Regulation of Fuels and Fuel Additives: Changes to Renewable Fuel Standard Program

Note: EPA no longer updates this information, but it may be useful as a reference or resource.
PDF Version (50 pp, 1031K, About PDF)

[Federal Register: May 26, 2009 (Volume 74, Number 99)]
[Proposed Rules]
[Page 25003-25052]
From the Federal Register Online via GPO Access [wais.access.gpo.gov]
[DOCID:fr26my09-23]

Regulation of Fuels and Fuel Additives: Changes to Renewable Fuel
Standard Program

[[Continued from page 25002]]

[[Page 25003]]

batches of ethanol.\163\ Thus, existing petroleum pipelines in some
areas of the country might play a role in the shipment of ethanol from
the points of production/importation to petroleum terminals.
---------------------------------------------------------------------------

    \159\ Stress corrosion cracking could lead to a pipeline leak.
The potential impacts on water from today's proposal are discussed
in Section X of today's preamble.
    \160\ Different grades of gasoline and diesel fuel are typically
shipped in multi-product pipelines in batches that abut each other.
To the extent possible, products are sequenced in a way to allow the
interface mixture between batches to be cut into one of the
adjoining products. In cases where diesel fuel abuts gasoline in the
pipeline, the resulting mixture must typically be reprocessed into
its component parts by distillation for resale as gasoline and diesel fuel.
    \161\ Gasoline-ethanol mixtures can be blended into finished gasoline.
    \162\ Association of Oil Pipelines: http://aopl.org/go/
searchresults/888/?q=ethanol&sd=&ed=.  ``Hazardous Liquid Pipelines
Transporting Ethanol, Ethanol Blends, and Other Biofuels'', Notice
of policy statement and request for comment, Pipeline and Hazardous
Materials Safety Administration, Department of Transportation,
August 10, 2007, 72 FR 45002.
    \163\ Article on shipment of ethanol in Kinder Morgan pipeline:
http://www.ethanolproducer.com/article.jsp?article_id=5149. 
---------------------------------------------------------------------------

    However, the location of ethanol plants in relation to existing
pipeline origination points will limit the role of existing pipelines
in the shipment of ethanol.\164\ Current corn ethanol production
facilities are primarily located in the Midwest far from the
origination points of most existing product pipelines and the primary
gasoline demand centers. We project that a substantial fraction of
future cellulosic ethanol plants will also be located in the Midwest,
although a greater proportion of cellulosic plants are expected to be
dispersed throughout the country compared to corn ethanol plants. The
projected locations for this subset of future cellulosic ethanol plants
more closely coincide with the origination points of product pipelines
in the Gulf Coast.\165\ Imported ethanol could also be brought into
ports near the origination point of product pipelines in the Gulf Coast
and the Northeast. Nevertheless, the majority of ethanol will continue
to be produced at locations distant from the origination points of
product pipelines and gasoline demand centers. The gathering of ethanol
from production facilities located in the Midwest and shipment by barge
down the Mississippi for introduction to pipelines in the Gulf Coast is
under consideration. However, the additional handling steps to bring
the ethanol to the pipeline origin points in this manner could negate
any potential benefit of shipment by existing petroleum pipelines
compared to direct shipment by rail.
---------------------------------------------------------------------------

    \164\ Some small petroleum product refineries are currently
limited in their ability to ship products by pipeline because their
relatively low volumes were not sufficient to justify connection to
the pipeline distribution system.
    \165\ A discussion of the projected location of cellulosic
ethanol plants is contained in Section 1.5 of the DRIA.
---------------------------------------------------------------------------

    Evaluations are also currently underway regarding the feasibility
of constructing a new dedicated ethanol pipeline from the Midwest to
the East Coast.\166\ Under such an approach, ethanol would be gathered
from a number of Midwest production facilities to provide sufficient
volume to justify pipeline operation. To the extent that ethanol
production would be further concentrated in the Midwest due to the
siting of cellulosic ethanol plants, this would tend to help justify
the cost of installing a dedicated ethanol pipeline. Substantial issues
would need to be addressed before construction on such a pipeline could
proceed, including those associated with securing new rights-of-ways
and establishing sufficient surety regarding the return on the several
billion dollar investment.
---------------------------------------------------------------------------

    \166\ Magellan and Poet joint assessment of dedicated ethanol
pipeline: http://www.magellanlp.com/news/2009/20090316_5.htm. 
---------------------------------------------------------------------------

