or alkylate. In this case,
their emissions are not likely to change substantially.
3.2.2 Biodiesel
Like ethanol, we base our emission factors for biodiesel production distribution on the
estimates contained in the GREET model. Table 3.2-2 shows the emission factors associated
with soybean farming, soy oil production and esterification, and biodiesel distribution. We also
include emissions related to distributing the biodiesel and take credit for reduced emissions
related to distributing displaced diesel fuel. These emissions are summarized in Table 3.2-2.
Table 3.2-2.
Well-to-Pump Emissions for Producing and Distributing Biodiesel from Soybeans
(grams per gallon biodiesel)
Pollutant
VOC
CO
NOx
PM10
SOx
Total:
Soybean
Farming and
Transportation
3.1
14.5
24.4
1.0
3.7
Biodiesel
Production
38.3
10.6
19.4
0.5
3.8
Biodiesel
Transportation
0.5
0.2
1.1
0.0
0.1
Diesel Fuel
Transportation
Credit
-0.4
-0.1
-0.7
0.0
-0.1
Total
Emissions
41.5
25.1
44.3
1.5
7.5
At the same time, areas with refineries might experience reduced emissions, not
necessarily relative to current emission levels, but relative to those which would have occurred in
120
-------
the future had biodiesel use not risen. However, to the degree that increased biodiesel use
reduces imports of diesel fuel, as opposed to the domestic production of diesel fuel, these
reduced refinery emissions will occur overseas and not in the U.S.
121
-------
Chapter 3: Appendix
Fuel Property Tables and Summary of Predicted Emissions Changes
122
-------
Table 3A-1. CRC E-67 Study Test Fuel Properties
Inspection
API Gravity
Relalive Density
DVPE
Oxygenales-04815
MTBE
ETBE
BOH
02
Sulfur Conlenl
D 86 Disiillalion
IBP
5% Evaporated
10% Evaporated
20% Evaporated
30% Evaporated
40% Evaporated
50% Evaporated
60% Evaporated
70% Evaporated
80% Evaporated
90% Evaporated
95% Evaporated
EP
Recovery
Residue
Loss
Dnveability Index
E200
E300
Units
"API
60/60 °F
psi
vol %
vol %
vol %
wt %
ppm
T
•F
°F
°F
8F
°F
'F
°F
5F
°F
op
°F
°F
vol %
vol %
vol %
vol %
vol %
Fuel A
62.1
0.7310
7.74
0.03
0.02
0.02
0.02
18.8
94.2
126.3
136.0
148.6
163.6
179.8
194.7
209.0
224.2
243.4
294.3
327.4
351.2
97.0
1.8
1.2
1082,4
53.6
90.9
FuelB
59.9
0.7393
7.84
0.03
0.02
5.62
2.10
16.7
107.6
127.2
133.2
140.8
154.1
176.1
190.9
203.2
219.3
240.9
289,8
325.9
352.0
97.9
1.1
1.0
1075.8
57.6
91.5
FueIC
57.6
0.7482
7.70
0.13
0.01
10.37
3,84
19.0
104.3
124.6
130.5
138.8
146.6
153.7
192,7
223.5
245.7
281.5
329.2
343.4
374.0
97.7
1.2
1.1
1128,0
52.1
84
FuelD
61.4
0.7337
7.65
0.03
0.01
0.00
0,01
18.2
88.8
123.2
133.3
147.6
164.1
182.3
199,5
216.9
237.9
274.3
355.0
367.3
392.0
97.9
0.8
1.3
1153,3
50.6
83.6
FuelE
56.7
0.7519
7.80
0.11
0.01
10.26
3,78
17,2
106.3
124.3
130.5
139.5
147.2
153.8
197,7
226.2
259.2
299.7
351,7
364.9
385.4
97.4
1.4
1.2
1165.1
50,6
80.0
FuelF
60.1
0.7387
7.62
0.08
0.08
0.00
0,03
18.1
94.2
121.6
135.0
154.7
177.0
200.2
216.8
227.6
238,2
254.7
295.0
324.0
361.2
97.2
1.7
1.1
1148,0
40.0
90.9
FueIG
57.1
0.7502
7.78
0.13
0.04
10.15
3.76
17,5
103.7
125.3
133.2
143.7
152.9
163.4
212.2
226.7
237,0
251.7
290.7
327.8
365.4
96.7
1.5
1.8
1151,2
47,4
79.5
FuelH
60.6
0.7366
7.85
0.09
0.01
0.05
0.04
18,6
94.2
122.7
134.0
151.6
173.3
197.0
216,3
230.4
245.9
273.7
326,9
343.7
374.4
98.0
1.0
1.0
1176,8
41.7
85.2
Fuel I
57.2
0.7498
7.68
0.16
0.01
5.94
2,22
16.8
100.7
124.0
130.2
139.0
150.8
191.0
215,9
235.9
260.9
311.3
3542
366.6
391.8
97.9
1.1
1.0
1211.5
431
77.8
FueiJ
56.6
0.7525
757
013
001
5.90
219
191
102.6
1260
1344
1466
1755
220.5
2366
251.5
2719
305.2
3292
3387
3658
97.6
1.2
1.3
12549
35,2
78.4
FuelK
59.3
07416
771
016
002
0.00
003
219
93.9
1179
1297
1484
1744
208.5
2361
255.2
2796
319.1
3555
3686
3903
981
1.0
09
12582
376
75. 2
FuelL
54.4
0.761 1
7.69
016
001
1049
383
206
1061
1294
1400
1524
1588
202.1
2327
248.7
2735
3077
3491
3674
389.6
973
1.0
1 7
12823
394
780
123
-------
Table 3A-2. Summary of EPA E-67 vs. EPA Predictive Model Effects of E10 and MTBE Use Relative to CG and RFG
Base Fuel: EO: AAM CG EO: AAM CG E10: AAM E10:AAM E10:AAM 11% MTBE: Phase 2 RFG Class C RFG E10:
AAM Summer RVP -1 psi RVP-2psi Summer Avg Summer avg, Summer avg, Fuel props RFG: Non- MTBE: 1993 1993 region 2
avg. non-oxy delta delta T50 limited to T50 T90 02 are deltas oxy, from region 2 data, data (L),T50
L1 95°F only for EPA from AAM CG 1 993 region 2 low RVP & T90 delta
model class C Data, from AAM
(T50=195°F) low RVP data, low RVP
Fuel Parameters
RVP (psi)
T50 (°F)
T90 (°F)
Aromatics (vol %)
Olefins (vol %)
Oxygen (wt%)
Sulfur (ppm)
8.7
218
332
32
7.7
0
30
7.8
218
332
32
7.7
0
30
6.8
218
332
32
7.7
0
30
9.7
186
325
27
6.1
3.5
30
9.7
195
325
27
6.1
3.5
30
8.7
195
325
32
7.7
3.5
30
8.7
206
324
25.5
7.7
2.1
30
6.7
214
325
25.48
13.1
0
30
6.7
212
321
25.48
13.1
2.1
30
6.7
194
322
25.48
13.1
3.5
30
Predicted Emissions Changes
EPA Predictive Models (% change)
NOx
NMHC
EPA E-67 Model (% change)
NOx
NMHC
0.0
0.0
0.0
0.0
-1.0
-1.0
0.0
0.0
-2.1
-2.0
0.0
0.0
7.7
-7.4
9.2
-5.7
7.3
-7.0
6.7
-6.5
7.7
-7.5
6.7
-6.5
2.6
-9.2
-1.9
-4.7
-1.7
-7.7
-0.7
-3.2
2.4
-11.1
-2.2
-1.9
6.3
-12.9
8.8
-7.2
124
-------
Chapter 4: National Emission Inventory Impacts
This chapter describes the methods used to develop national emissions inventories under
the Renewable Fuel Standard (RFS) program. This chapter also presents and discusses these
inventories.
4.1 Impact of Ethanol Use
This section describes the methods used to develop national emissions inventories with
respect to ethanol consumption. This section also presents and discusses these inventories.
These inventories reflect only emissions from vehicles and equipment operating on ethanol-
blend gasoline, from both onroad and off-road sources. The off-road sources do not include
nonroad diesel, locomotive, or marine applications.
4.1.1 Overview of Cases
We consider five cases for the future use of ethanol-blend gasoline: a reference case,
then four control scenarios for increased ethanol use. The main difference between the cases is
our assumption about how much ethanol will be used and where it will go. While Chapter 2
discusses our methods for determining how much ethanol will go to each state in each case and
how fuel properties will be affected, this section of the DRIA uses those distributions to derive
estimates of the impact on national emissions inventories.
We evaluate each case by predicting fuel quality in each county of the U.S. in 2012. This
2012 fuel matrix is then used for all inventory and air quality assessments. The five ethanol use
cases are summarized in Table 4.1-1. The Reference Case represents our estimate of fuel quality
by county which existed in 2004. The remaining four cases represent increased levels of ethanol
use. Two of these assume that 7.2 billion gallons of ethanol will be consumed nationally, while
the other two assume a level of 9.6 billion gallons of ethanol use. For both the 7.2 and 9.6 billion
gallon volumes, we consider two alternative cases of minimum and maximum use of ethanol in
RFG. This min/max use of ethanol in RFG is reflected in the naming conventions used for the
cases in Table 4.1-1.
125
-------
Table 4.1-1. Overview of Cases for Future Ethanol Use and Distribution
Reference
7.2 Min
7.2 Max
9.6 Min
9.6 Max
Fuel quality is estimated for 2012, based on data fuel properties and programs in 2004
(the last year for which appropriate data were available).
Fuel quality is based on allocation of 7.2 billion gallons of renewable fuel in 20 12, with
a minimum amount of ethanol allocated to RFG areas.
Fuel quality is based on allocation of 7.2 billion gallons of renewable fuel in 20 12, with
a maximum amount of ethanol allocated to RFG areas.
Fuel quality is based on allocation of 9.6 billion gallons of renewable fuel in 2012, with
a minimum amount of ethanol allocated to RFG areas.
Fuel quality is based on allocation of 9.6 billion gallons of renewable fuel in 2012, with
a maximum amount of ethanol allocated to RFG areas.
4.1.2 National Emissions Inventory Estimation Procedure
Having approximated the effects of adding ethanol and removing MTBE on fuel
properties (see Chapter 2), the next step was to use the EPA's National Mobile Inventory Model
(NMIM)70 to calculate emissions inventories for gasoline fueled motor vehicles and nonroad
equipment in years 2012, 2015, and 2020. For all three years, we ran NMIM for January and
July, assuming that each was representative of winter and summer conditions, respectively. We
estimate annual emission inventories by summing the two monthly inventories and multiplying
by six. This was done in order to reduce the amount of time needed to actually run the model.
We chose 2012 because it is the year of full RFS program implementation. We chose
2015 and 2020 to illustrate how the emissions will change over time as the fleet changes. We
increased ethanol consumption beyond 2012 only by volumes required to maintain the same
proportion to gasoline that existed in 2012, and not by growth predicted in EIA estimates. By
restricting ethanol growth in this way, the same fuel quality that existed in 2012 would apply to
2015 and 2020, which would better highlight the effects of fleet turnover.
We ran NMIM separately for onroad and nonroad emissions, as each set of emissions
required a different set of adjustments subsequent to the model runs. For onroad emissions, the
effect of fuel quality on exhaust VOC and NOx emissions contained in the model (i.e., those in
MOBILE6.2) had to be replaced with those from the EPA Predictive Model. The effect of
ethanol on permeation emissions also had to be added to the onroad emission estimates. This
required a series of post-processing steps which are described below.
For nonroad emissions, NMIM was run using two simplified county-specific fuel
matrices. One represented no ethanol use nationally, while the other represented ethanol use
nationwide. We then interpolated between the two sets of county-specific emission estimates
based on the actual level of ethanol use expected in each county under the relevant ethanol use
case.
These steps for calculating emissions inventories are described in the following sections.
A summary of the models used and fundamental post-processing steps are listed below.
126
-------
Onroad Exhaust Emissions: Modeled using NMIM, which runs MOBILE6.2. Post-processed
model output to (1) replace VOC and NOx fuel effects for Tier 0 vehicles from
MOBILE6.2 with fuel effects from EPA Predictive Model (no fuel effect for Tier 1 and
later vehicles); (2) adjust exhaust air toxics emissions based on adjusted exhaust VOC
emissions. Conducted sensitivity analysis by applying fuel effects from EPA Predictive
Model to all vehicles.
Onroad Non-Exhaust Emissions: Modeled using NMIM, which runs MOBILE6.2. Post-
processed model output to account for permeation effects.
Nonroad Exhaust Emissions: Modeled using NMIM, which runs NONROAD2005 (which was
modified to account for hose permeation). Post-processed model output to interpolate
between the no-ethanol and ethanol cases using the county ethanol level.
Nonroad Non-Exhaust Emissions: Modeled using NMIM, which runs NONROAD2005
(which was modified to account for hose permeation). Post-processed model output to
interpolate between the no-ethanol and ethanol cases using RVP.
4.1.2.1 Onroad Emission Estimation Procedures
We ran NMIM to estimate county-specific emissions from gasoline motor vehicles for
January and July in years 2012, 2015, and 2020. For each month and year combination, we ran
the five onroad cases (Base, 7.2 Min, 7.2 Max, 9.6 Min, and 9.6 Max). The NMIM model
utilizes the MOBILE6.271 model to estimate motor vehicle emissions, as well as the effect of
fuel quality on emissions. As discussed in Chapter 3, the EPA Predictive Model contains more
recent estimates of the impact of fuel quality on exhaust VOC and NOx emissions. Therefore,
we removed the impact of fuel quality on exhaust VOC and NOx emissions as estimated by
MOBILE6.2 and replaced these impacts with those of the EPA Predictive Model. As also
discussed in Chapter 3, MOBILE6.2 does not include the impact of ethanol on permeation
emissions. Therefore, we added these emissions to those estimated by NMIM. Finally, we
arrived at annual emissions estimates by summing the January and July results, then multiplying
by six. The procedures for making these changes are discussed below.
4.1.2.1.1 Onroad Exhaust Emissions
MOBILE6.2 performs most of its emission estimation procedures for a non-oxygenated
8.7 RVP gasoline. The effect of differing fuel quality is represented by a set of adjustment
factors, which can vary by vehicle type, model year, and whether the vehicle is properly
operating or not (i.e., is a low or high emitter). Because the mix of vehicle types, model years,
and low and high emitters varies by county and calendar year, it is infeasible to estimate the net
impact of each fuel parameter on emissions outside of the model. In Section 3.1.1.1.1.4 of
Chapter 3, we describe a process whereby we performed linear regressions on the exhaust
emissions estimated by NMEVI in order to determine the average effect of RVP, ethanol content
and MTBE content on exhaust VOC and NOx emissions. Also in Section 3.1.1.1.1.4, we
describe these same impacts using the EPA Predictive Model. We combined these fuel-emission
127
-------
effects with the fuel quality expected to exist in each county under each ethanol use case to
estimate the adjustment which NMIM had applied to exhaust VOC and NOx emissions. This
NMIM adjustment for fuel quality was removed and replaced by one based on the EPA
Predictive Models. In our primary analysis, the fuel-emission effects from the EPA Predictive
Models were only applied to the fraction of exhaust VOC and NOx emissions which are emitted
by Tier 0 vehicles. In our sensitivity analysis, the fuel-emission effects from the EPA Predictive
Models were applied to all exhaust VOC and NOx emissions.
The following equation illustrates conceptually how we made this adjustment.
Adjusted NMIM (1 + EPA Predictive Model Fuel-
Exhaust = Exhaust -H (1 + NMIM Fuel-Emission Effect) x Emission Effect x Tier 0 Emission
Emissions Emissions Percentage)
Table 4.1-2 shows the values for "Tier 0 Fraction"; i.e., the fraction of VOC and NOx
emissions from vehicles with Tier 0 emissions characteristics. Note that the fraction drops as
time progress, reflecting the attrition of such vehicles in the national fleet. In the sensitivity
analysis, the Tier 0 vehicle emission fraction is 1.0 for all years and pollutants.
Table 4.1-2. Fraction of In-Use Exhaust Emissions Attributable to
Vehicles with Tier 0 Emissions Characteristics
Calendar Year
2012
2015
2020
VOC
0.339
0.183
0
NOx
0.162
0.065
0
After adjusting exhaust VOC and NOx according to the methods described above, we
adjusted the four exhaust toxic emissions: benzene, 1,3-butadiene, formaldehyde, and
acetaldehyde. MOBILE6.2 estimates exhaust toxic emissions by first estimating the fraction of
exhaust VOC emissions represented by each toxic based on fuel quality. The model then applies
this fraction to exhaust VOC emissions to estimate absolute emissions of air toxics. Since we
adjusted exhaust VOC emissions, it was necessary to adjust exhaust toxic emissions, as well, by
the ratio of the change in exhaust VOC emissions.
4.1.2.1.2 Onroad Non-Exhaust Emissions
The only adjustment to the non-exhaust emission estimates from NMIM was to add
county-specific estimates of the increase in permeation emissions due to ethanol use. In Section
3.1.1.1.4 of Chapter 3, we determined that a 10 vol% ethanol blend increased permeation
emissions by 0.8 grams per day at 95 F. We also concluded there that permeation emissions
double with every increase in temperature of 18 F. Because of this temperature relationship,
permeation effects were only accounted for in the July emission estimate since emissions during
the winter months could be at least four times lower, and thus negligible.
Permeation emissions occur whether a vehicle is being used or is parked. Therefore, the
average hourly emission factor in each county in July is determined by adjusting the 0.8 gram
128
-------
per day emission rate for the average fuel tank temperature occurring in that hour of the day in
each county in July and multiplying by the market share of E10 fuel in that county. Total
monthly emissions in each county were determined by summing across hours of the day,
multiplying by 31 days and multiplying by the number of vehicles estimated to reside in that
county.
The average fuel tank temperature is a function of the average ambient temperature at
that hour of the day, adjusted to account for the increase in fuel tank temperature for those
vehicles which are operating or which are still cooling down from operating. We obtain
estimates of these latter two factors from EPA's Draft MOVES2006 model. These are shown in
Table 4.1-3. The fuel tank temperature of vehicles which have been parked some time tend to
lag the ambient temperature both when the latter is rising and falling. We assume here that the
fuel tank temperature of these parked vehicles is equal to the ambient temperature, which is true
on average for the day.
Table 4.1-3. Increase in Fuel Tank Temperature Relative to Ambient
Hour of the Day
Midnight
1:00 AM
2:00 AM
3:00 AM
4:00 AM
5:00 AM
6:00 AM
7:00 AM
8:00 AM
9:00 AM
10:00 AM
11:00 AM
Noon
1:00 PM
2:00 PM
3:OOPM
4:00 PM
5:00 PM
6:00 PM
7:00 PM
8:00 PM
9:00 PM
10:00 PM
11:OOPM
Vehicles Operating or in Hot Soak
2.6%
2.8%
1.2%
0.9%
0.8%
2.6%
6.6%
12.3%
14.0%
10.0%
11.1%
12.5%
15.6%
16.0%
17.2%
21.0%
23.7%
28.5%
30.0%
25.7%
18.7%
13.5%
10.6%
7.8%
Average Tank Temperature Rise (F)
10.0
6.9
6.1
4.9
3.1
3.0
3.7
4.6
3.5
3.8
3.8
4.9
4.8
5.5
6.6
7.7
8.6
8.3
8.8
9.2
8.3
7.6
8.0
8.4
129
-------
The total number of gasoline vehicles in the U.S. in 2004 is estimated to be 228
million.72 We increased this figure by 1.9% per yearKK to derive estimates of the gasoline
vehicle fleet in 2012, 2015 and 2020. This produced estimates for the fleet of gasoline vehicles
in the U.S. of 265, 281 and 308 million vehicles in 2012, 2015 and 2020, respectively. These
vehicles were allocated to each county based on the county-specific distribution of national VMT
by gasoline vehicles contained in NMIM.
4.1.2.2 Nonroad Emissions
NMEVI is capable of utilizing any one of a series of EPA's NONROAD emission models.
We chose to use the NONROAD200573 model to estimate emissions from nonroad equipment
here, as it reflects EPA's latest estimates of emission factors for nonroad equipment. EPA has
also recently developed a set of emission factor inputs for the NONROAD model which include
the effect of ethanol on permeation emissions from a number of types of nonroad equipment (see
Chapter 3). However, the model is not able to select these ethanol related emission factors based
on the fuel quality inputs to the model. When run for multiple counties via NMIM, as is being
done here, the permeation emission factors either reflect EO or E10 fuel in all counties. It was
infeasible to run NMIM one county at a time. Therefore, in order to be able to access these
recent estimates of ethanol-related permeation emissions, we developed a methodology which
would include these permeation emissions appropriately while limiting the number of NMIM
runs to a reasonable number. Namely, we ran NMIM for two extreme ethanol use cases and
used these results to estimate emissions for the five ethanol use cases which are the focus of this
proposed rule.
The first case, called "No Oxygen," assumed the market share for ethanol-blend fuel was
zero nationwide. The second case, called "All Oxygen," assumed the market share for ethanol-
blend fuel was 100% nationwide. For both cases, we set the market share of all other oxygenates
to zero. The effects of ethanol use on other fuel properties, such as aromatics and olefins, were
calculated using the same methods described in Chapter 2. The only difference here is that the
changes were more extreme, as ethanol use (in the form of E10) always increased from zero to
100%. A commingling RVP effect of 0.1 psi was included in the 100% Oxygen case in order to
increase the spread of RVP between the two runs and to ensure that we were always interpolating
(and never extrapolating) during our subsequent processing of the results.
This approach is capable of simulating all the relevant effects of fuel quality on nonroad
emissions due to the way these effects are represented in the NONROAD model. Exhaust
emissions are only a function of gasoline oxygen content. Thus, we could use the overall oxygen
content of gasoline projected to be sold in a particular county (which is a function of both the
MTBE and ethanol content) under each of the five ethanol use scenarios to interpolate between
the zero and 3.5 wt% oxygen contents of the two extreme ethanol cases modeled, as described in
the following equation:
KK Annual growth rate in gasoline consumption on an energy basis per EIA Annual Energy Outlook, 2006
(therefore it applies regardless of future ethanol use scenario). Assumes constant annual mileage per vehicle over
this timeframe.
130
-------
Exhaust _ No Oxygen (All Oxygen Emissions - No Oxygen Emissions) x
Emissions Emissions (County Ethanol Fraction / 10)
For nonroad toxic exhaust emissions, the toxic emissions factors for nonroad equipment
are based on very limited data. Therefore, we base the nonroad inventories on ratios derived
from onroad results. We calculated the ratios of onroad ethanol control case emissions to the
reference case for each county. Then, we applied these ratios to the nonroad reference case to
derive the nonroad control case emissions.
Non-exhaust emissions are only a function of RVP and ethanol content. In our two runs,
these two fuel parameters varied together (i.e., were perfectly co-linear), as the county-specific
RVP level in the No Oxygen case was always increased by 1.1 psi in the 100% ethanol case.
Thus, we could use either the county-specific RVP or ethanol content of the gasoline projected to
be sold in a particular county under each of the five ethanol use scenarios to interpolate between
the zero and 3.5 wt% oxygen contents of the two extreme ethanol cases modeled. RVP was used
in order to account for the presence of a commingling RVP effect in some counties and not in
others, as described in the following equation:
= No Oxygen All Oxygen _ No Oxygen (County RVP - No Oxygen RVP)
and Refueling Emissions Emissions Emissions (All Oxygen RVP - No Oxygen RVP)
Emissions
4.1.3 National Emissions Inventory Projections
4.1.3.1 Emission Inventories: Primary Analysis
This section provides the national emissions inventories for the primary case analyses.
Criteria pollutant inventories are included, along with a brief discussion of the trends. A short
discussion of air toxics inventories is also included. See Tables 4A-1 through 4A-7 in the
Chapter 4 Appendix for complete primary-case inventories on air toxics and criteria pollutants,
as well as the percent changes in inventories from the Base case.
Table 4.1-4 shows ethanol impacts on VOC inventories for each of the five cases of
renewable fuel use in years 2012, 2015, and 2020. In any given year, the data suggest that total
VOC emissions will increase as ethanol use increases. The largest increase is seen in the 9.6
Min case, where the increase is less than 2% of the Base inventory.
Our analysis indicates that this increase is a result of VOC non-exhaust emissions, such
as those from evaporation or permeation. While VOC exhaust emissions decrease, they do not
decrease enough to counteract the increase from non-exhaust emissions.
131
-------
Table 4.1-4.
National VOC Emissions from Gasoline Vehicles and Equipment:
Reference Case Inventory and Change in Inventory for Control Cases (Tons/Year)
Primary Case
Total
Reference Inventory
7.2 Min (Change)
7.2 Max (Change)
9.6 Min (Change)
9.6 Max (Change)
Onroad
Reference Inventory
7.2 Min (Change)
7.2 Max (Change)
9.6 Min (Change)
9.6 Max (Change)
Nonroad
Reference Inventory
7.2 Min (Change)
7.2 Max (Change)
9.6 Min (Change)
9.6 Max (Change)
2012
5,837,000
31,000
8,000
57,000
29,000
2012
3,412,000
20,000
12,000
39,000
29,000
2012
2,425,000
10,000
-4,000
18,000
-1,000
2015
5,536,000
33,000
11,000
61,000
34,000
2015
3,270,000
21,000
14,000
40,000
32,000
2015
2,266,000
13,000
-3,000
21,000
2,000
2020
5,316,000
57,000
18,000
91,000
51,000
2020
3,257,000
24,000
18,000
44,000
37,000
2020
2,059,000
33,000
0
47,000
14,000
Table 4.1-5 shows ethanol impacts on CO inventories for each of the five cases of
renewable fuel use in years 2012, 2015, and 2020. In any given year, the data suggest that total
CO emissions will decrease as ethanol use increases. The largest reduction is seen in the 9.6
Max case; this decrease is still less than 4% of the Reference inventory.
132
-------
Table 4.1-5.
National CO Emissions from Gasoline Vehicles and Equipment:
Reference Case Inventory and Change in Inventory for Control Cases (Tons/Year)
Primary Case
Total
Reference Inventory
7.2 Min (Change)
7.2 Max (Change)
9.6 Min (Change)
9.6 Max (Change)
Onroad
Reference Inventory
7.2 Min (Change)
7.2 Max (Change)
9.6 Min (Change)
9.6 Max (Change)
Nonroad
Reference Inventory
7.2 Min (Change)
7.2 Max (Change)
9.6 Min (Change)
9.6 Max (Change)
2012
64,799,000
-843,000
-1,229,000
-1,971,000
-2,319,000
2012
37,671,000
-202,000
-234,000
-381,000
-402,000
2012
27,128,000
-642,000
-995,000
-1,590,000
-1,918,000
2015
64,328,000
-818,000
-1,231,000
-1,953,000
-2,330,000
2015
36,237,000
-173,000
-209,000
-328,000
-354,000
2015
28,090,000
-645,000
-1,021,000
-1,625,000
-1,975,000
2020
64,827,000
-36,000
-1,119,000
-992,000
-1,980,000
2020
35,921,000
-114,000
-167,000
-212,000
-271,000
2020
28,906,000
78,000
-952,000
-780,000
-1,709,000
Table 4.1-6 shows ethanol impacts on NOx inventories for each of the five cases of
renewable fuel use in years 2012, 2015, and 2020. In any given year, the data suggest that total
NOx emissions will increase as ethanol use increases. The largest increase is seen in the 9.6 Min
case, which is around 2% of the Reference inventory.
Our analysis also indicates that nonroad NOx emissions increase much greater than
onroad emissions. While onroad inventories increase less than one percent in Control cases,
nonroad inventories increase upwards of 15% in the 9.6 Min case.
133
-------
Table 4.1-6.
National NOx Emissions from Gasoline Vehicles and Equipment:
Reference Case Inventory and Change in Inventory for Control Cases (Tons/Year)
Primary Case
Total
Reference Inventory
7.2 Min (Change)
7.2 Max (Change)
9.6 Min (Change)
9.6 Max (Change)
Onroad
Reference Inventory
7.2 Min (Change)
7.2 Max (Change)
9.6 Min (Change)
9.6 Max (Change)
Nonroad
Reference Inventory
7.2 Min (Change)
7.2 Max (Change)
9.6 Min (Change)
9.6 Max (Change)
2012
2,576,000
19,000
20,000
40,000
39,000
2012
2,345,000
5,000
4,000
10,000
9,000
2012
231,000
14,000
16,000
30,000
30,000
2015
2,180,000
17,000
18,000
35,000
35,000
2015
1,935,000
2,000
1,000
3,000
3,000
2015
245,000
15,000
17,000
32,000
32,000
2020
1,856,000
21,000
20,000
41,000
38,000
2020
1,594,000
0
0
0
0
2020
262,000
21,000
20,000
41,000
37,000
Table 4.1-7 shows ethanol impacts on air toxic emissions for each of the five cases of
renewable fuel use in 2012. The impacts in 2015 and 2020 are shown in the Appendix to this
chapter.
For all air toxics shown, the most extreme changes occur in the 9.6 Min case. The data
suggest that, in 2012, total benzene emissions will decrease by as much as 6% due to decreases
in both onroad and nonroad emissions. Total formaldehyde emissions increase by up to 2% due
to increases in both onroad and nonroad emissions. Total acetaldehyde emissions increase by as
much as 48% due to increases in both onroad and nonroad emissions. Total 1,3-butadiene
emissions decrease by about 4% due to decreases in both onroad and nonroad emissions.
Generally, the trends in 2015 and 2020 parallel those of 2012. Benzene maintains a drop
of up to about 6% with increased ethanol use. Formaldehyde continues to increase with ethanol
use, by as much as 3%. Acetaldehyde also increases with greater ethanol use, by as much as
roughly 50%. Finally, 1,3-butadiene maintains a decrease in emissions with ethanol use, by as
much as 4%.
Again, we emphasize that the toxics inventories are based on very limited data, especially
when it comes to emissions from nonroad equipment.
134
-------
Table 4.1-7.
National Toxic Emissions from Gasoline Vehicles and Equipment in 2012:
Reference Case Inventory and Change in Inventory for Control (Tons/Year)
Primary Case
Benzene
1,3-Butadiene
Formaldehyde
Acet aldehyde
Total
Reference Inventory
7.2 Min (Change)
7.2 Max (Change)
9.6 Min (Change)
9.6 Max (Change)
177,000
-6,000
-3,000
-11,000
-8,000
18,200
-500
-300
-800
-600
40,200
300
0
800
500
19,800
6,200
5,000
9,600
8,500
Onroad
Reference Inventory
7.2 Min (Change)
7.2 Max (Change)
9.6 Min (Change)
9.6 Max (Change)
124,000
-5,000
-3,000
-8,000
-6,000
11,600
-400
-200
-600
-400
29,900
300
0
600
400
15,500
4,900
3,700
7,400
6,400
Nonroad
Reference Inventory
7.2 Min (Change)
7.2 Max (Change)
9.6 Min (Change)
9.6 Max (Change)
53,000
-1,000
0
-3,000
-2,000
6,700
-200
-200
-400
-300
10,200
200
100
300
200
4,300
1,300
1,300
2,200
2,100
4.1.3.2
Emission Inventories: Sensitivity Analyses
This section provides the national emissions inventories for the sensitivity case analyses.
Criteria pollutant inventories are included, along with a brief discussion of the trends. See
Tables 4A-1 through 4A-7 in the Chapter 4 Appendix for complete sensitivity-case inventories
on air toxics and criteria pollutants, as well as the percent changes in inventories from the
Reference case.
Table 4.1-8 shows ethanol impacts on VOC inventories for each of the five cases of
renewable fuel use in years 2012, 2015, and 2020. Where the primary analysis showed total
VOC emissions increasing with ethanol use in all cases, the sensitivity analysis shows that total
VOC emissions decrease in some cases. Onroad emissions decrease in all cases, while nonroad
trends across the years are mixed.
135
-------
Table 4.1-8.
National VOC Emissions from Gasoline Vehicles and Equipment:
Reference Case Inventory and Change in Inventory for Control Cases (Tons/Year)
Sensitivity Case
Total
Reference Inventory
7.2 Min (Change)
7.2 Max (Change)
9.6 Min (Change)
9.6 Max (Change)
Onroad
Reference Inventory
7.2 Min (Change)
7.2 Max (Change)
9.6 Min (Change)
9.6 Max (Change)
Nonroad
Reference Inventory
7.2 Min (Change)
7.2 Max (Change)
9.6 Min (Change)
9.6 Max (Change)
2012
5,775,000
4,000
-8,000
14,000
-5,000
2012
3,350,000
-6,000
-4,000
-4,000
-4,000
2012
2,425,000
10,000
-4,000
18,000
-1,000
2015
5,459,000
-1,000
-10,000
6,000
-9,000
2015
3,193,000
-13,000
-7,000
-15,000
-11,000
2015
2,266,000
13,000
-3,000
21,000
2,000
2020
5,218,000
10,000
-10,000
17,000
-6,000
2020
3,159,000
-23,000
-11,000
-30,000
-20,000
2020
2,059,000
33,000
0
47,000
14,000
Table 4.1-9 shows ethanol impacts on CO inventories for each of the five cases of
renewable fuel use in years 2012, 2015, and 2020. These figures are the same as those presented
for the primary analysis, since the EPA Predictive Models do not address CO emissions. In any
given year, the data suggest that total CO emissions will decrease as ethanol use increases. The
largest reduction is seen in the 9.6 Max case; this decrease is still less than 4% of the Reference
inventory for total emissions.
136
-------
Table 4.1-9.
National CO Emissions from Gasoline Vehicles and Equipment:
Reference Case Inventory and Change in Inventory for Control Cases (Tons/Year)
Sensitivity Case
Total
Reference Inventory
7.2 Min (Change)
7.2 Max (Change)
9.6 Min (Change)
9.6 Max (Change)
Onroad
Reference Inventory
7.2 Min (Change)
7.2 Max (Change)
9.6 Min (Change)
9.6 Max (Change)
Nonroad
Reference Inventory
7.2 Min (Change)
7.2 Max (Change)
9.6 Min (Change)
9.6 Max (Change)
2012
64,799,000
-843,000
-1,229,000
-1,971,000
-2,319,000
2012
37,671,000
-202,000
-234,000
-381,000
-402,000
2012
27,128,000
-642,000
-995,000
-1,590,000
-1,918,000
2015
64,328,000
-818,000
-1,231,000
-1,953,000
-2,330,000
2015
36,237,000
-173,000
-209,000
-328,000
-354,000
2015
28,090,000
-645,000
-1,021,000
-1,625,000
-1,975,000
2020
64,827,000
-36,000
-1,119,000
-992,000
-1,980,000
2020
35,921,000
-114,000
-167,000
-212,000
-271,000
2020
28,906,000
78,000
-952,000
-780,000
-1,709,000
Table 4.1-10 shows ethanol impacts on NOx inventories for each of the five cases of
renewable fuel use in years 2012, 2015, and 2020. In any given year, the data suggest that total
NOx emissions will increase as ethanol use increases. The largest increase is seen in the 9.6 Min
case, where the increase in total emissions is as high as 4.7% of the Reference inventory.
As in the primary analysis, nonroad NOx emissions increase much greater than onroad
emissions. While onroad inventories increase up to 2.9%, nonroad inventories increase upwards
of 15% in the 9.6 Min case.
137
-------
Table 4.1-10.
National NOx Emissions from Gasoline Vehicles and Equipment:
Reference Case Inventory and Change in Inventory for Control Cases (Tons/Year)
Sensitivity Case
Total
Reference Inventory
7.2 Min (Change)
7.2 Max (Change)
9.6 Min (Change)
9.6 Max (Change)
Onroad
Reference Inventory
7.2 Min (Change)
7.2 Max (Change)
9.6 Min (Change)
9.6 Max (Change)
Nonroad
Reference Inventory
7.2 Min (Change)
7.2 Max (Change)
9.6 Min (Change)
9.6 Max (Change)
2012
2,610,000
49,000
45,000
95,000
89,000
2012
2,379,000
35,000
29,000
65,000
59,000
2012
231,000
14,000
16,000
30,000
30,000
2015
2,211,000
44,000
40,000
87,000
81,000
2015
1,966,000
29,000
24,000
55,000
49,000
2015
245,000
15,000
17,000
32,000
32,000
2020
1,883,000
46,000
40,000
88,000
79,000
2020
1,621,000
25,000
20,000
47,000
42,000
2020
262,000
21,000
20,000
41,000
37,000
Table 4.1-11 shows ethanol impacts on air toxic emissions for each of the five cases of
renewable fuel use in 2012. The impacts in 2015 and 2020 are shown in the Appendix to this
chapter.
138
-------
Table 4.1-11.
National Toxic Emissions from Gasoline Vehicles and Equipment in 2012:
Reference Case Inventory and Change in Inventory for Control Cases (Tons/Year)
Sensitivity Case
Benzene
1,3-Butadiene
Formaldehyde
Acet aldehyde
Total
Reference Inventory
7.2 Min (Change)
7.2 Max (Change)
9.6 Min (Change)
9.6 Max (Change)
175,000
-9,000
-5,000
-14,000
-10,000
17,900
-600
-400
-1,100
-800
39,300
0
-200
300
0
19,200
5,800
4,700
9,000
8,000
Onroad
Reference Inventory
7.2 Min (Change)
7.2 Max (Change)
9.6 Min (Change)
9.6 Max (Change)
121,000
-6,000
-3,000
-10,000
-7,000
11,300
-500
-300
-800
-600
29,100
-100
-200
100
-100
14,900
4,500
3,400
6,800
5,900
Nonroad
Reference Inventory
7.2 Min (Change)
7.2 Max (Change)
9.6 Min (Change)
9.6 Max (Change)
53,000
-1,000
0
-3,000
-2,000
6,700
-200
-200
-400
-300
10,200
200
100
300
200
4,300
1,300
1,300
2,200
2,100
As in the primary analysis, the most extreme changes in the sensitivity analysis tend to
occur in the 9.6 Min case. The data suggest that, in 2012, total benzene emissions will decrease
by as much as 8% due to decreases in both onroad and nonroad emissions.LL Total
formaldehyde emissions may either increase or decrease, but the either way the change is less
than 1%. Nonroad formaldehyde emissions tend to increase, while onroad emissions tend to
decrease, except for the 9.6 Min case. Total acetaldehyde emissions increase by as much as 47%
due to increases in both onroad and nonroad emissions. Total 1,3-butadiene emissions decrease
by about 6% due to decreases in both onroad and nonroad emissions.
4.1.3.3
Local and Regional VOC and NOx Emissions in July 2012
We also estimate the percentage change in July 2012 for VOC and NOx emissions from
gasoline fueled motor vehicles and equipment in those areas which actually experienced a
significant change in ethanol use. Specifically, we focused on areas where the market share of
ethanol blends was projected to change by 50 percent or more. We also focused on summertime
emissions, as these are most relevant to ozone formation as discussed in Chapter 5. Finally, we
developed separately estimates for: 1) RFG areas, including the state of California and the
portions of Arizona where their CBG fuel programs apply, 2) low RVP areas (i.e., RVP
standards less than 9.0 RVP, and 3) areas with a 9.0 RVP standard. This set of groupings helps
LL Just prior to publication of the NPRM, we discovered an error in the MOBILE6 algorithms for estimating the
impact of oxygenate use on non-benzene emissions. This error led to an over-estimation of the reduction in non-
exhaust benzene emissions due to increased ethanol use. We will fix this error for the FRM analysis. We believe
that the size of the error is small (i.e., the 8% reduction in benzene emissions may drop to 7% with the correction).
139
-------
to highlight the emissions impact of increased ethanol use in those areas where emission control
is most important.
Table 4.1-12 presents our primary analysis estimates of the percentage change in VOC
and NOx emission inventories for these three types of areas when compared to the 2012
reference case. While ethanol use is going up in the vast majority of the nation, ethanol use in
RFG areas under the "Minimum Use in RFG" cases is actually decreasing compared to the 2012
reference case. This is important to note in order to understand the changes in emissions
indicated.
Table 4.1-12.
Change in Emissions from Gasoline Vehicles and Equipment in Counties Where Ethanol
Use Changed Significantly, July 2012, Primary Analysis
Ethanol Use
Ethanol Use in RFG
7.2 Billion Gallons
Minimum
Maximum
9.6 Billion Gallons
Minimum
Maximum
RFG Areas
Ethanol Use
VOC
NOx
Down
1.6%
-5.2%
Up
0.4%
2.4%
Down
1.6%
-5.2%
Up
0.4%
2.4%
Low RVP Areas
Ethanol Use
VOC
NOx
Up
3.1%
4.1%
Up
3.2%
6.0%
Up
4.1%
4.8%
Up
3.5%
4.4%
Other Areas
Ethanol Use
VOC
NOx
Up
4.1%
4.6%
Up
4.1%
6.0%
Up
5.4%
5.8%
Up
4.4%
4.8%
As expected, increased ethanol use tends to increase NOx emissions. The increase in low
RVP and other areas is greater than in RFG areas, since the RFG in the RFG areas included in
this analysis all contained MTBE. Also, increased ethanol use tends to increase VOC emissions,
indicating that the increase in non-exhaust VOC emissions exceeds the reduction in exhaust
VOC emissions. This effect is muted with RFG due to the absence of an RVP waiver for ethanol
blends. See Chapter 2 for a discussion of how ethanol levels will change at the state-level.
Table 4.1-13 presents the percentage change in VOC and NOx emission inventories
under our sensitivity analysis (i.e., when we apply the emission effects of the EPA Predictive
Models to all motor vehicles).
140
-------
Table 4.1-13.
Change in Emissions from Gasoline Vehicles and Equipment in Counties Where Ethanol
Use Changed Significantly, July 2012, Sensitivity Analysis
7.2 Bgal Min
7.2 Bgal Max
9.6 Bgal Min
9.6 Bgal Max
RFG Areas
Ethanol Use
voc
NOx
Down
2.6%
-9.0%
Up
0.2%
4.7%
Down
2.6%
-9.0%
Up
0.2%
4.7%
Low RVP Areas
Ethanol Use
VOC
NOx
Up
2.1%
8.2%
Up
2.1%
10.6%
Up
3.1%
9.8%
Up
2.5%
8.9%
Other Areas
Ethanol Use
VOC
NOx
Up
3.4%
8.4%
Up
3.4%
10.1%
Up
4.6%
10.3%
Up
3.7%
8.8%
Directionally, the changes in VOC and NOx emissions in the various areas are consistent
with those from our primary analysis. The main difference is that the increases in VOC
emissions are smaller, due to more vehicles experiencing a reduction in exhaust VOC emissions,
and the increases in NOx emissions are larger.
4.2 Impact of Biodiesel Use
Increased biodiesel use is expected to have a small effect on diesel emissions. As
discussed in Chapter 1, biodiesel use totaled 25 million gallons in 2004 and is projected to
increase to 300 million gallons in 2012. As the vast majority of the limited emission data on
biodiesel use was obtained from onroad engines and vehicles, we assume here that all biodiesel
fuel is used in onroad vehicles. This is unlikely to be the case, as farmers in particular, seem
likely to use some in their agricultural equipment. However, given the lack of data with which to
support a projected emissions impact on nonroad diesel emissions, it is more consistent with the
emissions data to assume biodiesel fuel will be used in onroad vehicles.
Total diesel fuel use in onroad diesels in 2004 was roughly 39.4 billion gallons and is
expected to grow to 47.5 billion gallons per year by 2012.MM The volumes of biodiesel
produced represent 0.06% and 0.6% of onroad diesel fuel consumption in 2004 and 2012,
respectively. In Chapter 3, we presented the emission impacts for a 20 percent biodiesel blend
(B20). In terms of B20, these biodiesel volumes represent 0.3% and 3.2% of onroad diesel fuel
consumption in 2004 and 2012, respectively.
MM Based on linear interpolation between estimate for 2001 from Table 7.1.2-1 and that for 2014 from Table 7.1.3-
4, both from the 2010 Nonroad FRM Final RIA, EPA420-R-04-007, May 2004, available in EPA Docket OAR-
2003-0012.
141
-------
We based the impact of biodiesel use on the CAIR emission inventories for onroad
diesels. CAIR inventories are not available for 2012, only for 2015 and 2020. Therefore, we
adjusted the 2015 inventories using year-by-year inventories developed for the EPA FRM
establishing new PM and NOx emission standards for 2007 and later model year heavy-duty
diesels.NN This analysis did not address CO emissions, so we assumed that CO emissions were
changing in the same way as VOC emissions. Table 4.2-1 shows the expected emission
reductions associated with the increase in biodiesel use.
Table 4.2-1.
Annual Emissions Nationwide from Onroad Diesels in 2012
VOC
NOx
CO
Fine PM
2015 CAIR
Inventory
(tons per
year)
128,000
1,033,000
336,000
25,000
Ratio of 20 12
to 2015
Emissions
1.049
1.385
1.049
1.182
20 12 Reference
Case Inventory
(tons per year)
135,000
1,430,000
353,000
27,000
Change in
Inventory Due to
Biodiesel (tons
per year)
-800
800
-1,100
-100
As can be seen, due to the low volume of biodiesel fuel use, the emission effects are very
small, essentially 1000 tons per year or less for any of the four pollutants.
4.3 Impact of Renewable Fuel Production and Distribution
4.3.1 Ethanol
In Chapter 2, we estimated that 3.5 billion gallons of ethanol was produced for use in
motor fuel in 2004. Based on growth in overall gasoline demand between 2004 and 2012,°° this
would represent 3.9 billion gallons of ethanol in 2012. Here, we estimate the increase in
emissions which will occur with an increase in ethanol production and distribution from 3.9
billion gallons to either 7.2 or 9.6 billion gallons per year.
We describe the emissions associated with producing and distributing ethanol on a per
gallon basis in Chapter 3.2.1, where the emissions factors were obtained from DOE's GREET
model, version 1.6. Here, we simply multiply those emission factors by the volume of ethanol
being used in each scenario. Table 4.3-1 shows estimates of annual emissions expected to occur
nationwide due to increased production of ethanol.
m From Figures II-B-8 thru 10 in the 2007 Onroad Diesel FRM Final RIA, EPA420-R-00-026, December 2000,
available in Public Docket No. A-99-06.
00 EIA projects gasoline demand of 16.93 and 18.84 quadrillion Btu in 2004 and 2012, respectively. This
represents overall growth between these two years of 11.3%.
142
-------
Table 4.3-1.
Annual Emissions Nationwide from Ethanol Production and Transportation: 2012
(tons per year)
voc
NOx
CO
PM10
SOx
Reference
Inventory
16,000
19,000
47,000
26,000
31,000
Increase in Emissions
7.2 Billion gallons of ethanol
13,000
16,000
39,000
23,000
26,000
9.6 Billion gallons of ethanol
22,000
28,000
68,000
39,000
45,000
As can be seen, the potential increases in emissions from ethanol production and
transportation are of the same order of magnitude as those from ethanol use, with the exception
of CO emissions. Generally, ethanol plants are not located in ozone non-attainment areas, so the
ozone impact of the increased VOC and NOx emissions should be minimal.
4.3.2 Biodiesel
In Chapter 1, we estimated that 25 million gallons of biodiesel were produced for use in
motor fuel in 2004. Based on growth in overall diesel fuel demand between 2004 and 2012,pp
this would represent the equivalent of 28 million gallons of biodiesel in 2012 for our reference
case. Here, we estimate the increase in emissions which will occur with an increase in biodiesel
production and distribution from 28 million gallons to 300 million gallons per year.
We describe the emissions associated with producing and distributing biodiesel on a per
gallon basis in Chapter 3. Here, we simply multiply those emission factors by the volume of
biodiesel being used in each scenario. Table 4.3-2 shows estimates of annual emissions expected
to occur nationwide due to increased production of biodiesel.
Table 4.3-2.
Annual Emissions Nationwide from Biodiesel Production and Transportation: 2012
(tons per year)
VOC
NOx
CO
PM10
SOx
Reference Inventory:
28 mill gal biodiesel per year
1,300
1,400
800
50
200
Increase in Emissions:
300 mill gal biodiesel per year
12,700
13,600
7,200
1,000
1,800
As can be seen, the potential increases in emissions from biodiesel production and
transportation are of the same order of magnitude as those from biodiesel use, with the exception
pp EIA projects gasoline demand of 16.93 and 18.84 quadrillion Btu in 2004 and 2012, respectively. This represents
overall growth between these two years of 11.3%. Source: Annual Energy Outlook 2006, DOE/EIA-0383(2006),
Reference Case Table 2, available in docket EPA-HQ-OAR-2005-0161.
143
-------
of CO emissions. Generally, biodiesel plants are not located in ozone non-attainment areas, so
the ozone impact of the increased VOC and NOx emissions should be minimal.
144
-------
Chapter 4: Appendix
145
-------
Table 4A-1. VOC Emission Inventories under Various Ethanol Use Cases
Tons/Year
Change from Reference
%Change from Reference
PRIMARY CASE
Total
Reference
7.2 Min
7.2 Max
9.6 Min
9.6 Max
Onroad
Reference
7.2 Min
7.2 Max
9.6 Min
9.6 Max
Nonroad
Reference
7.2 Min
7.2 Max
9.6 Min
9.6 Max
2012
5,837,000
5,868,000
5,845,000
5,894,000
5,866,000
2012
3,412,000
3,432,000
3,424,000
3,451,000
3,441,000
2012
2,425,000
2,435,000
2,421,000
2,443,000
2,424,000
2015
5,536,000
5,569,000
5,547,000
5,597,000
5,570,000
2015
3,270,000
3,291,000
3,284,000
3,310,000
3,302,000
2015
2,266,000
2,279,000
2,263,000
2,287,000
2,268,000
2020
5,316,000
5,373,000
5,334,000
5,407,000
5,367,000
2020
3,257,000
3,281,000
3,275,000
3,301,000
3,294,000
2020
2,059,000
2,092,000
2,059,000
2,106,000
2,073,000
2012
~
31,000
8,000
57,000
29,000
2012
-
20,000
12,000
39,000
29,000
2012
~
10,000
-4,000
18,000
-1,000
2015
~
33,000
11,000
61,000
34,000
2015
~
21,000
14,000
40,000
32,000
2015
~
13,000
-3,000
21,000
2,000
2020
~
57,000
18,000
91,000
51,000
2020
~
24,000
18,000
44,000
37,000
2020
~
33,000
0
47,000
14,000
2012
~
0.5%
0.1%
1.0%
0.5%
2012
~
0.6%
0.4%
1.1%
0.8%
2012
—
0.4%
-0.2%
0.7%
0.0%
2015
~
0.6%
0.2%
1.1%
0.6%
2015
~
0.6%
0.4%
1.2%
1.0%
2015
—
0.6%
-0.1%
0.9%
0.1%
2020
~
1.1%
0.3%
1.7%
1.0%
2020
~
0.7%
0.6%
1.4%
1.1%
2020
—
1.6%
0.0%
2.3%
0.7%
SENSITIVITY CASE
Total
Reference
7.2 Min
7.2 Max
9.6 Min
9.6 Max
Onroad
Reference
7.2 Min
7.2 Max
9.6 Min
9.6 Max
Nonroad
Reference
7.2 Min
7.2 Max
9.6 Min
9.6 Max
2012
5,775,000
5,779,000
5,767,000
5,789,000
5,770,000
2012
3,350,000
3,344,000
3,346,000
3,346,000
3,346,000
2012
2,425,000
2,435,000
2,421,000
2,443,000
2,424,000
2015
5,459,000
5,458,000
5,449,000
5,465,000
5,450,000
2015
3,193,000
3,180,000
3,186,000
3,178,000
3,182,000
2015
2,266,000
2,279,000
2,263,000
2,287,000
2,268,000
2020
5,218,000
5,228,000
5,208,000
5,235,000
5,212,000
2020
3,159,000
3,136,000
3,148,000
3,129,000
3,139,000
2020
2,059,000
2,092,000
2,059,000
2,106,000
2,073,000
2012
~
4,000
-8,000
14,000
-5,000
2012
~
-6,000
-4,000
-4,000
-4,000
2012
~
10,000
-4,000
18,000
-1,000
2015
~
-1,000
-10,000
6,000
-9,000
2015
~
-13,000
-7,000
-15,000
-11,000
2015
~
13,000
-3,000
21,000
2,000
2020
~
10,000
-10,000
17,000
-6,000
2020
~
-23,000
-11,000
-30,000
-20,000
2020
~
33,000
0
47,000
14,000
2012
~
0.1%
-0.1%
0.2%
-0.1%
2012
~
-0.2%
-0.1%
-0.1%
-0.1%
2012
—
0.4%
-0.2%
0.7%
0.0%
2015
~
0.0%
-0.2%
0.1%
-0.2%
2015
~
-0.4%
-0.2%
-0.5%
-0.3%
2015
—
0.6%
-0.1%
0.9%
0.1%
2020
~
0.2%
-0.2%
0.3%
-0.1%
2020
~
-0.7%
-0.3%
-0.9%
-0.6%
2020
—
1.6%
0.0%
2.3%
0.7%
146
-------
Table 4A-2. CO Emission Inventories under Various Ethanol Use Cases
Tons/Year
Change from Reference
%Change from Reference
PRIMARY CASE
Total
Reference
7.2 Min
7.2 Max
9.6 Min
9.6 Max
Onroad
Reference
7.2 Min
7.2 Max
9.6 Min
9.6 Max
Nonroad
Reference
7.2 Min
7.2 Max
9.6 Min
9.6 Max
2012
64,799,000
63,956,000
63,570,000
62,828,000
62,480,000
2012
37,671,000
37,469,000
37,437,000
37,290,000
37,269,000
2012
27,128,000
26,486,000
26,133,000
25,538,000
25,210,000
2015
64,328,000
63,510,000
63,097,000
62,375,000
61,998,000
2015
36,237,000
36,064,000
36,028,000
35,909,000
35,883,000
2015
28,090,000
27,445,000
27,069,000
26,465,000
26,115,000
2020
64,827,000
64,791,000
63,708,000
63,835,000
62,847,000
2020
35,921,000
35,807,000
35,754,000
35,709,000
35,650,000
2020
28,906,000
28,984,000
27,954,000
28,126,000
27,197,000
2012
-
-843,000
-1,229,000
-1,971,000
-2,319,000
2012
-
-202,000
-234,000
-381,000
-402,000
2012
-
-642,000
-995,000
-1,590,000
-1,918,000
2015
~
-818,000
-1,231,000
-1,953,000
-2,330,000
2015
~
-173,000
-209,000
-328,000
-354,000
2015
~
-645,000
-1,021,000
-1,625,000
-1,975,000
2020
~
-36,000
-1,119,000
-992,000
-1,980,000
2020
~
-114,000
-167,000
-212,000
-271,000
2020
~
78,000
-952,000
-780,000
-1,709,000
2012
~
-1.3%
-1.9%
-3.0%
-3.6%
2012
~
-0.5%
-0.6%
-1.0%
-1.1%
2012
—
-2.4%
-3.7%
-5.9%
-7.1%
2015
~
-1.3%
-1.9%
-3.0%
-3.6%
2015
~
-0.5%
-0.6%
-0.9%
-1.0%
2015
—
-2.3%
-3.6%
-5.8%
-7.0%
2020
~
-0.1%
-1.7%
-1.5%
-3.1%
2020
~
-0.3%
-0.5%
-0.6%
-0.8%
2020
—
0.3%
-3.3%
-2.7%
-5.9%
SENSITIVITY CASE
Total
Reference
7.2 Min
7.2 Max
9.6 Min
9.6 Max
Onroad
Reference
7.2 Min
7.2 Max
9.6 Min
9.6 Max
Nonroad
Reference
7.2 Min
7.2 Max
9.6 Min
9.6 Max
2012
64,799,000
63,956,000
63,570,000
62,828,000
62,480,000
2012
37,671,000
37,469,000
37,437,000
37,290,000
37,269,000
2012
27,128,000
26,486,000
26,133,000
25,538,000
25,210,000
2015
64,328,000
63,510,000
63,097,000
62,375,000
61,998,000
2015
36,237,000
36,064,000
36,028,000
35,909,000
35,883,000
2015
28,090,000
27,445,000
27,069,000
26,465,000
26,115,000
2020
64,827,000
64,791,000
63,708,000
63,835,000
62,847,000
2020
35,921,000
35,807,000
35,754,000
35,709,000
35,650,000
2020
28,906,000
28,984,000
27,954,000
28,126,000
27,197,000
2012
-
-843,000
-1,229,000
-1,971,000
-2,319,000
2012
-
-202,000
-234,000
-381,000
-402,000
2012
-
-642,000
-995,000
-1,590,000
-1,918,000
2015
~
-818,000
-1,231,000
-1,953,000
-2,330,000
2015
~
-173,000
-209,000
-328,000
-354,000
2015
~
-645,000
-1,021,000
-1,625,000
-1,975,000
2020
~
-36,000
-1,119,000
-992,000
-1,980,000
2020
~
-114,000
-167,000
-212,000
-271,000
2020
~
78,000
-952,000
-780,000
-1,709,000
2012
~
-1.3%
-1.9%
-3.0%
-3.6%
2012
~
-0.5%
-0.6%
-1.0%
-1.1%
2012
—
-2.4%
-3.7%
-5.9%
-7.1%
2015
~
-1.3%
-1.9%
-3.0%
-3.6%
2015
~
-0.5%
-0.6%
-0.9%
-1.0%
2015
—
-2.3%
-3.6%
-5.8%
-7.0%
2020
~
-0.1%
-1.7%
-1.5%
-3.1%
2020
~
-0.3%
-0.5%
-0.6%
-0.8%
2020
—
0.3%
-3.3%
-2.7%
-5.9%
147
-------
Table 4A-3. NOx Emission Inventories under Various Ethanol Use Cases
Tons/Year
Change from Reference
%Change from Reference
PRIMARY CASE
Total
Reference
7.2 Min
7.2 Max
9.6 Min
9.6 Max
Onroad
Reference
7.2 Min
7.2 Max
9.6 Min
9.6 Max
Nonroad
Reference
7.2 Min
7.2 Max
9.6 Min
9.6 Max
2012
2,576,000
2,595,000
2,596,000
2,616,000
2,615,000
2012
2,345,000
2,350,000
2,349,000
2,355,000
2,354,000
2012
231,000
245,000
247,000
261,000
261,000
2015
2,180,000
2,197,000
2,198,000
2,215,000
2,215,000
2015
1,935,000
1,937,000
1,936,000
1,938,000
1,938,000
2015
245,000
260,000
262,000
277,000
277,000
2020
1,856,000
1,877,000
1,876,000
1,897,000
1,894,000
2020
1,594,000
1,594,000
1,594,000
1,594,000
1,594,000
2020
262,000
283,000
282,000
303,000
299,000
2012
—
19,000
20,000
40,000
39,000
2012
~
5,000
4,000
10,000
9,000
2012
—
14,000
16,000
30,000
30,000
2015
—
17,000
18,000
35,000
35,000
2015
~
2,000
1,000
3,000
3,000
2015
—
15,000
17,000
32,000
32,000
2020
—
21,000
20,000
41,000
38,000
2020
~
0
0
0
0
2020
—
21,000
20,000
41,000
37,000
2012
—
0.7%
0.8%
1.6%
1.5%
2012
~
0.2%
0.2%
0.4%
0.4%
2012
—
6.1%
6.9%
13.0%
13.0%
2015
—
0.8%
0.8%
1.6%
1.6%
2015
~
0.1%
0.1%
0.2%
0.2%
2015
—
6.1%
6.9%
13.1%
13.1%
2020
—
1.1%
1.1%
2.2%
2.0%
2020
~
0.0%
0.0%
0.0%
0.0%
2020
—
8.0%
7.6%
15.6%
14.1%
SENSITIVITY CASE
Total
Reference
7.2 Min
7.2 Max
9.6 Min
9.6 Max
Onroad
Reference
7.2 Min
7.2 Max
9.6 Min
9.6 Max
Nonroad
Reference
7.2 Min
7.2 Max
9.6 Min
9.6 Max
2012
2,610,000
2,659,000
2,655,000
2,705,000
2,699,000
2012
2,379,000
2,414,000
2,408,000
2,444,000
2,438,000
2012
231,000
245,000
247,000
261,000
261,000
2015
2,211,000
2,255,000
2,251,000
2,298,000
2,292,000
2015
1,966,000
1,995,000
1,990,000
2,021,000
2,015,000
2015
245,000
260,000
262,000
277,000
277,000
2020
1,883,000
1,929,000
1,923,000
1,971,000
1,962,000
2020
1,621,000
1,646,000
1,641,000
1,668,000
1,663,000
2020
262,000
283,000
282,000
303,000
299,000
2012
—
49,000
45,000
95,000
89,000
2012
~
35,000
29,000
65,000
59,000
2012
—
14,000
16,000
30,000
30,000
2015
—
44,000
40,000
87,000
81,000
2015
~
29,000
24,000
55,000
49,000
2015
—
15,000
17,000
32,000
32,000
2020
—
46,000
40,000
88,000
79,000
2020
~
25,000
20,000
47,000
42,000
2020
—
21,000
20,000
41,000
37,000
2012
—
1.9%
1.7%
3.6%
3.4%
2012
~
1.5%
1.2%
2.7%
2.5%
2012
—
6.1%
6.9%
13.0%
13.0%
2015
—
2.0%
1.8%
3.9%
3.7%
2015
~
1.5%
1.2%
2.8%
2.5%
2015
—
6.1%
6.9%
13.1%
13.1%
2020
—
2.4%
2.1%
4.7%
4.2%
2020
~
1.5%
1.2%
2.9%
2.6%
2020
—
8.0%
7.6%
15.6%
14.1%
148
-------
Table 4A-4. Benzene Emission Inventories under Various Ethanol Use Cases
Tons/Year
Change from Reference
%Change from Reference
PRIMARY CASE
Total
Reference
7.2 Min
7.2 Max
9.6 Min
9.6 Max
Onroad
Reference
7.2 Min
7.2 Max
9.6 Min
9.6 Max
Nonroad
Reference
7.2 Min
7.2 Max
9.6 Min
9.6 Max
2012
177,000
171,000
174,000
166,000
169,000
2012
124,000
119,000
121,000
116,000
118,000
2012
53,000
52,000
53,000
50,000
51,000
2015
175,000
169,000
172,000
165,000
167,000
2015
124,000
119,000
122,000
116,000
119,000
2015
51,000
49,000
50,000
48,000
49,000
2020
180,000
173,000
176,000
169,000
172,000
2020
132,000
126,000
129,000
123,000
126,000
2020
48,000
47,000
48,000
46,000
46,000
2012
—
-6,000
-3,000
-11,000
-8,000
2012
~
-5,000
-3,000
-8,000
-6,000
2012
—
-1,000
0
-3,000
-2,000
2015
—
-6,000
-3,000
-10,000
-8,000
2015
~
-5,000
-2,000
-8,000
-5,000
2015
—
-2,000
-1,000
-3,000
-2,000
2020
—
-7,000
-4,000
-11,000
-8,000
2020
~
-6,000
-3,000
-9,000
-6,000
2020
—
-1,000
0
-2,000
-2,000
2012
—
-3.4%
-1.7%
-6.2%
-4.5%
2012
~
-4.0%
-2.4%
-6.5%
-4.8%
2012
—
-1.9%
0.0%
-5.7%
-3.8%
2015
—
-3.4%
-1.7%
-5.7%
-4.6%
2015
~
-4.0%
-1.6%
-6.5%
-4.0%
2015
—
-3.9%
-2.0%
-5.9%
-3.9%
2020
—
-3.9%
-2.2%
-6.1%
-4.4%
2020
~
-4.5%
-2.3%
-6.8%
-4.5%
2020
—
-2.1%
0.0%
-4.2%
-4.2%
SENSITIVITY CASE
Total
Reference
7.2 Min
7.2 Max
9.6 Min
9.6 Max
Onroad
Reference
7.2 Min
7.2 Max
9.6 Min
9.6 Max
Nonroad
Reference
7.2 Min
7.2 Max
9.6 Min
9.6 Max
2012
175,000
166,000
170,000
161,000
165,000
2012
121,000
115,000
118,000
111,000
114,000
2012
53,000
52,000
53,000
50,000
51,000
2015
172,000
164,000
168,000
159,000
162,000
2015
121,000
114,000
117,000
110,000
113,000
2015
51,000
49,000
50,000
48,000
49,000
2020
175,000
167,000
171,000
161,000
165,000
2020
127,000
120,000
123,000
116,000
119,000
2020
48,000
47,000
48,000
46,000
46,000
2012
—
-9,000
-5,000
-14,000
-10,000
2012
~
-6,000
-3,000
-10,000
-7,000
2012
—
-1,000
0
-3,000
-2,000
2015
—
-8,000
-4,000
-13,000
-10,000
2015
~
-7,000
-4,000
-11,000
-8,000
2015
—
-2,000
-1,000
-3,000
-2,000
2020
—
-8,000
-4,000
-14,000
-10,000
2020
~
-7,000
-4,000
-11,000
-8,000
2020
—
-1,000
0
-2,000
-2,000
2012
—
-5.1%
-2.9%
-8.0%
-5.7%
2012
~
-5.0%
-2.5%
-8.3%
-5.8%
2012
—
-1.9%
0.0%
-5.7%
-3.8%
2015
—
-4.7%
-2.3%
-7.6%
-5.8%
2015
~
-5.8%
-3.3%
-9.1%
-6.6%
2015
—
-3.9%
-2.0%
-5.9%
-3.9%
2020
—
-4.6%
-2.3%
-8.0%
-5.7%
2020
~
-5.5%
-3.1%
-8.7%
-6.3%
2020
—
-2.1%
0.0%
-4.2%
-4.2%
149
-------
Table 4A-5. Acetaldehyde Emission Inventories under Various Ethanol Use Cases
Tons/Year
Change from Reference
%Change from Reference
PRIMARY CASE
Total
Reference
7.2 Min
7.2 Max
9.6 Min
9.6 Max
Onroad
Reference
7.2 Min
7.2 Max
9.6 Min
9.6 Max
Nonroad
Reference
7.2 Min
7.2 Max
9.6 Min
9.6 Max
2012
19,800
26,000
24,800
29,400
28,300
2012
15,500
20,400
19,200
22,900
21,900
2012
4,300
5,600
5,600
6,500
6,400
2015
19,900
26,300
25,100
29,700
28,500
2015
15,800
21,000
19,700
23,600
22,400
2015
4,100
5,300
5,400
6,100
6,100
2020
20,900
27,900
26,400
31,500
30,100
2020
17,000
22,900
21,400
25,700
24,400
2020
3,900
5,000
5,000
5,800
5,700
2012
—
6,200
5,000
9,600
8,500
2012
~
4,900
3,700
7,400
6,400
2012
—
1,300
1,300
2,200
2,100
2015
—
6,400
5,200
9,800
8,600
2015
~
5,200
3,900
7,800
6,600
2015
—
1,200
1,300
2,000
2,000
2020
—
7,000
5,500
10,600
9,200
2020
~
5,900
4,400
8,700
7,400
2020
—
1,100
1,100
1,900
1,800
2012
—
31.3%
25.3%
48.5%
42.9%
2012
~
31.6%
23.9%
47.7%
41.3%
2012
—
30.2%
30.2%
51.2%
48.8%
2015
—
32.2%
26.1%
49.2%
43.2%
2015
~
32.9%
24.7%
49.4%
41.8%
2015
—
29.3%
31.7%
48.8%
48.8%
2020
—
33.5%
26.3%
50.7%
44.0%
2020
~
34.7%
25.9%
51.2%
43.5%
2020
—
28.2%
28.2%
48.7%
46.2%
SENSITIVITY CASE
Total
Reference
7.2 Min
7.2 Max
9.6 Min
9.6 Max
Onroad
Reference
7.2 Min
7.2 Max
9.6 Min
9.6 Max
Nonroad
Reference
7.2 Min
7.2 Max
9.6 Min
9.6 Max
2012
19,200
25,000
23,900
28,200
27,200
2012
14,900
19,400
18,300
21,700
20,800
2012
4,300
5,600
5,600
6,500
6,400
2015
19,200
25,100
23,900
28,300
27,100
2015
15,100
19,800
18,600
22,100
21,100
2015
4,100
5,300
5,400
6,100
6,100
2020
20,000
26,300
25,000
29,600
28,300
2020
16,200
21,400
19,900
23,800
22,600
2020
3,900
5,000
5,000
5,800
5,700
2012
—
5,800
4,700
9,000
8,000
2012
~
4,500
3,400
6,800
5,900
2012
—
1,300
1,300
2,200
2,100
2015
—
5,900
4,700
9,100
7,900
2015
~
4,700
3,500
7,000
6,000
2015
—
1,200
1,300
2,000
2,000
2020
—
6,300
5,000
9,600
8,300
2020
~
5,200
3,700
7,600
6,400
2020
—
1,100
1,100
1,900
1,800
2012
—
30.2%
24.5%
46.9%
41.7%
2012
~
30.2%
22.8%
45.6%
39.6%
2012
—
30.2%
30.2%
51.2%
48.8%
2015
—
30.7%
24.5%
47.4%
41.1%
2015
~
31.1%
23.2%
46.4%
39.7%
2015
—
29.3%
31.7%
48.8%
48.8%
2020
—
31.5%
25.0%
48.0%
41.5%
2020
~
32.1%
22.8%
46.9%
39.5%
2020
—
28.2%
28.2%
48.7%
46.2%
150
-------
Table 4A-6. Formaldehyde Emission Inventories under Various Ethanol Use Cases
Tons/Year
Change from Reference
%Change from Reference
PRIMARY CASE
Total
Reference
7.2 Min
7.2 Max
9.6 Min
9.6 Max
Onroad
Reference
7.2 Min
7.2 Max
9.6 Min
9.6 Max
Nonroad
Reference
7.2 Min
7.2 Max
9.6 Min
9.6 Max
2012
40,200
40,500
40,200
41,000
40,700
2012
29,900
30,200
29,900
30,500
30,300
2012
10,200
10,400
10,300
10,500
10,400
2015
39,900
40,400
40,000
41,000
40,600
2015
30,200
30,600
30,300
31,000
30,700
2015
9,700
9,800
9,800
10,000
9,900
2020
41,300
41,900
41,400
42,600
42,100
2020
32,300
32,800
32,400
33,300
32,900
2020
9,000
9,100
9,000
9,300
9,200
2012
—
300
0
800
500
2012
~
300
0
600
400
2012
—
200
100
300
200
2015
—
500
100
1,100
700
2015
~
400
100
800
500
2015
—
100
100
300
200
2020
—
600
100
1,300
800
2020
~
500
100
1,000
600
2020
—
100
0
300
200
2012
—
0.7%
0.0%
2.0%
1.2%
2012
~
1.0%
0.0%
2.0%
1.3%
2012
—
2.0%
1.0%
2.9%
2.0%
2015
—
1.3%
0.3%
2.8%
1.8%
2015
~
1.3%
0.3%
2.6%
1.7%
2015
—
1.0%
1.0%
3.1%
2.1%
2020
—
1.5%
0.2%
3.1%
1.9%
2020
~
1.5%
0.3%
3.1%
1.9%
2020
—
1.1%
0.0%
3.3%
2.2%
SENSITIVITY CASE
Total
Reference
7.2 Min
7.2 Max
9.6 Min
9.6 Max
Onroad
Reference
7.2 Min
7.2 Max
9.6 Min
9.6 Max
Nonroad
Reference
7.2 Min
7.2 Max
9.6 Min
9.6 Max
2012
39,300
39,300
39,100
39,600
39,300
2012
29,000
28,900
28,800
29,100
28,900
2012
10,200
10,400
10,300
10,500
10,400
2015
38,800
38,900
38,700
39,100
38,900
2015
29,100
29,000
28,900
29,200
29,000
2015
9,700
9,800
9,800
10,000
9,900
2020
39,900
39,900
39,700
40,200
40,000
2020
30,900
30,800
30,700
30,900
30,800
2020
9,000
9,100
9,000
9,300
9,200
2012
—
0
-200
300
0
2012
~
-100
-200
100
-100
2012
—
200
100
300
200
2015
—
100
-100
300
100
2015
~
-100
-200
100
-100
2015
—
100
100
300
200
2020
—
0
-200
300
100
2020
~
-100
-200
0
-100
2020
—
100
0
300
200
2012
—
0.0%
-0.5%
0.8%
0.0%
2012
~
-0.3%
-0.7%
0.3%
-0.3%
2012
—
2.0%
1.0%
2.9%
2.0%
2015
—
0.3%
-0.3%
0.8%
0.3%
2015
~
-0.3%
-0.7%
0.3%
-0.3%
2015
—
1.0%
1.0%
3.1%
2.1%
2020
—
0.0%
-0.5%
0.8%
0.3%
2020
~
-0.3%
-0.6%
0.0%
-0.3%
2020
—
1.1%
0.0%
3.3%
2.2%
151
-------
Table 4A-7. 1,3-Butadiene Emission Inventories under Various Ethanol Use Cases
Tons/Year
Change from Reference
%Change from Reference
PRIMARY CASE
Total
Reference
7.2 Min
7.2 Max
9.6 Min
9.6 Max
Onroad
Reference
7.2 Min
7.2 Max
9.6 Min
9.6 Max
Nonroad
Reference
7.2 Min
7.2 Max
9.6 Min
9.6 Max
2012
18,200
17,700
17,900
17,400
17,600
2012
11,600
11,200
11,400
11,000
11,200
2012
6,700
6,500
6,500
6,300
6,400
2015
18,000
17,500
17,700
17,200
17,400
2015
11,700
11,300
11,500
11,100
11,300
2015
6,300
6,200
6,200
6,100
6,100
2020
18,500
18,000
18,200
17,700
17,900
2020
12,500
12,100
12,300
12,000
12,100
2020
6,000
5,900
5,900
5,800
5,800
2012
—
-500
-300
-800
-600
2012
~
-400
-200
-600
-400
2012
—
-200
-200
-400
-300
2015
—
-500
-300
-800
-600
2015
~
-400
-200
-600
-400
2015
—
-100
-100
-200
-200
2020
—
-500
-300
-800
-600
2020
~
-400
-200
-500
-400
2020
—
-100
-100
-200
-200
2012
—
-2.7%
-1.6%
-4.4%
-3.3%
2012
~
-3.4%
-1.7%
-5.2%
-3.4%
2012
—
-3.0%
-3.0%
-6.0%
-4.5%
2015
—
-2.8%
-1.7%
-4.4%
-3.3%
2015
~
-3.4%
-1.7%
-5.1%
-3.4%
2015
—
-1.6%
-1.6%
-3.2%
-3.2%
2020
—
-2.7%
-1.6%
-4.3%
-3.2%
2020
~
-3.2%
-1.6%
-4.0%
-3.2%
2020
—
-1.7%
-1.7%
-3.3%
-3.3%
SENSITIVITY CASE
Total
Reference
7.2 Min
7.2 Max
9.6 Min
9.6 Max
Onroad
Reference
7.2 Min
7.2 Max
9.6 Min
9.6 Max
Nonroad
Reference
7.2 Min
7.2 Max
9.6 Min
9.6 Max
2012
17,900
17,300
17,500
16,800
17,100
2012
11,300
10,800
11,000
10,500
10,700
2012
6,700
6,500
6,500
6,300
6,400
2015
17,600
17,000
17,200
16,600
16,800
2015
11,300
10,800
11,000
10,500
10,700
2015
6,300
6,200
6,200
6,100
6,100
2020
18,000
17,300
17,600
16,900
17,200
2020
12,000
11,500
11,700
11,100
11,400
2020
6,000
5,900
5,900
5,800
5,800
2012
—
-600
-400
-1,100
-800
2012
~
-500
-300
-800
-600
2012
—
-200
-200
-400
-300
2015
—
-600
-400
-1,000
-800
2015
~
-500
-300
-800
-600
2015
—
-100
-100
-200
-200
2020
—
-700
-400
-1,100
-800
2020
~
-500
-300
-900
-600
2020
—
-100
-100
-200
-200
2012
—
-3.4%
-2.2%
-6.1%
-4.5%
2012
~
-4.4%
-2.7%
-7.1%
-5.3%
2012
—
-3.0%
-3.0%
-6.0%
-4.5%
2015
—
-3.4%
-2.3%
-5.7%
-4.5%
2015
~
-4.4%
-2.7%
-7.1%
-5.3%
2015
—
-1.6%
-1.6%
-3.2%
-3.2%
2020
—
-3.9%
-2.2%
-6.1%
-4.4%
2020
~
-4.2%
-2.5%
-7.5%
-5.0%
2020
—
-1.7%
-1.7%
-3.3%
-3.3%
152
-------
Chapter 5: Air Quality Impacts
5.1 Ozone
We performed ozone air quality modeling simulations for the eastern United States using
the ozone Response Surface Model (RSM) to estimate the effects of the projected changes in
emissions from gasoline vehicles and equipment for the 7.2 billion gallon ethanol use case. The
ozone RSM is a screening-level air quality modeling tool that allows users to quickly assess the
estimated air quality changes over the modeling domain. The ozone RSM is a model of a full-
scale air quality model and is based on statistical relationships between model inputs and outputs
obtained from the full-scale air quality model. In other words, the ozone RSM uses statistical
techniques to relate a response variable to a set of factors that are of interest, e.g., emissions of
precursor pollutants from particular sources and locations. The following section describes the
modeling methodology, including the development of the multi-dimensional experimental design
for control strategies and implementation and verification of the RSM technique. Additional
detail is available in the Air Quality Modeling Technical Support Document (AQMTSD) that
was drafted for the Mobile Source Air Toxics Rule Proposal (published March 29, 2006).74
The foundation for the ozone response surface metamodeling analyses was the CAMx
modeling done in support of the final Clean Air Interstate Rule (CAIR). The CAIR modeling is
fully described in the CAIR Air Quality Modeling Technical Support Document, but a brief
description is provided below.75 The modeling procedures used in the CAIR analysis (e.g.,
domain, episodes, meteorology) have been used for several EPA rulemaking analyses over the
past five years and are well-established at this point.
The ozone RSM uses the 2015 controlled CAIR emissions inventory as its baseline,
assuming future fuel quality remains unchanged from pre-Act levels, which serves as the
baseline for the analysis of the proposed standards.76 We then compare these baseline emissions
to the emissions which would have occurred in the future if fuel quality had remained unchanged
from pre-Act levels to those which will occur with fuel quality reflecting the increased renewable
fuel use projected in the future. This approach differs from that traditionally taken in EPA
regulatory impact analyses. Traditionally, we would have compared future emissions with and
without the requirement of the Act. However, as described in Chapter 1, we expect that total
renewable fuel use in the U.S. in 2012 to exceed 7.5 billion gallons even in the absence of the
Renewable Fuel Standard (RFS). Thus, a traditional regulatory impact analysis would have
shown no impact on emissions or air quality.
The modeling simulations that comprised the metamodeling were conducted using
CAMx version 3.10. It should be noted that because the ozone RSM is built from CAMx air
quality model runs, it therefore has the same strengths and limitations of the underlying model
and its inputs. CAMx is a non-proprietary computer model that simulates the formation and fate
of photochemical oxidants including ozone for given input sets of meteorological conditions and
emissions. The gridded meteorological data for three historical episodes were developed using
the Regional Atmospheric Modeling System (RAMS), version 3b.77 In all, 30 episode days
153
-------
were modeled using frequently-occurring, ozone-conducive, meteorological conditions from the
summer of 1995. Emissions estimates were developed for the evaluation year (1995) as well as a
future year (2015).
The CAMx model applications were performed for a domain covering all, or portions of,
37 States (and the District of Columbia) in the Eastern U.S., as shown in Figure 5.1-1. The
domain has nested horizontal grids of 36 km and 12 km. However, the output data from the
metamodeling is provided at a 12 km resolution (i.e., cells from the outer 36 km cells populate
the nine finer scale cells, as appropriate). Although the domain of the ozone RSM is the 37
Eastern states, the expanded use of ethanol in fuel is expected to occur nationwide. Chapter 4
describes the nationwide inventory impacts associated with the proposed standards.
Figure 5.1-1. Map of the CAMx Domain Used for RFS Ozone Metamodeling
The ozone RSM used for assessing the air quality impacts of expanded ethanol use in fuel
was developed broadly to look at various control strategies with respect to attaining the 8-hour
ozone NAAQS. The experimental design for the ozone RSM covered three key areas: type of
precursor emission (NOx or VOC), emission source type (i.e., onroad vehicles, nonroad vehicles,
area sources, electrical generating utility (EGU) sources, and non-utility point sources), and
location in or out of a 2015 model-projected residual ozone nonattainment area. This resulted in
a set of 14 emissions factors.
The 14 emission factors were randomly varied and used as inputs to CAMx. The
experimental design for these 14 factors was developed using a Maximin Latin Hypercube
method. Based on a rule of thumb of 10 runs per factor, we developed an overall design with
154
-------
140 runs (a base case plus 139 control runs). The range of emissions reductions considered
within the metamodel ranged from 0 to 120 percent of the 2015 CAIR emissions. This
experimental design resulted in a set of CAMx simulations that serve as the inputs to the ozone
response surface metamodel.
To develop a response surface approximation to CAMx, we used a multidimensional
kriging approach, implemented through the MIXED procedure in SAS. We modeled the
predicted changes in ozone in each CAMx grid cell as a function of the weighted average of the
modeled responses in the experimental design. A response-surface was then fit for the ozone
design value metric. Validation was performed and is summarized in the AQMTSD. The
validation exercises indicated that the ozone RSM replicates CAMx response to emissions
changes very well for most emissions combinations and in most locations.
The ozone RSM limits the number of geographically distinct changes in VOC and NOx
emissions which can be simulated. Emissions from motor vehicles and nonroad equipment can
be varied separately. Distinct percentage changes in either the motor vehicle or nonroad
inventories can also be applied in ozone nonattainment and attainment areas. However, distinct
emission impacts cannot be simulated in various ozone nonattainment areas (e.g., Chicago and
Houston or New York and Kansas City). This limits our ability to simulate the impact of
increased ethanol use in a couple of ways. First, ethanol use is not geographically uniform
across the U.S., either currently or in the future. Thus, the emission impacts resulting from
changes in ethanol use also varies geographically. Second, the emission impacts of ethanol use
are not uniform. Ethanol use in RFG and other areas which do not grant ethanol blends a 1.0 psi
RVP waiver will not experience as much of an increase in VOC emissions with increased
ethanol use as areas which grant ethanol blends an RVP waiver.
We developed a methodology which would best approximate the impact of changes in
local emissions on the ozone level in each local area, while maintaining as much of the impact of
ozone transport from other areas as possible given the above mentioned limitations. We do this
by running the ozone RSM twice for each scenario and drawing the resultant ozone impact from
the run which best matched the emission impact expected in a particular local area, considering
both the change in emissions modeled for that particular local area, as well as that occurring in
upwind areas.
First, as mentioned above, ethanol use is expected to change dramatically in some areas,
but not at all in others. Averaging the emission impacts across these two types of areas and
estimating the associated ozone impact would be very misleading. No area would be likely to
experience the ozone impact predicted. Some areas would experience a much greater impact,
while others would experience no impact. Therefore, the first step in using the Ozone RSM to
predict the ozone impacts related to the RFS is to estimate the change in VOC and NOx
emissions in those areas ethanol blend market share changed significantly. As was done in the
analysis of local emission impacts presented in Section 4.1.3.3 above, we defined a significant
change in ethanol blend market share as a change of 50% or more. This focuses the change in
emissions in those areas where the change is likely to occur.
155
-------
As discussed in Chapters 3 and 4, the effect of ethanol use on emissions differs
depending on the baseline fuel quality and the applicable RVP standards. In particular, ethanol
use has significantly different impacts on emissions in RFG, low RVP and 9 RVP areas.
Therefore, in order to better predict the ozone impact likely to occur in specific areas, we
estimate the change in VOC and NOx emissions separately for RFG, low RVP and 9 RVP areas
(per above, only for those areas in each case where ethanol blend market share changed by 50%
or more).
The Ozone RSM only covers the 37 easternmost states in the U.S. Therefore, we limited
the calculation of emission impacts to only those states. The Ozone RSM also only projects
ozone impacts for the years 2015, 2020, and 2030. Since we develop most of our impacts of the
RFS for the year 2012, we chose to run the Ozone RSM for the closest year to 2012, or 2015.
Finally, the Ozone RSM is designed to accept emission changes in terms of total onroad and total
nonroad sources, respectively. Both emission categories include diesels. Therefore, we included
estimates of VOC and NOx emissions from diesel vehicles and equipment in 2015 in our
calculation of the emission impacts. These diesel emissions do not change between the various
RFS scenarios. However, they do reduce the effective percentage change in VOC and NOx
emissions which is projected to occur. Overall, these analyses are very similar to those described
in Section 4.1.3.3 above, with the exceptions of the limitation to 37 states, the inclusion of diesel
emissions and the focus on 2015 instead of 2012. The results of these calculations are shown in
Table 5.1-1.
Table 5.1-1.
Emission Inputs to Ozone Modeling: Change in Mobile Sources Emissions in 37 Eastern
States where Ethanol Use Changes Significantly, July 2015 (percent change)
Case
RFG
Nonattainment
VOC
NOx
Low RVP
Nonattainment
VOC
NOx
Attainment
VOC
NOx
Primary Analysis
Onroad
7.2 Min
7.2 Max
-1.8
3.4
-0.2
0.1
6.0
6.7
0.2
0.3
7.1
7.6
0.2
0.2
Nonroad
7.2 Min
7.2 Max
4.7
-2.0
-3.5
1.4
0.5
0.5
3.6
6.3
2.3
2.3
2.4
2.7
Sensitivity Analysis
Onroad
7.2 Min
7.2 Max
0.6
2.8
-3.2
1.7
3.6
3.7
3.7
4.3
5.1
5.3
3.0
3.3
Nonroad
7.2 Min
7.2 Max
4.7
-2.0
-3.5
1.4
0.5
0.5
3.6
6.3
2.3
2.3
2.4
2.7
Our category of 9 RVP areas is very similar to the set of attainment areas in the Ozone
RSM. Therefore, the application of the emission impacts expected in 9 RVP areas in the Ozone
RSM was straightforward. However, both RFG and low RVP areas together generally comprise
156
-------
the set of nonattainment areas in the Ozone RSM. As seen in Table 5.1-1, the expected emission
impacts of the various RFS scenarios differ significantly depending on whether the area has RFG
or low RVP fuel. Both sets of emission impacts could not be run in the Ozone RSM at the same
time. Therefore, we ran the Ozone RSM twice. The first run applied the emission impacts
estimated for Low RVP areas to the ozone nonattainment areas in the Ozone RSM and applied
the emission impacts for 9.0 RVP areas to the ozone attainment areas in the Ozone RSM. This
run should produce satisfactory projections of ozone impacts for all areas except those areas with
RFG, as well as those areas where ethanol use is not expected to change.
The second run applied the emission impacts estimated for RFG areas to the ozone
nonattainment areas in the Ozone RSM and applied the emission impacts for 9.0 RVP areas to
the ozone attainment areas in the Ozone RSM. This run should produce satisfactory projections
of ozone impacts for all areas except those areas with low RVP, as well as those areas where
ethanol use is not expected to change.
We chose to use the first Ozone RSM run as our base for further adjustment. The main
reason for doing so involved the possibility that some nonattainment areas as defined in the
Ozone RSM might have 9 RVP fuel. The percentage changes in VOC and NOx emissions for
Low RVP areas were more similar to those for 9 RVP areas than those for RFG areas. Thus, the
first run more closely represented the emission impacts expected in these 9 RVP, nonattainment
areas than the second run.
The first adjustment made to the first Ozone RSM run was to set the predicted change in
ozone to zero in those counties not expected to experience a significant change in ethanol use.
This ignores any impact from ozone transport from other areas where ethanol use did change.
However, we believe that the ozone impacts due to transport are much smaller than those
associated with changes in local emissions. This is particularly true in this case, where the
percentage change in emissions would be the same in both the local and upwind areas.
The second adjustment made to the first Ozone RSM run was to substitute the projected
ozone impacts for RFG areas with those from the second run. Clearly, the second run does a
better job of representing the emission changes expected in RFG areas. Since the areas upwind
of RFG areas tend to have 9 RVP fuel, as opposed to Low RVP fuel, the second run should also
reasonably represent ozone transport from non-RFG areas. We present the ozone impacts of
increased ethanol use resulting from this methodology in the following section.
5.1.1 Ozone Response Surface Metamodel Results
This section summarizes the results of our modeling of ozone air quality impacts in the
future with and without the expanded use of ethanol in fuel. The impact of increased ethanol use
on the 8-hour ozone design values in 2015 are presented in Table 5.1-2. The Chapter 5
Appendix presents the impacts of increased ethanol use on a number of alternative measures of
ambient ozone concentration.
157
-------
Table 5.1-2.
Impact of Increased Ethanol Use on 8-hour Design Value Ozone Levels in 2015 (ppb):
7.2 Billion Gallons of Ethanol Use Scenario
Minimum Use in
RFC
Maximum Use
in RFC
Primary Analysis
Minimum Change
Maximum Change
Average Change Across 37 States
Population- Weighted Change Across
Average Change Where Ethanol Use
Population- Weighted Change Where
Significantly States
37 States
Changed Significantly States
Ethanol Use Changed
-0.030
0.395
0.062
0.057
0.137
0.134
-0.025
0.526
0.047
0.055
0.171
0.129
Sensitivity Analysis
Minimum Change
Maximum Change
Average Change Across 37 States
Population- Weighted Change Across
Average Change Where Ethanol Use
Population- Weighted Change Where
Significantly States
37 States
Changed Significantly States
Ethanol Use Changed
-0.180
0.637
0.134
0.114
0.294
0.268
0.000
0.625
0.088
0.106
0.318
0.250
As can be seen, ozone levels generally increase to a small degree with increased ethanol
use. This is likely due to the projected increases in both VOC and NOx emissions. Some areas
do see a small decrease in ozone levels. In our primary analysis, where exhaust emissions from
Tier 1 and later onroad vehicles are assumed to be unaffected by ethanol use, the population-
weighted increase in ambient ozone levels is 0.055-0.057 ppb. Since the 8-hour ambient ozone
standard is 0.08 ppm (85 ppb), this increase represents about 0.07 percent of the standard, a very
small percentage*2*2. While small, this figure includes essentially zero changes in ozone in areas
where ethanol use did not change. When we focus just on those areas where the market share of
ethanol blends changed by 50 percent or more, the population-weighted increase in ambient
ozone levels rises to 0.129-0.134 ppb. This increase represents about 0.16 percent of the
standard, still a very small percentage.
In our sensitivity analysis, where exhaust emissions from Tier 1 and later onroad vehicles
are assumed to respond to ethanol like Tier 0 vehicles, the population-weighted increase in
ambient ozone levels across the entire 37 state area is roughly twice as high, or 0.106-0.114 ppb.
This increase represents about 0.13 percent of the standard, still a very small percentage. When
we focus just on those areas where the market share of ethanol blends changed by 50 percent or
QQ
Appendix I of 40 CFR Part 50.
158
-------
more, the population-weighted increase in ambient ozone levels rises to 0.250-0.268 ppb. This
increase represents about 0.32 percent of the standard.
As we show in Table 5.1-2, the expanded use of ethanol in fuel is projected to result in a
small population-weighted net increase in future ozone. For much of the ozone RSM domain,
the net increase is generally so small as to be rendered insignificant when presenting design
values. Nonetheless, there are areas where the ozone increase is more significant. For the
primary analysis, we present the counties with the largest increases in the ozone design value in
Table 5.1-3 and Table 5.1-4. Each table presents the county level ozone design value results of
the minimum use in RFG areas scenario and the maximum use in RFG areas scenario,
respectively. It is important to note that the results of this ozone response surface metamodeling
exercise is meant for screening-level purposes only and does not represent the results that would
be obtained from full-scale photochemical ozone modeling. It is also important to note that the
ozone RSM results indicate that the counties which are projected to experience the greatest
increase in ozone design values are generally counties that are projected to have ambient
concentrations well below the 0.08 ppm ozone standard in the 2015 baseline.
159
-------
Table 5.1-3. 2015 Ozone Response Surface Metamodeling Results": Primary RFS 7.2
Billion Gallons of Ethanol Scenario, Minimum Use in RFG, Counties with Largest
Increases in the Ozone 8hr Design Valueb (ppb) Due to Increased Use of Ethanol
State Name
Illinois
Wisconsin
Indiana
Maine
Indiana
Missouri
Wisconsin
Illinois
Wisconsin
Ohio
Ohio
Ohio
Maine
Maine
Indiana
Ohio
Indiana
Michigan
New York
Ohio
County Name
Cook Co
Walworth Co
Lake Co
Hancock Co
Shelby Co
Cedar Co
Rock Co
Du Page Co
Sheboygan Co
Geauga Co
Clinton Co
Stark Co
Penobscot Co
York Co
Marion Co
Mahoning Co
Porter Co
Ingham Co
Westchester Co
Summit Co
2015 Baseline
(Post-CAIR)c
81.1
70.1
80.7
76.8
76.2
68.6
69.1
66.1
83.6
82.5
75.7
71.7
69.5
77.6
74.6
74.7
78.6
69.0
83.1
77.4
2015 With Ethanol
Use (minimum use in
RFG)
81.5
70.5
81.1
77.1
76.5
68.9
69.4
66.4
83.9
82.8
76.0
72.0
69.8
77.9
74.9
75.0
78.9
69.3
83.4
77.7
Effect of Expanded
Ethanol Use (ppb)
0.4
0.4
0.4
0.3
0.3
0.3
0.3
0.3
0.3
0.3
0.3
0.3
0.3
0.3
0.3
0.3
0.3
0.3
0.3
0.3
2015
Population
5,362,932
109,939
490,796
55,606
47,904
14,634
170,498
1,076,917
121,785
108,600
50,635
384,672
152,896
210,006
889,645
248,545
176,761
290,178
950,661
557,892
a Note that the results of the ozone response surface metamodeling (Ozone RSM) exercise is meant for screening-
level purposes only and do not represent results that would be obtained from full-scale photochemical ozone
modeling. There are a number of important caveats concerning these estimates: 1) The emission effects of
adding ethanol to gasoline are based on extremely limited data for recent vehicles and equipment; 2) The Ozone
RSM does not account for changes in CO emissions. Ethanol use should reduce CO emissions significantly,
directionally reducing ambient ozone levels in areas where ozone formation is VOC-limited; and, 3) The Ozone
RSM also does not account for changes in VOC reactivity. With additional ethanol use, the ethanol content of
VOC should increase. Ethanol is less reactive than the average VOC. Therefore, this change should also reduce
ambient ozone levels in a way not addressed by the Ozone RSM, again in areas where ozone formation is VOC-
limited.
b A design value is the mathematically determined pollutant concentration at a particular site that must be reduced
to, or maintained at or below the National Ambient Air Quality Standard to assume attainment. The 8-hour
ozone design value is the 3-year average of the fourth-highest daily maximum 8-hour average ozone
concentrations measured at each monitor within an area over each year, which must not exceed 0.08 ppm (85
ppb).
0 The Clean Air Interstate Rule (CAIR) modeling is fully described in the CAIR Air Quality Modeling Technical
Support Document (Docket EPA-HQ-OAR-2005-0036).
160
-------
Table 5.1-4. 2015 Ozone Response Surface Metamodeling Results": Primary RFS 7.2
Billion Gallons of Ethanol Scenario, Maximum Use in RFG, Counties with Largest
Increases in the Ozone 8hr Design Valueb (ppb) Due to Increased Use of Ethanol
State Name
Wisconsin
Wisconsin
Wisconsin
Michigan
Michigan
Michigan
Indiana
Indiana
Wisconsin
Michigan
Wisconsin
Indiana
Indiana
Indiana
Indiana
Maryland
Massachusetts
Wisconsin
Wisconsin
Virginia
County Name
Walworth Co
Sheboygan Co
Rock Co
Mason Co
Benzie Co
Ingham Co
Shelby Co
Allen Co
Winnebago Co
Huron Co
Door Co
Elkhart Co
Delaware Co
Marion Co
Posey Co
Frederick Co
Middlesex Co
Fond Du Lac
Co
Kewaunee Co
Henrico Co
2015 Baseline
(Post-CAIR)c
70.1
83.6
69.1
74.7
74.0
69.0
76.2
72.0
66.3
71.9
77.9
65.8
70.4
74.6
70.5
74.2
75.8
65.9
75.7
75.5
2015 With Ethanol
Use (maximum use in
RFG)
70.6
84.0
69.5
75.0
74.3
69.3
76.5
72.3
66.6
72.2
78.2
66.1
70.7
74.9
70.8
74.5
76.1
66.2
76.0
75.7
Effect of Expanded
Ethanol Use (ppb)
0.5
0.4
0.4
0.3
0.3
0.3
0.3
0.3
0.3
0.3
0.3
0.3
0.3
0.3
0.3
0.3
0.3
0.3
0.3
0.2
2015
Population
109,939
121,785
170,498
32,204
18,857
290,178
47,904
362,480
175,935
37,530
32,597
202,845
119,183
889,645
28,544
254,989
1,498,849
104,289
20,715
308,516
a Note that the results of the ozone response surface metamodeling (Ozone RSM) exercise is meant for
screening-level purposes only and do not represent results that would be obtained from full-scale photochemical
ozone modeling. There are a number of important caveats concerning these estimates: 1) The emission effects
of adding ethanol to gasoline are based on extremely limited data for recent vehicles and equipment; 2) The
Ozone RSM does not account for changes in CO emissions. Ethanol use should reduce CO emissions
significantly, directionally reducing ambient ozone levels in areas where ozone formation is VOC-limited; and,
3) The Ozone RSM also does not account for changes in VOC reactivity. With additional ethanol use, the
ethanol content of VOC should increase. Ethanol is less reactive than the average VOC. Therefore, this change
should also reduce ambient ozone levels in a way not addressed by the Ozone RSM, again in areas where ozone
formation is VOC-limited.
b A design value is the mathematically determined pollutant concentration at a particular site that must be
reduced to, or maintained at or below the National Ambient Air Quality Standard to assume attainment. The 8-
hour ozone design value is the 3-year average of the fourth-highest daily maximum 8-hour average ozone
concentrations measured at each monitor within an area over each year, which must not exceed 0.08 ppm (85
ppb).
0 The Clean Air Interstate Rule (CAIR) modeling is fully described in the CAIR Air Quality Modeling Technical
Support Document (Docket EPA-HQ-OAR-2005-0036).
The Wisconsin Department of Natural Resources recently performed a similar study of
the impact of increased ethanol use on ozone.78 They estimated that the conversion of gasoline
outside of RFG areas in Wisconsin to E10 blends would increase ozone in these areas on the
order of 1 ppb to as much as 2 ppb. (RFG areas in Wisconsin already contain 10 vol% ethanol.)
161
-------
This ozone increase was due to the predicted increase in NOx emissions associated with ethanol
use, since the non-RFG areas in Wisconsin are generally NOx limited for ozone formation.
Of the two ethanol use scenarios which we examined for ozone impacts, the 7.2 Max
scenario is closer to that examined by the State of Wisconsin. In the 7.2 Max scenario, RFG in
the Midwest continues to contain 10 vol% ethanol in the future, as assumed by Wisconsin. It is
interesting that the three counties with the highest predicted ozone increase in the 37 state area
covered by the Ozone RSM (and five out of the top eleven counties) are located in Wisconsin
(see Table 5.1-4). However, the ozone impacts are well below 1-2 ppb, ranging from 0.4-0.5
ppb. One difference between the estimates made here and those made by the State of Wisconsin
is that it appears that Wisconsin performed there modeling for calendar year 2003, while those
described above are for 2015. Emission standards applicable to new vehicles and equipment are
continually reducing emissions from these sources over time. Per the emission models used here
and by the State of Wisconsin (NONROAD and MOBILE6), the effect of fuel quality is
generally assumed to be proportional to the base emission level. As emissions from vehicles and
equipment decrease over time, the absolute impact of fuel quality changes decreases at the same
rate. Thus, the absolute emission changes predicted here for 2015 are likely to be much lower
than those predicted by Wisconsin for 2003.
There are a number of important caveats concerning these estimates. The Ozone RSM
does not account for changes in CO emissions. As shown in Chapter 4, ethanol use should
reduce CO emissions significantly, directionally reducing ambient ozone levels in areas where
ozone formation is VOC-limited. Accounting for the reduction in CO emissions in NOx-limited
areas, however, may have little impact on the ozone impact of ethanol use. For example, the
Wisconsin study cited above states that ozone formation in rural Wisconsin is NOx-limited. This
includes those Wisconsin counties listed in Tables 5.1-3 and 5.1-4 above. The inability of the
Ozone RSM to account for changes in CO emissions, therefore, may have little impact on the
direction and magnitude on the predicted level of ambient ozone concentrations in these
Wisconsin, and other NOx-limited, counties.
The Ozone RSM also does not account for changes in VOC reactivity. With additional
ethanol use, the ethanol content of VOC should increase. Ethanol is less reactive than the
average VOC. Therefore, this change should also reduce ambient ozone levels in a way not
addressed by the Ozone RSM. Again, like the impact of reduced CO emissions, this effect
applies to those areas where ozone formation is VOC-limited. Another limitation is the RSM's
inability to simulate the spatial distribution of emission impacts associated with the proposed
standard. Instead, we are forced to make simplifying assumptions about the geographic
uniformity of RFS emissions impacts, explained above. The caveats and limitations associated
with the RSM highlight the fact that it should only be used as a screening-level tool to
characterize broad trends associated with changes in different source categories of ozone
precursors.
Keeping these limitations in mind, the expanded use of ethanol will impact the national
emissions inventory of precursors to ozone, such as VOCs and NOx, as described in Chapter 4.
Exposure to ozone has been linked to a variety of respiratory effects including premature
162
-------
mortality, hospital admissions and illnesses resulting in school absences. Ozone can also
adversely affect the agricultural and forestry sectors by decreasing yields of crops and forests.