    Due to the uncertainties regarding the degree to which pipelines
will be able to participate in the transportation of ethanol, we
assumed that ethanol will continue to be transported by rail, barge,
and truck to the terminal where it will be blended into gasoline. The
distribution by these modes can be further optimized primarily through
the increased shipment by unit train and installation of additional hub
delivery terminals that can accept large volumes of ethanol for further
distribution to satellite terminals. To the extent that pipelines do
eventually play a role in the distribution of ethanol, this could tend
to reduce distribution costs and improve reliability in supply.
    USDA estimated that in 2005 approximately 60% of ethanol was
transported by rail, 30% was transported by tank truck, and 10% was
transported by barge.\167\ Denatured ethanol is shipped from
production/import facilities to petroleum terminals where it is blended
with gasoline. When practicable, shipment by unit train is the
preferred method of rail shipment rather than shipping on a manifest
rail car basis. The use of unit trains, sometimes referred to as a
virtual pipeline, substantially reduces shipping costs and improves
reliability. Unit trains are composed entirely of 70-100 ethanol tank
cars, and are dedicated to shuttle back and forth to large hub
terminals.\168\ Manifest rail car shipment refers to the shipment of
ethanol in rail tank cars that are incorporated into trains which are
composed of a variety of other commodities. Unit trains can be
assembled at a single ethanol production plant or if a group of plants
is not large enough to support such service individually, can be formed
at a central facility which gathers ethanol from a number of producers.
The Manly Terminal in Iowa, which is the first such ethanol gathering
facility, accepts ethanol from a number of nearby ethanol production
facilities for shipment by unit train. Regional (Class 2) railroad
companies are an important link bringing ethanol to gathering
facilities for assembly into unit trains for long-distance shipment by
larger (Class 1) railroads. Ethanol is sometimes carried by multiple
modes before finally arriving at the terminal where it is blended into
gasoline. For example, some ethanol is currently shipped from the
Midwest to a hub terminal on the East Coast by unit train where a
portion is further shipped to satellite terminals by barge or tank truck.
---------------------------------------------------------------------------

    \167\ ``Ethanol Transportation Backgrounder, Expansion of U.S.
Corn-based Ethanol from the Agricultural Transportation
Perspective'', USDA, September 2007, www.ams.usda.gov/tmd/
TSB/EthanolTransportationBackgrounder09-17-07.pdf.
    \168\ Hub ethanol receipt terminals can be located at large
petroleum terminals or at rail terminals.
---------------------------------------------------------------------------

    Ethanol is blended into gasoline at either 10 or 85 volume percent
at terminals (to produce E10 and E85) for delivery to retail and fleet
facilities by tank truck. Special retail delivery hardware is needed
for E85 which can be used in flexible fuel vehicles only.\169\ The
large volume of ethanol that we project will be used by 2022 means that
more ethanol will need to be used than can be accommodated by blending
to the current legal limit of 10% in all of the gasoline used in the
country. This will require the installation of a substantial number of
new E85 refueling facilities and the addition of a substantial number
of flex-fuel vehicles to the fleet. Concerns have been raised regarding
the inducements that would be necessary for retailers to install the
needed E85 facilities and for consumers to purchase E85.\170\ As
discussed in Section V.D. of today's preamble, this is prompting many
to evaluate whether a mid-level ethanol blend (e.g. E15) might be
allowed for use in existing (non-flex-fuel) vehicles. Current refueling
equipment (not designed for E85) is only certified for ethanol blends
up to 10 volume percent (E10).\171\ Hence, if a mid-level ethanol blend
were to be introduced, fuel retail facilities would need to ensure that
the equipment used to store/dispense mid-level ethanol

[[Page 25004]]

blends is compatible with the mid-level ethanol blend.\172\
Underwriters Laboratories has one certification standard for fuel
retail equipment that covers ethanol blends up to 10%, and a separate
certification standard for equipment that dispenses ethanol blends
above 10% (including E85).\173\
---------------------------------------------------------------------------

    \169\ The cost of retail dispensing hardware which is tolerant
to ethanol blends greater than E10 is discussed in Section VIII.B.
of today's preamble and discussed in more detail in Section 4.2 of the DRIA.
    \170\ See Section V.D of today's preamble for a discussion of
issues related to use of the projected volumes of ethanol that would
be produced to comply with the RFS2 standards.
    \171\ Underwriters Laboratory certifies retail refueling
equipment. UL stated that they have data which indicates that the
use of fuel dispensers certified for up to E10 blends to dispense
blends up to a maximum ethanol content of 15 volume percent would
not result in critical safety concerns (http://www.ul.com/newsroom/
newsrel/nr021909.html ). Based on this, UL stated that it would
support authorities having jurisdiction who decide to permit legacy
equipment originally certified for up to E10 blends to be used to
dispense up to 15 volume percent ethanol. The UL announcement did
address the compatibility of underground storage tank systems with
greater than E10 blends.
    \172\ Although it has yet to be established, most underground
steel storage tanks themselves would likely be compatible with
ethanol blends greater than 10 percent. The compatibility of piping,
submersed pumps, gaskets, and seals associated with these tanks with
ethanol blends greater than 10% would also need to be evaluated.
Some fiberglass tanks are incompatible and would need to be
replaced. It is difficult and sometimes impossible to verify the
suitability of underground storage tanks and tank-related equipment
for E85 use. The State of California prohibits the conversion of
underground storage tanks to E85 use. Significant changes to
dispensers, including hoses, nozzles, and other miscellaneous
fittings would be needed to ensure they are compatible with ethanol
blends greater than 10 percent.
    \173\ Joint UL/DOE Legacy System Certification Clarification
http://www.ul.com/global/eng/documents/offerings/industries/
chemicals/flammableandcombustiblefluids/development/UL_DOE_
LegacySystemCertification.pdf. 
---------------------------------------------------------------------------