We estimate that the measurable changes in VOC and NOx which are a result of
increased ethanol use will, on average, result in small increases in ambient ozone formation. As
we discussed above, the ozone modeling results in a net increase in the average population
weighted ozone design value metric measured within the modeled domain (37 Eastern states and
the District of Columbia). In Appendix A, we also present the impacts of increased ethanol use
on a number of alternative measures of ambient ozone concentration used in the calculation of
ozone-related health impacts. The changes in these statistics are also very small, and would
likely lead to negligible monetized impacts. We therefore do not estimate and monetize ozone
health impacts here due to the magnitude of this change and the uncertainty present in the air
quality modeling. We acknowledge, however, that to the extent it occurs, increased future levels
of ambient concentrations of ozone related to the increased use of ethanol may result in
detrimental health and welfare effects due to ozone.
5.2 Particulate Matter
5.2.1 Impact of Changes in Direct PM Emissions
The amount of data evaluating the impact of ethanol and MTBE blending on direct
emissions of PM from gasoline-fueled vehicles is extremely limited, as discussed in Chapter 3.
Most studies do not test PM emissions from vehicles fueled with unleaded gasoline, because the
level of PM emissions from properly operating vehicles is usually very low, less than 0.1 g/mi.
Two studies indicate that the addition of ethanol might reduce direct PM emissions from
gasoline vehicles79 '80. However, both studies were performed under wintertime conditions and
one at high altitude. One of the studies only consisted of three vehicles. The available data
indicate that ethanol blending might reduce exhaust PM emissions under very cold weather
conditions (i.e., 0 F or less), particularly at high altitude. There is no indication of PM emission
reductions at higher temperatures and under warmed up conditions. Thus, the data are certainly
too limited to support a quantitative estimate of the effect of ethanol on PM emissions.
5.2.2 Impact of Changes in Secondary PM Formation
In addition to being emitted directly from a combustion source, fine particles can be
formed through a series of chemical reactions in the atmosphere when SO2, NOx, and VOC
oxidize or otherwise react to form a wide variety of secondary PM. For example, SO2 oxidizes
to SO3 and sulfuric acid and NOx oxidizes to NO3 and nitric acid which, in turn, react with
ammonia in the atmosphere to form ammonium sulfate and ammonium nitrate. Particles
generated through this gas to particle conversion are referred to as secondary aerosols (SA) and
represent a significant portion of ambient fine particulate matter. Studies have shown that as
much as 70% of the total organic carbon in urban parti culate matter can be attributed to
secondary organic aerosol (SOA) formation although the amount can also be less.81 Secondary
PM tends to form more in the summer with higher temperatures and more intense sunlight.
163
-------
Source-receptor modeling studies conducted in the Los Angeles area is 1993 by Schauer
et al82 indicate that as much as 67% of the fine particulate matter collected could not be
attributed to primary sources. The authors concluded that much of this unidentifiable organic
matter is secondary organic aerosol formed in the atmosphere. This is consistent with previous
studies conducted by Turpin and Huntzicker in 1991 who concluded that 70% of the total
organic carbon in urban PM measurements made in southern California can be attributed to
SOA.
Gas phase VOCs are oxidized by OH, NO2, peroxyacetylnitrate (PAN), and ozone in the
atmosphere, but their propensity to condense in the particle phase is a function of two factors:
volatility and reactivity. To accumulate as an aerosol, a reaction product must first be formed in
the gas phase at a concentration equal to its saturation concentration. This requirement will not
be met if the relevant gas-phase reactions of the VOC are too slow or if the vapor pressure of the
reaction product is higher than the initial concentration of its VOC precursor.83 Limited data for
reaction rate constants determined both experimentally and estimated by structural relationships
are available in the published literature. However, the atmospheric chemistry behind SOA
reaction rates and the estimated aerosol yield is highly complex and carries with it a great deal of
uncertainty. Research in this area is ongoing and thus the capacity to quantitatively model SOA
formation is not yet a straightforward process.
In general, all reactive VOC are oxidized by OH or other compounds. Additionally,
alkenes, cycloalkenes, and other olefinic compounds can react with ozone and NO2 to form
secondary aerosols. In fact, ozone is responsible for nearly all the SOA formation from olefins,
while OH plays little or no role at all (Grosjean and Seinfeld, 1989; Izumi and Fukuyama, 1990).
Many VOC, however, will never form secondary organic aerosol under atmospheric conditions
regardless of their reactivity. This is because the products of reactions of these compounds have
vapor pressures that are too high to form aerosols at atmospheric temperatures and pressures.
These include all alkanes and alkenes with up to 6 carbon atoms, benzene and many low-
molecular weight carbonyls, chlorinated compounds and oxygenated solvents (Grosjean, 1992).
The VOC that have the greatest propensity to form SOA include aromatic hydrocarbons
(such as toluene but even including benzene), higher molecular weight olefins and cyclic olefins,
and higher molecular weight paraffins. Kleindienst et al suggest that a high fraction of SOA is
due to aromatic hydrocarbon precursors. Furthermore, "aromatic products having a single alkyl
group on the aromatic ring were found to represent a 'high-yield' family (e.g., toluene,
ethylbenzene); compounds having multiple methyl groups (e.g., m-xylene, 1,2,4-
trimethylbenzene) were found to represent a 'low-yield' family" (Kleindienst, 269). All of the
above mentioned VOC precursors are important either because there are large amounts of these
particular VOC emitted per day, or because a large fraction of the VOC reacts, or a combination
of the two. Based on VOC emissions inventory data collected in the Los Angeles area, the most
important aerosol precursors (in the LA area using 1982 VOC emissions inventories) are listed in
Table 5.2-1 below:
164
-------
Table 5.2-1. Predicted In Situ SOA Formation
During a Smog Episode in Los Angeles
VOC Functional Group kg emitted daily*
Aromatics
Olefins
Paraffins
Alkenes
Cyclic Olefins
Terpenes
Alkanes
Cycloalkanes
223985
31163
3220
6000
140493
37996
Secondary PM %
Produced (kg)* yield
3061
608
144
626
368
96
1.37
1.95
4.47
10.43
0.26
0.25
'Source: Grosjean et al, 1992
These predictions are a function of input data collected in the Los Angeles area, and
assume ambient levels of [ozone] = 100 ppb, [OH]=1.0xl06 molecules/cm3, and [NO3]=0 with 6
hours of reaction time. Aromatics are the largest functional group in terms of the absolute
quantities of VOC emitted daily, and thus they eventually form the most SOA. Likewise, many
high molecular weight paraffins (alkanes) form SOA on a significant scale simply because their
emissions are high. However, the relative fraction of paraffins that react is less than that of
aromatics in smog chamber experiments simulating SOA formation in the atmosphere. For
olefins, the alkenes exhibit a combination of both relatively high emissions, and a high fraction
of VOC reacted to form SOA. Cyclic olefins, in contrast, are emitted in relatively low levels, but
a high fraction of these VOC react and the end result is a proportionally higher SOA yield than
with the alkenes. Lastly, there are several "miscellaneous" compounds and terpenes that are
emitted on a relatively small scale (in southern California), but that produce a substantial amount
of secondary organic aerosol.
Researchers at EPA recently completed a field study in the Raleigh/Durham area of
North Carolina that investigated the contribution of various sources to ambient PM 2.5
concentrations.84 In the study they identified toluene as an SOA precursor. They estimate that
mobile sources contribute nearly 90% of the total toluene emissions in that region based on a
chemical mass balance approach. At the same time, however, SOA attributable to non-fuel-
related VOC (i.e., biogenic emissions) was found to be an even larger contributor to SOA (i.e.,
toluene was not likely the dominant source of SOA in this area). This study is currently
undergoing peer review and will be published shortly. Qualitatively, however, this information
is still quite useful since the study identifies a contributing source of SOA that is attributable
almost entirely to mobile sources.
VOC reaction rates increase with increasing ambient temperature and sunlight intensity,
so the level of SOA formed is much higher in summer than in winter. Even in the more
temperate coastal climates of southern CA, studies have found the summertime concentration of
SOA calculated through Chemical Mass Balance models show SOA formation to be anywhere
from 2-5 times higher in summer than winter. In a study conducted at both urban and rural
locations in the southeastern United Sates, the concentration of SOA in the summer and early fall
QC
was roughly 2-3 times that of colder months .
165
-------
As mentioned in Chapters 2 and 3, the addition of ethanol should reduce aromatics in
gasoline, which will in turn reduce the aromatics emitted in the exhaust. However, quantifying
the emission reduction is not possible at this time due to a lack of speciated exhaust data for
newer vehicles running on ethanol blends. In addition, increased NOx emissions resulting from
the increased use of ethanol could increase the formation of nitrate PM.
Overall, we expect that the decrease in secondary PM from organic aromatic
hydrocarbons is likely to exceed the increase in secondary nitrate PM. In 1999, NOx emissions
from gasoline-fueled vehicles and equipment comprised about 20% of national NOx emissions
from all sources86. In contrast, gasoline-fueled vehicles and equipment comprised over 60% of
all national gaseous aromatic VOC emissionsRR. The percentage increase in national NOx
emissions due to increased ethanol use should be smaller than the percentage decrease in national
emissions of gaseous aromatics. In most urban areas, ambient levels of secondary organic PM
exceed those of secondary nitrate PM. Thus, directionally, we expect a net reduction in ambient
PM levels due to increased ethanol use. However, we are unable to quantify this reduction at
this time.
There are numerous uncertainties associated with these predictions. These uncertainties
arise from uncertainty in the emissions inventory of a given area, from uncertainty in the kinetic
VOC reaction rate calculations, as well as from the uncertainty in the aerosol yield found
experimentally in smog chamber studies and their use in providing mechanisms in models.
While these predictions shed light on the VOC functional groups that play the most important
role in SO A formation, these estimates are too uncertain to base quantitative estimates of the
impact of gaseous VOC emissions on ambient levels of SOA. EPA ORD scientists are currently
carrying out a wide variety of laboratory studies to refine the SOA chemistry mechanisms for use
in the next version of the CMAQ model, which is expected to be completed in 2007 and
submitted for peer review. Therefore, this updated information will not be available in time for
the RFS final rulemaking, but should be available in time for the comprehensive study of the
Act's fuel requirements which is due in 2008.ss
Given these uncertainties and gaps in the available data, we are unable to estimate the
cumulative impact that an increase in the future use of ethanol in fuel will have on PM2.5
formation. EPA currently utilizes the CMAQ model to predict ambient levels of PM as a
function of gaseous and PM emissions. This model includes mechanisms to predict the
formation of nitrate PM from NOx emissions. However, it does not currently include any
mechanisms addressing the formation of secondary organic PM. EPA is currently developing a
model of secondary organic PM from gaseous toluene emissions for incorporation into the
CMAQ model in 2007, as mentioned in section 5.2.2. The impact of other aromatic compounds
will be added as further research clarifies their role in secondary organic PM formation.
Therefore, we expect to be able to quantitatively estimate the impact of decreased toluene
emissions and increased NOx emissions due to increased ethanol use as part of the
comprehensive analysis of U.S. fuel requirements required by Congress in 2008. As we stated
above, however, reductions in the aromatic content of fuel, and the related reduction in
1111 Based on internal analyses of emissions inventories.
ss Subject to funding.
166
-------
secondarily formed PM2.5, are expected to offset increases in secondarily formed PM2.5 as a result
of increased emissions of NOx.
167
-------
Chapter 5: Appendix
168
-------
Table 5A-1. 2015 Ozone Response Surface Metamodeling Summary Statistics for the RFS
Rule"'b; Primary Scenario, 7.2 Billion Gallons of Ethanol
Shour Design Value (ppb)
Statistic
Minimum Change
Maximum Change
Average Change
Standard Deviation
Population- Weighted Change
Minimum Use in RFC
-0.030
0.395
0.062
0.087
0.057
Maximum Use in RFG
-0.025
0.526
0.047
0.089
0.055
24hr Average (ppb)
Statistic
Minimum Change
Maximum Change
Average Change
Standard Deviation
Population- Weighted Change
Minimum Use in RFG
-0.105
0.308
0.010
0.018
0.025
Maximum Use in RFG
-0.128
0.084
0.005
0.014
0.007
Ihr Maximum (ppb)
Statistic
Minimum Change
Maximum Change
Average Change
Standard Deviation
Population- Weighted Change
Minimum Use in RFG
-0.056
0.393
0.019
0.033
0.037
Maximum Use in RFG
-0.070
0.153
0.011
0.027
0.021
Average 9-to-5 (ppb)
Statistic
Minimum Change
Maximum Change
Average Change
Standard Deviation
Population- Weighted Change
Minimum Use in RFG
-0.108
0.575
0.023
0.041
0.050
Maximum Use in RFG
-0.145
0.211
0.014
0.035
0.026
Average 10-to-3 (ppb)
Statistic
Minimum Change
Maximum Change
Average Change
Standard Deviation
Population- Weighted Change
Minimum Use in RFG
-0.081
0.353
0.015
0.027
0.032
Maximum Use in RFG
-0.076
0.167
0.008
0.021
0.014
aNote that the statistics presented here represent ethanol use changes across the entire 37-state ozone RSM
domain.
169
-------
b A design value is the mathematically determined pollutant concentration at a particular site that must be
reduced to, or maintained at or below, the National Ambient Air Quality Standard to assume attainment.
The 8-hour ozone design value is the 3-year average of the fourth-highest daily maximum 8-hour average
ozone concentrations measured at each monitor within an area over each year, which must not exceed 0.08
ppm (85 ppb). The other ozone metrics (24hr average, Ihr maximum, average 9-to-5, and average 10-to-3)
are calculated at the ozone RSM grid cell level (based on the CAMx model grid). Air quality metrics are
daily values calculated from daily observations (or modeled estimates), or through mathematical
manipulations of hourly observations (or modeled estimates). The ozone metrics presented here are those
typically used in ozone-related health impact functions.
170
-------
Table 5A-2. 2015 Ozone Response Surface Metamodeling Summary Statistics for the RFS
Rulea'b; Sensitivity Scenario, 7.2 Billion Gallons of Ethanol
Shour Design Value (ppb)
Statistic
Minimum Change
Maximum Change
Average Change
Standard Deviation
Population- Weighted Change
Minimum Use in RFC
-0.180
0.637
0.134
0.164
0.114
Maximum Use in RFG
0.000
0.625
0.088
0.152
0.106
24hr Average (ppb)
Statistic
Minimum Change
Maximum Change
Average Change
Standard Deviation
Population- Weighted Change
Minimum Use in RFG
-0.113
0.345
0.022
0.038
0.042
Maximum Use in RFG
-0.139
0.132
0.011
0.028
0.015
Ihr Maximum (ppb)
Statistic
Minimum Change
Maximum Change
Average Change
Standard Deviation
Population- Weighted Change
Minimum Use in RFG
-0.121
0.635
0.039
0.066
0.073
Maximum Use in RFG
-0.118
0.364
0.020
0.050
0.041
Average 9-to-5 (ppb)
Statistic
Minimum Change
Maximum Change
Average Change
Standard Deviation
Population- Weighted Change
Minimum Use in RFG
-0.155
0.577
0.031
0.052
0.062
Maximum Use in RFG
-0.150
0.229
0.015
0.038
0.027
Average 10-to-3 (ppb)
Statistic
Minimum Change
Maximum Change
Average Change
Standard Deviation
Population- Weighted Change
Minimum Use in RFG
-0.154
0.584
0.031
0.053
0.063
Maximum Use in RFG
-0.149
0.225
0.015
0.039
0.028
aNote that the statistics presented here reflect the impact of ethanol use changes across the entire eastern
U.S. 37-state ozone RSM domain.
171
-------
b A design value is the mathematically determined pollutant concentration at a particular site that must be
reduced to, or maintained at or below, the National Ambient Air Quality Standard to assume attainment.
The 8-hour ozone design value is the 3-year average of the fourth-highest daily maximum 8-hour average
ozone concentrations measured at each monitor within an area over each year, which must not exceed 0.08
ppm (85 ppb). The other ozone metrics (24hr average, Ihr maximum, average 9-to-5, and average 10-to-3)
are calculated at the ozone RSM grid cell level (based on the CAMx model grid). Air quality metrics are
daily values calculated from daily observations (or modeled estimates), or through mathematical
manipulations of hourly observations (or modeled estimates). The ozone metrics presented here are those
typically used in ozone-related health impact functions.
172
-------
Chapter 6: Lifecycle Impacts on Fossil Energy and Greenhouse
Gases
6.1 Lifecycle Modeling
Lifecycle modeling is an established methodology for accounting for all energy and
emissions from a production process. It is meant to incorporate the material aspects, input and
output, of each step in a product system. This methodology allows you to identify key processes
and emission sources in the process, and to equitably compare the impacts of varying products
and processes on the consumption of natural resources, pollutant generation and environmental
burden. It is important to note that lifecycle modeling provides only general comparisons, based
on industry-wide estimates and assumptions. The results of this type of analysis are highly
dependent upon the input data used, the variables considered, and the assumptions made.
Nevertheless, within these limitations, it can be an extremely useful tool for evaluating the true
environmental impacts of products and processes.
For transportation fuels, lifecycle modeling considers all steps in the production of the
fuel. This includes production of the fuel feedstock, transportation of the fuel feedstock to a
processing facility, fuel processing, and distribution of the fuel to the retail outlet. If the analysis
considers only the finished product, it is sometimes called a 'well-to-pump' analysis; if the fuel
combustion emissions are included, it can be called a 'well-to-wheel' analysis. There are
advantages to both approaches, in this work we have considered 'well-to-wheel' impacts.
However, we are not addressing the issues of vehicle technology and energy efficiency, since we
are making the assumption that the vehicle issues will not affected by the presence of renewable
fuels (i.e., efficiency of combusting one Btu of renewable fuel is equal to the efficiency of
combusting one Btu of conventional fuel).
To put this type of analysis into perspective, consider the example of gasoline. The fuel
feedstock is crude oil. The lifecycle analysis accounts for the energy used to extract the oil from
the ground and any associated emissions, such as the natural gas that is flared at the well head.
Next you evaluate transportation of the crude oil to the refinery. If it is domestic crude oil, it
may be delivered by pipeline and/or barge. The analysis takes into account national trends for
domestic oil transportation, and apportions energy used and emissions generated to each type of
transportation. For foreign crude oil, the energy and emissions from ocean tankers is included,
with an estimate of the average distance traveled by these tankers. Next is an estimation of the
energy use and emissions from the refinery. Because gasoline is not the only product produced
at the refinery, only a portion of the energy and emissions is allocated to gasoline production.
There are different methods for making this allocation, based on the value of the co-products or
an engineering assessment of the energy use and emissions from the various units in the refinery.
You then evaluate the energy use and emissions from transporting the gasoline to market, via
pipeline and truck, based on national average distances. Finally, vehicle energy use and
emissions are estimated. Figure 6.1-1 illustrates this process.
173
-------
Figure 6.1-1. Lifecycle Production Process, 'Well-to-Wheel', for Gasoline
Refueling
Grade Crude
Productiojvansport
via pipeline
Grade Refining
Fuel
Transport
via pipeline
Fuel
Use
Lifecycle modeling has been a useful tool in evaluating the environmental benefits of
various alternative transportation fuels. It allows the replacement fuel to be fairly compared
against the conventional transportation fuels - gasoline and diesel fuel. There have been several
significant lifecycle analyses of transportation fuels done in the last decade. The most widely
known is a model developed by the Department of Energy (DOE) Argonne National Laboratory
(ANL) called the Greenhouse Gases, Regulated Emissions, and Energy Use in Transportation
(GREET) model. This is the model used in the analysis for the Renewable Fuel Standard (RFS)
program.
There have been other well-respected lifecycle models and analyses of transportation
fuels, but none are as comprehensive and user-friendly as GREET. Most of these analyses use
similar lifecycle methodology. The differences in results are often due to differences in the
assumptions and projections made throughout the model. One result that has been debated
recently is the net energy balance of corn-based ethanol fuel. Some analysts have suggested that
there is actually a negative energy balance for corn ethanol, meaning that it takes more energy to
produce the ethanol than is contained in the resulting fuel, making it an unattractive
transportation fuel. However, this work was based on out-dated farming and ethanol production
data, included data not normally considered in lifecycle analysis for fuels, and did not follow the
standard methodology for lifecycle analysis. This study emphasizes the importance of the input
data and methodology when using lifecycle analysis. It also shows how dependent this type of
analysis is on the assumptions made throughout the model. For this reason, EPA has reviewed
and modified GREET somewhat to reflect the data and assumptions appropriate for the RFS.
These modifications are discussed further in section 6.1.2.
6.1.1 Overview of GREET
The lifecycle model used in the evaluation of the impacts of the RFS program is the fuel-
cycle model developed by DOE's Argonne National Laboratory. For this work, EPA used the
most recent version of this model, GREET 1.7. GREET is the most widely known and used
model of this type for transportation fuels. It has been reviewed, used and referenced by a wide
174
-------
variety of analysts, including General Motors, National Corn Growers Association, several fuel
industry organizations, and a wide variety of academic institutions. It is the most comprehensive
and user-friendly model of its type. It has been under development for over 10 years, with input
from EPA, USD A, DOE laboratories, and industry representatives. The model addresses the full
lifecycle for an exhaustive number of alternative transportation fuels and automotive
technologies. For these reasons, EPA felt it was the best tool for evaluating the energy and
emission impacts of the RFS program.
The GREET model has been developed to calculate per-mile energy use and emission
rates of various combinations of vehicle technologies and fuels for both fuel cycles and total
energy cycles. The model actually consists of three components: GREET 1.x, which calculates
fuel cycle energy use and emissions, GREET 2.x, which calculates light-duty vehicle cycle
energy use and emissions, and GREET 3.x, which calculates heavy-duty vehicle cycle energy
use and emissions. All discussion here refers to GREET 1.7, the most recent version of the fuel
component of GREET.
To estimate fuel cycle energy use and emissions, GREET first estimates energy use and
emissions for a given upstream stage. The model then combines the energy use and emissions
from all upstream stages for a fuel cycle, to estimate total upstream fuel cycle energy use and
emissions. Inputs are national-average energy usage rates, efficiencies and emission factors for
each stage. The model calculates total energy use, fossil energy use, and emission rates for the
regulated pollutants and greenhouse gases, reported as grams per mile or grams per million Btu.
These results allow comparison of transportation fuels, based on energy use and/or emissions.
6.1.2 Modifications to GREET
EPA chose to use GREET 1.7 to evaluate the lifecycle impacts of the RFS program.
GREET 1.7 is the most recently released version of the GREET model. However, this version of
the model does not reflect the potential impacts on transportation fuel industries as a result of the
RFS program. In addition, for this regulation our intent was to evaluate the impact of
incremental renewable fuel production resulting from the RFS program and not a current
industry average, as assumed by GREET. Therefore, EPA has modified some of the input
variables and assumptions made in the GREET model. We will continue to evaluate these
factors in preparation for the final rulemaking. As shown in Section 6.2.3 the fuel pathways
from GREET evaluated for this proposal include:
• Petroleum-based gasoline (conventional and RFG blendstock)
• Petroleum-based low sulfur diesel
• Corn ethanol
• Cellulosic ethanol (herbaceous and woody biomass feedstock)
• Soybean-based biodiesel
In the timeframe available for developing this proposal, we chose to concentrate our
efforts on those GREET input values that had significant influence on the lifecycle emissions or
energy estimates and that were likely to be based on outdated information. We reviewed the
input values only for corn-based ethanol, since this fuel is likely to continue to dominate the
175
-------
renewable fuel pool through at least 2012. For cellulosic ethanol and biodiesel the GREET
default values were used in this proposal. However, we have also initiated a contract with ANL
to investigate a wider variety of GREET input values, including those associated with the
following fuel/feedstock pathways:
• Ethanol from corn
• Ethanol from cellulosic materials (hybrid populars, switchgrass, and corn stover)
• Biodiesel from soybean oil
• Methanol from renewable sources
• Natural gas from renewable sources
• Renewable diesel formulations
The contract focuses on the potential fuel production developments and efficiency
improvements that could occur within the time-frame of the RFS program. The GREET input
value changes resulting from this work are not expected to be available in time for this proposal,
but they will be incorporated into revised lifecycle assessments for the final rule.
We did not investigate the input values associated with the production of petroleum-
based gasoline or diesel fuel in the GREET model for this proposal. However, the refinery
modeling discussed in Chapter 7 will provide some additional information on the process energy
requirements associated with the production of gasoline and diesel under a renewable fuels
mandate. We will use information from this refinery modeling for the final rule to determine if
any gasoline or diesel fuel GREET input values should be changed.
A summary of the GREET corn ethanol input values we investigated for this proposal is
given below.
6.1.2.1 Wet Mill versus Dry Mill Ethanol Plants
As described in Chapter 1, there are two processes for producing ethanol from corn: wet
milling and dry milling. The GREET 1.7 model assumes that 70% of existing ethanol plants are
dry mill, and 30% are wet mill. For this analysis, we only consider new or incremental ethanol
production and it was assumed that essentially all new ethanol plants will be dry mill operations.
That has been the trend in the last few years as the demand for ethanol has grown, and our
analysis of ethanol plants under construction and planned for the near future has verified this.
This trend is discussed in more detail in Chapter 1.
6.1.2.2 Coal versus Natural Gas in Ethanol Plants
The type of fuel used within the ethanol plant for process energy, to power the various
components used in ethanol production (dryers, grinders, heating, etc.) can vary among ethanol
plants. The type of fuel used has an impact on the energy usage, efficiency, and emissions of the
plant, and is primarily determined by economics. Most new plants built in the last few years
have used natural gas. However, based on specific situations and economics, some new plants
are using coal. In addition, EPA is promoting the use of combined heat and power, or
cogeneration, in ethanol plants to improve plant energy-efficiency and to reduce air emissions.
176
-------
This technology, in the face of increasing natural gas prices, has the potential to make ethanol
plants more efficient, and to make coal a more attractive energy source for new ethanol plants.
GREET 1.7 assumes that 20% of existing dry mills use coal and 80% use natural gas.
For wet mills, the model assumes 40% of existing plants use coal while 60% use natural gas.
GREET default factors are meant to represent the average percentage of fuel use for the entire
industry, and may not reflect the recent growth in the industry as outlined in Chapter 1. Our
analysis for this rule is based on new or incremental production. Therefore, for the current
analysis, it was assumed that 10% of all future dry mill plants will use coal for process energy.
This is based on detailed analysis of the ethanol industry, near-term plant construction and
expansion plans, and projected costs for coal and natural gas. This analysis was discussed in
detail in Chapter 1. Future work in preparation for the final rule will evaluate the potential trends
for combined heat and power and coal as process fuel.
6.1.2.3 Ethanol Plant Process Efficiency
The ethanol plant process energy use values assumed in GREET 1.7 are 36,000
Btu/gallon of ethanol produced by the dry milling process and 49,950 Btu/gallon of ethanol
produced by the wet milling process. The values were selected based upon a review of current
scientific and technical literature, including U.S. government estimates from National Renewable
Energy Laboratory87 (NREL), Argonne National Laboratory88 (ANL), U.S. Department of
Agriculture (USD A) ethanol plant survey data89, and other USD A studies90.
A 1999 ANL report predicted 2005 ethanol process efficiencies of 36,900 Btu/gallon (dry
mill) and 34,000 Btu/gallon (wet mill)91. Additionally, the average of then-current ethanol plant
process efficiencies from the late-nineties92 yields ethanol process efficiencies of 41,705
Btu/gallon (dry mill) and 47,918 Btu/gallon (wet mill), values which, when projected to the
present, coincide with currently achieved ethanol plant process efficiencies selected for use in
this analysis.
These process efficiency estimates also coincide with those used in GREET Version 1.793
as well as 1997 projected process efficiency cited in GREET Version 1.5 for year 200594.
Therefore, we believe that the default values in GREET are reasonable.
6.1.2.4 Corn Transport Distances
Corn transport distances selected for use in this analysis are 100 miles round trip. Corn
used in the ethanol production process is assumed to travel from corn fields to ethanol
production facilities in a two-step process; first, corn is transported from outlaying farms to
centrally-located collection facilities, such as county elevators. Second, this corn is transported
from the collection facilities to the ethanol production facilities. The first leg of the corn
transport process is assumed to be a 20-mile round trip and the second leg is assumed to be an
80-mile round trip. These assumptions coincide with those used in GREET95 Version 1.7 and
GREET Version 1.5.
177
-------
Corn transport data is limited, however; GraboskiTT found that the average one-way
hauling distance for corn from fields to county elevators was 7.5 miles and from county elevators
to ethanol processing facilities was 49.7 miles for an effective average round-trip corn transport
distance of 74.6 miles. Similarly, Gervais and Baumel96 found that average one-way corn
transport distances for the 1994-1995 Iowa growing season was 37.2 miles for semi-trucks
(35.8%), 4.9 miles for wagons (33.3%), and 9.1 miles for single and tandem axel vehicles
(30.9%). Several Minnesota corn mills indicated that the maximum radius of supply for their
mills was 65 to 80 miles (values apparently cited in the same study).
The available data on corn transport distances does not provide a clear indication that the
default values in GREET are unreasonable. Also, a sensitivity analysis indicates that changing
these values will not have a significant impact on the results (halving or doubling the
transportation distances changes results by ~1 percent). Therefore, we retained the GREET
default values for our analysis.
6.1.2.5
Ethanol Transportation Distances and Modes
The default values in GREET for ethanol transportation and modes are shown in Table
6.1-1. These values correspond to numbers in a USDA study on the energy balance of corn
ethanol.97
Table 6.1-1. GREET Ethanol Transportation Input Data
Mode
Rail
Barge
Truck
Plant to Terminal
%
40%
40%
20%
Distance (miles)
800
520
80
Terminal to Station
%
0%
0%
100%
Distance (miles)
30
The GREET default values are consistent with the analysis we performed on ethanol
distribution infrastructure. Chapter 1 of this document discusses current ethanol transportation
and distribution and indicates that if ethanol facilities are located within 100-200 miles of a
terminal, trucking is preferred. Rail and barge are used for longer distances. Pipelines are not
currently used to transport ethanol and are not projected to play a role in ethanol transport in the
future time frame considered.
We also discuss in Chapter 1 future ethanol transportation and distribution needs based
on the increased amounts of renewable fuels used as a result of this rule. We concluded that
most new ethanol freight volumes will be handled by rail and that ethanol transport by inland
waterway will remain constant.
A recent USDA Cost of Ethanol Production report also provides information on ethanol
distribution distances and modes.98 The report includes 2002 data from a survey of 21 dry mill
TT The authors assume that the corn payload weight is equal to the transport vehicle weight, that the vehicle returns
empty, and the effective average round-trip vehicle distance can be estimated as being one and a half times the one-
way travel distance (1.5 times 49.7 miles = 74.6 miles); Graboski, 2002, Fossil Energy Use in the Manufacture of
Com Ethanol, Colorado School of Mines, (Prepared for the National Corn Growers Association).
178
-------
ethanol plants. The survey collected data on modes and distances traveled for ethanol transport
from the facilities. The report concluded that 46 percent of the ethanol produced at the surveyed
plants in 2002 was shipped by truck an average one way distance of 93 miles, with a range of 30
to 250 miles. The remaining 54 percent of ethanol produced was shipped by rail an average one
way distance of 1,163 miles, with a range of 800 to 2,500 miles. However, this data is for a
subset of existing plants, for example, there is no barge transportation listed, and also does not
take into account the increased demand for ethanol projected by this rule.
Comparing the GREET default values to these other sources indicates that the GREET
defaults values for percent of ethanol transported by rail may be low. However, different studies
will produce different results depending on the underlying assumptions, for example, current or
future ethanol production scenario, location of ethanol production and use, etc. Due to lack of
precise data on future ethanol transportation by mode, we concluded that the current GREET
default values for percent of ethanol transported by mode are appropriate for the RFS analysis.
The GREET default values for miles shipped by mode fall within the range of values
listed in the USDA survey data of existing plants. The USDA survey data indicate higher than
average transportation distances; however the data is not comprehensive enough, only
representing a small fraction of total and projected ethanol production capacity, thus not
warranting a change to the default GREET values. Therefore, the default values shown in Table
6.1-1 were used in this analysis.
6.1.2.6
Corn Yield and Related Inputs
GREET includes a collection of energy use and material inputs to corn farming per
bushel (bu) of corn produced. Corn farming input data parameters and default values provided in
GREET version 1.7 are shown in Table 6.1-2.
Table 6.1-2. GREET Corn Farming Input Data
Input Parameter
Energy Use for Corn Farming
- Energy use from diesel fuel
- Energy use from gasoline
- Energy use from natural gas
- Energy use from LPG
- Energy use from purchased electricity
Nitrogen Fertilizer (as N)
Phosphate Fertilizer (as P2Os)
Potash Fertilizer (as K^O)
Herbicide Use:
Insecticide Use:
Default Value
22,500 Btu/bu
38.3%
12.3%
21.5%
18.8%
9.0%
460 g/bu
165 g/bu
205 g/bu
8.1 g/bu
0.68 g/bu
The default GREET input values for corn farming shown in Table 6.1-2 are based on
farm energy use and material inputs per acre divided by an assumed corn yield in bu/acre.
179
-------
Therefore, while corn yield is not a direct input in GREET, it is a critical part of the calculation
of corn energy and material input requirements. Although corn yields have been generally
raising over time, see Figure 6.1-2, the annual variation is very volatile".
Figure 6.1-2. U.S. Average Corn Yield
180 n
60
1990
1995
2000
2005
2010
We examined data on farm energy use, material input, and yield data to determine if the
GREET default values needed to be updated. The lifecycle modeling conducted for the RFS
program is based on future predictions. Unfortunately, no good projections of future energy use
associated with corn farming are available. USD A does list projections for corn yield. The 2012
projected U.S. average corn yield is 158.5 bu/acre.100 However, as corn yield is not a direct
input into GREET, and corn yield is linked to farm energy and material inputs, we used historic
corn yield data for the GREET default parameter analysis. The following USDA information
sources were compared to the GREET default values.
• The USDA Agricultural Resource Management Survey (ARMS) provides data from
selected States on fuel, electricity, natural gas, and seed corn used per acre on the farm
and activities of moving farm products to initial storage facilities.
• The USDA National Agricultural Statistics Service (NASS) produces annual reports
listing quantities of fertilizers and chemicals used per acre of corn.
• The NASS also produces annual data on crop production including yields per acre and
total production of corn by state.
USDA NASS data on fertilizer and chemical use and corn yields and production values
are provided annually. However, the ARMS is only conducted every five years. The three most
recent years of the ARMS are 1991, 1996, and 2001. Table 6.1-3 lists corn yield and input data
for the three years of the ARMS study.101'102
180
-------
Table 6.1-3. Farm Energy Use and Input Data per Acre
Input
Seed
Fertilizer:
- Nitrogen
- Potash
- Phosphate
- Lime
Energy:
- Diesel
- Gasoline
-LPG
- Electricity
- Natural Gas
Chemicals
Units
Kernels/acre
pounds/acre
pounds/acre
pounds/acre
pounds/acre
Gallons/acre
Gallons/acre
Gallons/acre
kWh/acre
Cubic ft/acre
c
9-State Weighted Average Values
1991
25,571
126.07
53.26
59.12
246.39a
6.75
3.46
3.52
31.92
255.72
22.90
1996
25,577
131.06
60.28
48.62
15.71
8.65
3.06
6.59
78.97b
206.59
26.37
2001
28,882
133.78
79.03
57.72
15.67
5.77
1.63
4.80
36.38
192.05
2.74
a Historic lime use data is highly uncertain. We currently do not include lime use data in the lifecycle
modeling. We are examining this issue through work with ANL and may include lime use as part of the
analysis for the final rule.
bHigh energy use in the 1996 survey is due to increased corn drying requirements. See the discussion after
Table 6. 1-5.
c Chemicals data for 1991 and 1996 are in terms of dol./acre, data for 2001 is in terms of pounds/acre
Although USDA corn data is available for every state that produces corn, the data
documented in Table 6.1-3 is for nine major corn producing States: Illinois, Indiana, Iowa,
Minnesota, Nebraska, Ohio, Michigan, South Dakota, and Wisconsin. In 2005, these nine States
accounted for 80 percent of U.S. corn production. In 2001 these nine States represented 92
percent of U.S. ethanol production, and are projected to represent 82 percent of ethanol
production in 2012. The data in Table 6.1-3 are weighted based on corn production data for each
of the nine States from the NASS (three year average corn production data is used for
weighting).
Values in Table 6.1-3 on energy use and material inputs are divided by corn yield in
bu/acre to get corn input parameters on a per bushel basis to compare to GREET default values.
The energy use values listed in Table 6.1-3 were converted to Btu based on the lower heating
values of the fuels as listed in the GREET model.
Due to the annual variability in corn yield, as shown in Figure 6.1-2, we used a three year
average for corn yield instead of the average yield for the survey year.
weighted average corn yields used.
103
Table 6.1-4 shows the
181
-------
Table 6.1-4. Corn Yields (Bu/acre)
9-State Weighted Average Values
1991
(1990- 1992 Yield)
123.47
1996
(1995-1997 Yield)
124.62
2001
(2000-2002 Yield)
139.94
We adjusted corn yield data to account for seed corn energy use. We assumed that
growing seed corn requires 1.5 times the energy and material inputs to grow than corn104.
Therefore, seed corn input is converted to bushels (80,000 kernels/bu) and multiplied by 1.5.
This value is then subtracted from the yield data to get an adjusted yield per acre. This
effectively reduces corn yields by about 0.5 bu/acre for each of the three years. Some recent
studies have suggested that the energy use required for seed corn production is actually 4.7 times
that of corn.105'106 However this is not thought to have a significant impact on results as it
represents a decrease of only about 1 percent in corn yields, and the 1.5 values was used in this
analysis.
Table 6.1-5 shows corn farming energy use and material inputs on a per bushel basis.
Table 6.1-5. Farm Energy Use and Input Data per Bushel
Input
Total Energy Use
Diesel Fuel Use
Gasoline Use
Natural Gas Use
LPGUse
Purchased Elec.
Nitrogen
Phosphate
Potash
Chemicals
Unit
Btu/bu
%
%
%
%
%
g/bu
g/bu
g/bu
a
9-State Weighted Average Values
1991
15,674
45.0%
20.8%
13.0%
15.5%
5.6%
465
218
196
0.19
1996
20,124
44.5%
14.2%
8.1%
22.4%
10.8%
479
178
220
0.21
2001
11,846
44.9%
11.5%
11.4%
24.7%
7.5%
435
188
257
8.93
a Chemicals data for 1991 and 1996 are in terms of dol./bu, data for 2001 is in terms of g/bu
It can be seen from Table 6.1-5 that there is substantial variation in the three years of
survey data, especially on energy use. Several factors can influence corn farming energy use.
For example, it was reported that 1991 was a dry year, lowering the moisture content of the corn
crop and thus requiring less energy to dry the corn, whereas the 1996 crop was reported to have a
higher moisture content and thus require more energy to dry resulting in the high energy use
values for 1996. Farm diesel use is also dependent on tillage type and soil conditions, wetter soil
requiring more diesel use, and decreased tillage requirements (e.g., no till) reducing diesel use.107
Comparing the average of the three years of data in Table 6.1-5 with GREET default
values for farm energy input show that GREET energy use is about 40 percent more than the
182
-------
survey average. However, while the GREET energy use default value is higher than the average
of the historic data there is significant variation in the historic data evaluated ranging from
11,846 to 20,124 Btu/bu of corn produced vs. the GREET default of 22,500 Btu/bu. There is
also some uncertainty in the historic values evaluated. The USDA survey data includes a farm
input category of custom work (in $/acre). This custom work includes dollars spent on
contracted energy use for corn drying, or associated with planting, fertilizing, or harvesting.
Depending on the amount of custom work done, this will lead to an increase in corn farming
energy input. The potential increase in seed corn energy discussed previously will also increase
corn farming energy input. Considering this added energy use, we conclude that the GREET
default value falls within the range of historic data available and represents a conservative case
for farm energy input. Therefore, due to the uncertainty and variability in the historic data and
lack of projections for future energy use, the current GREET default value for energy use was
used in the RFS analysis. We will examine this issue as part of the analysis for the final rule.
Comparing the average of the three years of data in Table 6.1-5 with GREET default
values for farm material input show that GREET nitrogen fertilizer use is similar, phosphate and
potash fertilizer use are about 15 and 10 percent lower in GREET respectively, and total
chemical use is about 2 percent lower in GREET. GREET default nitrogen and potash fertilizer
use fall within the range of historic values while phosphate use is slightly below the range of
historic values. Of all the material inputs, nitrogen fertilizer is the most critical both in terms of
upstream energy use to produce the fertilizer and the on-farm N2O emissions associated with
nitrogen fertilizer use. As GREET default nitrogen and potash fertilizer use fall within the range
of historic data, and due to uncertainty and variability in the data, we concluded that the current
GREET default values are a good representation of farm material inputs and were used for the
RFS analysis. We also concluded that the GREET default input for chemical use was an
accurate representation based on the historic data available and the default was used in the RFS
analysis.
The GREET default value for phosphate input is slightly lower than the range of historic
data available, 165 g/bu in GREET vs. 178-218 g/bu for the historic values. However, due to
variability in the historic data and lack of projections for future usage, there is no clear better
value to use for phosphate input, and the current GREET default value was used in the RFS
analysis. Lime data is currently not an input to the GREET model for corn farming. This is
something we are evaluating through the contract with ANL and may be available for the final
rule.
Another corn farming input included in the GREET model is a default factor for CC>2
emissions associated with land use change. The factor is based on the assumption that increased
corn demand for ethanol production will require currently idle crop / pastureland to be converted
into corn production. This land use conversion is assumed to result in net CC>2 emissions. The
GREET default factor for emissions associated with land use change of 195 g CCVbu corn was
used in this analysis. This value represents approximately 2 percent of corn farming GHG
emissions.
6.1.2.7 Ethanol Production Yield
183
-------
Several data sources have reported that 1 bushel of corn yields approximately 2.7 gallons
of ethanol108. However, the development of new enzymes, as well as other technology
developments, continue to increase the potential ethanol yield, for example, research done by
NREL109, it is projected that yields for dry milling could soon reach 2.85 gal/bu. The cost
modeling in Chapter 7 assumed a dry mill ethanol plant yield of 2.77 with 2% denaturant. The
GREET model input is based on pure ethanol, so to be consistent with the cost modeling of
Chapter 7, we used an ethanol yield of 2.71 gal/bu in our analysis.
6.1.2.8 Byproduct Allocation
As mentioned previously, there are a number of byproducts made during the production
of ethanol. In lifecycle analyses, the energy consumed and emissions generated by an ethanol
plant must be allocated not only to ethanol, but also to each of the by-products. There are a
number of methods that can be used to estimate by-product allocations. These include methods
based on the economic value of each by-product, or on energy usage, based on engineering
analysis of the actual processes related to each product. The method preferred by EPA is called
the displacement method. This method most accurately accounts for these by-products by
calculating the lifecycle emissions of the products that will be displaced by them. For example,
the DDGS produced by the ethanol plant is a replacement for corn and soybean animal feed. The
ethanol receives a credit for the lifecycle emissions of corn and soybean animal feed, since a
quantity of that feed type is no longer needed and is displaced by DDGS.
The displacement method for by-product allocation is the default for the GREET model.
EPA supports that approach and continues to use that method in this analysis. However, in
preparation for the Final Rule, EPA will evaluate the other by-product allocation methods to
determine the impact this assumption has on the overall results of the analysis.
6.2 Methodology
The results of the lifecycle modeling were used to determine the impacts of increased
renewable fuel use on overall U.S. transportation sector, and nationwide fossil energy and GHG
emissions. As described below, lifecycle reductions from renewable fuel use were compared to
sector wide inventories to show the overall impact of increased renewable fuel use. The GREET
model provides estimates on a national average, per fuel unit basis, such as the amount of fossil
fuel use per million Btus of ethanol produced, and the same for petroleum fuels. The model
could be used to generate estimates of absolute fossil fuel and emissions savings of replacing a
given amount of gasoline with ethanol. However, the model does not provide estimates of
energy consumed and emissions generated in total, such as the total amount of fossil fuel use in
the U.S. transportation sector in a given year. Therefore, we could not use GREET directly to
estimate the transportation sector or nationwide inventories needed for the analysis.
To be consistent between our modeling of savings and overall sector inventories, we used
GREET instead to generate comparisons between renewable fuels and the petroleum-based fuels
that they displace. These comparisons allowed us to develop displacement indexes which
represent the percent of lifecycle GHGs or fossil fuel reduced when a Btu of renewable fuel
184
-------
replaces a Btu of gasoline or diesel. In this way GREET was used to generate percent reductions
and not absolute values. These percent reductions or displacement values were then applied to
the same gasoline and diesel fuel inventories used to generate transportation sector and
nationwide inventories. This ensured that savings and sector wide inventories in terms of
absolute values were calculated in a consistent manner.
In order to estimate the impacts of increased use of renewable fuels on fossil energy and
greenhouse gases, we first determined how much gasoline and diesel would be replaced as a
result of this rule. We then combined lifecycle percent reductions from GREET with lifecycle
inventories and petroleum consumption values for gasoline and diesel fuel use to get the amounts
of fossil energy and greenhouse gases reduced. For example, to estimate the impact of corn-
ethanol use on GHGs, these factors were combined in the following way:
5GHG,corn ethanol
'in ethanol
xLC
gasoline
xDI
GHG,corn ethanol
where:
oHG,corn ethanol
:orn ethanol
H-'
Dl
GHG,corn ethanol
= Lifecycle GHG emission reduction over the reference case associated with
use of corn ethanol (million metric tons of GHG)
= Amount of gasoline replaced by corn ethanol on an energy basis (Btu)
= Lifecycle emissions associated with gasoline use (million metric tons of
GHG per Btu of gasoline)
= Displacement Index for GHGs and corn ethanol, representing the percent
reduction in gasoline lifecycle GHG emissions which occurs when a Btu
of gasoline is replaced by a Btu of corn ethanol
Variations of the above equation were also generated for impacts on all four endpoints of
interest (fossil fuel consumption, petroleum consumption emissions of CC>2, and emissions of
GHGs) as well as all three renewable fuels examined (corn-ethanol, cellulosic ethanol, and
biodiesel). These values are then compared to the total U.S. transportation sector and nationwide
inventories of fossil energy and greenhouse gases to get the overall impacts of the rule.
In this regard, the impact on overall transportation sector GHG emissions due to the
increased use of renewable fuels can be described mathematically as follows:
TSectoro/0jGHG = S,
oHG,corn ethanol
oHG,cell ethanol
oHG,biodiesel
TSectorGHG
where:
TSectoro/0;GHG =
Percent reduction in overall transportation sector GHG emissions resulting
from the use of renewable fuels (%)
185
-------
.com ethanoi = Lifecycle GHG emission reduction over the reference case associated with
use of corn ethanoi (million metric tons of GHG)
SGHG,ceii ethanoi = Lifecycle GHG emission reduction over the reference case associated with
use of cellulosic ethanoi (million metric tons of GHG)
SoHG.biodiesei = Lifecycle GHG emission reduction over the reference case associated with
use of biodiesel (million metric tons of GHG)
TSectorGHG = Overall transportation sector GHG emissions in 2012 (million metric tons
of GHG)
We used the same approach to estimate fossil energy, petroleum energy, and CC>2
reductions in the transportation sector. We also used the same approach to estimate nationwide
reductions.
Section 6.2.1 describes how we estimated the amount of gasoline and diesel fuel replaced
as a result of this rule. Section 6.2.2 describes the lifecycle emissions and energy associated with
gasoline and diesel fuel use. In Section 6.2.3 below, we outline how we generated displacement
indexes using GREET. Section 6.2.4 outlines how we developed the overall transportation
sector and nationwide fossil energy and greenhouse gas emissions.
6.2.1 Modeling Scenarios
In general, the volume fraction (R) represents the amount of conventional fuel no longer
consumed - that is, displaced - as a result of the use of the replacement renewable fuel. Thus R
represents the incremental amount of renewable fuel used under each of our renewable fuel
volume scenarios, in units of Btu. As mentioned in Section 6.1, we make the assumption that
vehicle energy efficiency will not be affected by the presence of renewable fuels (i.e., efficiency
of combusting one Btu of ethanoi is equal to the efficiency of combusting one Btu of gasoline).
Our analysis of the GHG and fossil fuel consumption impacts of renewable fuel use was
conducted using three volume scenarios. The first scenario was a reference case representing
2004 renewable fuel production levels, projected to 2012. This scenario provided the point of
comparison for the other two scenarios. The other two renewable fuel scenarios for 2012
represented the RFS program requirements and the volume projected by EIA. In both scenarios,
we assumed that the biodiesel production volume would be 0.3 billion gallons based on an EIA
projection, and that the cellulosic ethanoi production volume would be 0.25 billion gallons based
on the Energy Act's requirement that 250 million gallons of cellulosic ethanoi be produced
starting in the next year, 2013. The remaining renewable fuel volumes in each scenario would be
ethanoi made from corn. The total volumes for all three scenarios are shown in Table 6.2-1. For
the purposes of calculating the R values, we assumed the ethanoi volumes are 5% denatured, and
the volumes were converted to total Btu using the appropriate volumetric energy content values
(76,000 Btu/gal for ethanoi, and 118,000 Btu/gal for biodiesel).
186
-------
Table 6.2-1. Volume Scenarios in 2012 (billion gallons)
Corn-ethanol
Cellulosic ethanol
Biodiesel
Total volume
Reference Case
3.9
0.0
0.028
3.928
Required volume:
7.5 bill gal
6.95
0.25
0.3
7.5
Projected Volume:
9.9 bill gal
9.35
0.25
0.3
9.9
Since the impacts of increased renewable fuel use were measured relative to the 2012
reference case, the value of R actually represented the incremental amount of renewable fuel
between the reference case and each of the two other scenarios. The results are shown in Table
6.2-2. The results shown in Table 6.2-2 are direct reductions in fuel use and do not represent
lifecycle savings.
Table 6.2-2. Direct Conventional Fuel Replaced in 2012 (quadrillion Btu)
Gasoline Replaced by Corn-ethanol
Gasoline Replaced by Cellulosic ethanol
Diesel Fuel Replaced by Biodiesel
Total energy
Required
volume:
7.5 bill gal
0.220
0.018
0.032
0.270
Projected
Volume:
9.9 bill gal
0.394
0.018
0.032
0.444
6.2.2 Lifecycle Impacts of Conventional Fuel Use
In order to determine the lifecycle impact that increased renewable fuel volumes may
have on any particular endpoint (fossil fuel consumption or emissions of GHGs), we also needed
to know the conventional fuel inventory on a lifecycle basis. Since available sources of GHG
emissions are provided on a direct rather than a lifecycle basis, we converted these direct
emission and energy estimates into their lifecycle counterparts.
To do this, we used GREET to develop multiplicative factors for converting direct
(vehicle-based) emissions of GHGs, or direct (vehicle-based) consumption of petroleum, into
full lifecycle factors. GREET output was used to generate the conversion factors shown in Table
6.2-3.
187
-------
Table 6.2-3.
Direct (wheel only) Conversion Factors to
Well-to-Wheel (lifecycle) Emissions or Energy Use
Petroleum
Fossil fuel
GHG
C02
Gasoline
1.11
1.22
1.26
1.23
Diesel
1.10
1.21
1.25
1.21
The factors in Table 6.2-3 were applied to gasoline and diesel fuel inventories of
emissions or energy consumption at the consumer level (i.e. direct emissions or energy) to
convert them into alternative inventories representing full lifecycle contributions.