    Should other biofuels be introduced that do not require
differentiation from diesel fuel or gasoline in place of some of the
volume of ethanol that we project would be used under the RFS2
standards, this may tend to reduce the need for changes at fuel retail
facilities and the need for flex-fuel vehicles. Concerns about the
difficulties/costs associated with expanding the ethanol distribution
infrastructure and adding a sufficient number of vehicles capable of
using 10% ethanol to fleet is generating increased industry interest in
renewable diesel and gasoline which would be more transparent to the
existing fuel distribution system.
2. Overview of Biodiesel Distribution
    Biodiesel is currently transported from production plants by truck,
manifest rail car, and by barge to petroleum terminals where it is
blended with petroleum-based diesel fuel. Unblended biodiesel must be
transported and stored in insulated/heated containers in colder climes
to prevent gelling. Insulated/heated containers are not needed for
biodiesel that has been blended with petroleum-based diesel fuel (i.e.,
B2, B5). Biodiesel plants are not as dependent on being located close
to feedstock sources as are corn and cellulosic ethanol plants.\174\
Biodiesel feedstocks are typically preprocessed to oil prior to
shipment to biodiesel production facilities. This can substantially
reduce the volume of feedstocks shipped to biodiesel plants relative to
ethanol plants, and has allowed some biodiesel plants to be located
adjacent to petroleum terminals. Biodiesel production facilities are
more geographically dispersed than ethanol facilities and the
production volumes also tend to be smaller than ethanol
facilities.\175\ These characteristics in combination with the smaller
volumes of biodiesel that we project will be used under the RFS2
standards compared to ethanol allow relatively more biodiesel to be
used within trucking distance of the production facility. However, we
project that there will continue to be a strong and growing demand for
biodiesel as a blending component in heating oil which could not be
satisfied alone by local sources of production. It is likely that state
biodiesel mandates will also need to be satisfied in part by out-of-
state production. Fleets are also likely to continue to be a
substantial biodiesel user, and these will not always be located close
to biodiesel producers. Thus, we are assuming that a substantial
fraction of biodiesel will continue to be shipped long distances to
market. Downstream of the petroleum terminal, B2 and B5 can be
distributed in the same manner as petroleum diesel.
---------------------------------------------------------------------------

    \174\ Biodiesel feedstocks are typically preprocessed to oil
prior to shipment to biodiesel production facilities. This can
substantially reduce the volume of feedstocks shipped to biodiesel
plants relative to ethanol plants.
    \175\ Section 1.2 contains a discussion of our projections
regarding the location of biodiesel production facilities.
---------------------------------------------------------------------------

    Concerns remain regarding the shipment of biodiesel by pipeline
(either by batch mode or in blends with diesel fuel) related to the
contamination of other products (particularly jet fuel), the solvency
of biodiesel, and compatibility with pipeline gaskets and seals.\176\
The smaller anticipated volumes of biodiesel and the more dispersed and
smaller production facilities relative to ethanol also make biodiesel a
less attractive candidate for shipment by pipeline. Due to the
uncertainties regarding the suitability of transporting biodiesel by
pipeline, we assumed that biodiesel which needs to be transported over
long distance will be carried by manifest rail car and to a lesser
extent barge. Due to the relatively small plant size and dispersion of
biodiesel plants, we anticipate the volumes of biodiesel that can be
gathered at a single location will continue to be insufficient to
justify shipment by unit train. To the extent that pipelines do
eventually play a role in the distribution of biodiesel, this could
tend to reduce distribution costs and improve reliability in supply.
---------------------------------------------------------------------------

    \176\ Industry evaluations are currently underway to resolve these concerns.
---------------------------------------------------------------------------

3. Overview of Renewable Diesel Distribution
    We believe that renewable diesel fuel will be confirmed to be
sufficiently similar to petroleum-based diesel fuel blendstocks with
respect to distribution system compatibility. Hence, renewable diesel
fuel could be treated in the same manner as any petroleum-based diesel
fuel blendstock with respect to transport in the existing petroleum
distribution system. Approximately two-thirds of renewable diesel fuel
is projected to be produced at petroleum refineries.\177\ The transport
of such renewable diesel fuel would not differ from petroleum-based
diesel fuel since it would be blended to produce a finished diesel fuel
before leaving the refinery. The other one-third of renewable diesel
fuel is projected to be produced at stand-alone facilities located more
closely to sources of feedstocks. We anticipate that such renewable
diesel fuel would be shipped by tank truck to nearby petroleum
terminals where it would be blended directly into diesel fuel storage
tanks. Because of its high cetane and value, we anticipate that all
renewable diesel fuel would likely be blended with petroleum based
diesel fuel prior to use. Downstream of the terminal, renewable/
petroleum diesel fuel mixtures would be distributed the same as
petroleum diesel.
---------------------------------------------------------------------------

    \177\ Either co-processed with crude oil or processed in
separate units at the refinery for blending with other refinery
diesel blendstocks.
---------------------------------------------------------------------------

4. Changes in Freight Tonnage Movements
    To evaluate the magnitude of the challenge to the distribution
system up to the point of receipt at the terminal, we compared the
growth in freight tonnage for all commodities from the AEO 2007
reference case to the growth in freight tonnage under the RFS2
standards in which ethanol increases, as does the feedstock (corn) and
co-products (distillers grains). We did not include a consideration of
the distribution of cellulosic ethanol feedstocks on freight tonnage
for the proposal. We intend to evaluate this in the final rule. For
purposes of this analysis, we focused on only the ethanol portion of
the renewable fuel goals for ease of calculation and because ethanol
represents the vast majority of the total volume of biofuel. The
resulting calculations serve as an indicator of changes in freight
tonnages associated with increases in renewable fuels. We calculated
the freight tonnage for the total of all modes of transport as well as
the individual cases of rail, truck, and barge.

[[Page 25005]]

    In calculating the reference case percent growth rate in total
freight tonnage, we used data compiled by the Federal Highway
Administration to calculate the tonnages associated with these
commodities.\178\ We then calculated the growth in freight tonnage for
2022 under the RFS2 standards and compared the difference with the
reference case. The comparisons indicate that across all transport
modes, the incremental increase in freight tonnage of ethanol and
accompanying feedstocks and co-products associated with the increased
ethanol volume under the RFS2 standards are small. The percent increase
for total freight across all modes (including pipeline) by 2022 is 0.9
percent. Because pipelines currently do not carry ethanol, and the
increase in the volume of ethanol used in motor vehicles displaces a
corresponding volume of gasoline, pipelines showed a decrease in the
total tonnage carried due to a decrease in the volume of gasoline
carried by pipeline. The displaced gasoline also resulted in some
decrease in tonnage in other modes that slightly reduced the overall
increases in tonnage reflected in the totals.
---------------------------------------------------------------------------

    \178\ http://www.ops.fhwa.dot.gov/freight/freight_analysis/faf/
index.htm.
---------------------------------------------------------------------------

    To further evaluate the magnitude of the increase in freight
tonnage under the RFS2 standards, we calculated the portion of the
total freight tonnage for rail, barge, and truck modes made up of
ethanol-related freight for both the 2022 and control cases. The
freight associated with ethanol constitutes only a very small portion
of the total freight tonnage for all commodities. Specifically, ethanol
freight represents approximately 0.5% and 2.5% of total freight for the
reference case and RFS2 standards case, respectively. The results of
this analysis suggest that it should be feasible for the distribution
infrastructure upstream of the terminal to accommodate the additional
freight associated with this RFS2 standards especially given the lead
time available. Specific issues related to transportation by rail,
barge, and tank truck are discussed in the following sections. We
intend to incorporate the results of a recently completed study by Oak
Ridge National Laboratory (ORNL) on the potential constraints in
ethanol distribution into the analysis for the final rule.\179\ The
ORNL study concluded that the increase in ethanol transport would have
minimal impacts on the overall transportation system. However, the ORNL
study did identify localized areas where significant upgrades to the
rail distribution system would likely be needed.
---------------------------------------------------------------------------

    \179\ ``Analysis of Fuel Ethanol Transportation Activity and
Potential Distribution Constraints'', prepared for EPA by Oak Ridge
National Laboratory, March 2009.
---------------------------------------------------------------------------

5. Necessary Rail System Accommodations
    Many improvements to the freight rail system will be required in
the next 15 years to keep pace with the large increase in the overall
freight demand. Improvements to the freight railroad infrastructure
will be driven largely by competition in the burgeoning inter-model
transport sector. As inter-model freight represents the vast majority
of all freight hauled by these railroads, the biofuels transport sector
can be expected to benefit from the infrastructure build-out resulting
from inter-model transport sector competition. As such, most of the
needed upgrades to the rail freight system are not specific to the
transport of renewable fuels and would be needed irrespective of
today's proposed rule. We also expect that the excess rail capacity
associated with inter-model build-out to be adequately large to absorb
potential increases in truck transport associated with fuel cost
increases. The modifications required to satisfy the increase in demand
include upgrading tracks to allow the use of heavier trains at faster
speeds, the modernization of train braking systems to allow for
increased traffic on rail lines, the installation of rail sidings to
facilitate train staging and passage through bottlenecks.
    Some industry groups \180\ and governmental agencies in discussions
with EPA, and in testimony provided for the Surface Transportation
Board (STB) expressed concerns about the ability of the rail system to
keep pace with the large increase in demand even under the reference
case (27% by 2022). For example, the electric power industry has had
difficulty keeping sufficient stores of coal in inventory at power
plants due to rail transport difficulties and has expressed concerns
that this situation will be exacerbated if rail congestion worsens. One
of the more sensitive bottleneck areas with respect to the movement of
ethanol from the Midwest to the East coast is Chicago. The City of
Chicago commissioned its own analysis of rail capacity and congestion,
which found that the lack of rail capacity is ``no longer limited to a
few choke points, hubs, and heavily utilized corridors.'' Instead, the
report finds, the lack of rail capacity is ``nationwide, affecting
almost all the nation's critically important trade gateways, rail hubs,
and intercity freight corridors.''
---------------------------------------------------------------------------