The direct petroleum energy for gasoline and diesel fuel is just the energy content of the
fuels used. Consistent with U.S. EPA National Inventory calculations110, we converted energy
use values for gasoline and diesel fuel to direct CC>2 emissions by multiplying by a carbon
content coefficient, a carbon oxidation factor, and converting the resulting carbon emissions into
CC>2. The CC>2 emissions were then scaled up by assuming a fraction increase to the CC>2
emissions to account for non-CC>2 GHGs (CH4 and N2O). The fraction increase was based on the
U.S. EPA National Inventory 2004 values for both CO2 and total GHG emissions. Table 6.2-4
shows the total lifecycle petroleum and GHG emissions associated with direct use of a Btu value
of gasoline or diesel fuel. These values represented factor LC in the equation described above.
Table 6.2-4.
Lifecycle Emissions and Energy (LC Values)
Petroleum (Btu/Btu)
Fossil fuel (Btu/Btu)
GHG (Tg-CO2-eq/QBtu)
CO2 (Tg-CCVQBtu)
Gasoline
1.11
1.22
99.4
94.2
Diesel
1.10
1.21
94.5
91.9
6.2.3 Displacement Indexes
In order to permit a quantitative evaluation of the degree to which a renewable fuel
reduces lifecycle fossil fuel consumption or GHG emissions, several metrics have been
developed. Three of the most prominent metrics are shown in Table 6.2-5
188
-------
Table 6.2-5. Metrics Used to Measure Lifecycle Impacts of Renewable Fuels
Metric
Net energy balance
Energy efficiency
Displacement index
Calculation
Renewable energy out - fossil energy in
Fossil energy in + renewable energy out
(or alternatively renewable energy out + fossil
% reduction in emissions or energy compared
that it replaces
energy in)
to the fuel
Although the net energy balance and energy efficiency metrics are used most often by
researchers in summarizing their lifecycle analyses, they can be misleading. For instance, a
negative net energy balance, or an energy efficiency of more than 1.0, is generally interpreted to
mean that lifecycle fossil fuel consumption negates the benefits of the "renewable" fuel.
However, because these metrics do not involve a direct comparison to the conventional gasoline
or diesel that the renewable fuel is replacing, even in these cases there may be an overall
reduction in fossil fuel use.
As an example, if 81,000 Btu of fossil fuels were required to make, transport, and store
one gallon of ethanol, then the energy efficiency would be calculated as follows:
Energy efficiency = 81,000 Btu/gal - 76,000 Btu/gal = 1.07
This result would imply that ethanol cannot be labeled "renewable," since one gallon of
ethanol contains less energy than was required to make that one gallon. However, the use of
ethanol may still reduce overall fossil fuel use even in this case. If, for example, 18,000 Btu of
fossil fuels were required to make one ethanol-equivalent gallon of gasoline (i.e. 76,000 Btu of
gasoline), then a total of 94,000 Btu of fossil fuel energy would be consumed whenever 76,000
Btu of gasoline energy was combusted in a conventional vehicle. Since 81,000 Btu is less than
94,000 Btu, the use of ethanol would result in less fossil fuel consumption than the use of
gasoline, even though the energy efficiency is greater than 1.0. The energy content of ethanol
(76,000 Btu) is not considered fossil energy and therefore not included in the comparison with
gasoline calculation above.
Because of this potential for the net energy balance and energy efficiency metrics to
provide misleading information, we have chosen to use the displacement index. The
displacement index provides the most direct measure of the impacts of replacing conventional
gasoline or diesel with a renewable fuel, and is also better suited to describing impacts of
renewable fuel use on fossil fuel consumption and GHGs.
The displacement index (DI) represents the percent reduction in GHG emissions or fossil
fuel energy brought about by the use of a renewable fuel in comparison to the conventional
gasoline or diesel that the renewable fuel replaces. The formula for calculating the displacement
index depends on which fuel is being displaced (i.e. gasoline or diesel), and which endpoint is of
interest (e.g. petroleum energy, GHG). For instance, when investigating the CC>2 impacts of
ethanol used in gasoline, the displacement index is calculated as follows:
189
-------
DIco2 = 1 - lifecycle CO? emitted for ethanol in g/Btu
lifecycle CC>2 emitted for gasoline in g/Btu
The units of g/Btu ensure that the comparison between the renewable fuel and the
conventional fuel is made on a common basis, and that differences in the volumetric energy
content of the fuels is taken into account. The denominator includes the CCh emitted through
combustion of the gasoline itself in addition to all the CO2 emitted during its manufacturer and
distribution. The numerator, in contrast, includes only the CC>2 emitted during the manufacturer
and distribution of ethanol, not the CC>2 emitted during combustion of the ethanol.
The combustion of biomass-based fuels, such as ethanol from corn and woody crops,
generates CC>2. However, in the long run the CC>2 emitted from biomass-based fuels combustion
does not increase atmospheric CC>2 concentrations, assuming the biogenic carbon emitted is
offset by the uptake of CC>2 resulting from the growth of new biomass111. As a result, CC>2
emissions from biomass-based fuels combustion are not included in their lifecycle emissions
results and are not used in the CC>2 displacement index calculations shown above. Net carbon
fluxes from changes in biogenic carbon reservoirs in wooded or crop lands are accounted for
separately in the GREET model.
When calculating the GHG displacement index, however, the CH4 and N2O emitted
during biomass-based fuels combustion are included in the numerator. Unlike CC>2 emissions,
the combustion of biomass-based fuels does result in net additions of CH4 and N2O to the
atmosphere. We assume that combustion CH4 and N2O emissions are not offset by carbon
uptake of renewable biomass production. As shown in Table 6.3-1, CH4 and N2O emissions
contribute to the total GHG impact. Therefore, combustion CH4 and N2O emissions are included
in the lifecycle GHG emissions results for biomass-based fuels and are used in the GHG
displacement index calculations.
Using GREET, we calculated the lifecycle values for energy consumed and GHGs
produced for corn-ethanol, cellulosic ethanol, and soybean-based biodiesel, as well as the
gasoline and diesel fuel that would be displaced. For both renewable and conventional fuels, we
summed the lifecycle results for both the feedstock and the fuel. The results are shown in Table
6.2-6.
190
-------
Table 6.2-6. Output from GREET Used to Develop Displacement Indexes
Units
Gasoline3
Corn
ethanol
Cellulosic
ethanolb
LS
Diesel
Biodiesel
Well-to-Pump
Fossil energy
Petroleum energy
CH4
N2O
C02
CO2-eq
Btu/mmBtu
Btu/mmBtu
g/mmBtu
g/mmBtu
g/mmBtu
g/mmBtu
224,136
107,292
107
0.29
17,893
20,435
732,712
85,202
114
53
52,894
71,204
49,440
80,389
4
30
-9,531
-686
207,011
98,649
105
0.28
16,629
19,134
629,122
169,688
86
8
40,719
45,011
End point combustion
Fossil energy
Petroleum energy
CO2 combustion0
CO2-eq combustiond
Btu/mmBtu
Btu/mmBtu
g/mmBtu
g/mmBtu
1,000,000
1,000,000
76,419
79,015
74,755
77,351
74,755
77,351
1,000,000
1,000,000
77,570
77,669
79,388
79,487
a Volume-weighted average of conventional gasoline (65%), RFG blendstock (25%), and CaRFG blendstock (10%).
Straight average of results for herbaceous and woody biomass.
° Taken from an OTAQ Fact Sheet
Based on assuming an increase over CO2 emissions, percent increase from the U.S. EPA National Inventory for
CO2 and GHG emissions from on-road sources.
We used the values from the table above to calculate the displacement indexes. The
results are shown in Table 6.2-7.
Table 6.2-7. Displacement Indexes Derived from GREET
DIpetroleum
Dlpossil Fuel
DIoHG
DIc02
Corn ethanol
92.3%
40.1%
25.8%
43.9%
Cellulosic ethanol
92.7%
96.0%
98.1%
110.1%
Biodiesel
84.6%
47.9%
53.4%
56.8%
The displacement indexes in this table represent the impact of replacing a Btu of gasoline
or diesel with a Btu of renewable fuel. Thus, for instance, for every Btu of gasoline which is
replaced by corn ethanol, the total lifecycle GHG emissions that would have been produced from
that Btu of gasoline would be reduced by 25.8 percent. For every Btu of diesel which is replaced
by biodiesel, the total lifecycle petroleum energy that would have been consumed as a result of
burning that Btu of diesel fuel would be reduced by 84.6 percent.
Note that our DI estimates for cellulosic ethanol assume that the ethanol in question was
in fact produced from a cellulosic feedstock, such as wood, corn stalks, or switchgrass.
However, the definition of cellulosic biomass ethanol given in the Energy Act also includes
ethanol made from non-cellulosic feedstocks if 90 percent of the process energy used to operate
the facility is derived from a renewable source. In the context of our cost analysis, we
191
-------
determined that this latter version of cellulosic ethanol is more likely to be produced to meet the
Act's requirement of a minimum of 250 million gallons beginning in 2013. Therefore, for cost
estimation purposes we have assumed that cellulosic ethanol would actually be made from corn,
but at a plant where 90 percent of the process energy has come from a renewable source. Further
discussion of this issue can be found in Section 1.2.2. However, for this analysis it was assumed
that cellulosic ethanol was in fact produced from cellulosic feedstockuu.
6.2.4 Transportation Sector and Nationwide Inventories
For our analysis described above, we need estimates of transportation sector and
nationwide fossil energy and GHG emissions to determine the percent reduction impacts of the
program (e.g., TSectorGHG factor in the equation above). These inventories are direct not
lifecycle and are needed for 2012 to compare to the projected renewable fuel savings in 2012.
6.2.4.1 Fossil Fuel Inventory
The transportation sector and nationwide fossil fuel inventory is just the energy content
of the fuels used. Fossil fuel use in the transportation sector includes gasoline and diesel as well
as other petroleum fuels, such as residual oil and LPG. It also includes other fossil energy use in
the form of natural gas and the fossil portion of electricity used. Inherent with the assumptions
on the amounts of renewable fuels use projected to 2012, there are also assumed values for
gasoline and diesel fuel use. Values for energy use of the different transportation fuels other
than gasoline and diesel (e.g., jet fuel, natural gas, etc.) were taken directly from the 2006
Annual Energy Outlook.
The nationwide fossil fuel inventory includes petroleum, natural gas, and coal energy use.
The direct fossil fuel inventory values are shown in Table 6.2-8.
Table 6.2-8.
Direct Fossil Fuel Inventories (QBtu)
Nationwide
Transportation Sector
2012
94.53
31.41
6.2.4.2 Petroleum Inventory
As with fossil energy, the transportation sector and nationwide petroleum inventory is
just the energy content of the fuels used. The transportation sector petroleum inventory includes
gasoline and diesel as well as other petroleum fuels, such as residual oil and LPG.
The nationwide petroleum inventory includes petroleum use in the transportation sector
as well as other sectors. The direct petroleum inventory values are shown in Table 6.2-9.
1111 There are indications that facilities producing ethanol from cellulosic feedstocks will be online by as early as
2007. For example, Xethanol Corporation announced recently that it plans to build a full-scale cellulosic ethanol
plant in Augusta, Georgia, by mid-2007.
192
-------
Table 6.2-9.
Direct Petroleum Inventories (QBtu)
Nationwide
Transportation Sector
2012
43.87
30.47
6.2.4.3
COi Inventories
We calculated direct CC>2 emissions for the transportation sector in 2012 by applying
carbon emissions factors to the projected amount of fuels used in those years.
Direct CC>2 emissions from the transportation sector as a whole are calculated in the same
way as direct gasoline and diesel emissions are calculated as described in Section 6.2.2. We
converted energy use values for transportation sector fuels to direct CC>2 emissions by
multiplying by a carbon content coefficient, a carbon oxidation factor, and converting the
resulting carbon emissions into CC>2. Emissions from electricity use in the transportation sector
(rail) are calculated based on the U.S. average mix of fossil fuels used to generate electricity.
Consistent with the EPA inventory report we made an adjustment to diesel fuel, jet fuel
and residual oil use to subtract out the emissions associated with bunker fuel. The AEO values
include the energy use of bunker fuels, but the emissions of these fuels are not considered part of
the U.S. transportation sector emissions. This adjustment was done by decreasing emissions of
diesel fuel, jet fuel, and residual oil by the portion of emissions associated with bunker fuels as
determined in the EPA inventory report.
Direct nationwide CC>2 emissions are also calculated in the same way applying factors for
all fossil fuels used as reported by the 2006 Annual Energy Outlook. This type of analysis
results in a small understatement of total Nationwide CC>2 emissions as it does not capture other
industrial sources of CC>2 emissions for example CC>2 emissions from calcinations of limestone in
the cement industry. However, there are no projections of these other emissions sources for
2012, and they are a relatively small part of total Nationwide CC>2 emissions, representing only
6% of total CC>2 emissions in 2004 according to the EPA National Inventory values. Therefore,
while impacts of increased renewable fuel use as a percent of nationwide CO2 emissions may be
slightly overestimated the impacts on results are not thought to be significant. The results of
direct CC>2 emission calculations are shown in Table 6.2-10.
Table 6.2-10.
CO2 Direct Inventories (Tg CO2)
Nationwide
Transportation Sector
2012
6,406
2,108
6.2.4.4
GHG Inventories
Projections for direct GHG emissions can not be calculated directly from the energy
projections as was done for CC>2. The approach to estimating CC>2 emissions from mobile
193
-------
combustion sources varies significantly from the approach to estimating non-CO2 GHG
emissions (CH4 and N2O emissions). While CO2 can be reasonably estimated by applying an
appropriate carbon content and fraction of carbon oxidized factor to the fuel quantity consumed,
CH4 and N2O emissions depend largely on the emissions control equipment used (e.g., type of
catalytic converter) and vehicle miles traveled. Emissions of these gases also vary with the
efficiency and vintage of the combustion technology, as well as maintenance and operational
practices. Due to this complexity, a much higher level of uncertainty exists in the estimation of
CH4 and N2O emissions from mobile combustion sources, compared to the estimation of CO2
emissions.
Projections for direct transportation sector and nationwide GHG emission are done by
assuming a fraction increase to the CO2 emissions to account for non-CO2 GHGs. The fraction
110
increase was based on the U. S. EPA National Inventory 2004 values for both CO2 and total
GHG emissions. This same increase is applied to 2012 CO2 values. Table 6.2-11 shows the
fraction increase values for GHGs over CO2 emissions calculated from the U.S. EPA National
Inventory report.
Table 6.2-11. U.S. National Inventory 2004 CO2 and GHG Inventories
Nationwide
Transportation Sector
C02 (Tg-C02)
5,988
1,860
GHG (Tg-CO2-eq.)
7,074
1,960
Fraction Increase
1.1807
1.0538
The results of direct GHG emission calculations are shown in Table 6.2-12.
Table 6.2-12.
GHG Direct Inventories (Tg CO2-eq.)
Nationwide
Transportation Sector
2012
7,564
2,222
6.3 Impacts of Increased Renewable Fuel Use
We used the methodology described above to calculate impacts of increased use of
renewable fuels on consumption of petroleum and fossil fuels and also emissions of CO2 and
GHGs. This section describes our results.
6.3.1 Fossil Fuels and Petroleum
We used the S equation in Section 6.2 to calculate the reduction associated with the
increased use of renewable fuels on lifecycle fossil fuel and petroleum consumption. These
values are then compared to the total U.S. transportation sector and nationwide inventories to get
a percent reduction. The results are presented in Tables 6.3-1 and 6.3-2.
194
-------
Table 6.3-1.
Fossil Fuel Impacts of Increased Use of Renewable Fuels in 2012,
In Comparison to the Reference Case
Reduction (quadrillion Btu)
Percent reduction in Transportation
Sector Energy Use
Percent reduction in Nationwide
Energy Use
Required volume:
7.5 bill gal
0.2
0.5 %
0.2%
Projected Volume:
9.9 bill gal
0.3
0.8 %
0.3%
Table 6.3-2.
Petroleum Impacts of Increased Use of Renewable Fuels in 2012,
In Comparison to the Reference Case
Reduction (billion gal)
Percent reduction in Transportation
Sector Energy Use
Percent reduction in Nationwide
Energy Use
Required volume:
7.5 bill gal
2.3
1.0%
0.7%
Projected Volume:
9.9 bill gal
3.9
1.6%
1.1%
6.3.2 Greenhouse Gases and Carbon Dioxide
One issue that has come to the forefront in the assessment of the environmental impacts
of transportation fuels relates to the effect that the use of such fuels could have on the emissions
of greenhouse gases (GHGs). The combustion of fossil fuels has been identified as a major
contributor to the increase in concentrations of atmospheric carbon dioxide (CO2) since the
beginning of the industrialized era, as well as the build-up of trace GHGs such as methane (CH4)
and nitrous oxide (N2O). This lifecycle analysis evaluates the impacts of renewable fuel use on
greenhouse gas emissions.
The relative global warming contribution of emissions of various greenhouse gases is
dependant on their radiative forcing, atmospheric lifetime, and other considerations. For
example, on a mass basis, the radiative forcing of CFLt is much higher than that of CC>2, but its
effective atmospheric residence time is much lower. The relative warming impacts of various
greenhouse gases, taking into account factors such as atmospheric lifetime and direct warming
effects, are reported on a 'CCVequivalenf basis as global warming potentials (GWPs). The
GWPs used by GREET were developed by the UN Intergovernmental Panel on Climate Change
113
(IPCC) as listed in their Third Assessment Report , and are shown in Table 6.3-3.
195
-------
Table 6.3-3.
Global Warming Potentials for Greenhouse Gases
Greenhouse Gas
C02
CH4
N2O
GWP
1
23
296
Greenhouse gases are measured in terms of CO2-equivalent emissions, which result from
multiplying the GWP for each of the three pollutants shown in the above table by the mass of
emission for each pollutant. The sum of impacts for CH4, N2O, and CC>2, yields the total
effective GHG impact.
We used the S equation in Section 6.2 to calculate the reduction associated with the
increased use of renewable fuels on lifecycle emissions of CC>2. These values are then compared
to the total U.S. transportation sector and nationwide emissions to get a percent reduction. The
results are presented in Table 6.3-4.
Table 6.3-4.
COi Emission Impacts of Increased Use of Renewable Fuels in 2012,
In Comparison to the Reference Case
Reduction (million metric tons CC^)
Percent reduction in Transportation
Sector Emissions
Percent reduction in Nationwide
Emissions
Required volume:
7.5 bill gal
12.6
0.6 %
0.2%
Projected Volume:
9.9 bill gal
19.8
0.9 %
0.3%
Carbon dioxide is a subset of GHGs, along with CH4 and N2O as discussed above. It can
be seen from Table 6.2-7 that the displacement index of CC>2 is greater than for GHGs for each
renewable fuel. This indicates that lifecycle emissions of CH4 and N2O are higher for renewable
fuels that for the conventional fuels replaced as shown in Table 6.2-6. Therefore, reductions
associated with the increased use of renewable fuels on lifecycle emissions of GHGs are lower
than the values for CC>2. The results for GHGs are presented in Table 6.3-5.
Table 6.3-5.
GHG Emission Impacts of Increased Use of Renewable Fuels in 2012,
In Comparison to the Reference Case
Reduction (million metric tons CO2-eq.)
Percent reduction in Transportation
Sector Emissions
Percent reduction in Nationwide
Emissions
Required volume:
7.5 bill gal
9.0
0.4 %
0.1%
Projected Volume:
9.9 bill gal
13.5
0.6 %
0.2%
196
-------
6.4 Implications of Reduced Imports of Petroleum Products
6.4.1 Impacts on Imports of Petroleum Products
To assess the impact of the RFS program on petroleum imports, the fraction of domestic
consumption derived from foreign sources was estimated using results from the AEO 2006. We
describe in this section how fuel producers will change their mix of imports in response to a
decrease in fuel demand.
We do not expect the projected reductions in petroleum consumption (0.3 to 0.57 Quads)
to impact world oil prices by a measurable amount. We base this assumption on the overall size
of worldwide petroleum demand and analysis of the AEO 2006 cases. Domestic and
international crude oil production, facing the world oil price, would also be expected to remain
unchanged relative to the reference case. If petroleum demand changes were much larger and
international refinery operations were impacted, the market economics might be expected to be
different. However, this is outside the scope of this assessment which focuses solely on the RFS
impacts.
The displacement of domestic crude oil production, imports of crude oil, and imports of
finished products will depend on the marginal costs of each source. In general, it is financially
preferable for domestic refineries to eliminate the most expensive marginal cost sources. The
highest cost sources tend to be finished product imports followed by crude oil imports.114
Refineries prefer to refine crude oil as opposed to importing finished products because of the
higher margins involved with the former and the greater utilization of refining capacity. Crude
oil, as an international commodity, will be purchased at the market price by refineries. Thus,
while crude oil from abroad may be produced more cheaply than domestic production sources,
refineries that purchase from either source will pay the international market price for that specific
grade of crude oil based on specific gravity and sulfur content plus the cost of transport to the
U.S. Thus, we expect the domestic crude production to be only marginally affected.
The Energy Information Administration (EIA) has modeled the effects of the RFS in the
current AEO 2006.115 In addition, the EIA has conducted three separate analyses of
Congressional bills which include earlier forms of the renewable fuel standard. These separate
analyses however were based on earlier AEO versions and, in some instances, considered
numerous provisions in addition to an RFS which collectively, affected world oil prices. Thus,
we did not directly use these earlier analyses, rather opting to use only the results in the AEO
2006 to assess the RFS impacts on imports. Comparison of the AEO 2006 reference case against
the low macroeconomic growth case allowed us to evaluate how a decrease in demand would
affect the mix of imported finished products, imported crude oil, and domestic production.
Similar to the assumptions above, the price of crude oil remains the same between the AEO low
macroeconomic growth and reference cases. Comparison of the two cases show that with an
initial decrease in demand for petroleum products (approximately 300,000 barrels per day), net
imports will account for approximately 95% of the reductions.116 Both reduced domestic crude
production and natural gas plant liquids account for most of the remainder. Since imported
finished products are the highest marginal cost sources, they account for all the initial reductions
197
-------
in petroleum imports. If demand is reduced even further (over 860,000 barrels per day),
approximately 50% of the reductions come from imported products, 44% from imported crude
oil, and the remainder from reduced domestic, natural gas plant liquid (NGL) production, and
exports.
Note that there is uncertainty in quantifying how refineries will change their mix of
sources with a decrease in petroleum demand, particularly at the levels estimated for the RFS.
Changes in world oil price from the reference case could also significantly alter the mix of
sources from which refineries choose. For example, a comparison between the AEO low price
case (as opposed to low macroeconomic growth case) and the reference case would yield a 50-50
split between product and crude imports. We believe that the actual refinery response could
range between these two points, so that finished product imports would compose between 50 to
100% of the net import reductions, with crude oil imports making up the remainder. For the
purposes of this RIA, we show values for the case where net import reductions come entirely
from imports of finished products, as discussed below.
By using the petroleum reduction levels as discussed in 6.3.2 of the RIA, and comparing
these to the AEO 2006 results, we estimate that 95% of the lifecycle petroleum reductions will
be met through reductions in net petroleum imports. Table 6.4-1 shows the reductions in net
petroleum imports estimated for the RFS program. We expect that these import reductions will
be met exclusively from finished petroleum products rather than from crude oil, for the economic
reasons given above and consistent with the results of the AEO 2006 low growth case. As an
example calculation, we apportioned 95% of the total reductions in gasoline and diesel to
displaced finished product imports. By 2012, imports of finished products are estimated to be
reduced by 145,000 and 241,000 barrels per day, respectively, for the 7.5 and 9.9 cases (note that
both these cases account for the 7.2 and 9.5 billion gallons of ethanol plus the additional 0.3
billion gallons of biodiesel, as discussed earlier in the RIA section 6.2.1). We compare these
reductions in imports against the AEO projected levels of net petroleum imports. The range of
reductions in net petroleum imports are estimated to be between 1 to 2%, as shown in Table 6.4-
2.
Table 6.4-1.
Reductions in Imports of Finished Products
(barrels per day)
Cases
7.5
9.9
2012
145,454
240,892
Table 6.4-2.
Percent Reductions in Petroleum Imports
Compared to AEO2006 Import Projections
Cases
7.5
9.9
2012
1.1%
1.7%
198
-------
One of the effects of increased use of renewable fuel is that it diversifies the energy
sources used in making transportation fuel. To the extent that diverse sources of fuel energy
reduce the dependence on any one source, the risks, both financial as well as strategic, of
potential disruption in supply or spike in cost of a particular energy source is reduced.
In order to understand the energy security implications of the RFS, EPA will work with
Oak Ridge National Laboratory (ORNL). As a first step, ORNL will update and apply the
method used in the 1997 report Oil Imports: An Assessment of Benefits and Costs., by Leiby,
Jones, Curlee and Lee.117 This paper was cited and its results utilized in previous DOT/NHTSA
rulemakings, including the 2006 Final Regulatory Impact Analysis of CAFE Reform for Light
Trucks.118 This method is consistent with that used in the Effectiveness and Impact of Corporate
Average Fuel Economy (CAFE) Standards Report conducted by the National Research
Council/National Academy of Sciences in 2002. Both reports estimate the marginal benefits to
society, in dollars per barrel, of reducing either imports or consumption. This "oil premium"
approach emphasizes identifying those energy-security related costs that are not reflected in the
market price of oil, and which maybe change in response to an incremental change in the level of
oil imports or consumption.
Since the 1997 publication of this report changes in oil market conditions, both current
and projected, suggest that the magnitude of the "oil premium" may have changed. Significant
factors that should be reconsidered include: oil prices, current and anticipated levels of OPEC
production, U.S. import levels, potential OPEC behavior and responses, and disruption
likelihoods. ORNL will apply the most recently available careful quantitative assessment of
disruption likelihoods, from the Stanford Energy Modeling Forum's 2005 workshop series, as
well as other assessments.119 ORNL will also revisit the issue of the macroeconomic
consequences of oil market disruptions and sustained higher oil prices. Using the "oil premium"
calculation methodology which combines short-run and long-run costs and benefits, and
accounting for uncertainly in the key driving factors, ORNL will provide an updated range of
estimates of the marginal energy security implications of displacing oil consumption with
renewable fuels. The results of this work effort are not available for this proposal but will be
part of the assessment of impacts of the RFS in the final rule.
6.4.2 Impacts on Import Expenditures
The reductions in petroleum imports were discussed in Section IX.D of the preamble. As
noted in the preamble, we calculate the decreased expenditures on petroleum imports assuming
this would not result in any other changes consumer behavior that would be reflected in fuel use.
All reductions in petroleum imports are expected to be from finished petroleum products rather
than crude oil, as discussed in the prior section. The economic savings in petroleum product
imports was calculated by multiplying the reductions in gasoline and diesel imports by their
corresponding price. According to the EIA, the price of imported finished products is the market
price minus domestic local transportation from refineries and minus taxes.120 Since no published
forecasts are available for this price, an estimate was made by using the AEO 2006 gasoline and
199
-------
distillate price forecasts and subtracting the average federal and state taxes based on historical
data.
vv
As an example calculation, the RFS is expected to yield a reduction of 2.23 billion
gallons of gasoline in the year 2012 (7.5 case). 95% of these reductions, or 2.12 billion gallons,
are expected to come from imports of finished gasoline. Thus, the domestic refining sector would
avoid purchases of 2.12 billion gallons of gasoline at $1.58 per gallon (2004$), the forecasted
price using the method described. The avoided payments abroad total $3.2 billion. Using a
similar approach for imported diesel, we estimate that an additional $0.3 billion is saved, for a
total of $3.5 billion saved for 2012, as shown in Table 6.4-3.
We compare these avoided petroleum import expenditures against the projected value of
total U.S. net exports of all goods and services economy-wide. Net exports is a measure of the
difference between the value of exports of goods and services by the U.S. and the value of U.S.
imports of goods and services from the rest of the world. For example, according to the AEO
2006, the value of total import expenditures of goods and services exceeds the value of U.S.
exports of goods and services to the rest of the world by $695 billion for 2006 (for a net export
level of minus $695 billion) and by $383 billion for 2012 (for a net export level of minus $383
billion).ww In Table 6.4-3, we compare the avoided expenditures in petroleum imports versus the
total value of U.S. net exports of goods and services for the whole economy for 2012. Relative to
the 2012 projection, the avoided petroleum expenditures due to the RFS would represent 0.9 to
1.5% of economy-wide net exports.
Table 6.4-3.
Avoided Petroleum Import Expenditures for 2012 ($2004 billion)
AEO2006 Total
Net Exports
-$383
RFS Cases
7.5
9.9
Avoided
Expenditures in
Petroleum Imports
$3.5
$5.8
Percent versus Total
Net Exports
0.9%
1.5%
The average taxes per gallon of gasoline and diesel have stayed relatively constant. For 2000-2006, gasoline taxes
were $0.44/gallon ($2004) while for 2002-2006, diesel taxes were $0.49/gallon. The average was taken from
available EIA data (http://tonto.eia.doe.gov/oog/info/gdu/gasdiesel.asp).
** For reference, the U.S. Bureau of Economic Analysis (BEA, http://www.bea.gov/) reports that the 2005 import
expenditures on energy-related petroleum products totaled $235.5 billion (2004$) while petroleum exports totaled
$13.6 billion - for a net of $221.9 billion in expenditures. Net petroleum expenditures made up a significant fraction
of the $591.3 billion current account deficit in goods and services for 2005 (2004$).
200
-------
Chapter 7: Estimated Costs of Renewable Fuels, Gasoline and
Diesel
7.1 Ethanol
This section provides a description of the analysis we conducted for estimating the cost of
corn cellulosic ethanol. Our analysis indicates that corn ethanol will cost $1.20 per gallon to
produce (2004 dollars) in 2012. We also estimated that using cellulosic feedstock, the
production costs for ethanol would be approximately $1.65 per gallon (2004 dollars). By 2012
this cellulosic cost may decline with breakthroughs and advances in technology. Based on
reports from a variety of sources and discussions we held with members of academia as well as
those directly involved in the industry, we believe several roadblocks remain to the production of
large volumes of cellulosic derived ethanol. It appears that good progress has been achieved in
the laboratory, but this information must be validated in pilot or demonstration type plants.
7.1.1 Corn Ethanol
Of the new ethanol production capacity expected to be built, according to Section 1.2.2 of
this DRIA, less than three percent combined is expected to be produced from cellulosic
feedstocks or in plants that differ significantly from dry mill corn ethanol plants. Several plants
will be able to utilize other starchy feedstocks besides corn, such as milo, barley, wheat, and
sorghum. However, corn is the primary feedstock, and therefore, the following analyses will
focus on dry mill starch ethanol production.
7.1.1.1 Engineering and Construction Requirements for Corn Ethanol Plants
To meet a goal of 7.2 billion gallons per year (Bgal/y) in 2012 from the mid-2006
capacity of 4.9 Bgal/yr, 2.3 Bgal/yr of additional capacity will have to be constructed.^ If we
consider that it is likely that at least 9.6 Bgal/yr of capacity will come on-line by 2012, the
annual capacity increase is 4.7 Bgal/yr. Our industry characterization work considering plants
that are either under construction or are planned to be constructed in the next 2-3 years suggests
average size will be 72 million gallons per year (MMgal/yr) for new plants, or 68 MMgal/yr if
expansions are included.
Based on conversations with representatives from design-build firms working in this
field, as well as material from public sources, each new plant requires design engineering work
lasting about six months followed by construction lasting 12-14 months before plant startup is
possible, resulting in a total project timeline of 18-20 months. The design phase for a basic 50
MMgal/yr plant is expected to require the attention of about 12 engineers full time, and the
construction phase will employ an average of about 125-150 workers each day.
For details on current and expected ethanol capacity, refer to Section 1.2 of this DRIA.
201
-------
These figures provide a basis for estimating the personnel requirements of the total
volume needed to meet the expected volumes. Over the six-year build-up period, a maximum of
1,200 construction workers and 45 engineers would be required on a monthly basis. For the case
of 9.6 Bgal/yr of ethanol capacity being online by the end of 2012, these numbers increase to
2,100 and 90, respectively. These numbers only include those involved with the final
construction and startup of the plants, and do not account for additional work required to design
and fabricate vessels, control systems, and other equipment that will be delivered to the
construction site.
These figures simply estimate the number of workers required at the final assembly stage
of the plant, and do not capture many more personnel hours that will go into designing and
constructing vessels, pipe fittings, control systems, and other pieces of equipment that will be
installed and brought online by the plant construction crews. A report produced by one
consultant suggested that expansion of the ethanol industry was responsible for more than 65,000
construction jobs in 2005.121
7.1.1.2 Corn Ethanol Production Costs
Corn ethanol costs for our work were estimated using a model developed by USD A,
documented in a peer-reviewed journal paper on cost modeling of the dry-grind corn ethanol
process.122 It produces results that compare well with cost information found in surveys of
existing pi ants.123
The USDA model is for a forty million gallon per year dry mill (40 MMgal/yr) corn plant
producing ethanol with a primary co-product of distillers dried grains with solubles (DDGS).
The ethanol yield used in the model is 2.77 gallons per bushel with 2.0% gasoline denaturant.
The model is based on work done in chemical process simulation software to generate equipment
sizes, stream flowrates, and material and energy balances. These results were then put together
with feedstock, energy, and equipment cost information in a spreadsheet format to arrive at a
final per-gallon cost estimate. Although the model is current in terms of technology, yields, and
capital estimates, we made some modifications to allow estimation of costs for ethanol plants of
different sizes and operating under different energy and feedstock prices.
We estimate an average corn ethanol production cost of $1.20 per gallon in 2012 (2004
dollars) in the case of 7.2 Bgal/yr and $1.26 per gallon in the case of 9.6 Bgal/yr. The cost of
ethanol production is most sensitive to the prices of corn and the primary co-product, DDGS.
Utilities, capital, and labor expenses also have an impact, although to a lesser extent. Corn
feedstock minus DDGS sale credits represents about 50% of the final per-gallon cost, while
utilities, capital and labor comprise about 20%, 10%, and 5%, respectively. For this work, we
used corn price projections from USDA of $2.23 per bushel in 2012 for the 7.2 Bgal/yr case, and
an adjusted value of $2.31 per bushel for the 9.6 Bgal/yr case. Prices used for DDGS were $65
per ton in the 7.2 Bgal/yr case and $55 per ton in the 9.6 case. Figure 7.1-1 shows the cost
breakdown for production of a gallon of ethanol.
202
-------
Figure 7.1-1. Cost Breakdown of Corn Ethanol Production (2004$).
-------
as one additional operator per shift and 10% additional electric utility use. These figures should
be considered conservative estimates, and were based on information from press releases as well
as a conversation with staff of company that designs and builds ethanol plants. Using this
information, the cost savings is about $0.11 per gallon of ethanol for a coal-fired plant.
Under the Energy Act, starch ethanol can be counted as cellulosic if at least 90% of the
process energy is derived from animal wastes or other waste materials.125 It is expected that the
250 million gallons per year of cellulosic ethanol production required by 2013 will be made
using this provision. While we have been unable to develop a detailed production cost estimate
for ethanol from corn which meets cellulosic criteria, we assume that the costs will not be
significantly different from conventionally produced corn ethanol. We believe this is reasonable
because to the extent that these processes are utilized, we expect them to be in locations the very
low or zero cost of the feedstock or biogas itself will likely offset the costs of hauling the
material and the additional capital for handling and firing it. In addition, because the quantity of
ethanol produced using these processes is still expected to be a relatively small fraction of the
total ethanol demand, the sensitivity of the overall analysis to this assumption is also very small.
In general, energy prices used in the model were taken from historical EIA data for
2004126, and scaled according the ratios of 2004-2012 price forecasts published in the Annual
Energy Outlook 2006.127 The prices used in the modeling are shown in Table 7.1-1.
Table 7.1-1. Energy Prices Used for Ethanol Cost Modeling for 2012 (2004$)
Natural Gasa
$/MMBtu
6.16
Coaf
$/MMBtu
1.94
Electricitya
$/kWh
0.044
Gasolineb
$/gal
1.25
a Historical data based on averages for Iowa, Illinois, Minnesota, and Nebraska
bPADD 2 bulk gasoline prices, excluding taxes
To arrive at the corn price for the two volume cases in 2012, the nominal 2012 USD A
price was adjusted to 2004 dollars according to the GDP deflators given in the 2006 FAPRI
outlook report.128'129 This number represents corn price for the 7.2 Bgal/yr case, as the RFS
volume was known at the time of USDA's most recent modeling work. This figure was then
adjusted for the 9.6 Bgal/yr case by adding 7.7 cents per bushel, determined by interpolating
between two nearby cases in the EIA NEMS ethanol cost model. 13° That model generated cost
curves for corn and DDGS based on data from a 2005 report by FAPRI examining the effects of
different RFS ethanol volume scenarios on agricultural markets.131 Since USDA does not
project DDGS prices in its outlook report, those figures were taken from the FAPRI report, and
then adjusted using the same methodology as the corn prices. While we believe the use of
USDA and FAPRI estimates for corn and DDGS prices is reasonable, additional modeling work
is being done for the final rulemaking using the Forestry and Agricultural Sector Optimization
Model described further in Chapter 8 of this DRIA.
7.1.2 Cellulosic Ethanol
7.1.2.1 How Ethanol is Made from Cellulosic Feedstocks
204
-------
It is not clear when the first processes to produce ethanol from cellulosic biomass were
discovered. While ethanol produced from starch can be traced historically to ancient times,
cellulosic derived ethanol appears to have been investigated in the 1800's. Until recently, the
demand for fuel ethanol has been somewhat limited, and not sufficient to support a cost-
competitive, commercial process to convert cellulose into ethanol.
With the increasing demand for fuel ethanol, significant progress in making ethanol
from cellulosic feedstocks has been made during the past few years. Interest in ethanol has
continued to grow, initially fostered in part from EPA's reformulated gasoline (RFG)
regulations that required such gasoline to contain a minimum of 2 percent oxygen by weight in
the fuel. This minimum oxygen requirement has recently been revoked by EPA in response to
the Energy Act, which revised the Clean Air Act requirement for oxygen in RFG. The
Renewable Fuel Standard (RFS) continues to create a demand for ethanol. Likewise, there is
an increased incentive to produce cellulosic ethanol because the Energy Act mandates that,
starting in 2013, renewable fuels in gasoline must contain a minimum of 250 million gallons of
cellulosic derived ethanol.
To make ethanol from cellulosic feedstocks, pretreatment is necessary to hydrolyze
cellulosic and hemicellulosic polymers and break down the lignin sheath. In so doing, the
structure of the cellulosic feedstock is opened to allow efficient and effective enzyme hydrolysis
of the cellulose/hemicellulose to glucose and xylose. The central problem is that the p-linked
saccharide polymers in the cellulose/hemicellulose structure prevent the microbial fermentation
reaction. By comparison, the fermentation of the starch produced from corn kernels, which have
a-linked saccharide polymers, takes place much more readily. An acid hydrolysis process was
developed to pretreat cellulosic feedstocks (through hydrolysis which breaks up the p-links), but
it continues to be prohibitively expensive for producing ethanol.
Technologies that are currently being developed may solve some of the problems
associated with producing cellulosic ethanol. Specifically, one problem with cellulosic
feedstocks is that the hydrolysis reactions produce both glucose, a six-carbon sugar, and xylose,
a five-carbon sugar (pentose sugar, CsHioOs; sometimes called "wood sugar"). Early
conversion technology required different microbes to ferment each sugar. Recent research has
developed better cellulose hydrolysis enzymes and ethanol-fermenting organisms.YY Now,
glucose and xylose can be co-fermented—hence, the terminology, weak-acid enzymatic
hydrolysis and co-fermentation. In addition, at least one group is researching the use of recently
developed genome modifying technology to produce a variety of new or modified enzymes and
microbes that show promise for use in a process known as weak-acid, enzymatic-prehydrolysis
zz
YY "Purdue yeast makes ethanol from agricultural waste more effectively." Purdue News, June 28, 2004; Writer:
Emil Venere, (765) 494-4709, venere@purdue.edu; source: Dr. Nancy Ho, (765) 494-7046,
nwyho@ecn.purdue.edu.
zz DOE Genomics: GTL Roadmap, Systems Biology for Energy and Environment, U.S. Department of Energy,
Office of Science, Office of Biological and Environmental Research, Office of Advanced Scientific Computing
205
-------
7.1.2.2 Difficulties in Estimating Capital and Operating Costs for New or Pioneer
Process Plants
Many years ago the petroleum and chemical process industries learned that it can be
financially problematic to scale-up a bench or laboratory scale process to a full commercial sized
operation. There are simply too many process variables that act one way at a batch rate of one-
or two-gallons per day, or even 100-gallons per day, but then act a completely different way in a
continuous, 70,000 gallon per day operation. Under these, admittedly somewhat extreme
expectations, there is also absolutely no reasonable way to optimize a process. We expect that at
least pilot or demonstration size projects will be necessary before a fully commercial sized,
reasonably optimized plant can be constructed.
The petroleum and chemical process industries have also learned that if a different
feedstock, with similar, but at the same time quite different characteristics, becomes available, it
is nearly always necessary to make several pilot plant runs before the feedstock in introduced
into the process. There are a wide variety of potential cellulosic feedstocks, such as switch grass,
forest thinnings, municipal waste, wood chips, and corn stover (corn stalks). The physical
characteristics of these materials, such as size, composition, and density vary widely. As a
result, there could be significant differences in the process configurations required to convert
each of them into ethanol. Compositional and density variations may require different reactor
residence times for each feedstock, which will impact throughput. Many of the process streams
will actually be slurries of the feedstock. It is also quite likely that each slurry stream will have
its own flow and compositional characteristics. The flow characteristics of any slurry, under real
operating conditions, must be well understood in order to properly design an optimum system.
Additionally, valve and pump types, sizes, and materials of construction, as well as line sizes and
configurations, may vary. Apart from the various process issues, questions also remain
regarding which of the feedstocks is actually the best in terms of ethanol yield per dollar.
Consequently, we believe a good deal more process data is necessary before a reasonably
accurate cost to design, engineer, and build a commercial scale cellulosic based ethanol plant can
be expected. At the present time, there is only one cellulosic ethanol plant in North America
(logen132 a privately held company, based in Ottawa, Ontario, Canada). As far as we know, the
technology that logen employs is not yet fully developed or optimized. Consequently, there is no
proven process design or operating data which could be used to estimate how much it will cost to
produce cellulosic ethanol.
Generally, the industry seems to be moving toward a process that uses dilute acid
enzymatic prehydrolysis with simultaneous saccharification (enzymatic) and co-fermentation.
The model we used incorporates this type of process to estimate the cost of producing ethanol
from corn stover. We chose corn stover because it is ubiquitous and because of the likelihood it
will eventually be used as a feedstock.
Research, Germantown, MD 20874-1290, August 2005; DOEGenomesToLife.org/roadmap: downloadable as whole
or in sections.
206
-------
In 1999, the National Renewable Energy Laboratory (NREL) published a report outlining
its work with the USDA to design a computer model of a plant to produce ethanol from hard-
wood chips.133 Although the cellulosic model was originally prepared for hardwood chips, it
was meant to serve as a modifiable-platform for ongoing research using cellulosic biomass as
feedstock to produce ethanol. Their long-term plan was that various indices, costs, technologies,
and other factors would be regularly updated.
NREL modified the model in order to compare the cost of using corn-grain with the cost
of using corn stover to produce ethanol. We used the corn stover model from the second
NREL/USDA study for this analysis. Because there are no operating plants that could
potentially provide real world process design, construction, and operating data for processing
cellulosic ethanol, NREL had originally considered modeling the plant based on assumptions
associated with a pioneer plant. Such assumptions would likely result in costs significantly
higher than corn ethanol plants due to the higher level of uncertainty in both the design and
engineering as well as the final construction and operating costs. The literature indicates that
such models often underestimate actual costs since the high performance assumed for pioneer
process plants is generally unrealistic.
The NREL analysis assumed that the corn stover plant was an Nth generation plant, built
after the industry had been well enough established to provide verified costs. The corn stover
plant was normalized to the corn kernel plant, e.g., placed on a similar basis. Additional costs
for risk financing, longer start-ups, and other costs associated with first-of-a-kind or pioneer
plants were not included in the study.134 It is also reasonable to expect the cost of cellulosic
ethanol will be higher than corn ethanol because of the complexity of the cellulose conversion
process. During the recent past, process improvements and other advancements in corn
production have considerably reduced the cost of producing corn ethanol. We also believe it is
realistic to assume that cellulose-derived ethanol process improvements will be made and that
one can likewise reasonably expect that as the industry matures, the cost of producing ethanol
from cellulose will also decrease.
7.1.2.3 Methods, Data Sources, and Assumptions
For our analysis, we used the spreadsheet model that NREL developed for its comparison
of the costs of producing ethanol from corn grain and corn stover.AAA The NREL model used
the Aspen Plus™ process simulator to calculate the flows and the heat and material balances for
the process. We decided to use the NREL spreadsheet corn stover model, as is, since we did not
have access to the Aspen Plus™ model nor to all the input. Rather, we left the feedrate, yields,
AAA The first, woodchip-plant study was designed to produce 52.2 million gallons of ethanol per year from about
2,200 tons per day (350 operating days per year; 15 days for downtime, including turn-around) of woodchips. The
second study normalized the original woodchip plant into the corn stover plant to produce 25 million gallons of
ethanol per year (about 1,235 wet tons per day), in 1999 dollars. The adjustments included feedrate and feedstock
volume and cost adjustments; equipment sizes with adjustments to capital and installation costs, and the cost of
capital, labor, and process chemicals, including denaturant.
207
-------
and streams flows as they were, but adjusted equipment capital and installation costs, and utility,
chemical, and labor costs to 2004 dollars. We used the same indices used by NREL to update
their corn stover study; however, we used actual costs and indices for 2004 where possible. For
example, in their 2000 calculations, NREL had extrapolated the Chemical Engineering Plant
Cost Index and the Chemical Cost Index through 2012. However, we used actual 2004 data
rather than the extrapolated data.
We did not change the corn stover cost. Several issues remain to be settled regarding the
amount of stover that should be left in place and how it should be gathered, baled, and shipped.
We found cost ranges of from $25 per dry ton to $45 per dry ton. For purposes of this analysis
we used the $35 per dry ton that NREL assumed in its analysis.
For the analysis, we calculated the annual production cost in dollars per gallon of fuel
ethanol. The annual production cost includes equipment straight-line depreciation for the life of
the plant (10 years), and variable costs, labor, supplies and overhead, minus any by-product
credits. Gasoline for denaturant and diesel for bulldozers to move the stover were projected into
2012 prices using lEA's AEO 2006135 report. The market selling price minus the annual
production cost is the before-tax profit. We calculated variable operating costs using NREL's
best estimate of quantities of chemicals and additives based on their laboratory work. NREL
calculated fixed costs using industry standards for percentages of direct labor (indirect labor was
40% of direct labor and overhead was 60% of total labor); other operating supplies, insurance,
etc. totaled 3.25% of total installed cost. According to the analysis three major cost categories
made up the majority of the total production cost: feeds stock- 31.2%; fixed costs - 23.8%; and
depreciation (reflects installed capital cost of equipment) - 33.8%.
As previously stated, there remain several feedstock issues to be settled, not least being
which of the many available types will be the best or most efficient. We chose an average cost
of $35 per dry ton; we don't believe the cost will rise and there is reason to expect it to come
down a little, as a result of the research that is currently under way. On the other hand, several
researchers have indicated switch grass may be better than corn stover; others point to forest
wastes, etc. In the end, the best feedstock will likely be the one that is readily available and close
to the plant; gathering, baling, and hauling continue to be important issues that will definitely
impact the viability of a feedstock. Equipment cost reductions may have a significant impact on
future costs. For example, there appear to be reasons to expect significant savings from
purchasing enzymes rather than growing them onsite.
7.1.2.4 Results and Discussion
Given the limitations we've already discussed, and perhaps others, we determined that it
would have cost approximately $1.65 in 2012 (2004 dollars) to produce a gallon of ethanol using
corn stover as a cellulosic feedstock.
The provisions offering grants and shared financing included in Title XV of the Energy
Act136 will likely encourage process development work to generate the necessary construction
208
-------
and operating cost estimates. We assume the results produced by the above referenced NREL
study are accurate and reasonable given the state of our current knowledge. We solicit comment,
however, on whether and to what extent we should build in projections for lower costs due to
availability of grants, developments in enzymatic hydrolysis and improvements in process plant
engineering and design.
7.1.4 Ethanol's Blending Cost
Ethanol has a high octane value of 115 (R+M)/2 which contributes to its value as a
gasoline blendstock. As the volume of ethanol blended into gasoline increases from 2004 to
2012, refiners will account for the octane provided by ethanol when they plan their gasoline
production. This additional octane would allow them to back off of their octane production from
their other gasoline producing units resulting in a cost savings to the refinery. For this cost
analysis, the cost savings is expressed as a cost credit to ethanol added to the production cost for
producing ethanol.
We obtained gasoline blending costs on a PADD basis for octane from a consultant who
conducted a cost analysis for a renewable fuels program using an LP refinery cost model.137 LP
refinery models value the cost of octane based on the octane producing capacity for the
refinery's existing units, by added capital and operating costs for new octane producing capacity,
and based on purchased gasoline blendstocks. We used this projected octane values for ethanol
and the other gasoline blendstocks discussed below. The value of octane is expressed as a per-
gallon cost per octane value is summarized in Table 7.1-2.
Table 7.1-2.
Octane Value of Ethanol and Other Gasoline Blendstocks (cents/octane-gallon)
PADD1
PADD 2
PADD 3
PADD 4
PADD 5
CA
0.71
0.38
0.67
0.86
0.86
1.43
Octane is more costly in California because the Phase 3 RFG standards restriction
aromatics content which also reduces the use of a gasoline blendstock named reformate - a
relatively cheap source of octane. Also, California's Phase 3 RFG distillation restrictions tend to
limit the volume of eight carbon alkylate, another lower cost and moderately high octane
blendstock.
Another blending factor for ethanol is its energy content. Ethanol contains less heat
content per gallon than gasoline. Since refiners blend up their gasoline based on volume, they do
not consider the energy content of its gasoline, only its price. Instead, the consumer pay's for a
gasoline's energy density based on the distance that the consumer can achieve on a gallon of
gasoline. Since we try to capture all the costs of using ethanol, we consider this effect. Ethanol
contains 76,000 British Thermal Units (BTU) per gallon which is significantly lower than
gasoline, which contains an average of 115,000 BTUs per gallon. This lower energy density is
accounted for below in the discussion of the gasoline costs.
209
-------
7.2 Biodiesel and Renewable Diesel Production Costs
7.2.1 Overview of Analysis
We based our cost to produce biodiesel fuel on a range estimated from the use of
USDA's and NREL's biodiesel computer models. Both of these models represent the continuous
transesterification process for converting vegetable soy oil to esters, along with the ester
finishing processes and glycerol recovery. The models estimate biodiesel production costs using
prices for soy oil, methanol, chemicals and the byproduct glycerol. The models estimate the
capital, fixed and operating costs associated with the production of soy based biodiesel fuel,
considering utility, labor, land and any other process and operating requirements.