    \180\ Industry groups include the Alliance of Automobile
Manufacturers, American Chemistry Council, and the National
Industrial Transportation League; governmental agencies include the
Federal Railroad Administration (FRA), the Government Accountability
Office (GAO), and the American Association of State Highway
Transportation Officials (AASHTO). Testimony for the STB public
hearings includes Ex Parte No. 671, Rail Capacity and Infrastructure
Requirements and Ex Parte No. 672, Rail Transportation and Resources
Critical to the Nation's Energy Supply.
---------------------------------------------------------------------------

    Significant private and public resources are focused on making the
modifications to the rail system to cope with the increase in demand.
Rail carriers report that they typically invest $16 to $18 billion a
year in infrastructure improvements.\181\ Substantial government loans
are also available to small rail companies to help make needed
improvements by way of the Railroad Rehabilitation and Improvement
Finance (RRIF) Program, administered by Federal Railroad Administration
(FRA), as well as Section 45G Railroad Track Maintenance Credits,
offered by the Internal Revenue Service (IRS). The American Association
of State Highway Transportation Officials (AASHTO) estimates that
between $175 billion and $195 billion must be invested over a 20-year
period to upgrade the rail system to handle the anticipated growth in
freight demand, according to the report's base-case scenario.\182\ The
report suggests that railroads should be able to provide up to $142
billion from revenue and borrowing, but that the remainder would have
to come from other sources including, but not limited, to loans, tax
credits, sale of assets, and other forms of public-sector
participation. Given the reported historical investment in rail
infrastructure, it may be reasonable to assume that rail carriers would
be able to manage the $7.1 billion in annual investment from rail
carriers that AASHTO projects would be needed to keep pace with the
projected increase in freight demand.
---------------------------------------------------------------------------

    \181\ ``The Importance of Adequate Rail Investment'',
Association of American Railroads, www.aar.org/GetFile.asp?File_ID=150.
    \182\ AASHTO Freight-Rail Bottom-Line Report, 2003.
---------------------------------------------------------------------------

    However, the Government Accounting Office (GAO) found that it is
not possible to independently confirm statements made by Class I rail
carriers regarding future investment plans.\183\ In

[[Page 25006]]

addition, questions persist regarding allocation of these investments,
with the Alliance of Automobile Manufacturers, American Chemistry
Council, National Industrial Transportation League, and others
expressing concern that their infrastructural needs may be neglected by
the Class I railroads in favor of more lucrative intermodal traffic.
Moreover, the GAO has raised questions regarding the competitive nature
and extent of Class I freight rail transport. This raises some concern
that providing sufficient resources to facilitate the transport of
increasing volumes of ethanol and biodiesel might not be a first
priority for rail carriers. In response to GAO concerns, the Surface
Transportation Board (STB) agreed to undertake a rigorous analysis of
competition in the freight railroad industry.\184\
---------------------------------------------------------------------------

    \183\ The railroads interviewed by GAO were generally unwilling
to discuss their future investment plans with the GAO. Therefore,
GAO was unable to comment on how Class I freight rail companies are
likely to choose among their competing investment priorities for the
future, including those of the rail infrastructure, GAO testimony
Before the Subcommittee on Surface Transportation and Merchant
Marine, Senate Committee on Commerce, Science, and Transportation,
U.S. Senate, Freight Railroads Preliminary Observations on Rates,
Competition, and Capacity Issues, Statement of JayEtta Z. Hecker,
Director, Physical Infrastructure Issues, GAO, GAO-06-898T
(Washington, DC: June, 21, 2006).
    \184\ GAO, Freight Railroads: Industry Health Has Improved, but
Concerns about Competition and Capacity Should Be Addressed, GAO-07-
94 (Washington, DC: Oct. 6, 2006); GAO, Freight Railroads: Updated
Information on Rates and Other Industry Trends, GAO-07-291R Freight
Railroads (Washington, DC: Aug. 15, 2007). STB's final report,
entitled Report to the U.S. STB on Competition and Related Issues in
the U.S. Freight Railroad Industry, is expected to be completed
November, 1, 2008.
---------------------------------------------------------------------------

    Given the broad importance to the U.S. economy of meeting the
anticipated increase in freight rail demand, and the substantial
resources that seem likely to be focused on this cause, we believe that
overall freight rail capacity would not be a limiting factor to the
successful implementation of the biofuel requirements to meet the RFS2
standards. Evidence from the recent ramp up of ethanol use has also
shown that rail carriers are enthusiastically pursuing the shipment of
ethanol. Class 2 railroads have been particularly active in gathering
sufficient numbers of ethanol cars to allow Class 1 railroads to ship
ethanol by unit train. Likewise, we believe that that Class 2 railroads
and, to a lesser extent, the trucking industry, will play a key role in
the transportation of DDGs and other byproducts from regions with
concentrated ethanol production facilities to those with significant
livestock operations. Based on this recent experience, we believe that
ethanol will be able to compete successfully with other commodities in
securing its share of freight rail service.
    While many changes to the overall freight rail system are expected
to occur irrespective of today's proposed rule, a number of ethanol-
specific modifications will be needed. For instance, a number of
additional rail terminals are likely to be configured for receipt of
unit trains of ethanol for further distribution by tank truck or other
means to petroleum terminals. The placement of ethanol unit train
receipt facilities at rail terminals would be particularly useful in
situations where petroleum terminals might find it difficult or
impossible to install their own ethanol rail receipt capability. We
anticipate that ethanol storage will typically be installed at rail
terminal ethanol receipt hubs over the long run. We do not anticipate
that the rail industry will experience substantial difficulty in
installing such ethanol-specific facilities once a clear long term
demand for ethanol in the target markets has been established to
justify the investment. However, the need for long-term demand to be
established prior to the construction of such facilities will likely
mean that the needed facilities will, at best, come on-line on a just-
in-time basis. This may lead to use of less efficient means of ethanol
transport in the short term. The ability to rely on transloading while
ethanol storage facilities at rail terminal ethanol receipt hub
facilities are constructed will speed the optimization of the
distribution of ethanol by rail by allowing the construction of ethanol
storage at rail terminal hubs to be delayed.
    We estimate that a total of 44,000 rail cars would be needed to
distribute the volumes of ethanol and biodiesel that we project would
be used in 2022 to satisfy the RFS2 requirements.\185\ Our analysis of
ethanol and biodiesel rail car production capacity indicates that
access to these cars should not represent a serious impediment to
meeting the requirements under the RFS2 standards. Ethanol tank car
production has increased approximately 30% per year since 2003, with
over 21,000 tank cars expected to be produced in 2007. The volume of
these newly-produced tank cars, coupled with that of an existing tank
car fleet already dedicated to ethanol and biodiesel transport,
suggests that an adequate number of these tank cars will be in place to
transport the proposed renewable fuel volume requirements in the time
available.
---------------------------------------------------------------------------

    \185\ A discussion of how we arrived at the estimated number of
tank cars needed is contained in Section 4.2 of the DRIA.
---------------------------------------------------------------------------

    We request comment on the extent to which the rail system will be
able to deliver the additional volumes of ethanol and biodiesel that we
anticipate would be used in response to the RFS2 standards in a timely
and reliable fashion. A recently completed report by ORNL identifies
specific segments of the rail system which would likely see the most
significant increase in traffic due to increased shipments of ethanol
under the EISA.\186\
---------------------------------------------------------------------------

    \186\ ``Analysis of Fuel Ethanol Transportation Activity and
Potential Distribution Constraints'', prepared for EPA by Oak Ridge
National Laboratory, March 2009.
---------------------------------------------------------------------------

6. Necessary Marine System Accommodations
    The American Waterway's Association has expressed concerns about
the need to upgrade the inland waterway system in order to keep pace
with the anticipated increase in overall freight demand. The majority
of these concerns have been focused on the need to upgrade the river
lock system on the Mississippi River to accommodate longer barge tows
and on dredging inland waterways to allow for movement of fully loaded
vessels. We do not anticipate that a substantial fraction of renewable/
alternative fuels will be transported via these arteries. Thus, we do
not believe that the ability to ship ethanol/biodiesel by inland marine
will represent a serious barrier to the implementation of
implementation of the requirements under RFS2 standards. Substantial
quantities of the corn ethanol co-product dried distiller grains (DDG)
is expected to be exported from the Midwest via the Mississippi River
as the U.S. demand for DDG becomes saturated. We anticipate that the
volume of exported DDG would take the place of corn that would be
shifted from export to domestic use in the production of ethanol. Thus,
we do not expect the increase in DDG exports to result in a substantial
increase in river freight traffic. We request comment on the extent to
which marine transport may be used in the transport of cellulosic
ethanol feedstocks.
7. Necessary Accommodations to the Road Transportation System
    Concerns have been raised regarding the ability of the trucking
industry to attract a sufficient number of drivers to handle the
anticipated increase in truck freight.\187\ The American Trucking
Association projected the need for additional 54,000 drivers each year.
We estimate that the growth in the use of biofuels through 2022 due to
the RFS2 standards would result in the need for a total of approximately 3,000

[[Page 25007]]

additional trucks drivers. Given the relatively small number of new
truck drivers needed to transport the volumes of biofuels needed to
comply with the RFS2 standards through 2022 compared to the total
expected increase in demand for drivers over the same time period
(>750,000), we do not expect that the implementation of the RFS2
standards would substantially impact the potential for a shortage of
truck drivers. However, specially certified drivers are required to
transport ethanol and biodiesel because these fuels are classified as
hazardous liquids. Thus, there may be a heightened level of concern
about the ability to secure a sufficient number of such specially
certified tank truck drivers to transport ethanol and biodiesel. The
trucking industry is involved in efforts to streamline the
certification of drivers for hazardous liquids transport and more
generally to attract and retrain a sufficient number of new truck drivers.
---------------------------------------------------------------------------

    \187\ ``The U.S. Truck Driver Shortage: Analysis and
Forecasts'', Prepared by Global Insights for the American Trucking
Association, May 2005. www.truckline.com/NR/rdonlyres/
E2E789CF-F308-463F-8831-0F7E283A0218/0/ATADriverShortageStudy05.pdf.
---------------------------------------------------------------------------