Each model is based on a medium sized biodiesel plant that was designed to process raw
degummed virgin soy oil as the feedstock, yielding 10 MM gallons per year of biodiesel fuel.
USDA estimated the equipment needs and operating requirements for their biodiesel plant
through the use of process simulation software. This software determines the biodiesel process
requirements based on the use of established engineering relationships, process operating
conditions and reagent needs. To substantiate the validity and accuracy of their model, USDA
solicited feedback from major biodiesel producers. Based on responses, they then made
adjustments to their model. The NREL model is also based on process simulation software,
though the results are adjusted to reflect NREL's modeling methods. The output for both models
were provided in spreadsheet format, though, USDA updated the prices for a plant in year 2005,
while the NREL's model is based on prices in year 2002.
The production costs are based on an average biodiesel plant located in the Midwest
using soy oil and methanol, which are catalyzed into esters and glycerol by use of sodium
hydroxide. Because local feedstock costs, distribution costs, and biodiesel plant type introduce
some variability into cost estimates, we believe that using an average plant to estimate
production costs provides a reasonable approach. Therefore, we simplified our analysis and used
costs based on an average plant and average feedstock prices since the total biodiesel volumes
forecasted are not large and represent a small fraction of the total projected renewable volumes.
The production costs are a based on a plant that makes 10 MM gallons per year of biodiesel fuel.
Production costs for yellow grease derived biodiesel, were estimated with the models, using
adjustments for feedstock costs.
The model is further modified to use input prices for the feedstocks, byproducts and
energy prices to thus reflect the effects of the fuels provisions in the Energy Act. Based on the
USDA model, for soy oil-derived biodiesel we estimate a production cost of $2.06 per gallon in
2004 and $1.89 per gal in 2012 (in 2004 dollars) For yellow grease derived biodiesel, USDA's
model estimates an average production cost of $1.19 per gallon in 2004 and $1.10 in 2012 (in
2004 dollars). In order to capture a range of production costs, we compared these cost
projections to those derived from the NREL biodiesel model. With the NREL model, we
estimate biodiesel production cost of $2.11 per gallon for soy oil feedstocks and $1.28 per gallon
210
-------
for yellow grease in 2012, which are slightly higher than the USDA results. We present details
on the use of both models later in this section.
With the current Biodiesel Blender Tax Credit Program, producers using virgin vegetable
oil stocks receive a one dollar per gallon tax subsidy while yellow grease producers receive 50
cents per gallon (c/gal), reducing the net production cost to a range of 89 to 111 c/gal for soy oil
fuel and 60 to 78 c/gal for yellow grease biodiesel in 2012. This compares favorably to the
projected wholesale diesel fuel prices of 138 c/gal in 2012, signifying that the economics for
biodiesel are positive under the effects of the blender credit program, though, the tax credit
program expires in 2008 if not extended. Congress may later elect to extend the blender credit
program, though, following the precedence used for extending the ethanol blending subsidies.
Additionally, the Small Biodiesel Blenders Tax credit program and state tax and credit programs
offer some additional subsidies and credits, though the benefits are modest in comparison to the
Blender's Tax credit.
7.2.2 Inputs to and Results of USDA's Model
We used USDA's biodiesel model as a source to generate an estimate for the cost to
produce biodiesel fuel. The model is in spreadsheet format with inputs in 2005 dollars, and
contains all of the capital and operating costs for a plant to produce 10 million gallons per year of
biodiesel fuel.
7.2.2.1 Feedstock Costs
Feedstock prices are the largest component in generating production costs for biodiesel
fuel. For soy oil prices, we used prices based on USDA's 2006 Outlook, which has forecasted
soy oil prices considering production of biodiesel under the Energy Act. USDA's Outlook, is a
national forecasting analysis, that models the effects of demand for farm products and farm
product prices for soy beans, soy bean oil, corn and other farm commodities. The 2006 Outlook
estimated soy oil prices considering the demand of soy oil derived biodiesel fuel at
approximately 160 MM gallons per year in 2006 and 312 MM gallons a year in 2007138. This is
in close proximity to EIA's soy oil derived biodiesel volume projection of 135 MM, 265 MM in
2006 and 2007, respectively. We therefore used the soy oil prices from USDA's Outlook to
determine biodiesel production costs. The USDA does not forecast yellow grease prices, so we
assumed that yellow grease feedstocks costs would maintain the same relative historical pricing
differential to virgin soy oil. In the past, some analysis has shown that yellow grease has sold for
about half the price of soy oil139. The resulting feedstock costs to make a gallon of biodiesel
under projected volumes for RFS are in Table 7.2-1.
211
-------
Table 7.2-1. Projected Prices of Feedstock" (2004 Dollars per Gallon)
Marketing Year
2004
2012
Soy Oil
1.71
1.56
Yellow Grease
0.86
0.78
a Production of biodiesel assumed to consume 7.42 Ibs of soy oil per gallon. USDA prices in 2012 are adjusted to
2004 dollars to account for inflation, using GDP index of 109.7 in year 2004 and 130.8 in year 2012.
7.2.2.2
Capital Costs
For capital costs we used USDA's total installed capital cost of $10.66 MM for a 10 MM
gallon per year plant. This estimate was determined by the USDA, using a detailed analysis to
generate costs for equipment needs, installation, land, engineering and construction work,
buildings, utility needs, contingencies, startup costs etc. The USDA model is based on 2005
dollars, so we adjusted the numbers to 2004 values using the GDP index. Per the USDA method,
the total installed capital costs on a per gallon basis was amortized on a 10 year straight line
depreciation rate using a facility dependent cost of 10 percent times the capital costs.
Maintenance charges, insurances and facility supply costs were also calculated as percentages of
the capital. The total of all of these are equal to 16 c/gal.
7.2.2.3
Operating Costs
The total operating expenses were 20 and 18 c/gal for a soy based biodiesel plant in 2004
and 2012, respectively. The operating cost included a 4 cent per gallon offset from sale of the
glycerol product at a price of 5 cents/lb. The operating costs include values for utilities, feed
reagents, manpower and were based on the USDA's model. The components of the operating
costs are discussed below.
7.2.2.4
Utility and Labor Costs
We estimated utility costs using energy requirements from USDA's model and adjusted
the inputs to match the energy and electricity prices for the Midwest, using prices from EIA's
AEO. The cost for steam was estimated using the price of natural gas. Each pound of steam was
produced from heating water, which required 810 British Thermal Units (BTUs) per pound of
steam. Additionally, the steam costs are estimated assuming that the BTU's to make steam are
increased by a factor of two, to account for steam distribution efficiency losses, treatment of
boiler water to prevent fouling, maintenance and other miscellaneous costs. The utility
requirements per gallon of biodiesel and energy prices are presented in Tables 7.2-2 and 7.2-3
212
-------
Table 7.2-2. Utility Requirements per Gallon Biodiesel3
Medium Pressure Steam, Ibs
Electricity, kWh
Cooling Tower Water, Ibs
4.0
0.10
96.1
' Utilities per USD A model from the production of biodiesel from soy oil.
lable 7.2-3. Midwest Energy Prices per Year (
Year
Electricity, $/kWh
Natural gas, $/MM BTU
2004
0.046
7.16
in 2004 $)
2012
0.044
6.16
Labor costs include the salaries and benefits for personnel to operate a biodiesel plant.
This was estimated in the USDA model, though the labor costs were in 2005 dollars, which we
adjusted to 2004 dollars using the GDP price index. The resulting labor costs are 5 c/gal of
produced biodiesel fuel.
7.2.2.5
Chemical Reagents
Another operating expense, the production of biodiesel also requires the use of chemicals
and chemical reagents, as these act as a catalyst in the transesterification process. Additionally,
methanol is required as it is the feedstock that is chemically combined with soy oil and yellow
grease during the transesterification process, yielding the biodiesel product. The amount of
chemicals and methanol required to make a gallon of biodiesel are listed in Table 7.2-4.
Table 7.2-4. Reagent Requirements
Reagent
Water
Hydrochloric acid
Methanol
Sodium Methoxide
Sodium Hydroxide
Annual Requirement,
(Ibs per gallon of biodiesel)
0.0323
0.0185
0.8006
0.0231
0.0031
For the prices of chemical reagents, we used prices that were supplied in USDA's 2005
model and adjusted them to 2004 dollars. Additionally, since we have no forecasting
mechanism we assumed that the chemical reagent prices remained unchanged in 2012.
213
-------
However, we estimated methanol prices, as the cost for this feedstock is a significant component
of the total operating costs. For our analysis, we generated values by use of a correlation that
calculates methanol's price as a function of the price of natural gas140. In 2004 and 2012, using
Midwest natural gas prices, we estimated methanol prices of 13.1 and 11.6 cents per pound,
respectively. All other chemical prices, we assumed were constant over time and are in Table
7.2-5.
Table 7.2-5. Reagent Prices (in 2004 $)
Reagent
Hydrochloric acid
Sodium Methoxide
Sodium Hydroxide
Prices, $/lb
0.167
1.358
0.273
7.2.2.6
Glycerol Byproduct
The feedstock cost credit for the glycerin by product in our modeling work was 5 cents
per pound, based on recent pricing trends, assuming that additional glycerol generated from
expansion of biodiesel production will continue to keep prices low. The model, like many
biodiesel plants produces a crude 80% glycerine stream, which is usually sold to glycerine
refiners for purification. In the past, crude glycerine has sold for around $0.15 / pound. Because
of the increase in biodiesel production around the world, however, the crude glycerine market
has become saturated and the price is now around $0.05 / pound. As more biodiesel capacity
comes on line, this price may very well drop further, though other markets for the use of glycerol
are likely to develop because glycerol is a platform chemical used throughout industry. We
assumed that the current glycerin pricing environment will continue in the future. For our cost
estimation, the byproduct glycerin was sold at 5 cents per pound, reflecting current saturated
market and low pricing conditions. The income from sale of the byproduct glycerin lowered
biodiesel production costs by 2 percent and 4 percent for soy oil and yellow grease derived
biodiesel fuel, respectively.
The total biodiesel production costs derived using the USDA's model are summarized in
Table 7.2-6
214
-------
Table 7.2-6.
Projected Production Costs for Biodiesel by Feedstock per Gallon
(2004 Dollars)
Marketing Year
2004
2012
Soy Oil
2.06
1.89
Yellow Grease
1.19
1.10
7.2.3 Inputs to and Results of NREL's Biodiesel Model
We used NREL's biodiesel model as another source to generate an estimate for the cost
to produce biodiesel fuel. The model is in spreadsheet format, and contains all of the capital and
operating costs for a plant to produce 10 million gallons per year of biodiesel fuel. To make the
results directly comparable to USDA's model, we used energy costs in the Midwest.
Based on the results of the NREL model, we estimate that the total production costs to
make soy oil derived biodiesel fuel are $2.28 and $2.11 per gallon for years 2004 and 2012,
respectively. This is 22 cents more per gallon than the estimate derived from USDA's model.
The production costs for biodiesel fuel produced from yellow grease are estimated at $1.38 and
$1.28 per gallon for years 2004 and 2012, respectively. The components that make up our NREL
estimate are discussed in the sections that follow.
7.2.3.1
Feedstock Costs
The feedstock costs increase because the NREL model assumes 7.87 pounds of soy oil
are required to make a gallon of biodiesel fuel. This is slightly higher than the pounds required by
the USDA model, though the difference may be due to each model being based on soy oils with
differing chemical structures, i.e. more esters, differing densities. The higher amount of soy oil
required by the NREL model raises the production costs for biodiesel by about 10 c/gal for
feedstock costs alone, versus the USDA model. Similar to the USDA modeling analysis, the
prices for yellow grease is assumed to be half the cost of soy oil feedstock. The feedstock costs
are summarized in Table7.2-7.
Table 7.2-7. Projected Prices of Feedstock (2004 Dollars per Gallon)
Marketing Year
2004
2012
Soy Oil
1.81
1.65
Yellow Grease
0.91
0.83
a Production consumes 7.87 Ibs of soy oil per gallon of biodiesel. USDA prices in 2012
are adjusted to 2004 dollars to account for inflation, using GDP index of 109.7 in year
2004 and 130.8 in year 2012.
215
-------
7.2.3.2
Capital Costs
The total capital cost in the NREL model account for all of the costs for building a plant,
including but not limited to the expenses for equipment, tanks, installation costs, engineering,
tanks, construction, land and site development, start up and permitting charges. These costs do
not account for expenses incurred from maintenance, insurance and taxes, however. The total
capital coats for a plant are $14.8 million in 2002 dollars, which we adjusted to 2004 dollars
using the GDP price index. The capital costs were amortized assuming a seven percent return on
investment, resulting in a cost of 17 c/gal. All of the economic factors used for amortizing the
capital costs are summarized in Table 7.2-8.
Table 7.2-8.
Economic Factors Used in Deriving the Capital Cost Amortization Factor
Depreciation
Life
10 Years
Economic
and Project
Life
15 Years
Federal and
State Tax
Rate
0%
Return on
Investment
(ROI)
7%
Resulting
Capital
Amortization
Factor
0.11
7.2.3.3
Operating Costs
The total operating costs are 31 and 30 c/gal for years 2004 and 2012, respectively.
These costs are not directly comparable to those from the USDA model, as fixed operating cost
are included in the operating costs for the NREL model, while the USDA model accounts for
fixed costs in the capital estimate. The operating cost for the NREL analysis includes items for
utilities, reagents, manpower, insurance, taxes, general administration and maintenance costs,
though do not account for capital costs. Additionally, the sale of the glycerol byproduct (80%
strength) generated income of 4 c/gal of produced biodiesel, using glycerol price of 5 cents per
pound. The cost associated with insurance, taxes, general administration and supplies incur a
cost of 2.4 c/gal of biodiesel. The remaining components of operating costs for the NREL
modeling analysis are discussed below.
7.2.3.4
Utility and Labor
The utility costs were estimated using the energy requirements in the NREL model
along with the same prices for energy, steam and electricity, as those used in our USDA analysis.
The utility requirements per gallon of biodiesel fuel are listed in Table 7.2-9
216
-------
Table 7.2-9. Utility Requirements per Gallon of Biodiesel
Natural Gas, SCF
Medium Pressure Steam, Ibs
Electricity, kWh
Cooling Tower Water, Ibs
2.0
3.2
0.1
8.3
The NREL model accounts for the salaries of 4 employees per shift to run and maintain
the plant. In addition to salaries for these personnel, the labor expenses also accounted for
employee fringe benefits and the cost for a plant supervisor. The resulting labor costs are 6 cents
for each gallon of biodiesel.
7.2.3.5
Chemical Reagents
The NREL model also requires the use of the same chemicals and chemical reagents that
are used in the USD A model. The amount of chemical reagents in the NREL model, however,
reflect the use of diluted hydrochloric acid (HC1) and sodium methoxide for the biodiesel
production process. Hydrochloric acid is listed as being at 33 percent strength, which we
assumed also applied to the strength of sodium methoxide, since the amount of HC1 in the model
is reflective of about one third the value of the USDA's model. For the chemical and reagents
prices, we used the same pricing values as those in our USDA modeling analysis. The resulting
total chemical and reagent costs on a per gallon basis are about 17 cents for each gallon of
biodiesel fuel produced. All of the required chemicals and reagents for the production of
biodiesel are presented on an undiluted basis in Table 7.2-10.
Table 7.2-10. Reagent Requirements
Reagent
Water
HCLa
Methanol
NAOCH3a
Sodium Hydroxide
Annual Requirement,
(Ibs per gallon of biodiesel)
3.4646
0.0098
0.6037
0.0338
0.1901
a HC1 is Hydrochloric acid, NAOCH3 is sodium methoxide.
The total biodiesel production costs derived from the NREL model are summarized in Table
7.2-11.
217
-------
Table 7.2-11. Projected Production Costs for Biodiesel by Feedstock per Gallon
(2004 dollars)
Marketing Year
2004
2012
Soy Oil
2.28
2.11
Yellow Grease
1.38
1.28
a Production consumes 7.87 Ibs of soy oil per gallon of biodiesel. USDA prices in 2012 are adjusted to
2004 dollars to account for inflation, using GDP index of 109.7 in year 2004 and 130.8 in year 2012.
7.2.4 Biodiesel Blending Credit Programs
There are numerous credit and incentive programs that encourage the blending of
biodiesel. These programs reimburse blenders and producers for adding biodiesel to transport
diesel fuel, which acts to lower the production costs and makes the production of biodiesel more
economically competitive with petroleum derived diesel fuel. There are several
federal/nationwide biodiesel credit programs that offer subsidies for blending or use of biodiesel
as a transport diesel fuel which are discussed below.
The Commodity Credit Commission Bio-energy Program is an existing program that
expires at the end of fiscal year 2006, though due to a funding shortfall the program will
terminate on July 31, 2006. This program was administered by the USDA and pays biodiesel
producers grants when the economics to produce biodiesel are poor. The stipend is determined
based on available funding and the volume of renewable fuel that can receive the credit. For
historical purposes, the payments in 2004 and 2005 averaged about 107 and 50 c/gal of fuel
produced, respectively. For the first half of 2006, the credit on a per gallon basis is reduced
further, as the payment is diluted by increased production volume of fuels available to receive
the credit.
The Energy Act extended the Biodiesel Blenders Tax Credit program to the end of year
2008. This program was created under the American Jobs Creation Act of 2004 which created an
excise tax credit that can be claimed by anyone who blends biodiesel into transport diesel fuel.
Under this program, blenders may claim a credit against the applicable federal motor fuels excise
tax for blends containing biodiesel. According to IRS guidelines, the credit may be claimed by
anyone who adds biodiesel into diesel fuel at a level greater than 0.1 percent in the final blend.
The full credit for biodiesel made from virgin vegetable oils and animal fats is $1.0 per gallon,
while biodiesel derived from recycled grease receives 50 cents/gallon. A blender with more
excise tax credits than taxes owed can receive a refund from the IRS. Additionally, under the
current program, imported biodiesel and fuel made from imported feedstocks can also receive the
credit.
The Income Tax Credit Alternative is a program that is also available. This program does
not require any blending of biodiesel, though it does offers allow a similar excise tax credit as in
the blenders tax credit program. The excise tax can only be taken against actual income,
however, which makes the program less economically attractive than the blenders' credit
program.
218
-------
The Energy Act also created the Small Biodiesel Blenders Tax credit program. Under this
program, a credit of 10 c/gal is available to small producers who make biodiesel fuel from virgin
vegetable oils. This stipend is limited to companies with annual production volumes less than 60
MM gallons per year, using the aggregated capacity from all production sites for an individual
company. The maximum payment per company is capped at $15 MM per year and the program
is set to expire at the end of year 2008.
In addition to the federal programs, there are state and local programs that offer state fuel
tax exemptions, tax credits, and incentives that are more modest.
7.3 Distribution Costs
7.3.1. Ethanol Distribution Costs
There are two components to the costs associated with distributing the volumes of
ethanol necessary to meet the requirements of the Renewable Fuels Standard (RFS): 1) the
capital cost of making the necessary upgrades to the fuel distribution infrastructure system, and
2) the ongoing additional freight costs associated with shipping ethanol to terminals. The most
comprehensive study of the infrastructure requirements for an expanded fuel ethanol industry
was conducted for the Department of Energy (DOE) in 2002 .141 That study provided the
foundation for our estimates of the capital costs associated with upgrading the distribution
infrastructure system as well as the freight costs to handle the increased volume of ethanol
needed in 2012. Distribution costs are evaluated here for the case where just enough ethanol is
used to meet the requirements of the RFS in 2012 (7.2 billion gallons per year) and for a market-
driven case where the volume of ethanol used is 9.6 billion gallons per year. The 2012 reference
case against which we are estimating the cost of distributing the additional volume of ethanol
projected for 2012 is 3.9 billion gallons.BBB
7.3.1.1 Capital Costs to Upgrade the Ethanol Distribution System
The 2002 DOE study examined two cases regarding the use of renewable fuels. The first
case assumed that 5.1 Bgal/yr of ethanol would be used in 2010, and the second case assumed
that 10 Bgal/yr of ethanol would be used in 2015. We interpolated between these two cases to
provide an estimate of the capital costs to support the use of 7.2 Bgal/yr of ethanol in 2012. The
10 Bgal/yr case from the DOE study was used to represent the market-based case examined in
today's rule of 9.6 Bgal/yr of ethanol. For both the 7.2 Bgal/yr and 9.6 Bgal/yr cases, we
adjusted the results from the DOE study to reflect a 3.9 Bgal/yr 2012 ethanol use baseline. The
following Table 7.3-1 contains our estimates of the infrastructure changes and associated capital
costs for the two ethanol use scenarios examined in today's rule.
BBB See Chapter 1 of this DRIA regarding the 2012 ethanol use reference case.
219
-------
Table 7.3-1. Ethanol Distribution Infrastructure Capital Costs
Relative to a 3.9 Billion Gallon per Year Reference Case
New Terminal Blending Systems for Ethanol
Number of terminals
Capital cost
New Ethanol Storage Tanks at Terminals
Number of tanks
Capacity
Capital cost
Terminal Storage Tanks Converted to Ethanol
Number of tanks
Capacity
Capital cost
Terminals Using Ethanol for the First Timea
Number of terminals
Capital cost
New Rail Delivery Facilities at Terminals
Number of terminals
Capital cost
Retail Facilities Using Ethanol for First Timea
Number of retail facilities
Capital cost
New Tractor Trailer Transport Trucks
Number of Trucks
Capital Costs
New Barges
Number of new barges
Capital cost
New Rail Cars
Number of new rail cares
Capital cost
Total Capital Costs
Capital Costs Attributed to Terminal and Retail
(i.e. fixed) Facilities
Capital Costs Attributed to Mobile Facilities
(tank trucks, rail cars, & barges)
7.2 billion gallons
per year
287
$96,924,000
200
1,826,000 barrels
$26,208,000
50
362,000 barrels
$1,060,000
250
$5,005,000
36
$12,936,000
40,150
$23,689,000
245
$38,167,000
19
$29,988,000
1,735
$104,161,000
$317,207,000
$154,891,000
$162,316,000
9.6 billion
gallons per year
515
$164,663,000
370
3,415,000 barrels
$48,803,000
83
592,000 barrels
$1,739,000
453
$9,065,000
59
$20,867,000
74,820
44,146,000
435
$50,075,000
32
$51,974,000
2,690
$161,120,000
$542,319,000
$297,150,000
$263,169,000
a Terminal and retail facilities using ethanol for the first time will need to make various modifications to ensure the
compatibility of their systems with ethanol.
Amortized over 15 years, the total capital costs (of $317,207,000 under the 7.2 Bgal/yr
case and 542,319,000 Bgal/yr case) equate to an annual cost of approximately $34,830,000 under
the 7.2 Bgal/yr case and $59,544,000 under the 9.6 Bgal/yr case. This translates to
220
-------
approximately one cent per gallon of new ethanol volume of which 0.5 c/gal is attributed to fixed
facilities and 0.5 c/gal is attributed to mobile facilities.
We performed a sensitivity analysis to evaluate the impact on costs if a relatively greater
reliance on rail transport versus marine transport occurs than was assumed in the DOE study.ccc
The 2002 DOE study estimated that 53 percent of the increase in ethanol volume shipped
between PADDs would be carried by barge and 47 percent by rail. For the purposes of this
analysis, we assumed that 30 percent of the increase in ethanol shipments that were projected in
the DOE study to be carried by barge would instead be carried by rail. This equates to 37
percent of the increase in ethanol shipments being carried by barge and 63 percent by rail. To
provide a conservatively high estimate of the potential economic impact, we assumed that this
shift translates into a 30 percent increase in rail infrastructure costs. The actual increase in rail
infrastructure costs may be somewhat lower given improvements in the efficiency of ethanol
transport by rail. Under this scenario, a total of 2,260 new rail tank cars would be needed under
the 7.2 Bgal/yr case and 3,490 under the 9.6 Bgal/yr case. The overall effect of this increased
reliance on rail transport would increase the capital costs by approximately $26,133,000 under
the 7.2 Bgal/yr case and $39,004,000 under the 9.6 Bgal/yr case. This equates to an additional
0.1 c/gal of new ethanol production.
7.3.1.2 Ethanol Freight Costs
The 2002 DOE study contains estimated ethanol freight costs for each of the 5 PADDs.
These estimated costs are summarized in the following Table 7.3-2.
contained in Figure 7.3-1.
142
A map of the PADDs is
Table 7.3-2. Estimated Ethanol Freight Costs from the 2002 DOE Study
PADD
1
2
3
4
5
National Average
Annual ethanol use of 5.1 billion
gallons per year
(cents per gallon)
11.1
4.3
6.6
4.7
12.7
7.7
Annual ethanol use of 10.0 billion
gallons per year
(cents per gallon)
7.2
2.4
5.8
7.4
10.7
5.7
See chapter 1.5 in this DRIA regarding the modes of transportation used to distribute ethanol.
221
-------
Figure 7.3-1.
Petroleum Administration for Defense {PAD} Districts
\ /1
•
p I J *, ^-f
:> ." /"""• ... >^>rf.
The Energy Information Administration (EIA) translated the cost estimates from the 2002
DOE study to a census division basis.143 A summary of the resulting (EIA) ethanol distribution
cost estimates are contained in the following Table 7.3-3. A map of the census divisions is
contained in Figure 7.3-2.
222
-------
Table 7.3-3. EIA Estimated Ethanol Freight Costs (based on 2002 DOE Study)
Census Division
From
East North Central
East North Central
East North Central
East North Central
East North Central
East North Central
West North Central
West North Central
West North Central
West North Central
West North Central
West North Central
West North Central
West North Central
West North Central
To
New England
Middle Atlantic
East North Central
South Atlantic
East South Central
Pacific
New England
Middle Atlantic
East North Central
West North Central
South Atlantic
East South Central
West South Central
Mountain
Pacific
Freight Cost
(cents per gallon)
9.8
9.8
4
9.8
4.7
14.0
11.4
11.4
4
4
11.4
4.7
4.7
4.5
13.0
223
-------
Figure 7.3-2. Census Divisions
WEST
MIDWEST
Wed
North Central
East
North Central
NORTHEAST
England
We took the EIA projections and translated them into State-by-State ethanol freight costs.
In conducting this translation, we accounted for increases in the cost in transportation fuels used
to ship ethanol by truck, rail, and barge. For the purposes of this analysis, all ethanol was
assumed to be produced in the Midwest in the East and West North Central Census Divisions
(corresponding closely to PADD 2). Ethanol consumed within these census divisions was
assumed to be transported by truck, while distribution outside of these areas was assumed to be
by rail, ship, and/or barge. A single average distribution cost for each destination census division
was generated by weighting together the 2012 freight costs given for each mode in both source
census divisions according to their volume share. These cent per gallon figures were first
adjusted upward by 10 percent to reflect higher energy prices, and then additional adjustments
were applied to some individual states based on their position within the census division. In the
cases of Alaska and Hawaii, differences in ethanol delivery prices from the mainland were
inferred from gasoline prices. For some states, different freight costs for ethanol supplied to
large hub terminal versus small satellite terminals was estimated. The reasoning behind this is
224
-------
that large shipments of ethanol shipped from the Midwest by barge, ship, and/or unit train will be
initially unloaded at hub terminals for further distribution to satellite terminals. The estimated
additional freight cost of shipping ethanol from hub terminals to satellite terminals is contained
in the following Table 7.3-3. The largest adjustment was applied to the Rocky Mountain states
since they are generally large in area and additional expense is required to transport freight
through higher elevations and rugged terrain. Smaller adjustments were applied to states that are
smaller, flatter, or have access by navigable waterways. The states to which an adjustment was
not applied were generally in the Midwest. Given the large number of ethanol plants in the
Midwest, we do not believe that there are substantial differences in the cost of distributing
ethanol with the area.
Table 7.3-3.
Additional Freight Costs to Deliver Ethanol from Hub to Satellite Terminals
States
OH
AK, AL, AR, FL, GA, KY, LA, MD, ME, MS,
NC, NH, NY, OK, OR, PA, SC, TN, TX, VA,
VT, WA, WV
AZ, CO, ID, NM, NV, UT, WY
Additional Freight Costs to Deliver Ethanol
from a Hub to a Satellite Terminal
(cents per gallon)
1
2
3
Expressed on a national average basis, we estimate that the freight costs to transport
ethanol to terminals would be 9.2 c/gal of ethanol. This translates to an annual freight cost for
the additional volume of ethanol used in 2012 of $303,600,000 under the 7.2 Bgal/yr case, and
$524,400,000 under the 9.6 Bgal/yr case. Adding in the annualized capital costs associated with
modifying the distribution system to handle the increased volumes of ethanol results in a total
annual ethanol distribution cost in 2012 of $338,430,000 under the 7.2 Bgal/yr case and
$583,944,000 under the 9.6 Bgal/yr case.
Our estimates of the State-by-State ethanol freight costs are contained in the following
Table 7.3-4.
225
-------
Table 7.3-4. State-by-State Ethanol Freight Costs
State
Florida, Pennsylvania
Maine
New Hampshire, Vermont
Massachusetts
Rhode Island
Delaware, Georgia,
Maryland, New Jersey,
New York, North
Carolina, South Carolina,
Virginia, West Virginia
District of Columbia
Iowa
Illinois, Kansas,
Minnesota, Missouri,
Nebraska, Wisconsin,
South Dakota,
Indiana, North Dakota
Ohio
Kentucky
Tennessee
Michigan
Oklahoma
Mississippi
Alabama
Arkansas
Louisiana
Texas
New Mexico
Colorado
Wyoming
Utah, Montana
Idaho
Arizona
Nevada
California, Oregon,
Washington
Hawaii
Alaska
National Average
PADD
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
3
3
3
3
3
O
4
4
4
4
5
5
5
5
5
-
Ethanol Freight Cost: Hub Terminal / Satellite Terminal
(cents per gallons)
8.4/10.4
13.4/15.4
12.4/14.4
11.4/13.4
11.4/13.4
11.4/13.4
11.4
3.4
4.4
5.4
5.4/6.4
6.2/8.2
6.2/8.2
6.4
8.3/10.3
6.2/8.2
7.2/9.2
7.3/9.3
7.3/9.3
10.3/12.3
12.4/15.4
10.4/13.4
12.4/15.4
13.4/16.4
15.4/18.4
15.4/18.4
16.4/19.4
16.5/18.5
36.5/39.5
41.5/43.5
9.2
226
-------
7.3.2 Biodiesel Distribution Costs
The volume of biodiesel used by 2012 under the RFS is estimated at 300 million gallons
per year. The 2012 reference case against which we are estimating the cost of distributing the
additional volume of biodiesel needed to meet the requirements of the RFS is 28 million
gallons000
For the purposes of this analysis, we are assuming that to ensure consistent operations
under cold conditions all terminals will install heated biodiesel storage tanks and biodiesel will
be transported to terminals in insulated tank trucks and rail cars in the cold seasons.EEE The
capital costs associated with the distribution of biodiesel will be somewhat higher than those
associated with the distribution of ethanol. The cost to install the heated storage tanks and
blending equipment at terminals is estimated at $250,000 per terminal. We estimate that 180
terminals would need to add the capacity to blend biodiesel in order to meet the requirements of
the RFS for a total one time cost to terminals of $44,948,000. The cost to provide insulated tank
trucks and rail cars is estimated to add 10 percent to the cost of these vessels. We estimate that
17 new tank trucks and 25 new rail cars will be needed to distribute the additional volume of
biodiesel required at a cost of $3,163,000 and $1,650,000 respectively. Thus, the total capital
cost to prepare the distribution infrastructure system to handle the increase in the volume of
biodiesel under the RFS is estimated at $49,813,000. Amortized over 15 years, this equates to an
annual cost of $5,470,000 which translates to approximately 2 c/gal of new biodiesel volume.
Due to the developing nature of the biodiesel industry, specific information on biodiesel
freight costs is lacking. The need to protect biodiesel from gelling during the winter may
marginally increase freight costs over those for ethanol. Counterbalancing this is the likelihood
that biodiesel shipping distances may be somewhat shorter due to the more geographically
dispersed nature of biodiesel production facilities. In any event, the potential difference
between biodiesel and ethanol freight costs is likely to be small and the cost of distributing
biodiesel does not appreciably affect the results of our analysis. Therefore, we believe that
estimated freight costs for ethanol of 9.2 c/gal adequately reflects the freight costs for biodiesel.
The annual freight cost to distribute the additional volume of biodiesel projected to be used in
2012 is estimated at $25,020,000. Adding the annualized capital costs associated with modifying
the distribution system to handle the increased volume of biodiesel ($5,470,000) results in a total
annual distribution cost in 2012 for the additional biodiesel volume of $30,490,000.
7.4 Gasoline and Diesel Blendstock Costs
In sections above, we estimated the cost of producing and distributing ethanol and
biodiesel. This section summarizes the overall cost of several different changes to the gasoline
pool, including the increase in ethanol use, the phase out of MTBE and the reuse of the former
MTBE feedstock, isobutylene, to produce alkylate. We also estimate the cost for using more
biodiesel in the diesel fuel pool.
DDD See Chapter 1 of this DRIA regarding the 2012 reference case.
EEE See Section 1.5 of this DRIA regarding the special handling requirements for biodiesel under cold conditions.
227
-------
7.4.1 General Overview
To estimate the cost of increased use of renewable fuels, the phase out of MTBE and the
production of alkylate, EPA retained the services of a contractor to run its LP refinery model.
The contractor work to estimate the cost of these changes are underway, however, the work
could not be completed in time for this proposed rule. Thus, to provide a cost estimate for this
proposal, a simpler spreadsheet analysis was conducted. Although the spreadsheet analysis
cannot capture the more complex cost effects for the blending of ethanol that can be captured
with the LP refinery model, such as the economics of blending ethanol into premium versus
regular grade gasoline or estimate summer versus wintertime ethanol use, it can capture the
major cost factors that contribute to these changes.
The cost analysis is conducted by comparing a reference year without the Energy Act fuel
changes to a modeled year with the fuel changes. We used 2004 as the base year. We grew the
2004 gasoline demand to 2012 to develop our reference case assuming that MTBE is still used,
and ethanol is used proportional to their use in 2004. The sum of fuel changes, including the
phase-out of MTBE, the increased use of alkylate and increased use of biofuels, is all assumed to
be in place in 2012 and compared to the 2012 reference case. The analysis is conducted based
on EIA's forecast that average price of crude oil will drop down to $47 per barrel. Predicting
crude oil prices is difficult since so many factors can affect the price of crude oil. To capture the
near term effect of higher priced crude oil and the possibility that crude oil prices could remain
high into the more distant future, we conducted a sensitivity analysis assuming that crude oil is
priced at $70 per barrel.
7.4.2 RVP Cost for Blending Ethanol into Summertime RFG
The following subsection details our assessment of the means and cost for lowering
gasoline Reid Vapor Pressure (RVP)FFF of summertime RFG to accommodate the removal of
MTBE and the addition of ethanol. When MTBE is removed, it results in a reformulated
gasoline blendstock for oxygenate blending (RBOB) which is slightly lower in RVP than the
MTBE-reformulated gasoline blend. The subsequent blending of ethanol into gasoline, however,
causes about a 1 PSI increase in RVP. To end up with an ethanol-blended RFG which averages
the same RVP as the MTBE-blended RFG, which is necessary to comply with the RFG
hydrocarbon standards, some of the lighter hydrocarbons must be removed from the gasoline
blendstock increasing its production cost.
7.4.2.1 Magnitude of RVP Change
Estimating the change in RVP was based on the actual in-use RVP level of RFG. The in-
use RVP level for RFG was estimated by averaging the RVP levels of gasoline samples reported
in the Association of Automobile Manufactures (AAM) gasoline survey for the RFG cities
reported there. The RVP level of the RBOB minus MTBE would have to be about 6.7 RVP to
derive a finished MTBE-blended RFG of 6.85 - which assumes that MTBE has an RVP of 8.0
-g ^ pressure tjjat gasoiine generates when measured at a standardized condition using an American
Society of Testing Materials (ASTM) testing methodology. RVP is somewhat related to the true vapor pressure
generated by gasoline but tends to be somewhat higher.
228
-------
and is blended at 11 volume percent. Since ethanol-blended RFG would also be expected to
have a final RVP of 6.85, we estimate that the RVP of an RBOB for blending with ethanol would
need to be 5.60, assuming that ethanol would be blended into RFG at 10 volume percent and
would have a 20 PSI blending RVP. Thus, the RBOB RVP would have to be reduced by 1.1 psi
in a transition from MTBE to ethanol. These RVPs are summarized in Table 7.4-1.
Table 7.4-1.
RVP Levels for MTBE and Ethanol-blended Reformulated Gasoline
RVP Level
MTBE
RFG
6.85
RBOB
for
blending
with
MTBE
6.71
RBOB
for
blending
with
Ethanol
5.60
Ethanol
RFG
6.85
7.4.2.2
Means for Reducing RVP
Gasoline contains light, medium and heavy hydrocarbons. Medium and heavy
hydrocarbons, which make up the majority of the gasoline pool, have six or more carbon
molecules (C6+) while light hydrocarbon compounds have a carbon count less than six. The
light hydrocarbon components in gasoline are butanes (C4s) and pentanes (C5s)GGG. The
gasoline produced by more complex refineries is made up often or more different streams
produced by refinery processes or streams imported into the refinery. Some of these streams
contain significant levels of butanes and pentanes while others do not. A refiner's gasoline pool
is the volume of various hydrocarbon streams or components that are added to a refiner's
gasoline volume before shipment.
In gasoline, each hydrocarbon compound has its own pure vapor pressure. The
compounds usually contribute a different or modified vapor pressure when blended into the
gasoline pool due to its physical interaction with the other constituents in the pool. For ease of
making blending RVP calculations, the modified vapor pressure of a single compound is called
the blending RVP and we will be using blending RVP values in this analysis. The C7+
hydrocarbons in gasoline have relatively low blending RVP values ranging from 9 PSI to near
zero. Butane and pentane hydrocarbons have much higher blending RVP's; isobutane's and
normal butane's blending RVPs are 71 and 65, respectively, and isopentane's and normal
pentane's blending RVPs are 20 and 17, respectively. For gasoline, blending a high RVP stream
such as butanes into the gasoline pool will only be minimally reduced by blending in or diluting
with lower RVP blend stocks streams due to the physical nature of vapor pressure. Thus,
controlling the butane content of the gasoline pool will have the largest impact on pool RVP with
minimal impacts on volume.
GGG These molecules can have single and/or double bonds between their carbon atoms as well as be straight chain or
branched chain. For this cost analysis referral to butanes and pentanes means inclusion of both single and double
carbon bond type molecules and straight or branched chain molecules.
229
-------
Since butanes and pentanes have high blending RVP's, refiners control the amount
blended into their gasoline pool up to the RVP allowed by the applicable environmental or other
in-use gasoline standards. In the summer low RVP season, refiners are probably not adding
butane, but separating some of the butanes and blending back a portion to meet RVP
requirements. To accomplish a current RVP goal of say 9.0, refiners utilize existing distillation
columns such as light straight run naphtha splitters, reformate splitters, stabilizers and other
existing process distillation columns to remove butanes and pentanesHHH. These existing
distillation columns are limited in making significant reductions in pool RVP. This is because
the gasoline supply streams from these units contain only a portion of the amount of butanes and
pentanes which ends up into gasoline. After these existing methods and equipment for removing
light hydrocarbons from the gasoline pool are fully utilized, further lowering RVP could require
a refiner to add additional distillation column capacity to remove butanes and in some cases
pentanes.
Further control of RVP can be realized by reducing butanes or pentanes in their fluidized
catalytic cracker unit's (FCCUs) gasoline blendstock, which is also called FCC naphtha. To
accomplish this task, refiners would likely have to add a distillation column called a debutanizer
and perhaps another column called a depentanizer, to separate these light hydrocarbons from the
rest of the FCC gasoline blendstock. Debutanizers distill or separate butanes and lighter
hydrocarbons off the top of the distillation column while pentanes and heavier C6+ hydrocarbons
remain in the bottom and are subsequently blended into gasoline. In depentanizers, pentanes and
lighter hydrocarbons (the debutanized stream) are removed from the hydrocarbon feed and
drawn off the top of the column while the heavier C6+ hydrocarbon remain in the bottom of the
distillation column and are blended into gasoline. If a refiner has a FCC depentanizer the
"debutanized" FCC gasoline flows from the debutanizer to the depentanizer as hydrocarbon feed
where pentanes are then removed.
In the U.S., 103 of the total 115 refineries that produce gasoline have FCCUs. The
FCCU converts gas oil and residual fuel to gasoline, which is the heavy and light hydrocarbons
as defined above, and even lighter hydrocarbons, by reacting or cracking the gas oil over
fluidized, heated catalyst. The gasoline volume produced by the FCCU makes up to 35-50
volume percent of refiner's gasoline pool and is thus the largest contributor to the gasoline
pool.144 FCCU gasoline contains butanes, pentanes, and C6+ hydrocarbons with the amount of
these hydrocarbons being set by each refiner's FCC conversion rate and the FCCU's gasoline
distillation capability, as most of the butanes and lighter hydrocarbons are removed off of the top
of the debutanizer column.111 Typical ranges are 0 to 10 percent for butanes and 5 to 17 volume
percent for pentanes in the FCC naphtha pool.145 The higher percentage of butane is likely for a
9.0 RVP gasoline, while lower percentages are consistent with lower RVP gasoline. Each
HHH Distillation columns are the process equipment used to separate light from heavier hydrocarbons through the
process of vaporization and condensing. The addition and removal of heat to the column is what drives the
separation process. Heat is added to the column through a heat exchanger called a reboiler while heat is removed
from the top of the column with an exchanger called a condenser. The lighter hydrocarbons are vaporized and travel
up the column where they are removed as a product while the heavier hydrocarbons move down the column are
drawn off the bottom. In a distillation column, there are many distillation trays which provide the mechanism for
mixing and separation of the hydrocarbons.
111 FCC conversion can be defined as the amount of FCC charge that is cracked into gasoline and lighter
hydrocarbons.
230
-------
refiner's FCC conversion is set by many process parameters, including the type of FCC unit, the
FCC feedstock type, feed throughput, catalyst type, unit constraints, unit bottlenecks, catalyst
condition and operational mode. Higher amounts of butanes and pentanes are generated as the
FCCU conversion rate is increased with a typical conversion rate being 74 percent.146
It is important to determine the gasoline RVP level at which refiners will begin to remove
pentanes after the butanes have all been removed. Because butanes are more volatile than
pentanes, initial reductions in RVP are achieved by removal of butanes and at some point
achieving further reductions in RVP requires removal of pentanes from the pool. This is
important because, as described below, we estimate that reducing the gasoline pool RVP by one
RVP number requires a reduction of the equivalent of 2 volume percent of the gasoline pool in
butane, whereas, attaining the same RVP reduction requires a reduction of the equivalent of 10
percent of the gasoline pool in pentanes.
We used several different means for estimating the point where further RVP decreases
requires pentanes to be removed. We spoke to several distillation vendors who have helped
refiners make process changes to lower gasoline pool RVP to meet low RVP standards that were
instituted in the 1990's and year 2000. One vendor stated that most refiners currently producing
a reformulated federal or low RVP (7.0, 7.2 or lower) gasoline today made modifications to their
FCC debutanizers to meet the RVP specification. The modifications were achieved either
through revamping the existing debutanizer by installing new high capacity trays and heat
exchangers, or through the addition of a new debutanizer column. According to this vendor,
approximately 40% of refiners revamped their FCC debutanizer while 60% installed a new
debutanizer column. The vendor stated that a FCC gasoline RVP of about 6.7 to 7.0 is achieved
by most refiners when butanes are removed to less than 0.5 volume percent of the FCC gasoline
pool. He further stated that these low levels of butanes could typically be attained through FCC
debutanizer modifications. Obtaining a FCC gasoline RVP of 7.0 or below would probably
allow most refiners to produce their gasoline to a pool RVP of 7.0 or lower.
The distillation vendor also stated that half of the refiners that made debutanizer
modifications also installed new FCC depentanizers. Prior to lower RVP requirements, refiners
typically did not have depentanizers for depentanizing their FCC gasoline blendstock. The
vendor was not sure as to why the depentanizers were added but thought that refiners only
required a FCC debutanizer modification to meet lower RVP specification. The vendor also
stated that current refiners producing a 7.8 to 9.0 RVP pool cap may have original unmodified
debutanizers and typically do not have FCC depentanizers. The original unmodified
debutanizers were designed to remove butanes down to a 1.5 to 2.0 volume percent level in FCC
gasoline.
To understand this issue further, we contacted several refiners who make low RVP
gasoline or RFG to understand about how they reduced the RVP of their gasoline pool. The
refiners reported that they had to spend capital for FCC debutanizer modifications and that these
modifications allowed production of a 7.0 RVP gasoline by removing butanes to less than a 1.0%
level in the gasoline pool. One refiner operating their FCCU at a low conversion rate actually
made a 6.4 RVP FCC gasoline. Only one out of five refiners reported that during the
summertime production season they had to remove some pentanes to meet the 7.0 RVP
231
-------
specification for their pool. During the summer low RVP gasoline season, this refiner
intermittently had to remove about 20 percent of the refinery's pentanes from the gasoline pool.
The other refiners reported no need to remove pentanes to meet a 7.0 RVP spec. The
refiners reported that the new depentanizers the distillation vendor referred to may have been
installed for several reasons; to allow segregation of the heavier gasoline C6+ components for
sulfur sweetening, to remove pentanes to lower the pool RVP or to segregate the pentanes so that
the pentanes may be backblended back into the pool per RVP allowance/" Some refiners
produce several grades of gasoline with varying RVP specifications, thus segregating pentanes
and back blending would allow a refiner to more accurately control each pool's RVP.
Backblending of pentanes would be particularly important for refiners producing RBOB
(renewable blendstock for oxygenate blending) for blending with ethanol since that RBOB must
be very low in RVP to accommodate the RVP boost of ethanol. None of the refiners commented
on the operations of their FCC debutanizers/depentanizers, but one refiner reported that pentanes
would have to be removed from gasoline to get the pool below a 7.5 RVP specification.
We also evaluated information from several different refinery models in an attempt to
understand the breakpoint between butane and pentane reduction to reduce RVP. For this
analysis, we used a typical gasoline blend, which represents the gasoline quality for a notional
refinery for PADDs 1, 2 and 3. We used this gasoline blend because it seemed like a reasonable
mix of gasoline blendstocks. This gasoline blend is summarized in Table 7.4-2.
Table 7.4-2.
Baseline 9 RVP Gasoline Composition
Gasoline Blendstocks
Isobutanes
Normal Butane
C5s & Isom
Naphtha C5-160
Naphtha 160-250
Alkylate
Hydrocrackate
Full Range FCC Naphtha
Light Reform
Heavy Reform
MTBE
Total
RVPpsi
% Volume
1.3
4.1
5.8
3.5
3.7
12.1
4.0
38.1
5.3
21.6
0.5
100.0
8.5
We then applied the blending RVPs from different refinery models, which included
Mathpro's, Oak Ridge National Laboratory's (ORNL) and a refining industry consultant who
wished to remain anonymous, to the typical gasoline blend to estimate this butane/pentane
111 Send the C6+ hydrocarbons through a Merox or similar process were mercaptan sulfur molecules are converted to
meet odor and corrosion requirements.
232
-------
breakpoint in RVP. Before proceeding with the analysis, we needed to estimate the amount of
butane entrained in the gasoline pool.
Butanes remain entrained in the gasoline pool because distillation of hydrocarbons does
not allow a perfect cut between the various hydrocarbons which comprise gasoline and some
butanes would be expected to remain in refined streams after distillation to remove them. It is
important to know how the various refinery modelers set up the input tables of their refinery
models to account for this. Mathpro said that their gasoline blendstocks do not incorporate
entrained butane and that they put a lower limit on the amount of butane which can be removed
from the gasoline pool. We assumed a lower limit of 1.5 percent butanes in the gasoline blend
when using their gasoline blendstocks to evaluate this issue. Ensys, which has provided many of
the technical inputs to the Oak Ridge National Laboratory (ORNL) refinery model, stated that
the gasoline blendstocks in the ORNL refinery model were based on actual refinery streams, but
did not know how much butane was in those streams. Since the blendstock qualities were based
on actual refinery blendstocks, we presumed that the blendstocks did contain entrained butane.
The refinery industry consultant felt that their gasoline blendstocks contained entrained butane
and that they model removing all the butane in their low RVP refining studies and we did the
same. The blendstock blending RVP levels are summarized in Table 7.4-3.
Table 7.4-3.
Estimated Gasoline Component Vapor Pressures (psi RVP)
Component
Isobutanes
Normal Butane
C5s & Isomerate
Straight Run Naphtha
(C5-160F)
(160-250 F)
Alkylate
Hydrocrackate
Full Range FCC Naphtha
Light Reformate
Heavy Reformate
MTBE
MathPro
71
65
13.3
—
13
2.5
3.5
12.5
3.7
7.5
3.8
8
ORNL
71
65
13.3
—
12
3
6.5
14
6.9
6.9
3.9
8
Consultant X
71
65
13.8
8.8
___
___
4.9
7.2
7.1
6.4
o o
J.J
8
Our analysis here showed that applying the Mathpro blendstocks to the typical gasoline
blend and limiting butane reduction to 1.5 percent yielded a lower RVP limit of lowering butane
to 6.2 RVP. Applying the ORNL blendstocks to the typical gasoline blend and removing all the
butane yielded a lower RVP limit for lowering butane to 7.1 RVP. Applying the other refinery
industry consultant's blendstock qualities to the typical gasoline blend and removing all the
butane yielded a lower RVP limit for lowering butane to 6.5 RVP. Averaging these three values
yields 6.6 RVP as the lower limit for removing butane before pentanes would need to be
removed.
233
-------
We believe that there is a good explanation for why the butane-pentane breakpoint for
RVP reduction varies so much based on the people we spoke to and also on our refinery
modeling analysis. Each refiner has many differing types of gasoline production processes with
varying throughputs and gasoline yield capabilities. Also, each refiner processes a differing
crude oil slate, with a varying hydrocarbon composition which further contributes to each refiner
producing its own unique gasoline blend stocks. Thus, differing crude slates and process units
cause a refiner to yield different amounts of the light and heavy hydrocarbon components for
blending into its gasoline pool.
To take into account the various RVP values for the butane-pentane breakpoint based on
the low and high figures obtained from the aforementioned discussions with the vendors,
refiners, and consultants, and the refinery modeling study, we considered a range of values for
this analysis. Based on the above discussions and analyses, we believe that, after butanes have
been removed, pentanes would begin to be removed when a gasoline blend's RVP is lowered
below a range of values between 7.5 and 6.2 RVP. However, the analysis suggests that for most
refiners, the breakpoint is likely at an RVP level of 6.8, the average of summertime RFG.
Thus to accommodate the ethanol, the MTBE is removed from RFG, and the RVP of the
base gasoline is adjusted so that when the ethanol is added, the resulting ethanol-blended RFG
will have an RVP of 6.8. This occurs by removing pentanes from the gasoline pool because
nearly all the butanes are presumed to have been removed in forming the MTBE-blended RFG.