    Truck transport of biofuel feedstocks to production plants and
finished biofuels and co-products from these plants is naturally
concentrated on routes to and from these production plants. This may
raise concerns about the potential impact on road congestion and road
maintenance in areas in the proximity of these facilities. We do not
expect that such potential concerns would represent a barrier to the
implementation of the RFS2 standards. The potential impact on local
road infrastructure and the ability of the road network to be upgraded
to handle the increased traffic load is an inherent part in the
placement of new biofuel production facilities. Consequently, we expect
that any issues or concerns would be dealt with at the local level.
    We request comment on the extent to which satisfying the
requirements under the RFS2 standards might exacerbate the anticipated
shortage of truck drivers or lead to localized road congestion and
condition problems. Comment is further requested on the means to
mitigate such potential difficulties to the extent they might exist.
8. Necessary Terminal Accommodations
    Terminals will need to install additional storage capacity to
accommodate the volume of ethanol/biodiesel that we anticipate will be
used in response to the RFS2 standards. Questions have been raised
about the ability of some terminals to install the needed storage
capacity due to space constraints and difficulties in securing
permits.\188\ Overall demand for fuel used in spark ignition motor
vehicles is expected to remain relatively constant through 2022. Thus,
much of the demand for new ethanol and biodiesel storage could be
accommodated by modifying storage tanks previously used for the
gasoline and petroleum-based diesel fuels that would displaced by
ethanol and biodiesel. The areas served by existing terminals also
often overlap. In such cases, one terminal might be space constrained
while another serving the same area may be able to install the
additional capacity to meet the increase in demand. Terminals with
limited ethanol storage (or no access to rail/barge ethanol shipments)
could receive truck shipments of ethanol from terminals with more
substantial ethanol storage (and rail/barge receipt) capacity. The
trend towards locating ethanol receipt and storage capability at rail
terminals located near petroleum terminals is likely to be an important
factor in reducing the need for large volume ethanol receipt and
storage facilities at petroleum terminals. In cases where it is
impossible for existing terminals to expand their storage capacity due
to a lack of adjacent available land or difficulties in securing the
necessary permits, new satellite storage or new separate terminal
facilities may be needed for additional ethanol and biodiesel storage.
However, we believe that there would be few such situations.
---------------------------------------------------------------------------

    \188\ The Independent Fuel Terminal Operators Association
represents terminals in the Northeast.
---------------------------------------------------------------------------

    Another question is whether the storage tank construction industry
would be able to keep pace with the increased demand for new tanks that
would result from today's proposal. The storage tank construction
industry recently experienced a sharp increase in demand after years of
relatively slack demand for new tankage. Much of this increase in
demand was due to the unprecedented increase in the use of ethanol.
Storage tank construction companies have been increasing their
capabilities which had been pared back during lean times.\189\ Given
the projected gradual increase in the need for biofuel storage tanks,
it seems reasonable to conclude that the storage tank construction
industry would be able to keep pace with the projected demand.
---------------------------------------------------------------------------

    \189\ It currently may take 4 to 8 months to begin construction
of a storage tank after a contract is signed due to tightness in
construction assets and steel supply.
---------------------------------------------------------------------------

    The RFG and anti-dumping regulations currently require certified
gasoline to be blended with denatured ethanol to produce E85. The
gasoline must meet all applicable RFG and anti-dumping standards for
the time and place where it is sold. We understand that some parties
may be blending butanes and or pentanes into gasoline before it is
blended with denatured ethanol in order to meet ASTM minimum volatility
specifications for E85 that were set to ensure proper drivability,
particularly in the winter.\190\ If terminal operators add blendstocks
to finished gasoline for use in manufacturing E85, the terminal
operator would need to register as a refiner with EPA and meet all
applicable standards for refiners.
---------------------------------------------------------------------------

    \190\ ``Specification for Fuel Ethanol (Ed75-Ed85) for Spark-Ignition
Engines'', American Society for Testing and Materials standard ASTM D5798.
---------------------------------------------------------------------------

    Recent testing has shown that much of in-use E85 does not meet
minimum ASTM volatility specifications.\191\ However, it is unclear if
noncompliance with these specifications has resulted in a commensurate
adverse impact on drivability. This has prompted a re-evaluation of the
fuel volatility requirements for in-use E85 vehicles and whether the
ASTM E85 volatility specifications might be relaxed.\192\ For the
purpose of our analysis, we are assuming that certified gasoline
currently on hand at terminals can be used to make up the non-ethanol
portion of E85.\193\
---------------------------------------------------------------------------

    \191\ Coordinating Research Council (CRC) report No. E-79-2,
Summary of the Study of E85 Fuel in the USA Winter 2006-2007, May
2007. http://www.crcao.org/reports/recentstudies2007/E-79-2/E-79-
2%20E85%20Summary%20Report%202007.pdf. 
    \192\ CRC Cold Start and Warm-up E85 Driveability Program,
http://www.crcao.com/about/Annual%20Report/2007%20Annual%20Report/
Perform/CM-133.htm. 
    \193\ This is different from the approach taken in the refinery
modeling which assumed that special blendstocks would be used to
blend E85. A discussion of the refinery modeling can be found in
Section 4 of the DRIA.
---------------------------------------------------------------------------