Because a small amount of butanes remain entrained with the pentanes, the vapor pressure of the
pentanes is presumed to be higher than the pure blendstocks. For our estimates of the impact of
lowering pentane content on RVP, we presume that there is about 1 percent butane content in the
pentanes which would result in an average blending RVP for pentanes of about 20 RVP.
7.4.2.3 Cost of Reducing Gasoline RVP
The total cost of RVP control was identified as the combination of three separate cost
elements. First, capital and operating costs would be incurred through the installation of new
depentanizer columns. We assume that separating pentane from the rest of the gasoline pool
requires these investments. Then, the removed pentane is assumed to incur an opportunity cost
based on the next available price for these hydrocarbons on the open market compared to the
price of gasoline. Finally, the removal of these lighter hydrocarbons causes the gasoline pool to
increase in energy content. Thus, we determined the energy density change and estimated the
cost impact for the energy change based on the wholesale price for gasoline. The calculation of
each of these cost elements and the resulting total costs are summarized below.
Costs were developed for adding a new depentanizer distillation column for the removal
of pentanes from FCC gasoline in a typical-sized refinery. Capital and operating costs for a new
depentanizer were based on the capital and operating cost of a naphtha splitter from Mathpro for
cost work conducted for us for the Mobile Source Air Toxics Proposed Rule (MSAT2).147 The
costs for a naphtha splitter are expected to be similar to that of a depentanizer because it distills
pentanes out of the top of the column while not boiling the heavier compounds which pass
through the bottom of the distillation column. The cost information for this distillation column is
summarized in Table 7.4-4.
234
-------
Table 7.4-4.
Process Operations Information for New Depentanizer
Capacity (bbl/day)
Capital Cost, ISBL (MM$)
Electricity (kWh/bbl)a
Fuel Gas (foeb/bbl)b
Other Variable Operating Cost ($/bbl)c
Depentanizer
15000
6.6
2.8
0.01
0.01
kWh/bbl is kilowatt-hours of electricity per barrel of feed
bFoeb/bbl is fuel oil equivalent barrel of fuel gas per barrel of feed
0 $/bbl is dollars per barrel of feed
Capital Costs
Capital costs are the one-time costs incurred by purchasing and installing new hardware
in refineries. Capital costs for a particular processing unit are supplied by vendors or estimated
from other sources at a particular volume capacity, and these costs are adjusted to match the
volume of the particular case being analyzed using the "sixth tenths rule" as described by Gary
and Handewerk.148
The capital costs are adjusted to account for the off-site costs and differences in labor
costs relative to the Gulf Coast using Gary and Handewerk estimates.149 Off-sites costs were
assumed to be 1.25 times the onsite costs. Location factors for the refineries in each PADD were
assumed to be the same by PADD. Table 7.4-5 contains the location cost factors for each PADD
and for California.
Table 7.4-5.
Location Factors by PADD Used for Estimating Capital Costs'1
Factor
Location
PADD1
1.5
PADD 2
1.3
PADD 3
1.0
PADD 5
1.2
CA
1.2
PADD 4 is not included because PADD 4 does not use any RFG
The capital costs were estimated for the volume of FCC gasoline produced (see Table
7.4-8 below). For costing out the depentanizer, it was assumed that the column would remove
all the pentanes in the FCC naphtha, and any excess pentane removed would be reblended back
into gasoline. The capital costs were amortized on the yearly gasoline volume. The economic
factors used for amortizing the capital costs and the resultant capital cost factor are summarized
in Table 7.4-6.
235
-------
Table 7.4-6.
Economic Factors Used in Deriving the Capital Cost Amortization Factor
Amortization
Scheme
Societal Cost
Depreciation
Life
10 Years
Economic
and Project
Life
15 Years
Federal and
State Tax
Rate
0%
Return on
Investment
(ROI)
7%
Resulting
Capital
Amortization
Factor
0.11
Fixed Costs
Operating costs which are based on the cost of capital are called fixed operating costs.
Fixed costs are incurred to maintain the unit in good working order, insure the unit against
accidental damage, and for a number of other factors. These are fixed because the cost is
normally incurred even when the unit is temporarily shutdown. These costs are incurred each
and every year after the unit is installed.
Maintenance cost is estimated to be three percent of capital cost after adjusting for
location and offsites. This factor is typical and is based on the maintenance factor used in
previous refinery modeling studies. Other fixed operating costs include: 0.2 percent for land,
one percent for supplies which must be inventoried such as spare distillation trays, and two
percent for insurance. These factors sum to 6.2 percent, which is applied to the total capital costs
(after adjusting for offsite costs) to generate the fixed operating costs. Labor costs are very small
and are presumed included with the rest of the fixed operating costs.
Variable Operating Costs
Variable operating costs are the costs incurred to run the unit on a day-to-day basis and
are based completely on unit throughput. Thus, when the unit is not operating, variable
operating costs are not being incurred.
The electricity and natural gas costs are based on a simple arithmetic average of 2004
utility prices paid by industries for the states with refineries within the states. 15° 151 The 2004
average prices for each PADD are adjusted to represent estimated prices in year 2012 using the
ratio of projected 2012 prices to 2004 prices in the Annual Energy Outlook 2006.152 These
projected energy prices are summarized in Table 7.4-7.
Table 7.4-7. Summary of 2012 Utility Costs
Electricity
(eVkWh)
Natural Gas
($/MMbtu)
PADD1
7.18
7.77
PADD 2
4.34
6.71
PADD 3
5.63
5.51
PADD 5
8.34
8.35
CA
9.17
6.96
236
-------
For the removal of pentanes, costs developed for additional FCCU depentanizer capacity
per treated gallon of FCC gasoline were then amortized over the entire gasoline pool. The
volume of FCC naphtha in each PADD as well as the gasoline volume in each PADD was taken
from the refinery-by-refinery cost model used for estimating benzene control costs for the
MSAT2 proposed rulemaking.153 For each PADD, the PADD's FCC gasoline volume was
divided by the PADD's total refinery gasoline volume to determine the percent contribution of
FCC gasoline to the total gasoline pool. The FCC naphtha and total gasoline volume for each
PADD is summarized in Table 7.4-8.
Table 7.4-8.
Volume of FCC Naphtha Compared to Total Refinery Gasoline Production by PADD
(barrels per day for an average-sized refinery)
Factor
FCC Naphtha (bbl/day)
Total Refinery Gasoline
(bbl/day)
PADD1
34,700
80,700
PADD 2
24,300
69,400
PADD 3
33,800
82,700
PADD 5
exCA
7,000
22,300
CA
27,300
90,200
7.4.2.5
Cost Summary for RFC RVP Impacts
RVP control costs were developed by PADD for converting MTBE-blended RFG to
ethanol-blended RFG through the addition of new depentanizers. For the min RFG scenarios,
there are some situations where ethanol is removed from RFG. For these situations, refiners
would be expected to stop using their existing depentanizers resulting in the saving of the
pentanizer operating costs only. In Table 7.4-9 we provide the per-gallon capital and operating
costs by PADD for adding a new pentanizer, and the operating costs only for the situation that
existing pentanizers are shut down.
237
-------
Table 7.4-9.
Summary of the RVP Impacts for Blending More Ethanol or
Reducing the Volume of Ethanol into RFG (cVgal)
Pentane
Distillation Cost
With Capital
Costs
Pentane
Distillation Cost
Without Capital
Costs
PADD1
1.37
1.15
PADD2
1.04
0.87
PADD3
1.02
0.88
PADD4
1.07
0.83
PADD5
(ex CA)
1.22
1.01
CA
1.01
0.88
7.4.2.4
Other Costs of Summertime RFG Volatility Control
When butanes, and sometimes pentanes, are removed from the gasoline pool, they are
sold off in markets which bring a lower return than gasoline. The lost opportunity of blending
and selling these petroleum components in gasoline is called the opportunity cost. The
opportunity cost is merely the price difference between higher valued gasoline and the price for
these petroleum compounds on the open market. For this analysis, we assume that the removed
pentanes would be reblended into gasoline, most likely into the summertime CG pool, although
they could be stored up for blending into the wintertime CG or RFG pool. Either way, the
pentanes would likely not be lost from the gasoline pool. Instead, we assume that when the
pentanes are reblended into another portion of the gasoline pool, the appropriate volume of
butanes would be removed from that gasoline pool to balance the RVP of the gasoline pool. We
obtained 2004 prices for butane from Platts, and compared them to the gasoline price in 2004.154
Comparing the butane price with the gasoline price shows that the opportunity cost of removing
butanes is about 36 c/gal and we used this cost when assessing the cost for removing the butanes.
The 2004 prices for butane and gasoline are summarized in Table 7.4-10 and we apply this cost
below.
Table 7.4-10.
Prices for Butane and Gasoline in 2012 Used for Estimating the Opportunity Cost of
Removing Butanes from Gasoline (cents/gallon)
Butane price
94
Gasoline
130
The energy density of the removed butane is lower than gasoline which lowers the impact
of its removal. This lower energy density is accounted for below in the balancing of the gasoline
238
-------
pool's energy content. The energy contents of butane and gasoline is summarized in Table 7.4-
11.
Table 7.4-11.
Energy Content of Butane and Gasoline for Estimating the Fuel Economy
Impacts of Reducing the RVP of Gasoline (MMBtu/gal)
Butane
94,000
Gasoline
112,000
7.4.3 Cost savings for phasing out Methyl Tertiary Butyl Ether (MTBE)
The Energy Act rescinded the oxygen standard for RFG and when the provision took
effect, U.S. refiners stopped blending MTBE into gasoline. When MTBE use ended, the
operating costs for operating those plants also ceased. The total costs saved for not operating the
MTBE plants is calculated by multiplying the volume of MTBE no longer blended into gasoline
with the operating costs for the plants producing that MTBE.
The volume of MTBE blended into U.S. gasoline in 2004 is provided by EIA and for our
reference case is grown using the gasoline growth rate to 2012 as summarized in Section 2.1.3
above. The cost savings of phasing-out of MTBE is based on the reference case volumes to have
a single case to which we compare the control cases. These volumes are summarized again here
Table 7.4-12 by PADD.
Table 7.4-12.
MTBE Consumption in PADDs 1-5 and CA in 2004 (million gallons)
2004 MTBE
Consumption
Projected
MTBE
Consumption
in 2012
PADD1
1,360
1,510
PADD 2
2
2
PADDS
498
555
PADD 4
0
0
PADDS
(ex CA)
19
21
CA
0
0
USA
1,878
2,092
The operating cost for producing MTBE depends on the type of MTBE plant producing
it. There are 4 different types of plants producing MTBE, as well as imports. As MTBE was
phased-out under the state MTBE bans in California, Connecticut and New York, most or all of
the imported MTBE found a market elsewhere overseas, so this volume of MTBE and the plants
producing it was no longer relevant to the U.S. market. There are four different types of
domestic MTBE plants and what they have in common is that they react isobutylene with
purchased methanol to produce the MTBE.155 The primary difference between them is how they
obtain or produce the isobutylene, which is summarized by plant type:
239
-------
1) MTBE plants contained in refineries are called "captive" MTBE plants. These plants
use isobutylene produced from the refinery's fluidized catalytic cracker (FCC) unit.
2) Propylene Oxide based plants produce tert-butyl alcohol (TEA) as a byproduct. This
TEA is converted to isobutylene by a deoxidation reaction.
3) Ethylene crackers produce isobutylene in the process of cracking heavier weight
hydrocarbons. Thus, the cost for producing MTBE by these plants is similar to captive
units.
4) Merchant MTBE plants produce their isobutylene from the normal butanes or mixed
butanes from natural gas condensate. The normal butanes are first isomerized to
isobutane and then dehydrogenated to isobutylene. If starting from mixed butanes the
normal butane is separated from the branched chain butanes through distillation and are
isomerized and combined together with the branched chain butanes and fed to a
dehydrogenation reactor for converting to isobutylene.
EPA conducted a detailed analysis of the volume of MTBE produced by each of these
MTBE plants several years ago. Table 7.4-13 shows the estimated production by MTBE plant
type in the year 2000 (from Pace Consultants156).
Table 7.4-13.
Sources of MTBE Used in U.S. Gasoline in the Year 2000
Type of MTBE Plant
Captive refinery plants
Propylene Oxide (TEA) based
merchant plants
Ethylene based merchant
plants
Natural gas liquids (NGL)
based plants
Imports (NGL based)
Total
MTBE Production Volume
(barrels/day (bbl/day))
Physical Volume
79,000
45,000
21,000
67,000
51,000
263,000
Percent of non-imported
MTBE
37
21
10
32
-
-
240
-------
Since the year 2000 U.S. demand for MTBE has diminished to about one half of that of
the year 2000. The principal reason is that the states of California, New York and Connecticut,
which consume significant amounts of RFG, have banned the use of MTBE. However,
additional reductions occurred due to the phase-out of elective MTBE use in conventional
gasoline. We did not assess how the past MTBE plants might have changed due to the change in
MTBE use since the year 2000. It is likely that a portion of the captive MTBE plants have
already been shutdown in those refineries that serve RFG markets which are located in states
which have banned the use of MTBE. Of the year 2000 petrochemical and merchant MTBE
plant capacity, it is likely that a portion of that MTBE plant capacity is being used to export
MTBE. Lacking updated information, we assumed the mix of MTBE plants providing MTBE to
the U.S. market in 2004, the base year of our analysis, to be the same as that in 2000.
We estimated the costs saved by shutting down this MTBE plant capacity at the volumes
of MTBE blended into the projected 2012 reference case gasoline. For estimating the costs, the
volumetric feedstock demands and the operating costs factors for each of these MTBE plants
were based on an Ethermax MTBE plant found in literature.157 We could not locate any
information on the costs for deoxidizing TEA, so we used the costs for isomerization. The
volumetric operating costs for the reactions used to produce isobutylene and for producing
MTBE are summarized in Table 7.4-14.
241
-------
Table 7.4-14.
Volumetric Feedstock Demands and Operating Cost Factors for Producing MTBE
MTBE plant costs apply
for:
n-Butane and TEA
(BPSD)
Isobutane (BPSD)
Isobutylene (BPSD)
Methanol (BPSD)
MTBE (BPSD)
Capital Costs (million $)
Plant Size (kbbl/day)
Steam (Ibs/hr)
Electricity (Kwh)
Cooling water (gals/min)
Catalyst
Isomerization
Merchant and PO
(used for
deoxidation
of TEA)
1549
-
-
-
-
15.1
3800
18,900
117
545
0.19
Dehydrogenation
Merchant
-
1549
-
-
-
75.7
8260
63,000
10,000
15,800
3.1
MTBE Plant
Merchant, PO,
Ethylene Cracker
and Captive
-
-
1549
530
1560
9.3
1560
18,900
117
151
-
The feedstock prices are based on year 2004 average prices from Platts. [reference] A
price was not found for TEA, so it was set equal to isobutylene. They are adjusted to 2012 using
the same ratio used to estimate gasoline prices in 2012 as discussed below in subsection 7.4.6.
The operating cost factors are multiplied by the utility prices for each factor. We derived
the energy prices by averaging the year 2004 prices for Texas and Louisiana, the two primary
refining districts where most of the MTBE is manufactured. These prices are adjusted to 2012
by multiplying each price by the ratio of electricity and natural gas prices in 2012 to that from
2004 from AEO 2006, and are the same as the utility prices summarized in Table 7.4-7 above for
PADD 3. The estimated feedstock prices and energy prices for 2012 and are summarized in
Table 7.4-15.
Table 7.4-15.
Summary of Year 2012 Projected Feedstock and Utility Prices"
Mixed and Normal Butanes (c/gal)
Isobutylene (c/gal)
TEA (c/gal)
Methanol (c/gal)
Electricity (c/kWh)
Natural Gas ($/MMbtu)
93
105
105
77
5.63
5.51
a c/kWh is cents per kilowatt-hour, $/MMbtu is dollars per million British thermal units.
242
-------
Applying the 2012 feedstock and utility prices in Table 7.4-15 to the feedstock demands
and input cost factors in Table 7.4-14, results in the production costs for each MTBE production
type presented in Table 7.4-16. We applied the weighting factors from Table 7.4-13 above to
derive the weighted average MTBE production costs shown in Tables 7.4-16.
Table 7.4-16.
Cost Savings for phasing out MTBE Consumption in 2012
(cents/gallon)
MTBE Plant
Type
MTBE
Production
Cost
Captive and
Ethylene Cracker
140
Propylene
Oxide
148
Merchant
155
Weighted
Average
146
We also credited MTBE for its octane value. MTBE has a high octane value of 110
(R+M)/2 which partially offsets its production cost. The cost of octane is presented above and is
applied to the difference in octane value between MTBE and the average of the various gasoline
grades (89 (R+M)/2). MTBE's blending cost, which is the combined production and octane cost,
is summarized in Table 7.4-17.
Table 7.4-17.
MTBE Blending Cost for PADDs 1 - 5 and CA in 2012
(cents/gallon)
20 12 MTBE
Production
Cost
Octane Value
Net MTBE
Blending
Cost
PADD1
146
-15.8
131
PADD2
146
-8.4
138
PADDS
146
-14.7
132
PADD4
N/A
N/A
N/A
PADDS
(ex CA)
146
-18.9
127
CA
N/A
N/A
N/A
7.4.4 Production of Alkylate from MTBE Feedstocks
7.4.4.1
Overview of Converting MTBE Feedstocks
Discontinuing the blending of MTBE into U.S. gasoline is expected to result in the reuse
of most of one of the primary MTBE feedstocks, isobutylene, to be used to produce alkylate.
Alkylate is formed by reacting isobutylene together with isobutane.158 Prior to the oxygen
requirement for RFG, this isobutylene that is sourced from refinery FCCUs was, in most cases,
used to make alkylate. Another option would be for reacting isobutylene with itself to form
isooctene which would likely be hydrogenated to then form isooctane. There are several
differences between using isobutylene to form alkylate versus isooctane.
243
-------
One difference is that isooctane has a higher octane number (100 (R+M)/2), than alkylate
(93 (R+M)/2). A second difference is that an alkylate plant can produce twice the volume of fuel
compared to an isooctane plant. If an MTBE plant converts to alkylate production, it produces
80% more gasoline in terms of energy content than it did when producing MTBE. The gain in
energy comes from the fact that isobutane is combined with this isobutylene in the production of
alkylate, versus the addition of methanol in the production of MTBE. Isobutane contains more
energy than methanol, so the product does as well. If an MTBE plant converts to isooctane
production, it produces 15% less gasoline in terms of energy content than it did when producing
MTBE. The loss in energy comes from the fact that isobutylene is reacted with itself to form
isooctane (i.e., no other feedstock is combined with the isobutylene in the reaction). Thus, the
energy content of methanol is lost relative to MTBE production.
Alkylate and iso-octane both have low RVP (2-6 psi). Isooctane's RVP is particularly
low and alkylate RVP can be very low, though it tends to vary depending on operating condition
and feedstock quality. These RVPs are lower than MTBE's RVP of roughly 8 psi. Due to this
low RVP, the substitution of alkylate or isooctane for MTBE makes it slightly easier to add
ethanol to RFG and still meet the Phase 2 RFG VOC performance standards. Ethanol tends to
add approximately 1 RVP when added to gasoline, so the production of a blendstock with an
RVP in the range of 5.5 is needed to facilitate ethanol use in RFG. Both alkylate and isooctane
would help facilitate this.
Based on previous conversations with a contractor, the MTBE plants operating in the
U.S. have different possible fates for converting over to producing alkylate or isooctane
depending on the plant type.159 This is discussed by each plant type.
Captive Refinery MTBE Plants
Captive refinery plants would most likely redirect the isobutylene currently used to
produce MTBE to their alkylation unit if this unit has sufficient capacity or can be cost
effectively revamped to a higher capacity. Isobutylene was usually used to produce alkylate
prior to MTBE production and this would be the preferred route now, due to the higher volume
of gasoline produced with alkylate versus isooctane. However, if a refiner's current alkylation
unit does not have excess capacity and could not be inexpensively increased, the isobutylene that
was going to the MTBE unit could be converted to isooctane. Thus, as a lower volume limit it is
possible these units produce isooctane, and as an upper limit all these units will produce alkylate.
In no case will the MTBE production from these plants be completely lost as the isobutylene is
available at no cost and has no other high value market.
Propylene Oxide Based MTBE Plants
There are several options for the isobutylene produced by the propylene oxide based
MTBE plants. The TEA which is converted over to isobutylene, could be sold into the
chemicals market since it has other value as a chemical feedstock. Alternatively, the TEA could
still be converted to isobutylene, which is what these plants are doing now to produce MTBE,
and the isobutylene could be converted over to either alkylate or isooctane production. These
244
-------
plants are also very large and have the economies of scale to support conversion to isooctane or
alkylate.
Ethylene Based MTBE Plants
The ethylene based plants tend to be smaller than the other petrochemical MTBE plants
and tend to be co-located with refineries. For these reasons, the ethylene-based MTBE plants
would likely shutdown and send their isobutylene to their co-located refineries for conversion to
alkylate. Thus, while the MTBE plant itself is shut down, the isobutylene volume used to
produce MTBE today would not be lost. The main reason for the difference in fate for these
plants and the propylene oxide based plants is their size. As a lower limit, the isobutylene used
in these ethylene based plants could be used to produce isooctane in refineries, as was the case
for the captive refinery plants.
Natural Gas Liquid Based Plants
Merchant, natural gas liquids (NGL) based MTBE plants would face the greatest
challenge to stay in business. These plants produce the isobutylene they need to produce MTBE
from mixed, field butanes. Isobutane is produced by isomerization of normal butane or is
separated from mixed butanes. This isobutane is then dehydrogenated to form isobutylene.
Producing isobutylene in this way is more costly than using isobutylene already present within a
refinery or raffmate stream in an ethylene plant. It is also more costly than producing
isobutylene from tertiary butyl alcohol. The original mixed field butanes perhaps could be stored
until winter and then blended into gasoline. Thus, sufficient revenue must be obtained from
alkylate or isooctane production to cover the capital cost of the plant conversion plus the cost of
producing the isobutylene from mixed field butanes. If these plants were to convert, they would
be more likely to convert to alkylate than isooctane production. However, a review of the
historic alkylate price premiums suggests that these plants probably could not support conversion
to even alkylate production, which is contrary to today's alkylate prices which would likely
support such a conversion. Consequently, with the phase-out of MTBE, due to the uncertainty in
future alkylate premiums, in the worst case that all of these plants would shut down or export the
MTBE abroad, or in the best case they could convert to alkylate production.
7.4.4.2 Economics of Conversion of MTBE Feedstocks to Alkylate Plants
We assessed the economics for the conversion for MTBE feedstocks to produce alkylate
and isooctane. The costs for producing alkylate are based on an Exxon sulfuric acid alkylation
plant, and the costs for producing isooctane are based on a UOP catalytic condensation
process. 16° The feedstock volumes and operating cost factors are summarized in Tables 7.4-18
and 19.
245
-------
Table 7.4-18.
Feedstock Volumes and Operating Cost Factors for Producing Alkylate
Alkylate plant costs
n-Butane & TEA
(BPSD)
Isobutane (BPSD)
Isobutylene (BPSD
Mixed Butanes (BPSD)
Alkylate (BPSD)
Capital Costs (million $)
Plant Size (kbbl/day)
Steam (Ibs/hr)
Electricity (Kwh)
Cooling water (gals/min)
Catalyst ($MM/yr)
Other Costs ($/bbl)
Isomerization
Merchant and PO
5110
-
-
-
-
15.1
3800
18,900
117
545
0.19
-
Dehydrogenation
Merchant
-
5110
-
-
-
75.7
8260
63,000
10,000
15,800
3.1
-
Alkylate Plant
Merchant, PO,
Ethyl ene Cracker
and Captive
-
-
5110
4678
7500
16.0
7500
63,000
3345
10,900
-
1.1
Table 7.4-19.
Feedstock Volumes and Operating Cost Factors for Producing Isooctane
Isooctane plant costs
n-Butane & TEA
(BPSD)
Isobutane (BPSD)
Isobutylene (BPSD)
Iso-octane (BPSD)
Capital Costs (million $)
Plant Size (kbbl/day)
Hydrogen (scf/bbl)
Steam (Ibs/hr)
Electricity (Kwh)
Cooling water (gals/min)
Catalyst ($MM/yr)
Isomerization
Merchant and PO
5110
-
-
-
15.1
3800
-
18,900
117
545
0.19
Dehydrogenation
Merchant
-
5110
-
-
75.7
8260
-
63,000
10,000
15,800
3.1
Iso-octane Plant
Merchant, PO,
Ethyl ene Cracker
and Captive
-
-
5110
3745
13.1
3745
770
39,000
390
390
1.1
246
-------
Applying the 2012 feedstock and utility prices in Table 7.4-15 to the feedstock demands
and input cost factors in Tables 7.4-18 and 7.4-19 and based on the treatment of capital costs per
Tables 7.4-5 and 7.4-6, results in the total operating costs for converting MTBE plants over to
alkylate and isooctane presented in Table 7.4-20.KKK
Table 7.4-20.
Production Cost for Converting MTBE Feedstocks to Alkylate and Iso-octane in 2012
(cents per gallon)
MTBE Plant
Type
Alkylate
Iso-octane
Captive and
Ethylene Cracker
142
169
Propylene
Oxide
146
190
Merchant
152
-
There is a significant cost difference between producing alkylate and isooctane. The
octane blending benefit for each pathway helps to offset the production cost differences.
Isooctane would receive about a 5 to 17 c/gal credit for its high octane value depending on the
PADD, while alkylate would receive about a 2 to 4 c/gal credit. Despite this credit, the
economics for isooctane appear to be poorer. This leads use to conclude that the reuse of the
MTBE feedstocks would primarily be, if not exclusively, through the production of alkylate.
Also, the cost for conversion of the merchant plant feedstocks to alkylate seems to be too high to
support the conversion of these plants to alkylate. Even though the cost for production of
alkylate by merchant plants is only slightly higher than the 146 c/gal cost for producing MTBE,
MTBE benefits from a 8 to 19 c/gal octane blending cost credit compared to only the 2 to 7 c/gal
blending benefit for alkylate. Thus, we conclude that the merchant MTBE plants would either
shutdown or sell their MTBE elsewhere. We do not know the economics for whether the
propylene oxide plants would continue to react their TEA to isobutylene and then produce
alkylate, or if they would simply sell the TEA through the chemicals market. Due to this
uncertainty, we assumed that half of the TEA would be sold as TEA, and the other half would be
converted over to produce alkylate. The projected conversion of MTBE plants that we assumed
for our cost analysis is summarized in Table 7.4-21.
The cost for supplying steam is estimated by assigning each pound of steam 810 British Thermal Units (BTUs)
of heating the water to generate the steam. The cost estimated by applying the natural gas cost to the BTU's
required is increased by a factor of 2.0 to account for efficiency losses for steam distribution, for treating the boiler
water to prevent fouling and to account for maintenance and other miscellaneous costs (Chemical Engineering
Handbook, Perry and Chilton).
247
-------
Table 7.4-21.
Projected Fate of MTBE Feedstocks
Plant Type
Captive Plants
Propylene Oxide
Ethyl ene Cracker
Merchant Plants
Produce Alkylate
Vi sell TEA
l/2 produce Alkylate
Produce Alkylate
Shutdown or sell
MTBE Elsewhere
Assuming that the domestic MTBE plants will not all convert over to produce alkylate
changes the weighting factors provided in Table 7.4-13 above. The revised weighting factors are
shown in Table 7.4-22.
Table 7.4-22.
Revised Estimated MTBE Feedstocks to Alkylate by MTBE Plant Type
MTBE Plant Type
Captive Plants
Propylene Oxide
Ethyl ene Cracker
Merchant Plants
Total MTBE
Production
Capacity
(barrels/day)
79,000
45,000
21,000
67,000
Volume
Projected to be
Converting to
Alkylate
(barrels/day)
79,000
22,500
21,000
0
Revised
Weighting
Factors
0.65
0.18
0.17
0
Applying these revised weighting factors to the production cost of alkylate results in an
average alkylate production cost of 143 c/gal. The production cost is adjusted by PADD to
account for the blending octane of alkylate. The blending cost for alkylate is shown in Table
7.4-23.
248
-------
Table 7.4-23.
Alkylate Blending Cost in 2012
(cents/gallon)
2012
Alkylate
Production
Cost
Octane Value
Alkylate
Blending
Cost
PADD1
143
-3.6
139
PADD2
143
-1.9
141
PADD3
143
o o
-J.J
139
PADD4
N/A
N/A
N/A
PADD5
(ex CA)
143
-4.3
138
CA
N/A
N/A
N/A
7.4.5 Refinery Gasoline Volumes and Costs
In the sections above, we estimated the volume changes associated with the phase out of
MTBE, the subsequent conversion of much of the isobutylene to alkylate, the increase in ethanol
use and ethanol's impact on summertime RFG. In this section we estimate the volume of
refinery-produced gasoline that would change as a result of the aforementioned changes in
gasoline blendstocks.
To account for the changes in gasoline and diesel fuel volumes it was necessary to
establish a baseline from which to compare the various Energy Act control cases. As
summarized in Section 2.1-3, 2004 basecase volumes were established for each PADD
identifying the volumes of ethanol, MTBE and refinery produced gasoline or gasoline
blendstock. Subsequently, in Section 2.1-7 we describe how we grew the 2004 volumes to 2012
to derive reference case volumes from which to compare the various control cases. However,
because of the volumetric increase in low energy density ethanol as well as changes in other
gasoline blendstocks with varying energy density, it was necessary to match energy content of
the control cases to that of the reference case. We estimated the energy content of the gasoline
pool for each PADD of the reference case by assigning each gasoline blendstock an energy
content shown in Table 7.4-24.
Table 7.4-24.
Energy Content of Gasoline and Gasoline Blendstocks
(BTU/gallon)
Blendstock
Ethanol
MTBE
Alkylate
Butane
Gasoline
Energy Content
76,000
93,500
115,000
94,000
115,000
249
-------
Applying the gasoline blendstock energy contents to the volumes represented in the 2012
reference case results in the total energy content for the gasoline in each PADD, for California
and the total U.S. shown in Table 7.4-25.
Table 7.4-25.
Summary of 2012 Reference Case Volumes and Energy Content
Ethanol
MTBE
Gasoline
Total
Vol.
Total
Energy
Content
Volume (million gallons)
PADD1
735
1,514
52,538
54,787
PADD 2
1,800
1,646
41,399
43,201
PADD 3
88
555
22,317
22,959
PADD 4
93
0
4,966
5,059
PADD 5
exCA
232
20
8,566
8,820
CA
950
0
15,573
16,523
USA
3,898
2,091
145,359
151,349
Total Energy Content (1015BTUs)
6.24
4.90
2.62
0.58
1.00
1.86
17.21
We used the total energy content of the reference case regional gasoline pools as the basis
for estimating the volume of refinery-produced gasoline for each control case. The control cases
are: Maximum RFG, 7.2 billion gallons ethanol; Minimum RFG, 7.2 billion gallons ethanol;
Maximum RFG, 9.6 billion gallons ethanol; and Minimum RFG, 9.6 billion gallons ethanol. For
each control case the appropriate volume of ethanol from Table 2.1-14 is applied to each PADD.
The increase in ethanol volumes causes increases in gasoline vapor pressure which must be
accounted for by reductions in butane. In subsection 7.4.2, we discussed this effect for ethanol
blended into summertime RFG. However, the increased blending of ethanol into wintertime
gasoline may cause the reduction in wintertime butane content as well. The American Standard
for Testing Materials (ASTM) has established vapor pressure limits on wintertime gasoline to
ensure that the gasoline will not negatively impact motor vehicle driveability. The RVP limits in
midwinter are primarily 13.5 PSI in the South and 15 PSI in the North. There is also a
vapor/liquid standard which is designed to prevent vapor lock, and it is likely the more stringent
standard in the winter. According to a large refiner with refineries all across the U.S., and Jacobs
Engineering which is a refining industry consulting firm, refiners today are blending butane into
wintertime gasoline up to the ASTM standards and the ASTM standards prevents them from
blending more butanes available to them. Because refiners are controlled by these ASTM
standards today it suggests that the ethanol newly blended into wintertime gasoline will result in
a commensurate decrease in butane content to balance the RVP of those pools, thus we have
conducted our analysis based on this.
Yet we also learned that some states have put in place a 1 psi blending waiver for ethanol
blended into wintertime gasoline. It is possible that in response to the increased use of ethanol
that more states will put in place such waivers. It is also possible that some refiners are butane
short and thus could blend more butanes into their wintertime gasoline under the ASTM
standards which could allow them to absorb the vapor pressure increase in their wintertime
gasoline. Because of these uncertainties that could allow refiners to blend in ethanol into
250
-------
wintertime gasoline without having to remove butanes, we also conducted our analysis to capture
this possibility. Despite this potential flexibility, we still assume that the blending of ethanol into
summertime RFG will result in the ultimate removal of butane after the removal and reblending
of pentanes. This is because if waivers are the primary source for allowing more ethanol
blending into wintertime gasoline, these waivers would not provide any relief for the reblended
pentanes.
To understand the change in ethanol volume blended into wintertime gasoline for
estimating the change in butane blended into wintertime gasoline, we needed to know the
volume of ethanol blended into wintertime gasoline in the reference case. In the volume analysis
summarized in Sections 2.1-3 and 2.1-7 above, which established the base and reference cases
for ethanol consumption, the summer/winter split was not established for the volume of ethanol
in gasoline, so we estimated that split here. We assumed that rather than store up ethanol for use
during one season or the other, that ethanol is produced and used year-round. Thus, the
volumetric summer versus wintertime use of ethanol is determined by the relative volumes of
gasoline used during the two seasons. This split is assumed to be 55 percent used during the
winter, and 45 percent used during the summer. For estimating the volume of butane which must
be removed from the gasoline because of the addition of ethanol, we assumed that ethanol will
be blending into gasoline at 10 volume percent, except for California where it would continue to
be blended at 5.7 volume percent. For the gasoline blended with the ethanol and when we
assume that butanes will be removed, 2 volume percent of butanes would have to be removed to
accommodate the ethanol. Table 7.4-26 summarizes the summertime RFG and wintertime RFG
and CG volumes of ethanol and estimated change in both summertime and wintertime butane
volume blended into gasoline. For the min-RFG cases, ethanol is coming out of the
summertime RFG pools in many PADDs which can result in positive butane values which
indicates that butanes are being blended back into gasoline, while negative values indicates that
butanes are being withdrawn from the gasoline pool.
251
-------
Table 7.4-26. Estimated Changes in Summertime RFG and Wintertime Gasoline
Ethanol Volumes and Their Impact on Butane Blending into Gasoline
(million gallons in 2012)
Reference Case
Summertime RFG Ethanol
Wintertime RFG & CG
Ethanol
7.2 Bil Gals, Max RFG
Summertime RFG Ethanol
Wintertime RFG & CG
Ethanol
Change in Butane -
butanes removed in winter
Change in Butane -
butanes remain in winter
7.2 Bil Gals, Min RFG
Summertime RFG Ethanol
Wintertime RFG & CG
Ethanol
Change in Butane -
butanes removed in winter
Change in Butane -
butanes remain in winter
9.6 Bil Gals, Max RFG
Summertime RFG Ethanol
Wintertime RFG & CG
Ethanol
Change in Butane -
butanes removed in winter
Change in Butane -
butanes remain in winter
9.6 Bil Gals, Min RFG
Summertime RFG Ethanol
Wintertime RFG & CG
Ethanol
Change in Butane -
butanes removed in winter
Change in Butane -
butanes remain in winter
PADD
1
333
402
955
1,168
-250
-112
0
1,451
-128
60
955
1,594
-327
-112
0
2,137
-253
60
PADD
2
274
1,024
273
1,733
-128
0
137
2,223
-191
25
273
2,333
-236
0
137
2,400
-223
25
PADD
3
13
50
241
295
-85
-41
0
124
-11
2
241
346
-95
-41
0
362
-54
2
PADD
4
0
54
0
30
4
0
0
182
-23
0
0
192
-25
0
0
280
-41
0
PADD
5
exCA
105
127
31
175
5
13
0
235
0
19
31
236
-6
13
0
352
-22
19
CA
430
520
430
525
-2
0
107
525
57
58
430
528
-1
0
107
528
57
58
USA
1,155
2,178
1,932
3,926
-456
-140
244
4,739
-297
164
1,932
5,230
-690
-140
244
6,059
-535
164
252
-------
The volume of MTBE shown in each PADD in the reference case in Table 7.4-25 is
eliminated and is replaced with 0.84 gallons alkylate for each gallon of MTBE reduced. We
assumed that the replacement volume of alkylate would be used in the same PADD proportional
to the removed MTBE volume.
The volume of refinery-produced gasoline needed to balance each PADD's total gasoline
pool is determined by a BTU balance. For each PADD and for each control case, the volume
change (relative to the reference case) of each gasoline blendstock (ethanol, alkylate and butane)
is multiplied times the BTU content of the blendstock and subtracted from the total BTU content
of the reference case gasoline pool shown in Table 7.4-26 above. The estimated volume of
refinery-produced gasoline needed to make up the balance of the gasoline pool for each control
case is calculated by dividing the BTU content of gasoline, also shown in Table 7.4-24, into the
remaining BTU value calculated by subtracting the BTU content of the other blendstocks from
the reference case gasoline pool BTU content. The BTU-balanced gasoline pool volumes for
each PADD and control case are shown in Table 7.4-27. Also, the change in ethanol and
gasoline volume between each control case and the reference case is calculated and shown. We
also estimate the volumes for each case if butanes currently blended into gasoline are not
removed due to the blending of ethanol into wintertime gasoline. These volumes are shown in
Table 7.4-28.
253
-------
Table 7.4-27.
Estimated 2012 Volumes by PADD - Butanes Removed in Winter
(million gallons in 2012)
7.2 Bil Gals, Max RFC
Ethanol
Change in Ethanol
Change in MTBE
New Alkylate
Change in Butane
Gasoline
Change in Gasoline
Change in Gasoline (%)
7.2 Bil Gals, Min RFC
Ethanol
Change in Ethanol
Change in MTBE
New Alkylate
Change in Butane
Gasoline
Change in Gasoline
Change in Gasoline (%)
9.6 Bil Gals, Max RFC
Ethanol
Change in Ethanol
Change in MTBE
New Alkylate
Change in Butane
Gasoline
Change in Gasoline
Change in Gasoline (%)
9.6 Bil Gals, Min RFC
Ethanol
Change in Ethanol
Change in MTBE
New Alkylate
Change in Butane
Gasoline
Change in Gasoline
Change in Gasoline (%)
PADD
1
2,123
1,389
-1,514
1,277
-250
51,779
-759
-1.4
1,682
947
-1,514
1,277
-128
51,972
-566
-1.1
2,900
2,165
-1,514
1,277
-327
51,329
-1,209
-2.3
2,925
2,190
-1,514
1,277
-253
51,252
-1,286
-2.4
PADD
2
3,151
1,351
-1,646
1
-128
40,611
-789
-1.9
3,904
2,104
-1,646
1,388
-191
41,165
-1,234
-3.0
4,243
2,443
-1,646
1,388
-236
39,978
-1,422
-3.4
4,226
2,426
-1,646
1,388
-223
39,978
-1,421
-3.4
PADD
3
560
472
-555
468
-85
22,058
-259
-1.2
188
100
-555
468
-11
22,243
-74
-0.3
654
566
-555
468
-95
22,003
-313
-1.4
629
542
-555
468
-54
21,986
-331
-1.5
PADD
4
56
-36
0
0
4
4,987
20
0.4
334
241
0
0
-23
4,826
-141
-2.8
352
259
0
0
-25
4,815
-151
-3.0
511
418
0
0
-41
4,723
-243
-4.9
PADD
5
exCA
353
121
-20
17
5
8,482
-84
-1.0
459
227
-20
17
0
8,416
-150
-1.8
492
260
-20
17
-6
8,399
-167
-2.0
672
440
-20
17
-22
8,292
-274
-3.2
CA
955
0
0
0
-2
15,571
0
0
633
-317
0
0
57
15,736
163
1.0
960
0
0
0
-1
15,567
0
0
636
-314
0
0
57
15,734
161
1.0
USA
7,200
3,302
-2,091
1,764
-456
143,486
-1,873
-1.3
7,200
3,302
-2,091
1,764
-297
143,357
-2,002
-1.4
9,600
5,702
-2,091
1,764
-690
142,092
-3,267
-2.2
9,600
5,702
-2,091
1,764
-535
141,965
-3,394
-2.3
254
-------
Table 7.4-28.
Estimated 2012 Volumes by PADD - Butanes not Removed in Winter
(million gallons in 2012)
7.2 Bil Gals, Max RFC
Ethanol
Change in Ethanol
Change in MTBE
New Alkylate
Change in Butane
Gasoline
Change in Gasoline
Change in Gasoline (%)
7.2 Bil Gals, Min RFC
Ethanol
Change in Ethanol
Change in MTBE
New Alkylate
Change in Butane
Gasoline
Change in Gasoline
Change in Gasoline (%)
9.6 Bil Gals, Max RFC
Ethanol
Change in Ethanol
Change in MTBE
New Alkylate
Change in Butane
Gasoline
Change in Gasoline
Change in Gasoline (%)
9.6 Bil Gals, Min RFC
Ethanol
Change in Ethanol
Change in MTBE
New Alkylate
Change in Butane
Gasoline
Change in Gasoline
Change in Gasoline (%)
PADD
1
2,123
1,389
-1,514
1,277
-112
51,666
-871
-1.7
1,682
947
-1,514
1,277
60
51,818
-720
-1.4
2,900
2,165
-1,514
1,277
-112
51,154
-1,384
-2.6
2,925
2,190
-1,514
1,277
60
50,996
-1,542
-2.9
PADD
2
3,151
1,351
-1,646
1
0
40,506
-893
-2.2
3,904
2,104
-1,646
1,388
25
39,988
-1,444
-3.4
4,243
2,443
-1,646
1,388
0
39,785
-1,614
-3.9
4,226
2,426
-1,646
1,388
25
39,775
-1,624
-3.9
PADD
3
560
472
-555
468
-41
22,021
-295
-1.3
188
100
-555
468
2
22,232
-85
-0.4
654
566
-555
468
-41
21,960
-357
-1.6
629
542
-555
468
2
21,940
-376
-1.7
PADD
4
56
-36
0
0
0
4,990
24
0.5
334
241
0
0
0
4,806
-159
-3.2
352
259
0
0
0
4,795
-171
-3.4
511
418
0
0
0
4,690
-277
-5.6
PADD
5
exCA
353
121
-20
17
13
8,475
-91
-1.1
459
227
-20
17
19
8,400
-166
-1.9
492
260
-20
17
13
8,382
-183
-2.1
672
440
-20
17
19
8,259
-306
-3.6
CA
955
0
0
0
0
15,569
0
0
633
-317
0
0
58
15,734
162
1.0
960
0
0
0
0
15,566
0
0
636
-314
0
0
58
15,733
160
1.0
USA
7,200
3,302
-2,091
1,764
-140
143,228
-2,131
-1.5
7,200
3,302
-2,091
1,764
164
142,980
-2,379
-1.6
9,600
5,702
-2,091
1,764
-140
141,642
-3,717
-2.6
9,600
5,702
-2,091
1,764
164
141,394
-3,965
-2.7
255
-------
7.4.6 Overall Gasoline Costs
In the sections above, we estimated the costs for producing and distributing additional
volumes of ethanol, ending the use of MTBE and reusing the MTBE feedstock isobutylene for
producing alkylate, removing butanes, and for decreases in refinery produced gasoline. This
section pulls these individual parts together to estimate the overall costs for these fuel changes.
In addition to the costs for increasing and decreasing the volumes of these various gasoline
blendstocks, we account for their energy density and octane value.
The costs of these fuels changes are expressed three different ways. First, we express the
fuels costs based on the production costs for each gasoline blendstock, including ethanol, without
the ethanol consumption subsidies. Second, we express the cost with the ethanol consumption
subsidies included since this portion of the renewable fuels costs will be not be represented to the
consumer in its fuels costs, but instead is reflected in the federal and state tax payments. Third
we present the cost to refiners by assigning historical prices adjusted to 2012 for ethanol and the
other gasoline blendstocks.
The costs for each PADD of each control case are estimated by multiplying the change in
volume for each gasoline blendstock, relative to the reference case, times its production,
distribution and octane blending costs for the cost analyses, or times the projected prices for the
cost to refiners analysis. The production and octane blending costs for ethanol are summarized
above in Section 7.1. The distribution costs for ethanol are summarized in Section 7.3. The
ethanol distribution costs vary by PADD, and also based on whether the ethanol is being added
or withdrawn from the PADD. When ethanol is added to gasoline, the distribution costs include
both capital and operating costs, while when ethanol is withdrawn from gasoline, only the
operating portion of the distribution costs are subtracted. The ethanol blending costs for adding
ethanol to summertime RFG are from Table 7.4-9. The production and octane blending costs for
MTBE and alkylate are summarized in Tables 7.4-17 and 7.4-23 above. The distribution cost for
MTBE and alkylate are assumed to be 4 c/gal, the same as that for gasoline. The cost for
changes to butane content are based on the opportunity costs for butane which, based on Platts, is
36 c/gal less than gasoline as shown in Table 7.4-10. The cost of changes to refinery produced
gasoline is assumed to be represented by the bulk price of gasoline in each PADD from EIA's
2006 Petroleum Marketing Annual, plus 4 c/gal distribution costs.161162 The 2004 gasoline cost
is projected to 2012 costs based on the ratio of the wholesale gasoline price in 2012 to the
wholesale gasoline price in 2004 from AEO 2006. This ratio is 1.08. The cost to distribute
gasoline to terminals is assumed to remain the same in 2012 at 4 c/gal. These various estimated
costs, including production, blending and distribution costs, are summarized in Table 7.4-29.
256
-------
Table 7.4-29. Gasoline Blendstock Costs Used in Cost Analyses
Ethanol Cost
7.2 Bil Gals
Ethanol Cost
9.6 Bil Gals
MTBE Cost
Alkylate Cost
Butane Cost
Gasoline Cost
MaxRFG
MinRFG
MaxRFG
MinRFG
PADD
1
132
132
138
138
135
143
98
133
PADD
2
126
126
132
132
142
145
95
131
PADD
3
130
130
136
136
136
143
94
130
PADD
4
134
134
140
140
N/A
N/A
103
138
PADD
5
139
139
145
145
131
142
114
149
CA
139
139
145
145
N/A
N/A
118
153
7.4.6.1
Costs without Ethanol Consumption Subsidies
Tables 7.4-30 through 33 summarize the costs for each aspect of the fuels changes (i.e.,
adding ethanol, removing butane, removing MTBE...), the total costs, and the per-gallon costs
for each PADD and the U.S. for each of the four control scenarios.LLL These costs include all
fuel changes expected to occur between 2004 and 2012, including the elimination of the RFG
oxygen standard, the elimination of MTBE from gasoline in the U.S., and the dramatic expansion
in the use of ethanol. The costs represent the production, distribution and blending costs, but not
the ethanol consumption subsidies. The costs are presented assuming that butanes are removed,
or not removed, from the gasoline pool when ethanol is blended into wintertime gasoline. It was
not possible to isolate these gasoline blendstock changes from one another due to their
interrelationship. Consequently these costs cannot and should not be associated solely with the
Renewable Fuels Standard. Rather, they reflect a combination of the various impacts on fuel
quality discussed. To get at some idea of the costs of adding additional ethanol volume to the
gasoline pool, a comparison can be made between the 9.6 and 7.2 billion gallon cases as the
impacts of the MTBE removal and alkylate addition is made between the reference case and 7.2
billion gallon case and are not changing between the 7.2 and 9.6 billion gallon cases. Because of
the very large volume of ethanol being blended into gasoline, the per-gallon costs are indicated
for all the gasoline in each PADD, not just the gasoline volume blended with ethanol.
As can be seen in these tables, the aggregated costs of these various fuel changes is
estimated to cause a net cost without subsidy which ranges from 0.3 to 1 cent per gallon.
However, as shown in the following subsection, when the impact of the tax subsidy is included,
the cost of these fuel changes to the fuels industry and to consumers decreases dramatically
depending on the control case.
By looking at incremental costs from the 7.2 billion gallon ethanol cases to 9.6 billion
EPA typically assesses the social costs and benefits of its rulemakings. However, this analysis is more limited in
scope by evaluating the average cost of producing more ethanol or less gasoline without accounting for some of the
market distortions that impact the production costs. For example, some of the costs and cost savings of using more
ethanol could not be quantified. Some of these are discussed below in subsection 7.4.5.4 and also above in reference
to the farm subsidies impacting the price of corn. Similarly, there are government incentives for the production of
crude oil which could not be quantified which would affect the estimates of cost savings for the gasoline displaced
by the increased use of ethanol.
257
-------
gallon ethanol cases, the cost of adding in an additional 2.4 billion gallons of ethanol in isolation
becomes apparent.
Table 7.4-30.
Estimated Cost without Ethanol Subsidies for the 7.2 billion Gallon Ethanol Maximum
RFC Case
(million dollars per year and cents
Adding
Ethanol
RFG RVP Cost
Eliminating
MTBE
Adding
Alkylate
Changing
Butane
Volume
Change in
Gasoline
Production
Total Cost
Without
Subsidies
Per-Gallon
Cost Without
Subsidies
Changing
Butane
Volume
Change in
Gasoline
Production
Total Cost
Without
Subsidies
Per-Gallon
Cost Without
Subsidies
PADD1
1,567
62
-2,039
1,825
PADD2
1,568
0
-2
2
PADD3
529
15
-753
669
per gallon; Mv/bbl crude oil)
PADD4
-40
0
0
0
PADD5
140
-5
-27
25
CA
5
0
0
0
USA
3,769
72
-2,821
2,521
Butanes Removed in Winter
-245
-1,013
157
0.29
-122
-1,034
412
0.94
-80
-336
46
0.20
4
28
-7
-0.15
5
-126
11
0.13
-2
O
0
0
-439
-2,484
619
0.41
Butanes not Removed in Winter
-110
-1,163
141
0.26
0
-1,171
396
0.91
-39
-383
40
0.17
0
33
-7
-0.14
15
-137
11
0.12
0
-5
0
0
-133
-2,826
582
0.38
258
-------
Table 7.4-31.
Estimated Cost without Ethanol Subsidies for the 7.2 billion Gallon Ethanol Minimum
RFC Case
(million dollars
Adding
Ethanol
RFGRVPCost
Eliminating
MTBE
Adding
Alkylate
Changing
Butane
Volume
Change in
Gasoline
Production
Total Cost
Without
Subsidies
Per-Gallon
Cost Without
Subsidies
Changing
Butane
Volume
Change in
Gasoline
Production
Total Cost
Without
Subsidies
Per-Gallon
Cost Without
Subsidies
PADD1
1,067
-26
-2,039
1,825
per year and cents per gallon; Mv/bbl crude oil)
PADD2
2,442
-7
-2
2
PADD3
112
-1
-753
669
PADD4
268
0
0
0
PADD5
262
-7
-27
25
CA
-316
-33
0
0
USA
3,837
-74
-2,821
2,521
Butanes Removed in Winter
-126
-755
-54
-0.10
-182
-1,618
633
1.44
-10
-96
-78
-0.34
-24
-194
50
0.98
0
-224
28
0.31
67
250
-31
-0.19
-276
-2,638
548
0.36
Butanes not Removed in Winter
59
-961
-75
-0.14
24
-1,850
608
1.38
2
-110
-79
-0.35
0
-220
48
0.93
22
-248
26
0.29
68
249
-31
-0.19
174
-3,141
496
0.33
259
-------
Table 7.4-32.
Estimated Cost without Ethanol Subsidies for the 9.6 billion Gallon Ethanol Maximum
RFC Case
(million dollars
Adding
Ethanol
RFGRVPCost
Eliminating
MTBE
Adding
Alkylate
Changing
Butane
Volume
Change in
Gasoline
Production
Total Cost
Without
Subsidies
Per-Gallon
Cost Without
Subsidies
Changing
Butane
Volume
Change in
Gasoline
Production
Total Cost
Without
Subsidies
Per-Gallon
Cost Without
Subsidies
PADD1
2,573
62
-2,039
1,825
per year and cents per gallon; Mv/bbl crude oil)
PADD2
2,981
0
-2
2
PADD3
668
15
-753
669
PADD4
304
0
0
0
PADD5
316
-5
-27
25
CA
10
0
0
0
USA
6,852
72
-2,821
2,521
Butanes Removed in Winter
-320
-1,613
487
0.88
-225
-1,864
892
2.03
-89
-406
105
0.46
-26
-209
69
1.35
-7
-250
52
0.58
-2
-8
1
0
-668
-4,350
1,606
1.05
Butanes not Removed in Winter
-110
-1,848
462
0.84
0
-2,117
864
1.96
-39
-463
99
0.43
0
-237
67
1.30
15
-274
50
0.56
0
-10
0
0
-133
-4,948
1,543
1.01
260
-------
Table 7.4-33.
Estimated Cost without Ethanol Subsidies for the 9.6 billion Gallon Ethanol Minimum
RFC Case
(million dollars
Adding
Ethanol
RFGRVPCost
Eliminating
MTBE
Adding
Alkylate
Changing
Butane
Volume
Change in
Gasoline
Production
Total Cost
Without
Subsidies
Per-Gallon
Cost Without
Subsidies
Changing
Butane
Volume
Change in
Gasoline
Production
Total Cost
Without
Subsidies
Per-Gallon
Cost Without
Subsidies
PADD1
2,603
-26
-2,039
1,825
per year and cents per gallon; Mv/bbl crude oil)
PADD2
2,961
-7
-2
2
PADD3
640
-1
-753
669
PADD4
490
0
0
0
PADD5
535
-7
-27
25
CA
-332
-33
0
0
USA
6,897
-74
-2,821
2,521
Butanes Removed in Winter
-247
-1,717
398
0.72
-213
-1,863
877
1.99
-51
-458
77
0.33
-42
-336
112
2.16
-25
-409
92
1.02
67
247
-50
-0.31
-510
-4,507
1,506
0.99
Butanes not Removed in Winter
59
-2,058
363
0.66
24
-2,129
848
1.93
2
-488
70
0.30
0
-382
108
2.07
22
-459
88
0.99
68
246
-50
-0.31
174
-5,270
1,427
0.93
Crude oil prices are much higher today which decreases the relative cost of producing
and blending in more ethanol into gasoline. EIA predicts that crude oil prices will decrease in
the future and average $47 per barrel in 2012. However, continued tight supplies caused by
strong worldwide demand along with continued unrest in the Middle East could cause crude oil
prices to remain high. For this reason, we conducted a sensitivity analysis assuming that crude
261
-------
oil is priced at around $70 per barrel. For this sensitivity analysis we simply ratioed the gasoline
production costs, MTBE and alkylate feedstock costs upwards by a 1.38 multiplication factor to
adjust these prices. The factor was estimated based on the ratio of wholesale gasoline price
increase going from 2004 to 2012, which is 1.08 as discussed above, compared to the ratio of
crude oil price increase of the projected crude oil price in 2012, which is $47/bbl, over the
average crude oil price in 2004 which was $41 per barrel. Comparing these two ratios
established a relative price increase for gasoline of 0.54 c/gal for every 1 cent per gallon increase
in crude oil. The crude oil price ratio of $70 per barrel versus the 2004 value of $41 per barrel
was multiplied by 0.54 to establish the 1.38 gasoline and gasoline blendstock production cost
adjustment factor. We adjusted the gasoline, MTBE feedstocks (except for methanol which is
produced from natural gas and so is assumed to remain at the 2012 prices) and alkylate
feedstocks using this factor. We set butane prices at 36 c/gal lower than the newer gasoline
costs, thus maintaining the same relative butane opportunity cost. We did not adjust the
distribution costs, any of the utility costs, octane costs or ethanol prices based on the assumption
that these would change much less and for simplicity sakes, we kept these input costs the same as
our main analysis.
262
-------
Table 7.4-34.
Estimated Cost without Ethanol Subsidies for the 7.2 billion Gallon Ethanol Maximum
RFC Case with Crude Oil Priced at $70 per Barrel
(million dollars per year and cents per gallon)
Adding
Ethanol
RFG RVP Cost
Eliminating
MTBE
Adding
Alkylate
Changing
Butane
Volume
Change in
Gasoline
Production
Total Cost
without
Subsidies
Per-Gallon
Cost without
Subsidies
Changing
Butane
Volume
Change in
Gasoline
Production
Total Cost
without
Subsidies
Per-Gallon
Cost without
Subsidies
PADD1
1,567
62
-2,467
2,289
PADD2
1,568
0
O
O
PADD3
529
15
-909
839
PADD4
-40
0
0
0
PADD5
140
-5
-33
31
CA
5
0
0
0
USA
3,769
72
-3,412
3,162
Butanes Removed in Winter
-336
-336
-173
-0.32
-167
-167
85
0.19
-110
-110
-63
-0.27
6
6
2
0.04
7
7
-20
-0.23
-3
O
-2
-0.01
-603
-603
-171
-0.11
Butanes not Removed in Winter
-151
-1,480
-180
-0.33
0
-1,490
78
0.18
-53
-487
-65
-0.28
0
42
2
0.04
21
-174
-20
-0.23
0
-7
-2
0
-183
-3,595
-187
-0.12
263
-------
Table 7.4-35.
Estimated Cost without Ethanol Subsidies for the 7.2 billion Gallon Ethanol Minimum
RFC Case with Crude Oil Priced at $70 per Barrel
(million dollars per year and cents per gallon)
Adding
Ethanol
RFGRVPCost
Eliminating
MTBE
Adding
Alkylate
Changing
Butane
Volume
Change in
Gasoline
Production
Total Cost
without
Subsidies
Per-Gallon
Cost without
Subsidies
Changing
Butane
Volume
Change in
Gasoline
Production
Total Cost
without
Subsidies
Per-Gallon
Cost without
Subsidies
PADD1
1,069
-26
-2,467
2,289
PADD2
2,442
-7
-3
3
PADD3
112
-1
-909
839
PADD4
268
0
0
0
PADD5
262
-7
-33
31
CA
-316
-33
0
0
USA
3,837
-74
-3,412
3,162
Butanes Removed in Winter
-172
-961
-270
-0.49
-251
-2,059
125
0.28
-14
-122
-94
-0.41
-33
-247
-12
-0.22
-1
-286
-33
-0.37
91
318
61
0.37
-380
-3,556
-223
-0.15
Butanes not Removed in Winter
80
-1,223
-279
-0.51
32
-2,353
113
0.26
O
-140
-95
-0.42
0
-281
-12
-0.24
29
-316
-33
-0.37
93
317
-61
-0.37
238
-3,996
-245
-0.16
264
-------
Table 7.4-36.
Estimated Cost without Ethanol Subsidies for the 9.6 billion Gallon Ethanol Maximum
RFC Case with Crude Oil Priced at $70 per Barrel
(million dollars per year and cents per gallon)
Adding
Ethanol
RFGRVPCost
Eliminating
MTBE
Adding
Alkylate
Changing
Butane
Volume
Change in
Gasoline
Production
Total Cost
without
Subsidies
Per-Gallon
Cost without
Subsidies
Changing
Butane
Volume
Change in
Gasoline
Production
Total Cost
without
Subsidies
Per-Gallon
Cost without
Subsidies
PADD1
2,573
62
-2,467
2,289
PADD2
2,981
0
-3
3
PADD3
668
15
-909
839
PADD4
304
0
0
0
PADD5
316
-5
-33
31
CA
10
0
0
0
USA
6,851
72
-3,412
3,162
Butanes Removed in Winter
-438
-2,053
-35
-0.06
-309
-2,371
300
0.68
-122
-517
-25
-0.11
-35
-265
3
0.06
-10
-318
-19
-0.21
-2
-10
-2
0
-917
-5,535
222
0.15
Butanes not Removed in Winter
-150
-2,351
-45
-0.08
0
-2,693
288
0.65
-53
-589
-28
-0.12
0
-301
2
0.04
21
-349
-19
-0.21
0
-12
-2
0
-183
-6,295
196
0.13
265
-------
Table 7.4-37.
Estimated Cost without Ethanol Subsidies for the 9.6 billion Gallon Ethanol Minimum
RFC Case with Crude Oil Priced at $70 per Barrel
(million dollars per year and cents per gallon)
Adding
Ethanol
RFGRVPCost
Eliminating
MTBE
Adding
Alkylate
Changing
Butane
Volume
Change in
Gasoline
Production
Total Cost
without
Subsidies
Per-Gallon
Cost without
Subsidies
Changing
Butane
Volume
Change in
Gasoline
Production
Total Cost
without
Subsidies
Per-Gallon
Cost without
Subsidies
PADD1
2,602
-26
-2,467
2,289
PADD2
2,961
-7
-3
3
PADD3
640
-1
-909
839
PADD4
490
0
0
0
PADD5
535
-7
-33
31
CA
-332
-33
0
0
USA
6,897
-74
-3,412
3,162
Butanes Removed in Winter
-339
-2,185
-125
-0.23
-293
-2,370
291
0.66
-70
-545
-46
-0.20
-57
-428
5
0.10
-33
-521
-28
-0.32
91
315
41
0.25
-701
-5,734
138
0.09
Butanes not Removed in Winter
80
-2,618
-140
-0.25
32
-2,708
278
0.63
O
-621
-49
-0.21
0
-486
4
0.07
29
-584
-29
-0.32
93
313
41
0.25
238
-6,705
105
0.07
7.4.6.2
Gasoline Costs Including Subsidy
Tables 7.4-38 through 41 express the total and per-gallon gasoline costs for the four
control scenarios with the federal and state ethanol subsidies included. These subsidies reduce
the cost to fuel producers and to consumers seen "at the pump" for fuel purchases, while the rest
266
-------
of the costs are paid through taxes.MMM The federal tax subsidy is 51 c/gal for each gallon of
new ethanol blended into gasoline. The state tax subsidies are summarized above in Section
2.1.43 and the cost reduction to the fuel industry and consumers are estimated by multiplying
the subsidy times the volume of new ethanol estimated to be used in the state. We also show
how the subsidized costs change if crude oil is priced at $70 per barrel which is summarized in
Tables 7.4-42 through 7.4-45.
Table 7.4-38.
Estimated Cost Including Subsidies for the 7.2 billion Gallon Ethanol Maximum RFG Case
(million dollars per year and cents per gallon; $47/bbl crude oil)
Federal
Subsidy
State
Subsidies
Total Cost
Including
Subsidies
Per-Gallon
Cost
Including
Subsidies
Total Cost
Including
Subsidies
Per-Gallon
Cost
Including
Subsidies
PADD1
-708
0
PADD2
-689
-180
PADD3
-241
0
PADD4
19
0
PADD5
-62
0
CA
O
0
USA
-1,684
-180
Butanes Removed in Winter
-551
-1.00
-458
-1.05
-195
-0.85
11
0.22
-50
-0.56
-3
-0.02
-1,246
-0.82
Butanes not Removed in Winter
-567
-1.03
-473
-1.08
-201
-0.87
12
0.23
-51
-0.57
O
-0.02
-1,282
-0.84
The subsidy ensures that the ethanol's price set by the marketplace would surely be less than its production
cost, but it may not be as low as the subsidized production cost either. This analysis of subsidized costs sets a lower
bound on ethanol's price, and it is likely that ethanol's actual price would be somewhere inbetween the two
analyses. Additionally, other factors affect the price of ethanol and gasoline which are complicated and beyond the
scope of this analysis to project.
267
-------
Table 7.4-39.
Estimated Cost Including Subsidies for the 7.2 billion Gallon Ethanol Minimum RFG Case
(million dollars per year and cents per gallon; $47/bbl crude oil)
Federal
Subsidy
State
Subsidies
Total Cost
Including
Subsidies
Per-Gallon
Cost
Including
Subsidies
Total Cost
Including
Subsidies
Per-Gallon
Cost
Including
Subsidies
PADD1
-483
-7
PADD2
-1,073
-166
PADD3
-51
0
PADD4
-123
0
PADD5
-116
0
CA
162
0
USA
-1,684
-173
Butanes Removed in Winter
-543
-0.99
-606
-1.38
-129
-0.56
-73
-1.42
-87
-0.99
130
0.79
-1,308
-0.86
Butanes not Removed in Winter
-564
-1.03
-632
-1.44
-130
-0.57
-75
-1.47
-89
-1.01
130
0.79
-1,361
-0.89
268
-------
Table 7.4-40.
Estimated Cost Including Subsidies for the 9.6 billion Gallon Ethanol Maximum RFG Case
(million dollars per year and cents per gallon; $47/bbl crude oil)
Federal
Subsidy
State
Subsidies
Total Cost
Including
Subsidies
Per-Gallon
Cost
Including
Subsidies
Total Cost
Including
Subsidies
Per-Gallon
Cost
Including
Subsidies
PADD1
-1,104
-6
PADD2
-1,246
-183
PADD3
-289
0
PADD4
-132
0
PADD5
-133
0
CA
-4
0
USA
-2,908
-189
Butanes Removed in Winter
-623
-1.13
-537
-1.22
-183
-0.80
-63
-1.22
-81
-0.91
-4
-0.03
-1,492
-0.98
Butanes not Removed in Winter
-648
-1.17
-565
-1.28
-190
-0.82
-65
-1.27
-83
-0.93
-5
0
-1,555
-1.02
269
-------
Table 7.4-41.
Estimated Cost Including Subsidies for the 9.6 billion Gallon Ethanol Minimum RFG Case
(million dollars per year and cents per gallon; $47/bbl crude oil)
Federal
Subsidy
State
Subsidies
Total Cost
Including
Subsidies
Per-Gallon
Cost
Including
Subsidies
Total Cost
Including
Subsidies
Per-Gallon
Cost
Including
Subsidies
PADD1
-1,117
-6
PADD2
-1,237
-169
PADD3
-276
0
PADD4
-213
0
PADD5
-224
0
CA
160
0
USA
-2,908
-176
Butanes Removed in Winter
-725
-1.31
-529
-.20
-200
-0.87
-101
-1.95
-133
-1.48
110
0.67
-1,578
-1.03
Butanes not Removed in Winter
-760
-1.38
-558
-1.27
-206
-0.90
-106
-2.03
-136
-1.52
110
0.67
-1,657
-1.08
270
-------
Table 7.4-42.
Estimated Cost Including Subsidies for the 7.2 billion Gallon Ethanol Maximum RFG Case
with Crude Oil Priced at $70 per Barrel
(million dollars per year and cents per gallon)
Federal
Subsidy
State
Subsidies
Total Cost
Including
Subsidies
Per-Gallon
Cost
Including
Subsidies
Total Cost
Including
Subsidies
Per-Gallon
Cost
Including
Subsidies
PADD1
-708
0
PADD2
-689
-180
PADD3
-241
0
PADD4
19
0
PADD5
-62
0
CA
-3
0
USA
-1,684
-180
Butanes Removed in Winter
-881
-1.60
-785
-1.80
-304
-1.32
20
-0.40
-82
-0.93
-4
-0.03
-2,035
-1.34
Butanes not Removed in Winter
-888
-1.62
-791
-1.81
-306
-1.33
21
0.41
-82
-0.93
-4
-0.03
-2,051
-1.35
271
-------
Table 7.4-43.
Estimated Cost Including Subsidies for the 7.2 billion Gallon Ethanol Minimum RFG Case
with Crude Oil Priced at $70 per Barrel
(million dollars per year and cents
Federal
Subsidy
State
Subsidies
Total Cost
Including
Subsidies
Per-Gallon
Cost
Including
Subsidies
Total Cost
Including
Subsidies
Per-Gallon
Cost
Including
Subsidies
PADD1
-483
-6
PADD2
-1,073
-166
PADD3
-51
0
PADD4
-123
0
per gallon)
PADD5
-116
0
CA
162
0
USA
-1,684
-173
Butanes Removed in Winter
-759
-1.39
-1,115
-2.54
-145
-0.64
-135
-2.62
-149
-1.67
223
1.36
-2,080
-1.37
Butanes not Removed in Winter
-768
-1.40
-1,126
-2.56
-146
-0.64
-135
-2.63
-149
-1.67
222
1.36
-2,102
-1.38
272
-------
Table 7.4-44.
Estimated Cost Including Subsidies for the 9.6 billion Gallon Ethanol Maximum RFG Case
with Crude Oil Priced at $70 per Barrel
(million dollars per year and cents
Federal
Subsidy
State
Subsidies
Total Cost
Including
Subsidies
Per-Gallon
Cost
Including
Subsidies
Total Cost
Including
Subsidies
Per-Gallon
Cost
Including
Subsidies
PADD1
-1,104
-6
PADD2
-1,246
-183
PADD3
-289
0
PADD4
-132
0
per gallon)
PADD5
-133
0
CA
-4
0
USA
-2,908
-189
Butanes Removed in Winter
-1,145
-2.08
-1,128
-2.57
-314
-1.36
-129
-2.51
-151
-1.70
-7
-0.04
-2,875
-1.88
Butanes not Removed in Winter
-1,155
-2.09
-1,141
-2.59
-317
-1.38
-130
-2.52
-152
-1.70
-7
-0.04
-2,901
-1.90
273
-------
Table 7.4-45.
Estimated Cost Including Subsidies for the 9.6 billion Gallon Ethanol Minimum RFG Case
with Crude Oil Priced at $70 per Barrel
(million dollars per year and cents
Federal
Subsidy
State
Subsidies
Total Cost
Including
Subsidies
Per-Gallon
Cost
Including
Subsidies
Total Cost
Including
Subsidies
Per-Gallon
Cost
Including
Subsidies
PADD1
-1,117
-6
PADD2
-1,237
-169
PADD3
-276
0
PADD4
-213
0
per gallon)
PADD5
-224
0
CA
160
0
USA
-2,908
-176
Butanes Removed in Winter
-1,145
-2.08
-1,128
-2.57
-314
-1.36
-129
-2.51
-151
-1.70
-7
-0.04
-2,875
-1.88
Butanes not Removed in Winter
-1,263
-2.29
-1,129
-2.56
-325
-1.41
-210
-4.03
-253
-2.82
201
1.23
-2,978
-1.95
7.4.6.3
Costs to Refiners
Whether refiners choose to blend ethanol depends on the economic incentive to do so.
This in turn depends on the price they must pay for the ethanol not the production costs of the
ethanol. If they can produce a finished gasoline at a lower cost by purchasing and blending in
ethanol, than by refining crude oil, they will have an incentive to do so. Historically, the
subsidized price of ethanol has not been based on its octane value, which is very high, but
perhaps it is based on its impact on RVP. Prior to the year 2000, the subsidized price of ethanol
averaged about the same as the price of gasoline. After 2000, the subsidized price of ethanol
averaged about 12 c/gal lower than the price of gasoline.163 One possible reason for the relative
drop in ethanol prices starting in the year 2000 is that the Phase II RFG Program took effect.
The Phase RFG program required a much more stringent hydrocarbon standard. Perhaps the
relative price of ethanol to gasoline dropped to enable the ethanol manufacturers to participate in
the RFG markets, including the summer RFG market. Our analysis shows that the price of
ethanol would have to be lower by more than a dime to offset the cost of blending ethanol into
summertime RFG which correlates well with the historical price difference between gasoline and
ethanol. The cost analysis conducted above was reanalyzed based on a projected ethanol plant
gate price for ethanol set at 12 c/gal lower than PADD 2's bulk gasoline price (which is 4 c/gal
less than the PADD 2 gasoline price listed in Table 7.4-29 above). The price of ethanol in each
274
-------
PADD is based on the PADD ethanol plant gate price for ethanol plants located in the Midwest
plus the distribution costs. The prices of MTBE and alkylate were set based on the actual prices
listed by Platts for 2004, adjusted to 2012 using a ratio of 1.08 used to estimate the price of
gasoline in 2012 from its average price in 2004. These projected prices are summarized in Table
7.4-46.
Table 7.4-46.
Ethanol, MTBE and Alkylate Prices Used in the Cost to Refiners Analysis
Ethanol
7.2 &
9.6 Bil
Gals
MTBE
Alkylate
Max
RFG
Min
RFG
PADD1
127
127
149
147
PADD 2
121
121
149
147
PADD 3
125
125
149
147
PADD 4
129
129
-
-
PADD 5
-
134
149
147
CA
-
134
-
-
Another way that this analysis was conducted to model the cost to refiners was to balance
the volume of the gasoline pool of each control case using the final volume, not its BTU content.
This is appropriate because refiners typically ignore the BTU content of gasoline blendstocks
when blending up its gasoline. Instead the consumer usually absorbs the costs associated with
lower energy density gasoline. To set up the analysis to model the volume of gasoline produced,
the final gasoline pool volume of each PADD of each control case was matched to the same
gasoline pool volume of the same PADD of the reference case. Thus, the addition of each gallon
of ethanol caused a gallon decrease in gasoline (with similar assumptions made for the other
gasoline blendstocks). We believe that this better captures the refiners' perspective for the
blending of ethanol as well as the other fuel changes. The estimated costs to refiners for each of
the four control cases assuming the 12 cent price differential for ethanol is maintained are
summarized in Tables 7.4-47 through 50. This analysis suggests that refiners will willingly
make these changes including blending in more ethanol, even beyond the RFS minimum,
providing that crude oil prices remain above $47 per barrel, and that the relative pricing assumed
here also continues to hold true. If anything, ethanol's price to refiners may be lower than the 12
cents below the price of gasoline assumed here because the projected future crude oil price is
higher than of the higher crude oil price and the magnitude of the subsidy, further improving the
incentive to refiners for using ethanol.
275
-------
Table 7.4-47.
Estimated Cost to Refiners for the 7.2 billion Gallon Ethanol Maximum RFG Case
(million dollars per year and cents per gallon)
Adding
Ethanol
RFGRVP
Cost
Eliminating
MTBE
Adding
Alkylate
Changing
Butane
Volume
Change in
Gasoline
Production
Total Cost
Per-Gallon
Cost
Changing
Butane
Volume
Change in
Gasoline
Production
Total Cost
Per-Gallon
Cost
PADD1
1,767
62
-2,256
1,877
PADD2
1,640
0
-2
2
PADD3
591
15
-827
688
PADD4
-47
0
0
0
PADD5
162
-5
-31
26
CA
7
0
0
0
USA
4,121
72
-3,116
2,592
Butanes Removed in Winter
-244
-1,202
2
0
-122
-1,604
-85
-0.20
-80
-389
-1
-0.01
4
44
2
0.04
5
-183
-26
-0.29
-2
-5
0
0
-439
-3,339
-109
-0.07
Butanes not Removed in Winter
-110
-1,387
-47
-0.09
0
-1,771
-131
-0.30
-39
-446
-17
-0.07
0
50
3
0.07
15
-196
-29
-0.33
0
-8
-1
-0.01
-133
-3,758
-221
-0.15
276
-------
Table 7.4-48.
Estimated Cost to Refiners for the 7.2 billion Gallon Ethanol Minimum RFG Case
(million dollars per year and cents per gallon)
Adding
Ethanol
RFGRVP
Cost
Eliminating
MTBE
Adding
Alkylate
Changing
Butane
Volume
Change in
Gasoline
Production
Total Cost
Per-Gallon
Cost
Changing
Butane
Volume
Change in
Gasoline
Production
Total Cost
Per-Gallon
Cost
PADD1
1,205
-26
-2,257
1,877
PADD2
2,555
-7
-2
2
PADD3
125
-1
-827
688
PADD4
312
0
0
0
PADD5
303
-7
-31
26
CA
-423
-33
0
0
USA
4,078
-74
-3,116
2,592
Butanes Removed in Winter
-126
-775
-102
-0.19
-182
-2,508
-143
-0.33
-10
O
-28
-0.12
-24
-301
-13
-0.26
0
-333
-43
-0.48
67
399
11
0.07
-276
-3,521
-317
-0.21
Butanes not Removed in Winter
59
-1,027
-169
-0.31
24
-2,791
-220
-0.51
2
-20
-32
-0.14
0
-333
-21
-0.42
22
-362
-50
-0.56
68
398
11
0.06
174
-4,136
-481
-0.32
277
-------
Table 7.4-49.
Estimated Cost to Refiners for the 9.6 billion Gallon Ethanol Maximum RFG Case
(million dollars per year and cents per gallon)
Adding
Ethanol
RFGRVP
Cost
Eliminating
MTBE
Adding
Alkylate
Changing
Butane
Volume
Change in
Gasoline
Production
Total Cost
Per-Gallon
Cost
Changing
Butane
Volume
Change in
Gasoline
Production
Total Cost
Per-Gallon
Cost
PADD1
2,755
62
-2,257
1,877
PADD2
2,966
0
-2
2
PADD3
709
15
-827
688
PADD4
335
0
0
0
PADD5
348
-5
-31
26
CA
13
0
0
0
USA
7,125
72
-3,116
2,592
Butanes Removed in Winter
-320
-2,136
-19
-0.03
-225
-2,893
-152
-0.35
-89
-498
-2
-0.01
-26
-324
-14
-0.28
-7
-374
-44
-0.50
-2
-13
-1
-0.01
-668
-6,238
-232
-0.15
Butanes not Removed in Winter
-110
-2,422
-95
-0.17
0
-3,202
-236
-0.55
-39
-567
-21
-0.09
0
-358
-23
-0.45
15
-404
-51
-0.57
0
-15
-2
-0.01
-133
-6,969
-428
-0.28
278
-------
Table 7.4-50.
Estimated Cost to Refiners for the 9.6 billion Gallon Ethanol Minimum RFG Case
(million dollars per year and cents per gallon)
Adding
Ethanol
RFGRVP
Cost
Eliminating
MTBE
Adding
Alkylate
Changing
Butane
Volume
Change in
Gasoline
Production
Total Cost
Per-Gallon
Cost
Changing
Butane
Volume
Change in
Gasoline
Production
Total Cost
Per-Gallon
Cost
PADD1
PADD2
PADD3
PADD4
PADD5
CA
USA
Butanes Removed in Winter
2,786
-26
-2,257
1,877
2,947
-7
-2
2
678
-1
-827
688
541
0
0
0
589
-7
-31
26
-419
-33
0
0
7,122
-74
-3,116
2,592
Butanes Removed in Winter
-247
-2,269
-135
-0.25
-213
-2,888
-162
-0.38
-51
-519
-32
-0.14
-42
-522
-22
-0.44
-25
-621
-68
-0.78
67
395
10
0.06
-510
-6,424
-409
-0.27
Butanes not Removed in Winter
59
-2,686
-246
-0.45
24
-3,213
-250
-0.58
2
-592
-52
-0.23
0
-578
-37
-0.73
22
-681
-83
-0.94
68
393
10
0.06
174
-7,358
-659
-0.44
7.4.7 Overall Diesel Fuel Costs
Biodiesel fuel is added to highway and nonroad diesel fuel, which increases the volume
and therefore the supply of diesel fuel and thereby reduces the demand for refinery-produced
diesel fuel. In this section, we estimate the overall cost impact, considering how much refinery
based diesel fuel is displaced by the forecasted production volume of biodiesel fuel. The cost
impacts are evaluated considering the production cost of biodiesel with and without the subsidy
from the Biodiesel Blenders Tax credit program. Additionally, the diesel cost impacts are
quantified under two scenarios, with refinery diesel prices as forecasted by EIA's AEO 2006
with crude at $47 and with refinery diesel prices based on $70 per barrel crude oil.
279
-------
We estimate the net effect that biodiesel production has on overall cost for diesel fuel in
year 2012 using total production costs for biodiesel and diesel fuel. The costs are evaluated
based on how much refinery based diesel fuel is displaced by the biodiesel volumes as forecasted
by EIA, accounting for energy density differences between the fuels. The cost impact is
estimated from a 2004 year basis, by multiplying the production costs of each fuel by the
respective changes in volumes for biodiesel and estimated displaced diesel fuel. We further
assume that all of the forecasted bio-diesel fuel volume is used as transport fuel, neglecting
minor uses in the heating oil market.
For this analysis, the production costs for biodiesel fuel are based on our estimates based
on the USDA and NREL modeling costs derived in the preceding sections. We average these
results to developed costs for soy oil and yellow grease feedstocks. Additionally, the production
costs are based on EIA's projection in 2012, that half of the total biodiesel volume will be made
from soy oil feedstock with the remaining volume being produced from yellow grease. To these
estimates, we add distribution costs of 11.2 c/gal to the biodiesel production costs, reflecting the
distribution estimates derived in section 7.3.2. For the refinery diesel production costs in 2012,
we used the projected wholesale national average diesel price of 138 c/gal for the AEO 2006
analysis. For the scenario with crude at $70 per barrel, we used a wholesale refinery diesel price
of 175 c/gal. Distribution cost for refinery produced diesel fuel were assumed to be 4 c/gal, as
the AEO wholesale price projection does not include the costs associated with distribution, taxes
and marketing.
Our estimate for the reduction in refinery produced diesel fuel is based on EIA's forecast
for approximately 300 MM gallons of biodiesel in 2012, along with the 2004 year biodiesel
production volume of 25 MM gallons. With this and accounting for differences in energy
density between biodiesel and diesel fuel, we estimate in 2012 that the additional biodiesel
production reduces the need for 254 MM gallons of refinery produced diesel fuel. Table 7.4-51
contains the energy densities used in this analysis.
Table 7.4-51. Energy Content of Fuels per Gallon
Fuel
Biodiesel
Refinery Produced Diesel
LHV BTU
117
128
's/ Gallon3
,093
,700
a LHV is lower heating value.
For the AEO scenario, the net effect of biodiesel production on diesel fuel costs,
including the biodiesel blenders' subsidy, is a reduction in the cost of transport diesel fuel costs
by $90 MM per year, which equates to a fuel cost reduction of about 0.15 c/galNNN. Without the
subsidy, the transport diesel fuel costs are increased by $118 MM per year, or an increase of 0.20
Based on EIA's AEO 2006, the total volume of highway and off-road diesel fuel consumed in 2012 was
estimated at 58.9 billion gallons.
280
-------
c/gal for transport diesel fuel. These costs are summarized in Table 7.4-52. With crude at $70
per barrel, including the biodiesel blenders subsidy, results in a cost reduction of $184 MM per
year, or a reduction of 0.31 c/gal for the total transport diesel pool. Without the subsidy,
transport diesel costs are increased by $25 MM per year, or 0.04 c/gal. See Tables 7.4-53 and
7.4-54 for summaries of these costs.
Table 7.4-52.
Estimated Cost of Increased use of Biodiesel for AEO 2006 prices (2004 dollars)
Total Cost ($million/yr)
Per-Gallon Cost (cents/gallon)
Costs without Subsidy
118
0.20
Costs with Subsidy
-90
-0.15
Table 7.4-53.
Estimated Cost of Increased use of Biodiesel with Crude at $70 barrel (2004 dollars)
Total Cost ($million/yr)
Per-Gallon Cost (cents/gallon)
Costs without Subsidy
25
0.04
Costs with Subsidy
-184
-0.31
7.4.8 Summary of Gasoline and Diesel Fuel Costs
Tables 7.4-54 and 7.4-55 summarize the aggregate annual costs to gasoline and diesel
fuel in 2012 for the individual fuel changes as well as for the sum of all the fuel changes. The
costs are presented with and without the federal and state renewable fuel use subsidies. The
costs are presented for the case that crude oil is priced at $47 per barrel.
281
-------
Table 7.4-54.
Total Estimated Costs in 2012 not Including the Ethanol Consumption Subsidy
($47/bbl crude oil price and 2004 dollars)
Gasoline
Costs
Diesel Fuel
Costs
Total Costs
(Gasoline
and Diesel
Fuel)
Adding Ethanol
RFG RVP Control
MTBE Removal
Alkylate
Removed Butanes
Reduced Gasoline
Volume
Total Costs
Per-Gallon Cost
Total Costs
Per-Gallon Cost
Total Costs
7.2 Max
RFG
3,769
72
-2,821
2,520
-133 to -439
-2,484 to
-2,826
619 to 582
0.41 to 0.38
118
0.20
737 to 700
7.2 Min
RFG
3,837
-74
-2,821
2,520
-275 to 174
-2,63 8 to
-3,141
548 to 496
0.38 to 0.33
118
0.20
666 to 5 14
9.6 Max
RFG
6,852
72
2,821
2,520
-667 to -133
-4,3 50 to
-4,948
1,606 to
1,542
1.05 to 1.01
118
0.20
1,724 to
1,660
9.6 Min
RFG
6,897
-74
-2,821
2,520
-510 to 174
-4,507 to
-5,270
1,507 to
1,426
0.99 to 0.93
118
0.20
1,625 to
1,544
282
-------
Table 7.4-55.
Total Estimated Costs in 2012 Including the Ethanol Consumption Subsidy
($47/bbl crude oil price and 2004 dollars)
Gasoline
Costs
Diesel Fuel
Costs
Total Costs
(Gasoline
and Diesel
Fuel)
Total Costs
without
Subsidies
State Subsidies
Federal Subsidy
Total Cost with
Subsidies
Per Gallon Cost
with Subsidy
Total Cost
without Subsidy
Subsidy
Total Cost with
Subsidy
Per-Gallon Cost
with Subsidy
Total Costs
with Subsidy
7.2 Max RFC
619 to 582
-180
-1,684
-1,245 to
-1,282
-0.82 to
-0.84
118
-208
-90
-0.15
-1,335 to
-1,372
7.2 Min
RFC
548 to 496
-173
-1,684
-1,308 to
-1,361
-0.86 to
-0.89
118
-208
-90
-0.15
-1,398 to
-1,451
9.6 Max RFC
1,606 to 1,542
-189
-2,908
-1,491 to
-1,555
-0.98 to
-1.02
118
-208
-90
-0.15
-1,581 to
-1,645
9.6 Min
RFC
1,507 to
1,426
-176
-2,908
-1,578 to
-1,657
-1.03 to -1.08
118
-208
-90
-0.15
-1,668 to
-1,747
Throughout this analysis we conducted sensitivity analyses which attempt to capture
known uncertainties that could affect the costs for these fuel changes. The sensitivity analyses
conducted include variability in ethanol production or demand, variability in ethanol blended
into RFG, variability in whether wintertime gasoline RVP would remain fixed or be allowed to
increase by about 1 psi in response to additional ethanol blended into wintertime gasoline, and
analyzing a range of possible future crude oil prices. We believe these sensitivities evaluated the
most important uncertainties associated with this analysis. However, there are other
uncertainties such as the price of corn, the price of natural gas and the conversion percentage of
MTBE to other gasoline blendstocks which we did not evaluate.
7.4.9 Other Potential Economic Impacts not Quantified
The above discussion attempts to quantify the impact of expanded use of renewable fuels
on the cost of gasoline and diesel fuel. It does so by looking at the cost by itself, as well as in the
context of the state and federal tax subsidies for the renewable fuels which may lower the price
consumers pay at the pump, but which is still borne by consumers through tax payments. In
reality, there are many other economic impacts associated with the use of renewable fuels and
the fossil fuels they replace which go well beyond the scope of the analysis conducted for the
283
-------
RIA and we have not attempted to quantify them here. For example, there is a concern that
increased renewable fuel use may have adverse impacts on surface and ground water quality and
soil erosion. To quantify the economic impact associated with this would require extensive
analysis of the likely responses of farmers to the increased demand for renewable fuels, the cost
of actions taken to remedy the impacts, and the cost of any resulting health and welfare impacts.
At the same time, expanded renewable fuel use displaces fossil fuel production, distribution, and
use, which itself has its own impacts on surface and ground water quality. Thus, any economic
impacts would have to be assessed in a holistic manner looking at the impacts across the entire
fuel supply.
Similarly, the renewable fuel production costs assumed in our analysis may not reflect the
entire cost to society associated with the production of the corn and soybean feedstocks used in
their production due other state and federal agricultural policies. Direct payments,
countercyclical payments, marketing loans, and subsidized crop insurance are all examples of
policies outside of this rulemaking that impact the price of corn and soybeans that are not
reflected in the production cost for ethanol and biodiesel, but do impact costs borne by
consumers indirectly through taxes. Quantifying the incremental impacts of this rulemaking on
the effects of these pre-existing programs would represent a significant challenge. However, the
challenge is complicated even more by the direct and indirect economic support provided for the
production, supply, and distribution of the fossil fuels which would be replaced by these
renewable fuels. Again, any assessment of the overall costs to society for increase renewable use
would have to look at the economic support provided across the entire fuel supply. Such an
analysis is well beyond the scope of this RIA.
Despite our inability to fully capture all the potential impacts on the cost to society of
increased renewable fuel use, two potential impacts were touched on briefly in our analysis, and
these are discussed in this subsection.
Economic Impacts of Emission Changes
As discussed in Chapters 4.1 and 5.1, we estimate that there may be an increase in
emissions and a corresponding small increase in ozone resulting from the expanded use of
renewable fuels. Our vehicle and equipment emission estimates are highly uncertain, however,
given the lack of data in particular on vehicles and engines complying with the latest standards.
However, to the extent that there are emission and ozone increases resulting from the expanded
use of renewable fuels, there can be a cost associated with them. In some cases, areas that see an
increase in emissions resulting from renewable fuel use may be forced to take other actions to
offset these emission increases. In other cases, particularly in attainment areas, the impact, while
not affecting attainment, may adversely impact air quality and human health. It is extremely
difficult to provide any quantitative estimate of what the mitigation costs might be to offset
emission increases, or to quantify the health impacts resulting from the air quality impacts. Not
only are the emission and air quality impacts highly uncertain, but they are also very location
dependent. While we have made projections on where the ethanol use may rise or fall for the
purposes of estimating nationwide fuel cost impacts and potential emissions impacts, these
projections are much less reliable when trying to predict specific local air quality impacts.
284
-------
Despite all of the above caveats, we have attempted to provide an upper bound estimate
of the potential national-level cost impacts; we would expect most areas to have lower health
impact costs and certainly lower abatement costs. As a surrogate for total NOx control costs and
potential health impacts, we looked at the potential health costs associated with the secondary
nitrate PM resulting from the projected increases in NOx emissions. We note again that we
actually expect an overall decrease in ambient PM2 5 formation due to the increased use of
ethanol in fuel (See Chapter 5.2). However, inventory modeling suggests that an increase in the
use of ethanol will result in increased future emissions of NOx. These increased NOx emissions
will add secondary nitrate PM2.5 in the atmosphere and we can estimate the cost impacts
considering just this single effect.
In recent rulemakings we have monetized PM emission impacts, including those resulting
from changes in secondarily formed PM2.5 due to NOx emission changes. Using this information
as a guide, we provide a screening-level estimate of the monetized PM-related health impacts
associated with an increase in NOx emissions. This estimate is derived from dollar-per-ton
values based on recent benefits modeling derived from the 2007 Heavy-Duty Highway final rule
analysis164 which is based on REMSAD modeling conducted in 2000; and a dollar-per-ton
estimate for nonroad sources is derived from the Clean Air Nonroad Diesel rule165 and is based
on REMSAD modeling conducted in 2004. These dollar-per-ton values represent monetized
health impacts in 2015. Using the projected 2015 emission changes presented in Table 4.1-6, we
estimate that the potential PM2.5-related monetized impact associated with NOx emissions from
increased use of ethanol to be up to $150 million in 2015, assuming 7.2 billion gallons of ethanol
use in 2012. Note that this impact is based on monetized changes in health effects, including
changes in mortality risk, chronic bronchitis, nonfatal heart attacks, respiratory hospital
admissions, asthma attacks, and other minor health endpoints. This value is not intended to
reflect potential expenditures related to the control of NOx emissions. Rather, it is presented
here as the upper, conservative bound of the potential costs associated with an increase in NOx
emissions. It is also important to point out that this value does not represent the cumulative
monetized health impacts associated with the potential PM changes associated with the future
use of ethanol. Given the potential decrease in ambient PM2.5 due to the decrease in aromatic
fuel content offset by the increase in NOx, we can not say for certain in which direction the total
monetized PM-related health impact will be. In reality there may be an overall reduction in PM-
related health costs, despite the increase due to increased NOx emissions.000
This estimate is also subject to a number of additional caveats. The dollar-per-ton values
reflect specific geographic patterns of emissions reductions and specific air quality and benefits
modeling assumptions which are derived from previous analyses and will not match those
associated with increased ethanol use in fuel. Furthermore, use of these dollar-per-ton values to
estimate benefits associated with different emission control programs may lead to higher or
000 Overall, we expect that the decrease in secondary organic PM is likely to exceed the increase in secondary
nitrate PM. In 2006, NOx emissions from gasoline-fueled vehicles and equipment comprise about 37% of national
NOx emissions from mobile sources. In contrast, gasoline-fueled vehicles and equipment comprise almost 90% of
national gaseous aromatic VOC mobile source emissions. The percentage increase in national NOx emissions due
to increased ethanol use should be smaller than the percentage decrease in national emissions of gaseous aromatics.
Finally, in most urban areas, ambient levels of secondary organic PM exceed those of secondary nitrate PM. Thus,
directionally, we expect a net reduction in ambient PM levels due to increased ethanol use. However, we are unable
to quantify this reduction at this time.
285
-------
lower monetized estimates than if values were calculated based on direct air quality modeling.
Great care should also be taken when applying these estimates to emission reductions that occur
in any specific location, as the dollar-per-ton values are based on national emission reduction
programs and therefore represent average the dollars-per-ton over the entire U.S. The dollars-
per-ton for emission reductions in specific locations may be very different than the national
average.
Potential Fuel Economy Benefits
The assumption used in this analysis is that ethanol use does not change energy efficiency
during the combustion process, such that fuel economy is directly proportional to the energy
density of the fuel. Since the volumetric energy content of ethanol is approximately 33% less
than conventional gasoline, one would expect a fuel economy decrease that is proportional to the
percent of ethanol blended into the gasoline. Several studies have suggested, however, that this
decrease in fuel economy associated with 10 percent ethanol blends is less than the relative
decrease in volumetric energy content of the fuel. In other words, there is less of a fuel
consumption increase than the lower energy density of E10 would suggest.
Several studies point to a net efficiency increase of 1 percent for E10, although these
findings are often accompanied by a caveat that makes drawing a firm conclusion difficult. For
example, the 2006 CRC E-67 study (discussed in further detail in RIA chapter 4) found that 10
percent Ethanol tended to decrease volumetric heat content by 2.2 percent on average. Since one
would expect the amount of fuel consumed over a given distance to be directly proportional to
the energy content of the fuel, fuel economy should also decrease by 2.2 percent. The test
results, however, showed the fuel economy decrease to be only 1.4 percent on average, inferring
an efficiency increase of 0.8 percent. The CRC reminds us that the test program was designed to
provide independent variation of T50, T90, and ethanol content while holding the other
parameters constant. To maintain fixed distillation temperatures while increasing ethanol, for
example, heavier hydrocarbons were also added to offset the changes in T50 and T90 that would
ordinarily accompany ethanol addition. This changes the volumetric energy content of the fuel
to a larger degree than that dictated by the addition of ethanol alone, and complicates this
analysis to some extent.
Results from the Auto/Oil Air Quality Improvement Research Program166 showed a
volumetric fuel economy decrease of 2.63 percent ± 0.44 percent (error bars are 95 percent
confidence interval) for 10 percent ethanol despite a 3.3 percent decrease in theoretical energy
content. On an energy specific fuel economy basis, they found a 0.97 percent ± 0.44 percent
increase with E10 compared to the base fuel. These small, but statistically significant, changes
in energy specific fuel economy are difficult to explain. One Auto/Oil program hypothesis is
that this increase occurs during portions of the FTP when the vehicle is running open loop -
during hard accelerations or in the cold start portion of bag 1. They also speculate that the
feedback control in these 1983 - 1989 model year vehicles was not sophisticated enough to
compensate for the subtle changes seen in stoichiometric A/F with these low level oxygenate
blends.
286
-------
Insufficient data exists to confirm the validity of this slight increase in efficiency with
ethanol. Therefore we have maintained the assumption that there is no change in motor vehicle
efficiency when operated on gasoline blends with ethanol. However, if additional testing were to
confirm a benefit on today's vehicle fleet, it would lower the overall cost estimates of increase
ethanol use.
287
-------
Chapter 8: Agricultural Sector Impacts
Elsewhere in this rulemaking, we have estimated the costs of producing renewable fuel,
transporting it to its place of use and absorbing it within the gasoline and diesel fuel pool as a
blend stock. In this section we focus on some of the other economic impacts that are likely to
result from large expansions in renewable fuel production and use within the United States. In
particular, since the vast bulk of this renewable fuel is expected to be produced using feedstock
commercially grown in the U.S., we examine the impact of this increasing demand on the
agricultural sector.
8.1 Agricultural Sector Impacts
Due to the timing of this NPRM, we were not able to complete a rigorous analysis of the
impacts of renewable fuel expansion scenarios in time to be included in this notice. Subsection 1
below gives basic estimates for impacts of renewable fuel on crop and land use, while Subsection
2 outlines the more detailed modeling analysis we are undertaking to be done in time for the final
rulemaking package.
8.1.1 Estimates of Land Impacts Based on Available Data
8.1.1.1 Corn Ethanol Land Requirements
Using information from USDA and other sources, we made an estimate of corn and land
use requirements for recent years as well as 2012 with production of 7.2 billion gallons of
ethanol per year (7.2 BGY).PPP'167 We repeated the calculations for the case of 9.6 BGY, but due
to the fact that USDA's modeling does not consider this level of ethanol production, we did not
attempt to estimate any additional corn acreage that might planted. Section 8.1.3 contains more
discussion of agricultural sector modeling underway for the final rulemaking.
This work assumed corn was the only feedstock being used to produce ethanol, and
would contribute the sole land use impact. Net imports were the only other ethanol source we
considered, as others were not expected to have a significant impact on the results. To simplify
calculations, the split-year figures given in agricultural sources were assumed to equate to the
first calendar year of the pair (i.e. 2005/6 agricultural data is used directly with 2005 ethanol
data).
The total land area required for the annual ethanol production requirement was back-
calculated using the ethanol production yield in gallons per bushel, and the average corn yield
per acre. The figures used are given in Table 8.1-1. Ethanol yield was taken as 2.7 gallons of
ethanol per bushel. Before performing the calculation, the ethanol consumption figure was
ppp Through personal communication with USDA in July 2006 we learned that their projections assumed
7.5 BGY of ethanol production in 2012 rather than 7.2 BGY. However, we did not attempt to adjust the results
because the difference is small (4%) and because we were uncertain of the level of influence of ethanol volume on
corn production.
288
-------
reduced by subtracting net imports where available, taken from P.O. Licht.168 The results of this
analysis suggest that for the 7.2 BGY case, approximately 21% of corn will be used to produce
ethanol in 2012, up from 13% in 2005. Assuming no additional corn is planted in response to the
industry producing 9.6 BGY, we estimate that 28% of corn will be required for ethanol
production. This figure should be seen as an upper limit, since it is likely that additional acres
would be planted as more corn is demanded. More details of the results are given in Table 8.1-2.
The result for 2012 (7.2 BGY case) is in reasonable agreement with figures published by the
Farm and Agricultural Policy Research Institute (FAPRI), given in Table 8.1-3.
Table 8.1-1. Inputs to Land Use Calculations
Year
2004
2005
2006
2012
2012
Bushels per Acre3
160
148
148
159
159
Million Bushels
Produced3
11,807
11,032
10,810
12,315
12,315d
Net Imports
(million gals)
172b
127b
0 (not available)
300b
300b
Total Ethanol
(million gals)1
3,410
4,000
4,900
7,200
9,600
a USD A Baseline Report OCE-2006-1
b P.O. Licht, World Ethanol Markets - The Outlook to 2015 (2006)
c RFA website and Chapter lof this DRIA
d This calculation assumes no additional corn is planted for the 9.6 BGY case.
Table 8.1-2. Cropland Allocation Results
Year
2004
2005
2006
2012
2012
Total Ethanol
Production
(billion gals)
3.4
4.0
4.9
7.2
9.6
Total Corn
Acres Planted
(millions)
73.6
74.3
73.2
77.7
77.7b
Corn Acres Required for
Ethanol Production
(millions)3
7.5
9.7
12.3
16.1
21.7
Ethanol
Requirement
as % of Corn
Acres
10.2
13.0
16.8
20.8
28.0
a The 2012 volume is for 7.2 billion gallons minus imports of 300 million gallons.
b This calculation assumes no additional corn is planted for the 9.6 BGY case.
289
-------
Table 8.1-3. FAPRI Data on Corn Production and Use (Assumes 7.5 BGY RFS)
Year
2005
2006
2012
Ethanol
Production3
(billion gals)
3.9
4.6
7.7
Million
Bushels
Produced"
11,112
10,714
12,431
Million Bushels Used
for Ethanol
Production3
1,576
1,800
2,749
Ethanol Requirement
as % of Corn
Production
14.2
16.8
22.1
a Taken from FAPRI 2006 U.S. and World Agricultural Outlook, Staff Report 06-FSR-l, page 83.
USD A projects that the yield per acre remains essentially flat between now and 2012,
while the number of corn acres planted rises by about 6%, resulting in a similar increase in total
corn production. However, this increase in corn production is not sufficient to offset the demand
by ethanol plants, indicated by the steep increase in percentage of total corn being used for
ethanol production. Figure 8.1-1 below shows these trends. Between now and 2012, USDA also
forecasts corn prices to rise by about 20 cents per bushel (assuming 7.5 BGY of ethanol). Higher
corn price is a driving force for planting more acres, a trend that we expect will be further
enhanced at the 9.6 BGY ethanol production level. However, in that situation we still expect a
significantly higher percentage of all corn produced to go to ethanol production. In this work,
we did not specifically analyze whether fallow acres would be planted with corn as a result of
ethanol production, or whether land occupied by other crops would be planted with corn instead.
290
-------
Figure 8.1-1. Trends in Corn Harvest and Use for Ethanol Production
Corn Acres (million)
• % Acres for Ethanol
• % Acres for Ethanol
35
o
c
(0
£
£
15
10
2004 2005 2006 2007 2008 2009 2010 2011 2012
a Assumes no additional corn acres are planted for 9.6 billion gallon per year case.
8.1.1.2
Biodiesel Land Requirements
Based on EIA's current projected biodiesel volumes for years 2006-2012, biodiesel fuel
produced from virgin vegetable oils will have a negligible impact on the utilization of farm crop
land. EIA's projections forecast biodiesel demand to nearly quadruple by year 2007, based on
last years demand, and then remain fairly stable. While this is a large percentage increase, it is
still relatively small in terms of absolute volume. Furthermore, these projections assume that
about half of biodiesel will be produced from virgin soy bean oil and the remaining from yellow
grease feedstocks. New biodiesel plant announcements in the trade journals indicate, however,
that in addition to soy oil, other types of virgin vegetable oils may be used in the future. These
virgin stocks will probably only make up a small fraction of the feedstocks for making biodiesel.
For our analysis, we assume that soy bean oil is the primary virgin oil vegetable feedstock use to
manufacture biodiesel, along with yellow grease. Additionally, we use soy bean growth and oil
content projections based on current USD A forecast to determine the land acreage impacted by
biodiesel fuel derived from virgin vegetable oils.
Biodiesel yields are generally proportional to the oil content of the seeds used as a
feedstock. Soy beans have one of the lowest oil concentrations of all oil seeds, currently at about
18.9% by weight and projected to remain at this level in the future169 Even though the oil
291
-------
content is low, soy oil derived from crushing soy beans is still the primary feedstock used to
produce biodiesel fuel in the U.S.170 This is because U.S. farm policies encourage the growth of
soy beans, which are then processed to produce soy meal and soy oil. This increases the supply
of soy oil which makes the price economically attractive for producing biodiesel. Current
biodiesel yields are averaging about 1.6 gallons per bushel of soy beans, with yield values
unlikely to change in the future as the upper yield amount is limited by the soy bean seed oil
content. Additionally, for our analysis, a bushel of soy beans is assumed to contain 60 pounds of
soy beans. Using these criteria, along with EIA's volume of soy oil derived bio-diesel, we
project the total tons of soy beans needed to produce biodiesel volumes projected under the RFS
program, shown in Table 8.1-4.
Soy bean yields (bushels per acre) have experienced a slight upward trend in recent years,
though the level of improvement has leveled off. The USD A projects that the current yield level
will continue in future years under phase-in of the RFS program, see Table 8.1-5. Based on
current USDA soy bean yields per acre, and soy oil yields from soy beans, we estimate the
amount of land required to produce soy bean oil to satisfy biodiesel demand is 2.04 MM
(million) acres in 2006 and 2.27 MM acres in 2012, under the RFS scenario. (See Table 8.1-6)
These estimates are based on EIA projections of biodiesel produced from soy bean oil and do not
include biodiesel from yellow grease or other virgin vegetable oil stocks. Thus, the amount of
biodiesel generated under the RFS programs years, utilizes a modest amount of farm land and is
not expected to have a major impact.
Table 8.1-4. Tons of Soy Beans for Biodiesel
Year
2004a
2005a
2006
2012
2015
Soy bean
(MM tons/ year)
0.46
1.67
2.61
2.91
2.91
a From Table 7, Selected Supply , Use and Price for Major Field Crops, Baseline projections, page 35 USDA report.
Soy oil yield was assumed to be 18% in 2004 72005 and 18.9 % in years 2006 and later.
292
-------
Table 8.1-5. USDA Soy Bean Yields Per Acrea
Year
2001
2002
2003
2004
2005
2006
2012
2015
Yields, Bushels/Acre
39.6
38.0
33.9
42.2
42.7
42.7
42.7
43.9
"Years 2005 to 2015 from USDA Outlook 2006 Table 7, Selected Supply, Use and Price for Major Field Crops,
Baseline projections. Data foryears 2001-2004, from "Oil Crops Situation and Outlook Yearbook", by USDA,
Appendix Table 2 Soy beans Acreage Planted, Harvested, Yield, production and Loan rates, U.S., 1960-2005
Table 8.1-6. Cropland Allocation for Soy Beans
Year
2004
2005
2006
2012
2015
Total
Soy beans Acres, MM
75.2
72.2
73.5
71.0
70.5
Soybean Acres for
Biodiesel Production, MMa
0.4
1.3
2.04
2.27
2.27
Biodiesel Percent of Total
Soy bean Acres, %
0.5
1.8
2.8
3.2
3.2
aMM is million, 2012 and 2015 data based on EIA AEO 2006 projections for soy bean oil derived biodiesel. 2004-
2006 data from the National Biodiesel Board (NBB), for 2006, the estimate of 150 MM gallons is NBB's forecast
volume. Values for 2004-2006 assume that soy bean oil is the feedstock used to produce 90% of nations total
biodiesel supply.
8.1.3 Agricultural Sector Impact Modeling for Final Rulemaking
This section describes work underway to evaluate the impacts of renewable fuel
production on the U.S. agricultural sector for the final rulemaking. Here we will outline our
methodologies and critical assumptions, as well as some anticipated results.
The RFS program attempts to spur the increased use of renewable transportation fuels
made principally from agricultural feed stock produced in the U.S. As a result, there will be
impacts on the U.S. agricultural sector. Economic theory suggests that an increase in demand for
a good will likely increase both its supply and price. In the case of renewable fuels, production
293
-------
of ethanol from corn for example will create a new use for corn, expanding corn's market value
and likely expanding the supply of corn to meet the higher demand.
It is anticipated that almost all of the renewable fuel used to fulfill the requirements of the
RFS program will come from agricultural feedstock produced within the U.S. While it is feasible
that feedstock could be imported to supply domestic production facilities, it is likely to be more
economical to procure feedstock in the general location of the renewable fuel production facility
rather than incur substantial feedstock transportation and other costs in shipping feedstock in the
large quantities necessary to support a renewable fuel production facility. Furthermore, a joint
study by the USDA and DOE has estimated that there will be ample domestic supplies of
feedstock to meet the levels of renewable fuel production being evaluated in this rulemaking
(although the mix of feedstock sources may be constrained by the economically available land to
grow corn for corn-based ethanol and soy beans for soy-based biodiesel).171 Thus, for the
purpose of the analyses of impacts on the U.S. agricultural sector, given that renewable fuel
imports to the U.S. are predicted to be relatively small, we are assuming that the feedstock
necessary to meet the 7.5 billion gallons of renewable fuel required in 2012 by the RFS will be
made from feedstock grown within the U.S. borders. Similarly, for the 9.9 billion gallons of
renewable fuel predicted by EIA to be used in 2012, we will again model agricultural sector
impacts assuming the necessary feedstock is supplied by the U.S.
To analyze the impacts of the RFS on the U.S. agricultural sector, EPA has selected the
Forest and Agricultural Sector Optimization Model (FASOM) developed by Professor Bruce
McCarl, Texas A&M University and others over the past thirty years. FASOM is a dynamic,
nonlinear programming model of the agriculture and forestry sectors of the U.S. Its objective
function is to maximize the present discounted value of producer and consumer surplus across
the U.S. agricultural and forestry sectors. (For this analysis, we will be focusing upon the
agriculture portion of the model.) The model is constrained by land use balances (e.g., increased
corn production could require acres which would otherwise be used to produce soy beans), and
commodity competition across domestic consumption, processing, livestock feeding and exports
(e.g., increased use of corn for biofuels causes alterations in exports, livestock feeding and
livestock herd size). The strength of this model is its consideration of the full direct and indirect
impacts of a shift in production of an agricultural commodity. For example, the model assesses
not only the impacts of increased demand for corn on acres devoted to corn production but also
where the incremental corn will be produced, what other crops will be displaced and how corn is
allocated among competing uses. Shifts in crop production will likely impact the price of corn
and other crop prices. In addition, the model can estimate the impacts of increased renewable
fuel use on animal feed costs, animal production and costs to consumers. Similarly, FASOM can
estimate effects on U.S. farm employment and income (broken down by region, and farm sector
such as corn farmers versus soy bean producers versus the livestock industry, for example).
Corn prices and distillers dried grain values from FASOM and other farm factors will provide
inputs to the ethanol cost modeling which, in turn, will impact estimates derived from the
refinery model.
While the model has broad capability to estimate such parameters as feedstock prices, we
will constrain the model by fixing some of these external values. For feedstock prices, we will
use a model version that closely tracks USDA predictions from their report, "USDA Agricultural
294
-------
Baseline Projections to 2015"172 . To estimate the amounts of feedstock required for the various
renewable fuels, we are assuming in the RFS case the volumes of fuel estimated by EIA for
biodiesel from soy and biodiesel from yellow grease (to represent non-soy feedstock) totaling
300 million gallons for 2012. For ethanol, we are assuming the RFS will encourage 250 million
gallons of cellulose-based ethanol and 6.95 billion gallons of ethanol from corn in 2012. In a
second case, which represents the amount of renewable fuels projected by EIA, we are assuming
9.35 billion gallons of corn-derived ethanol, 250 million gallons of cellulosic ethanol and 300
million gallons for biodiesel from soy and biodiesel from yellow grease.
As mentioned above, the FASOM model has broad capability to assess impacts resulting
from changes to the U.S. agricultural sector. While we expect to have the model assess a very
wide set of possible outcomes, we realize that models are an abstraction of reality and will also
do independent analyses on a number of factors. At a minimum, however, we expect to be able
to assess the direct impact on U.S. farm income and farm employment, shifts in crop production
and the impacts on commodity prices, including animal feed costs and how changes in feed costs
may impact production levels and prices of beef, chicken and other food products within the U.S.
As part of this assessment, we will include the value of by-products of ethanol and biodiesel
production including, the value of distillers dried grains, corn gluten feed and corn gluten meal
which are valuable co-products of dry and wet mill ethanol production from corn and the soy
meal by-product resulting from soy oil extraction which is part of soy-based biodiesel
production. We will also estimate the potential impacts on exports of crops and animal products
as these represent significant sources of income to U.S. farmers.
295
-------
Chapter 9: Small-Business Flexibility Analysis
This chapter presents our Small Business Flexibility Analysis (SBFA) which evaluates
the potential impacts of the proposed rule on small entities. The Regulatory Flexibility Act, as
amended by the Small Business Regulatory Enforcement Fairness Act of 1996 (SBREFA),
generally requires an agency to prepare a regulatory flexibility analysis of any rule subject to
notice and comment rulemaking requirements under the Administrative Procedure Act or any
other statute unless the agency certifies that the rule will not have a significant economic impact
on a substantial number of small entities.
9.1 Requirements of the Regulatory Flexibility Act
When proposing and promulgating rules subject to notice and comment under the Clean
Air Act, we are generally required under the Regulatory Flexibility Act (RFA) to conduct a
regulatory flexibility analysis unless we certify that the requirements of a regulation will not
cause a significant impact on a substantial number of small entities. The key elements of the
RFA include:
• a description of and, where feasible, an estimate of the number of small entities to which
the proposed rule will apply;
the projected reporting, record keeping, and other compliance requirements of the
proposed rule, including an estimate of the classes of small entities which will be subject
to the requirements and the type of professional skills necessary for preparation of the
report or record;
an identification to the extent practicable, of all other relevant Federal rules which may
duplicate, overlap, or conflict with the proposed rule; and,
• any significant alternatives to the proposed rule which accomplish the stated objectives of
applicable statutes and which minimize any significant economic impact of the proposed
rule on small entities.
The RFA was amended by SBREFA to ensure that concerns regarding small entities are
adequately considered during the development of new regulations that affect them. Although we
are not required by the Clean Air Act to provide special treatment to small businesses, the RFA
requires us to carefully consider the economic impacts that our proposed rules will have on small
entities. Specifically, the RFA requires us to determine, to the extent feasible, our rule's
economic impact on small entities, explore regulatory options for reducing any significant
economic impact on a substantial number of such entities, and explain our ultimate choice of
regulatory approach.
In developing this proposed rule, we concluded that the RFS program under
consideration would not have a significant impact on a substantial number of small entities. We
based this on several criteria. First, the industry is expected to be overcomplying by a wide
margin independent of the standard, thus causing compliance costs to be minimal. Second, the
296
-------
Energy Policy Act of 2005 already provides relief from the renewable fuels standards until 2011
for the majority of the small entities; and lastly, we are extending this relief to the remaining
small entities. This is discussed further below.
9.2 Need for the Rulemaking and Rulemaking Objectives
A detailed discussion on the need for and objectives of this proposed rule are located in
the preamble to the proposed rule. As previously stated, EPA is required to promulgate
regulations implementing a renewable fuel program under Section 1501 of the Energy Policy
Act, which amended the Clean Air Act by adding a Section 21 l(o). The Energy Policy Act
requires EPA to establish a program to ensure that U.S. gasoline contains specific volumes of
renewable fuel for each calendar year beginning in 2006, to increase the amount of renewable
fuel used in vehicles and engines in the U.S.
9.3 Definition and Description of Small Entities
Small entities include small businesses, small organizations, and small governmental
jurisdictions. For the purposes of assessing the impacts of the proposed rule on small entities, a
small entity is defined as: (1) a small business that meets the definition for business based on the
Small Business Administration's (SBA) size standards (see Table 9-1); (2) a small governmental
jurisdiction that is a government of a city, county, town, school district or special district with a
population of less than 50,000; and (3) a small organization that is any not-for-profit enterprise
which is independently owned and operated and is not dominant in its field. Table 9-1 provides
an overview of the primary SBA small business categories potentially affected by this regulation.
Table 9.3-1. Small Business Definitions
Industry
Gasoline refiners
Defined as small entity by SBA if:
<1,500 employees and a crude
capacity of <125,000 bpcd
NAICS Codes a
324110
North American Industrial Classification System
9.4 Summary of Small Entities to Which the Rulemaking Will Apply
The refiners that are potentially affected by this proposed rule are those that produce
gasoline. For our recent proposed rule "Control of Hazardous Air Pollutants from Mobile
Sources" (71 FR 15804, Wednesday, March 29, 2006), we performed an industry
characterization of potentially affected gasoline refiners. Information about the characteristics of
refiners comes from sources including the Energy Information Administration within the U.S.
Department of Energy, and from Hoover's (a division of Dun and Bradstreet). The refining
industry is located primarily in NAICS code 324110.
The industry characterization was used to determine which refiners would also meet the
SBA definition of a small refiner under this proposal. From the industry characterization, we
297
-------
determined that there were 20 gasoline refiners that met the definition of a small refiner. It
should be noted that because of the dynamics in the refining industry (e.g., mergers and
acquisitions), the actual number of refiners that ultimately qualify for small refiner status under
the final RFS program could be different from this initial estimate.
Title XV the Energy Policy Act provides, at Section 1501(a)(2) [42 U.S.C.
7545(o)(9)(A)-(D)], special provisions for "small refineries", which include a temporary
exemption from the standards until calendar year 2011. Further, the Energy Policy Act states
that EPA must use the definition of "small refinery" and apply the special provisions provided
for small refineries in the RFS program. The Energy Policy Act defines the term "small
refinery" as ".. .a refinery for which the average aggregate daily crude oil throughput for a
calendar year.. .does not exceed 75,000 barrels."
A small refinery (as defined by the Energy Policy Act) is very different from a small
refiner (as defined in SBA's regulations at 13 CFR 121.201). Per 13 CFR 121.201, and stated
above in Table 9-1, a small refiner is a small business that employs less than or equal to 1,500
employees and has an annual crude capacity of less than or equal to 125,000 bpcd. A small
refinery, per the Energy Policy Act, is a small-capacity refinery and could be owned by a larger
refiner that exceeds SBA's small entity size standards; whereas small refiners generally only
own a few (and more often than not, only one) refineries.
In our analysis of the potentially affected small refiners, we found that 42 refineries met
the Energy Policy Act's definition of a small refinery. Of these, we determined that 17 of these
refineries were owned by small refiners. Therefore, 17 of the 20 small refiners owned refineries
that also met the Energy Policy Act's definition of a small refinery. As a result, all but three
small refiners would automatically be granted relief by implementing the provisions specified in
the Energy Policy Act.
9.5 Related Federal Rules
We are not aware of any area where the regulations under consideration would directly
duplicate or overlap with the existing federal, state, or local regulations; however, several small
refiners are also subject to the gasoline sulfur, highway diesel sulfur, and nonroad diesel sulfur
control requirements. In addition, some of these small refiners will also be subject to the
upcoming mobile source air toxics (MSAT2) requirements for benzene in gasoline.
9.6 Projected Reporting, Recordkeeping, and Other Compliance
Requirements
For any fuel control program, EPA must have the assurance that refiners meet the
applicable standards. Thus, requirements are imposed to ensure that compliance obligations are
met.
298
-------
The recordkeeping, reporting and compliance provisions of the proposed RFS program
are fairly consistent with those currently in place for other fuel programs, including the highway
and nonroad diesel and MSAT regulations. These provisions include:
Registration (the registration numbers will also be used in the RINs)
Submission of annual reports summarizing a refiner's annual gasoline production and a
demonstration of its compliance with the renewable fuels standard and submission of
annual report detailing and tracking a refiner's RINs
Recordkeeping consisting of the retention of all compliance documents (such as Product
Transfer Documents and all reports submitted to EPA) for at least five years
For a more detailed discussion of these provisions, please see section IV of the preamble
to this proposed rule.
9.7 Projected Effects of the Proposed Rulemaking on Small Entities
After considering the economic impacts of today's proposed rule on small entities, we do
not believe that this action will have a significant economic impact on a substantial number of
small entities.
While the Energy Policy Act provided for a temporary exemption for small refineries
from the requirements of today's proposed rule, these parties will have to comply with the
requirements following the exemption period. Therefore, we had to take into account the
economic effects of the program on small entities when they would need to comply with the
standard. As shown in Table III.D.3.C-2, located in the preamble to this proposed rule, the
annual projections of ethanol production are greater than the annual renewable fuel volumes
required by the Act. In 2011, when the Act's small refinery exemption ends, over one billion
gallons in excess RINs are projected to be available. Further, excess RINs are anticipated for
each year of the program. Due to this projected excess supply in comparison to the standard, the
cost of RINs should be very low— near the level of the transaction costs.
Due to the low cost to potentially affected small entities, and the projected RIN
availability, we do not believe that this program will impose a significant economic impact on a
substantial number of small entities.
9.8 Regulatory Alternatives
Though we do not believe that this proposed rule will have a significant economic impact
on a substantial number of small entities, we still believe that small refiners generally lack the
resources available to larger companies. As discussed in section XII.C of the preamble to the
proposed rule, we find it necessary to extend the small refinery temporary exemption, as set out
in the Energy Policy Act, to all small qualified small refiners. In addition, past fuels rulemakings
have included a provision that, to qualify for EPA's small refiner flexibilities, a refiner must
have no more than 1,500 total corporate employees and have a crude capacity of no more than
299
-------
155,000 bpcd (slightly higher than SBA's crude capacity limit of 125,000 bpcd). To be
consistent with these previous rules, we are also proposing to allow those refiners that meet these
criteria to be considered small refiners for this rulemaking. Lastly, we are proposing that small
refiners may separate RINs from batches and trade or sell these RINs prior to 2011 if the small
refiner operates as an oxygenate blender.
300
-------
Endnotes
1 Renewable Fuels Association (RFA), Ethanol Biorefinery Locations (Updated June 19, 2006).
2 Ethanol Producer Magazine (EPM), U.S. & Canada Fuel Ethanol Plant Map (Spring 2006).
3 International Fuel Quality Center (IFQC), Special Biofuels Report #75 (April 11, 2006).
4 Renewable Fuels Association, From Niche to Nation: Ethanol Industry Outlook 2006
5 Marketer information obtained from ethanol marketer websites, ethanol producer websites, and
conversations with ethanol marketers/producers.
6 Renewable Fuels Association (RFA), Ethanol Biorefinery Locations (Updated June 19, 2006).
7 Ethanol Producer Magazine (EPM), U.S. & Canada Fuel Ethanol Plant Map (Spring 2006).
8 International Fuel Quality Center (IFQC), Special Biofuels Report #75 (April 11, 2006).
9 "BioEnergy International Commences Site Work on 108 Million Gallon Ethanol Project in
Lake Providence, LA", Source: www.grainnet.com Press Release (May 8, 2006).
10 "Biodiesel Performance, Costs, and Use", Anthony Radich, EIA page 6.
11 NBB Survey April 28, 2006 "Commercial Biodiesel Production Plants".
12 From Independent Biodiesel Feasibility Group Presentation.
13 "Pipeline Considerations for Ethanol", Department of Agricultural Economics, Kansas State
University, August 2002. http://www.agmrc.org/NR/rdonlyres/4EEOE81C-C607-4C3F-BBCF-
B75B7395C881/0/ksupipelineethl.pdf. and "Shipping Ethanol Through Pipelines", American
Petroleum Institute, April 6, 2006, http://api-
ep.api.org/printerformat.cfm?ContentID=54FDlAll-95FA-4B7C-ACElD7C6F121FBlC.
14 Infrastructure Requirements for an Expanded Fuel Ethanol Industry, Downstream Alternatives
Inc., January 15,2002.
15 "Ethanol Industry's Interest in River Sites Grows", River Transport News, Vol. 15, No. 10,
May 22, 2006.
16 Infrastructure Requirements for an Expanded Fuel Ethanol Industry, Downstream Alternatives
Inc., January 15,2002.
17 Infrastructure Requirements for an Expanded Fuel Ethanol Industry, Downstream Alternatives
Inc., January 15,2002.
301
-------
18 P.O. Licht, "World Ethanol Markets, The Outlook to 2015" (2006).
19 EIA Monthly Energy Review, June 2006 (Table 10.1: Renewable Energy Consumption by
Source, Appendix A: Thermal Conversion Factors).
20 U.S. EPA Office of Transportation & Air Quality, List of Federal Reformulated Gasoline
Areas (Updated February 23, 2004).
21 U.S. EPA Office of Transportation & Air Quality, 2004 RFGFuel Survey Results
(http ://www. epa. gov/otaq/regs/fuel s/rfg/properf/rfgperf.htm)
22 U.S. EPA Office of Transportation and Air Quality, State Winter Oxygenated Fuel Program
Requirements for Attainment or Maintenance of CO NAAQS (November 2005)
23 EIA 2004 Petroleum Marketing Annually (Table 48: Prime Supplier Sales Volumes of Motor
Gasoline by Grade, Formulation, PAD District, and State).
24 EIA Historical RFG MTBE Usage (file received from EIA representative on March 9, 2006).
25 EIA Monthly Energy Review, June 2006 (Table 10.1: Renewable Energy Consumption by
Source, Appendix A: Thermal Conversion Factors).
26 FHWA Highway Statistics 2004: Estimated Use of Gasohol (April 2006).
27 U.S. EPA Office of Transportation & Air Quality, 2004 RFG Fuel Survey Results
(http ://www. epa. gov/otaq/regs/fuel s/rfg/properf/rf gperf.htm)
28 Alliance of Automobile Manufacturers (AAM) North American Fuel Survey 2004. The report
and data can be purchased through the AAM website,
http://autoalliance.org/fuel/fuel surveys.php, or by contacting the Alliance at 1401 Eye Street,
N.W., Suite 900, Washington, DC 20005, Tel: (202) 326-5500.
29 EIA Monthly Energy Review, June 2006 (Table 10.1: Renewable Energy Consumption by
Source, Appendix A: Thermal Conversion Factors).
30 EIA Historical RFG MTBE Usage (file received from EIA representative on March 9, 2006).
31 EIA 2004 Petroleum Marketing Annually (Table 48: Prime Supplier Sales Volumes of Motor
Gasoline by Grade, Formulation, PADD, and State).
32 EIA Monthly Energy Review June 2006 (Table 10.1: Renewable Energy Consumption by
Source, Appendix A: Thermal Conversion Factors).
33 EIA Annual Energy Outlook 2006 (Table 2: Energy Consumption by Sector and Source).
302
-------
34 EPAct Section 1541(c) Boutique Fuels Report to Congress, EPA420-R-06-901 (not yet
published)
35 EIA AEO 2006 Renewable Fuel Volumes (file received from EIA representative).
36 EIA Crude Oil Spot Pricing (www.eia.doe.gov).
37 EPAct Section 1504, promulgated on May 8, 2006 at 71 FR 26691.
38 "Analysis of the Production of California Phase 3 Reformulated Gasoline With and Without an
Oxygen Waiver", Math Proc, Inc., January 10, 2001.
39 EIA 2004 Petroleum Marketing Annually (Table 32: Conventional Motor Gasoline Prices by
Grade, Sales Type, PAD District, and State).
40
41
EIA NEMS Petroleum Market Model Documentation, Appendix I, Table 14.
American Coalition for Ethanol, STATUS: State by State Ethanol Handbook 2006
(supplemented by information obtained from state websites and conversations with state
government officials).
42 Alliance of Automobile Manufacturers North American Fuel Survey 2005. The report and data
can be purchased through the AAM website, http://autoalliance.org/fuel/fuel_surveys.php, or by
contacting the Alliance at 1401 Eye Street, N.W., Suite 900, Washington, DC 20005, Tel: (202)
326-5500. Information about the data and methods is contained in the report Survey Sampling
and Testing Details (May 2005), which may be downloaded from the AAM website at no
charge.
43 U.S. EPA, Final Regulatory Impact Analysis for Reformulated Gasoline, Table VI-A6,
EPA420-R-93-017, December 13, 1993. Available from RFG Final Rule, Docket No. A-92-12,
February 1994.
44 U.S. EPA, EPA 's National Mobile Inventory Model (NMIM), A Consolidated Emissions
Modeling System for MOBILE6 and NONROAD, EPA420-R-05-024, December 2005, version
NMEVI20060310 with county database NCD20060201 modified according to discussion in
Chapter 2.2.
45 U.S. EPA, Guide on Federal and State Summer RVP Standards for Conventional Gasoline
Only, EPA420-B-05-012, November 2005.
46 U.S. EPA website www. epa. gov/otaq/rfg/whereyoulive. htm , as of April 11, 2006.
47 U.S. EPA, A Comprehensive Analysis of Biodiesel Impacts on Exhaust Emissions, Draft
Technical Report, EPA420-P-02-001, October 2002.
303
-------
48 "Analysis of California's Request for Waiver of the Reformulated Gasoline Oxygen Content
Requirement for California Covered Areas." U.S. EPA, EPA420-R-01-016, June 2001.
49Durbin, T., Miller, J., Younglove, T., Huai, T., and Cocker, K. "Effects of Ethanol and
Volatility Parameters on Exhaust Emissions". CRC Project No. E-67. Coordinating Research
Council, Inc. 3650 Mansell Road, Suite 140. Alpharetta, GA 30022. January, 2006.
50 "Final Report, Effects of Gasoline Ethanol Blends on Permeation Emissions Contribution to
VOC Inventory from On-Road and Off-Road Sources," Air Improvement Resource, Inc., for the
American Petroleum Institute, March 3, 2005.
51 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.
52 Ragazzi, Ron and Ken Nelson, "The Impact of a 10% Ethanol Blended Fuel on the Exhaust
Emissions of Tier 0 and Tier 1 Light-Duty Gasoline Vehicles at 35 F," Colorado Department of
Public Health and Environment, March 26, 1999. For additional information or copies, contact
Gerald L. Gallagher or Margie M. Perkins.
53 Mulawa, Patricia. A. et. al. "Effect of Ambient Termperature and E-10 Fuel on Primary
Exhaust Particulate Mater Emissions from Light-Duty Vehicles," Environ. Sci. Technol. 1997,
31, 1302-1307.
54 Stump, F., et. al., Characterization of Emission from Malfunctioning Vehicles Fueled with
Oxygenated Gasoline - E10 Fuel, Part II and Part III, EPA RTF.
55 "Initial Mass Exhaust Emissions from Reformulated Gasolines". Technical Bulletin No. 1.
Auto/Oil Air Quality Improvement Research Program. Coordinating Research Council Inc.
Atlanta, Georgia. December 1990.
56 Hochhauser, A.M., et al. "Effects of Oxygenated Fuels and RVP on Automotive Emissions -
Auto/Oil Air Quality Improvement Research Program." Society of Automotive Engineers
publication No. 920326.
57 "Nonroad Evaporative Emission Rates," Table 2, U.S. Environmental Protection Agency,
EPA420-R-05-020, NR-012c, December 2005. This document is available in docket EPA-HQ-
OAR-2004-0008-0362.
58 "EPA's National Mobile Inventory Model (NMIM), A Consolidated Emissions Modeling
System for MOBILE6 and NONROAD," U.S. EPA, EPA420-R-05-024, December 2005.
304
-------
59 EPA Certification and Fuel Economy Information System, http://www.epa.gov/otaq/cfeis.htm.
60 Benson, J., Knepper, J., Reuter, R., Koehl,W., Leppard, W., Rippon, B., Burns, V., Painter, L.,
Rutherford, J., Hochhauser, A., and Rapp, L. "Emissions with E85 and Gasolines in
Flexible/Variable Fuel Vehicles-The Auto/Oil Air Quality Improvement Research Program."
SAE paper no. 952508. Warrendale, PA: SAE. October 1995.
61 Aakko, P. and Nylund, N.-O. Particle Emissions At Moderate and Cold Temperatures Using
Different Fuels. SAE Technical Paper 2003-01-3285.
62 McCormick R.L. et. al., "Fuel Additive and Blending Approaches to Reducing NOx Emissions
from Biodiesel", SAE 2002-01-1658.
63 Souligny M. et. al., "Heavy-Duty Engine Performance and Comparative Emission
Measurements for Different Biodiesel Blends Used in the Montreal Project", SAE 2004-01-1861.
64 Alam M. et. al., "Combustion and Emissions Performance of Low Sulfur, Ultra Low Sulfur
and Biodiesel Blends in a DI Diesel Engine", SAE 2004-01-3024.
65 McCormick R.L. et. al., "Regulated Emissions from Biodiesel Tested in Heavy-Duty Engines
Meeting 2004 Emission Standards", SAE 2005-01-2200.
66 Durbin, T.D., et. al., "Evaluation of the Effects of Biodiesel and Biodiesel Blends on Exhaust
Emission Rates and Reactivity - 2", Final Report, Center for Environmental Research and
Technology; College of Engineering; University of California, Riverside, August 2001.
67 Holden Bruce et. al., "Effect of Biodiesel on Diesel Engine Nitrogen Oxide and Other
Regulated Emissions", Naval Facilities Engineering Command (NAVFAC), Technical Report
TR-2275-ENV, May 2006.
68 Proc K., et. al. "RTD Biodiesel Transit Bus Evaluation: Interim Review Summary", Technical
Report NREL/TP-540-38364, 2005.
69 Frey Christopher, et. al., "Operational Evaluation of Emissions and Fuel Use of B20 Versus
Diesel Fueled Dump Trucks", NC State University/NC DOT Research Project No. 2004-18,
Final Report FHWA/NC/2005-07, 2005.
70 U.S. EPA, EPA 's National Mobile Inventory Model (NMIM), A Consolidated Emissions
Modeling System for MOBILE6 and NONROAD, EPA420-R-05-024, December 2005, version
305
-------
NMIM20060310 with county database NCD20060201 modified according to discussion in
Chapter 2.2.
71 U.S. EPA, User's Guide toMOBILE6.1 and MOBILE6.2 Mobile Source Emission Factor
Model, EPA420-R-03-010, August 2003, version MOBILE6.2.03 available at
http://www.epa.gov/otaq/m6.htm.
72 Federal Highway Administration, Highway Statistics 2004, Section V, Table VM-1 "Annual
Vehicle Distance Traveled in Miles and Related Data", October 2005.
73 U.S. EPA, User's Guide for the Final NONROAD 2005 Model, EPA420-R-05-013, December
2005, version NONROAD2005 available at http://www.epa.gov/otaq/nonrdmdl.htm.
74 U.S. EPA. 2006. Technical Support Document for the Proposed Mobile Source Air Toxics
Rule - Air Quality Modeling. This document is available in Docket EPA-HQ-OAR-2005-0036.
75 U.S. EPA. 2005. Technical Support Document for the Final Clean Air Interstate Rule - Air
Quality Modeling. Available from: http://www.epa.gov/cair/technical.html . This document is
available in Docket EPA-HQ-OAR-2003-0053.
76 U.S. EPA. 2005. Clean Air Interstate Rule Emissions Inventory Technical Support Document.
Available from: http://www.epa.gov/cair/technical.html. This document is available in Docket
EPA-HQ-OAR-2003-0053.
77 Pielke, R.A., W.R. Cotton, R.L. Walko, et al. 1992. "A Comprehensive Meteorological
Modeling System - RAMS." Meteor. Atmos. Phys., Vol. 49, pp. 69-91.
78 Ozone Air Quality Effect of a 10% Ethanol Blended Gasoline in Wisconsin, Wisconsin
Department of Natural Resources, September 6, 2005.
79 Ragazzi, R., Nelson, K. "The Impact of a 10% Ethanol Blended Fuel on the Exhaust
Emissions of Tier 0 and Tier 1 Light Duty Gasoline Vehicles at 35F." Colorado Department of
Public Health and Environment. March 26, 1999. For additional information or copies, contact
Gerald L. Gallagher or Margie M. Perkins.
80 Mulawa, P. A. "Effect of Ambient Temperature and E-10 on Primary Exhaust Particulate
Matter Emissions from Light-Duty Vehicles." Environmental Science and Technology 31, pp
1302-1307. 1997.
81 Grosjean, D. "In Situ Organic Aerosol Formation During a Smog Episode: Estimated
Production and Chemical Functionality." Atmospheric Environment. Vol. 26A. No. 6. pp. 953-
963: 1992.
306
-------
82 Schauer, J.J., Fraser, M.P., Cass, G.R., Simoneit, B.R.T., "Source Reconciliation of
Atmospheric Gas-Phase and Particle-Phase Pollutants during a Severe Photochemical Smog
Episode." Environmental Science and Technology 36, pp 3806-3814. 2002.
83 Pun, Betty K. and Christian Seigneur, "Investigative Modeling of New Pathways for
Secondary Organic Aerosol Formation," Final Report, CRC Project A-59, Coordinating
Research Council, March 2006.
84 Edney. RTF PM 2.5 field study. Currently undergoing peer review and will be published
shortly.
85 Zheng, M., Cass, G.R., Schauer, J.J., Edgerton, E.S. "Source apportionment of PM2.5 in the
Southeast United States using solvent-extractable organic compounds as tracers." Environmental
Science and Technology 36, pp. 2361-2371. 2002.
86 U.S. EPA, "National Emissions Inventory (NEI) Air Pollutant Emissions Trends Data, 1970 -
2002 Average annual emissions." Posted August 2005 on http://www.epa.gov/ttn/chief/trends/ ,
filename "NOX07182005.xls" (Accessed August 14, 2006). July 2005.
87 36,125 Btu/gallon (dry mill); Frank Taylor (USDA), personal communications, 2006.
88 41,400 Btu/gallon (dry mill) and 40,300 Btu/gallon (wet mill); Wang et al, 1999, Effects of
FuelEthanol Use on Fuel-Cycle Energy and Greenhouse Gas Emissions., (ANL/ESD-38), pp.
13.
89 34,800 Btu/gallon (dry mill); Shapouri and Gallagher, 2005,USDA 's 2002 Ethanol Cost-to-
Production Survey, Report Number 841, pp. 12.
90 35,418 Btu/gallon (dry mill); Kwiatkowski, McAloon, Taylor, and Johnston, 2006, Modeling
the Process and Costs of Fuel Ethanol Production by the Corn Dry-Grind Process, Industrial
Crops and Products, vol. 23, pp. 288.
91 Wang, Saricks, and Santini, 1999, Effects of Fuel Ethanol Use on Fuel-Cycle Energy and
Greenhouse Gas Emissions, (ANL/ESD-38), pp. 13.
92 Ethanol process efficiencies were averaged for the following articles and reports: Shapouri et
al. 1996, 46,879 Btu/gallon (dry mill) & 48,862 Btu/gallon (wet mill); Morris and Ahmed, 1992,
46,380 Btu/gallon (dry mill) & 46,380 Btu/gallon (wet mill); Stanley Consultants, 1996, 37,386
Btu/gallon (dry mill) & 54,977 Btu/gallon (wet mill); Welch, 1997, 51,000-53,000 Btu/gallon
(wet mill); Buckland, 1997, 39,415 Btu/gallon (dry mill); Wood, 1993, 53,261 Btu/gallon (dry
mill) & 53,089 Btu/gallon (wet mill); Wood, 1997, 45,000-50,000 Btu/gallon (wet mill);
McAloon, 1978, 38,096 Btu/gallon (dry mill); Minnesota Ethanol Commission, 1991, 39,000
Btu/gallon (dry mill); Minnesota Ethanol Commission, 1991, 34,000 Btu/gallon (wet mill);
Reeder, 1997, 34,000 Btu/gallon (wet mill); Madson, 1991, 39,600 Btu/gallon (dry mill);
307
-------
Merediz, 1997, 40,000-50,000 Btu/gallon (wet mill); and Shroff, 1997, <40,000 Btu/gallon (dry
mill).
93
94
36,000 Btu/gallon (dry mill), 45,950 (wet mill), 2005, GREET Version 1.7 (draft).
36,900 Btu/gallon (dry mill), 40,300 Btu/gallon (wet mill), Wang, Saricks, and Wu, 1997,
Fuel-Cycle Fossil Energy Use and Greenhouse Gas Emissions of Fuel Ethanol Produced from
U.S. Midwest Corn., pp. 15.
95 Wang, Saricks, and Wu, 1997, Fuel-Cycle Fossil Energy Use and Greenhouse Gas Emissions
of Fuel Ethanol Produced from U.S. Midwest Corn., pp. 13; cited in ANL/ESD-38, pp. 65.
96
Gervais and Baumel, 19XX, The Iowa Grain Flow Survey: Where and How Iowa Grain
Producers Ship Corn and Soybeans., CTRE, Iowa State University.
97 The Energy Balance of Corn Ethanol: An Update, Shapouri, H, J.A. Duffield, and M. Wang,
2002 AER-813, Washington DC: USDA Office of the Chief Economist.
98 USDA's 2002 Ethanol Cost-of-Production Survey, Shapouri, H, P Gallagher, Agricultural
Economic Report Number 841, July 2005.
99 The USDA National Agricultural Statistics Service (NASS) website, national statistics on field
corn: http://www.nass.usda.gov:8080/OuickStats/Create Federal All.jsp .
100 USDA Agricultural Baseline Projections to 2015, USDA Office of the Chief Economist,
World Agricultural Outlook Board, Baseline Report OCE-2006-1, Feb 2006.
101 The Energy Balance of Corn Ethanol: An Update, Shapouri, H, J.A. Duffield, and M. Wang,
2002 AER-813, Washington DC: USDA Office of the Chief Economist.
102 Shapouri, H., Duffield, J., Mcaloon, A.J. the 2001 Net Energy Balance of Corn-Ethanol.
2004. Proceedings of the Conference on Agriculture As a Producer and Consumer of Energy,
Arlington, VA., June 24-25.
103 The Energy Balance of Corn Ethanol: An Update, Shapouri, H, J.A. Duffield, and M. Wang,
2002 AER-813, Washington DC: USDA Office of the Chief Economist.
104 The Energy Balance of Corn Ethanol: An Update, Shapouri, H, J.A. Duffield, and M. Wang,
2002 AER-813, Washington DC: USDA Office of the Chief Economist.
105 Shapouri, H., Duffield, J., Mcaloon, A.J. the 2001 Net Energy Balance of Corn-Ethanol.
2004. Proceedings of the Conference on Agriculture As a Producer and Consumer of Energy,
Arlington, VA., June 24-25.
308
-------
106 Fossil Energy Use in the Manufacture of Corn Ethanol, Graboski, M., Report Prepared for the
National Corn Growers Association, August 2002.
107 Ibid
108 The Enhancement of Ethanol Yield from the Corn Dry Grind Process by Fermentation of the
Kernel Fiber Fraction. B.Dien, et. al. American Chemical Society Symposium Series,
December 1, 2004.
109 NREL Collaborative Research Program with major enzyme manufacturers.
110 Inventory of U.S. Greenhouse Gas Emissions and Sinks: 1990-2004, EPA 430-R-06-002,
April 2006.
111 Inventory of U.S. Greenhouse Gas Emissions and Sinks: 1990-2004, EPA430-R-06-002,
April 2006.
112 Inventory of U.S. Greenhouse Gas Emissions and Sinks: 1990-2004, EPA430-R-06-002,
April 2006.
113IPCC "Climate Change 2001: The Scientific Basis", Intergovernmental Panel on Climate
Change; J.T. Houghton, Y. Ding, DJ. Griggs, M. Noguer, P.J. van der Linden, X. Dai, C.A.
Johnson, and K. Maskell, eds.; Cambridge University Press. Cambridge, U.K. 2001.
114 Per communications with EIA staff, 6/1/06.
115 The EIA reports include: Renewable Fuels Legislation Impact Analysis, (July 2005), Impacts
of Modeled Provisions ofH.R. 6 EH: The Energy Policy Act of 2005 (July 2005), Summary
Impacts of Modeled Provisions of the 2003 Conference Energy Bill (February 2004). DOE EIA
Office of Integrated Analysis and Forecasting, (http://www.eia.doe.gov/oiaf/service_rpts.htm).
116 Net imports of petroleum include imports of crude oil, petroleum products, unfinished oils,
alcohols, ethers, and blending components minus exports of the same.
117 Leiby, Paul N., Donald W. Jones, T. Randall Curlee, and Russell Lee, Oil Imports: An
Assessment of Benefits and Costs, ORNL-6851, Oak Ridge National Laboratory, November 1,
1997. (http://pzl 1 .ed.ornl.gov/energvsecurity.html).
118 U.S. DOT, NHTSA 2006. "Final Regulatory Impact Analysis: Corporate Average Fuel
Economy and CAFE Reform for MY 2008-2011 Light Trucks," Office of Regulatory Analysis
and Evaluation, National Center for Statistics and Analysis, March.
(http://www.nhtsa.dot.gov/staticfiles/DOT/NHTSA/Rulemaking/Rules/Associated%20Files/200
6 FRIAPublic.pdf).
309
-------
119 Stanford Energy Modeling Forum, Phillip C. Beccue and Hillard G. Huntington, 2005. "An
Assessment of Oil Market Disruption Risks," FINAL REPORT, EMF SR 8, October 3, 2005.
(http ://www. Stanford. edu/group/EMF/publications/search.htm).
120 EIA (September 1997), "Petroleum 1996: Issues and Trends", Office of Oil and Gas,
DOE/EIA-0615, p. 71. (http://tonto.eia.doe.gov/FTPROOT/petroleum/061596.pdf).
121 Report entitled "Contribution of the ethanol industry to the economy of the United States,"
prepared by LECG, LLC February 21, 2006. The impact figures in the report were generated
using the U.S. Bureau of Economic Analysis Regional Input-Output Modeling System.
122 Kwaitkowski, J.R., McAloon, A., Taylor, F., Johnston, D.B., Industrial Crops and Products
23 (2006) 288-296.
123 Shapouri, H., Gallagher, P., USDA's 2002 Ethanol Cost-of-Production Survey (published
July 2005).
124 Gallagher, P.W., Brubaker, H., and Shapouri, H. "Plant Size: Capital Cost Relationships in
the Dry Mill Ethanol Industry," Biomass andBioenergy Vol. 28. 2005. Pp. 565-7.
125 Energy Policy Act of 2005, Section 1501(a)(2).
126 Historical data at http://tonto.eia.doe.gov/dnav/pet/pet_pri_allmg_d_nus_PTA_cpgal_m.htm
(gasoline), http://tonto. eia. doe. gov/dnav/ng/ng_pri sum dcu nus m. htm (natural gas),
http://www.eia.doe.gov/cneaf/electricity/page/sales_revenue.xls (electricity),
http://www.eia.doe.gov/cneaf/coal/page/acr/table28.html (coal).
127
EIA Annual Energy Outlook 2006, Tables 8, 12, 13, 15.
128 USDA Agricultural Baseline Projections to 2015, Report OCE-2006-1.
129 Farm and Agricultural Policy Research Institute 2006 U.S. and World Agricultural Outlook,
Report #06-FSR-1.
130 EIA NEMS model for ethanol production, updated for AEO 2006.
131 Food and Agricultural Policy Research Institute (FAPRI), "Implications of Increased Ethanol
Production for U.S. Agriculture", FAPRI-UMC Report #10-05.
132
http://www.iogen.ca/company/about/index.html.
133 Lignocellulosic Biomass to Ethanol Process Design and Economics Utilizing Co-Current
Dilute Acid Prehydrolysis and Enzymatic Hydrolysis Current and Futuristic Scenarios, Robert
Wooley, Mark Ruth, John Sheehan, and Kelly Ibsen, Biotechnology Center for Fuels and
Chemicals Henry Majdeski and Adrian Galvez, Delta-T Corporation; National Renewable
Energy Laboratory, Golden, CO, July 1999, NREL/TP-580-26157.
110
-------
134 Determining the Cost of Producing Ethanol from Corn Starch and Lignocellulosic
Feedstocks; A Joint Study Sponsored by: USD A and USDOE, October 2000 • NREL/TP-580-
28893 • Andrew McAloon, Frank Taylor, Winnie Yee, USD A, Eastern Regional Research
Center Agricultural Research Service; Kelly Ibsen, Robert Wooley, National Renewable Energy
Laboratory, Biotechnology Center for Fuels and Chemicals, 1617 Cole Boulevard, Golden, CO,
80401-3393; NREL is a USDOE Operated by Midwest Research Institute • Battelle • Bechtel;
Contract No. DE-AC36-99-GO10337.
135 Energy Information Administration, U.S. Department of Energy. See publications, Annual
Energy Outlook, 2006.
136
Energy Policy Act of 2005: TITLE XV—ETHANOL AND MOTOR FUELS, Subtitle A-
General Provisions, SEC. 1511, 1512, 1514.
137 Information provided by Jeff Kolb, Mathpro Inc., June 2006.
138
Per USDA phone discussion 6/22/06.
139Energy Information Administration NEMS Petroleum Marketing Model Documentation page
J-2.
140 Per EIA paper "MTBE Production Economics" Tancred C. M. Lidderdale, methanol price
cents per gallon = 15.79 + 0.099 *natural gas price ($ per million BTU).
141 Infrastructure Requirements for an Expanded Fuel Ethanol Industry, Downstream
Alternatives Inc., January 15, 2002.
142 Tables ES-9 and ES 10, Infrastructure Requirements for an Expanded Fuel Ethanol Industry,
Downstream Alternatives Inc, January 15, 2002.
143 Petroleum Market Model of the National Energy Modeling System, Part 2, March 2006,
DOE/EIA-059 (2006), http://tonto.eia.doe.gov/FTPROOT/modeldoc/m059(2006)-2.pdf.
144 Final Report: 1996 American Petroleum Institute/National Petroleum Refiners Association
Survey of Refining Operations and Product Quality, July 1997.
145 Estimate of the five carbon content of gasoline by a Refinery-by-Refinery cost model the
description of which is summarized in Chapter 9 of the Draft Regulatory Impact Analysis,
Control of Hazardous Air Pollutants From Mobile Sources - Proposed Rule; Docket # EPA-HQ-
OAR-2005-0036, March 29, 2006.
146 Final Report: 1996 American Petroleum Institute/National Petroleum Refiners Association
Survey of Refining Operations and Product Quality, July 1997.
Ill
-------
147 Table 9.2-1, Draft Regulatory Impact Analysis, Control of Hazardous Air Pollutants From
Mobile Sources - Proposed Rule; Docket* EPA-HQ-OAR-2005-0036, March 29, 2006.
148 Gary, James, H., Handewerk, Glenn E., Petroleum Refining: Technology and Economics,
Marcel Dekker, New York, 1994.
149 Gary, James, H., Handewerk, Glenn E., Petroleum Refining: Technology and Economics,
Marcel Dekker, New York, 1994.
150 Electric Power Annual 2004, Energy Information Administration, Department of Energy,
November 2005.
151 Natural Gas Annual 2004, Energy Information Administration, Department of Energy,
December 2005.
152 Annual Energy Outlook 2006, Energy Information Administration, Department of Energy,
February 2006.
153 Estimated volumes taken from refinery-by-refinery cost model, the description of which is
summarized in the Draft Regulatory Impact Analysis, Control of Hazardous Air Pollutants From
Mobile Sources - Proposed Rule; Docket* EPA-HQ-OAR-2005-0036, March 29, 2006.
154 Platts price information provided by Paulo Nery, Jacobs Engineering, June 2006.
155 White Paper discussing the conversion of MTBE plants to alkylate and isooctane, Pace
Consultants Incorporated, 2001.
156 White Paper discussing the conversion of MTBE plants to alkylate and isooctane, Pace
Consultants Incorporated, 2001.
157 Meyers, Robert A., Handbook of Petroleum Refining Processes, 2nd Edition, McGraw Hill,
1997.
158 Miller, K. Dexter Jr., DeWitt & Company Inc., Power Point presentation: Alkylates - Key
Components in Clean-Burning Gasoline, May 24, 1999.
159 White Paper discussing the conversion of MTBE plants to alkylate and isooctane, Pace
Consultants Incorporated, 2001.
160 Meyers, Robert A., Handbook of Petroleum Refining Processes, 2nd Edition, McGraw Hill,
1997.
161 Table 31, Petroleum Marketing Annual 2004, Energy Information Administration, U.S.
Department of Energy, August 2005.
112
-------
162 2006 Gas Price Kit, National Association of Convenience Stores, available at:
http://www.nacsonline.com/NACS/News/Press Releases/2006/pr020106.htm. February 2, 2006.
163 Piel, William, Tier Associates, Power Point Presentation which Summarizes Estimates of the
Oil Equivalent Cost of Ethanol, March 9, 2006.
164U.S. Environmental Protection Agency. December 2000. Clean Diesel Trucks, Buses, and
Fuel: Heavy-Duty Engine and Vehicle Standards and Highway Diesel Fuel Sulfur Control
Requirements (the "2007 Heavy-Duty Highway Rule"). Prepared by: Office of Air and
Radiation. EPA420-R-00-026. Available at http://www.epa.gov/otaq/highway-diesel/regs/2007-
heavy-duty-highway.htm . Accessed August 1, 2006.
165 U.S. Environmental Protection Agency. May 2004. Final Regulatory Analysis: Control of
Emissions from Nonroad Diesel Engines. Prepared by: Office of Air and Radiation. EPA420-R-
04-007. Available at http://www.epa.gov/nonroad-diesel/2004fr.htmtfdocuments . Accessed
August 1, 2006.
166 Hochhauser, A.M. et al. "Fuel Composition Effects on Automotive Fuel Economy - Auto/Oil
Air Quality Improvement Research Program," Society of Automotive Engineers, Paper No.
930138, 1993.
167 USDA Agricultural Baseline Projections to 2015, report OCE-2006-1.
168
F.O. Licht, World Ethanol Markets - The Outlook to 2015 (2006).
169 Phone/email discussion with USDA on 6/15/06.
170 "U.S. Biodiesel Development: New Markets for Conventional and Genetically Modified
Agriculture Fats and Oils" September 1998, page 13, USDA by James Duffield, Hosein
Shapouri, Michael Graboski, Robert McCormick and Richard Wilson.
171 "Biomass as Feedstock for a Bioenergy and Bioproducts Industry: The Technical Feasibility
of a Billion-Ton Annual Supply," Perlack, et al, DOE Report Number DOE/GO-102995-213 5,
April 2005.
172 "USDA Agricultural Baseline Projections to 2015," USDA Report Number OCE-2006-1,
February 2006.
------- |