    We request comment on the extent that this will be the case in
light of the projected outcome of the ASTM process. Comment is
requested on the fraction of terminals that currently have butane/
pentane blending capability and the logistical/cost implications of
adding such capability including sourcing and transportation issues
associated with supplying these blending components to the terminal for
the purpose of blending E85 to ASTM specifications. We also seek
comment on whether we should include a separate section in the RFS2
regulations to specify the requirements for producing E85, and whether
we should provide E85 manufacturers who use blendstocks to produce E85
with any flexibilities in complying with the refiner requirements.\194\
---------------------------------------------------------------------------

    \194\ Certain accommodations for butane blenders into gasoline
were provided in a direct final rule published on December 15, 2005
entitled, ``Modifications to Standards and Requirements for
Reformulated and Conventional Gasoline Including Butane Blenders and
Attest Engagements'', 70 FR 74552.

---------------------------------------------------------------------------

[[Page 25008]]

    A significant challenge facing terminals and one that is currently
limiting the volume of ethanol that can be used is the ability to
receive ethanol by rail. Only a small fraction of petroleum terminals
currently have rail receipt capability and a number likely have space
constraints or are located too far from the rail system which prevents
the installation of such capability. The trend to locate ethanol unit
train destinations at rail terminals will help to alleviate these
concerns. Petroleum terminals within trucking distance of each other
are also likely to cooperate so that only one would need to install
rail receipt capability. Given the timeframe during which the projected
volumes of ethanol ramp up, we believe that these means can be utilized
to ensure that a sufficient number of terminals have access to ethanol
shipped by rail although some will need to rely on secondary shipment
by truck from large ethanol hub receipt facilities. We request comment
on the current rail receipt capability at terminals and the extent to
which petroleum terminals can be expected to install such capability.
Comment is also requested on the extent to which the installation of
ethanol receipt facilities at rail terminals can help to meet the need
to bring ethanol by rail to petroleum terminals. Our current analysis
estimated that half of the new ethanol rail receipt capability needed
to support the use of the projected ethanol volumes under the EISA
would be installed at petroleum terminals, and half would be installed
at rail terminals. A recently completed study by ORNL estimated that
all new ethanol rail receipt capability would be installed at existing
rail terminals given the limited ability to install such capability at
petroleum terminals.\195\ We intend to review our estimates regarding
the location of the additional ethanol rail receipt facilities for the
final rule in light of the ORNL study.
---------------------------------------------------------------------------

    \195\ ``Analysis of Fuel Ethanol Transportation Activity and
Potential Distribution Constraints'', prepared for EPA by Oak Ridge
National Laboratory, March 2009.
---------------------------------------------------------------------------

9. Need for Additional E85 Retail Facilities
    We estimate that an additional 24,250 E85 retail facilities would
be needed to facilitate the consumption of the additional amount of
ethanol that we project would be used by 2022 in response to the
requirements under the RFS2 standards.\196\ On average, this equates to
approximately 1,960 new E85 facilities that would need to be added each
year from 2009 through 2022 in order to satisfy this goal. This is a
very ambitious timeline given that there are less than 2,000 E85 retail
facilities in service today. Nevertheless, we believe the addition of
these numbers of new E85 facilities may be possible for the industries
that manufacture and install E85 retail equipment. Underwriters
Laboratories recently finalized its certification requirements for E85
retail equipment.\197\ Equipment manufactures are currently evaluating
the changes that will be needed to meet these requirements.\198\
However, we anticipate the needed changes will not substantially
increase the difficulty in the manufacture of such equipment compared
to equipment which is specifically manufactured for dispensing E85 today.
---------------------------------------------------------------------------

    \196\ See Section 1.6 of the DRIA for a discussion of the
projected number of E85 refueling facilities that would be needed.
There would need to be a total of 28,750 E85 retail facilities,
4,500 of which are projected to have been placed in service absent
the RFS2 standards.
    \197\ See http://ulstandardsinfonet.ul.com/outscope/0087A.html. 
    \198\ All dispenser equipment except the hose used to dispense
fuel to the vehicle has been evaluated by UL. Once suitable hoses
have been evaluated, a complete E85 dispenser system can be
certified by UL.
---------------------------------------------------------------------------

    We estimate that the cost of installing E85 refueling equipment
will average $122,000 per facility which equates to $3 billion by
2022.\199\ These costs include the installation of an underground
storage tank, piping, dispensers, leak detection, and other ancillary
equipment that is compatible with E85.\200\ Our E85 facility cost
estimates are based on input from fuel retailers and other parties with
familiarity in installing E85 compatible equipment. We understand that
a certification has yet to be finalized by Underwriters Laboratories
for a complete equipment package necessary to store/dispense E85 at a
retail facility.\201\ Thus, there is some uncertainty regarding the
type of equipment that will be needed for compliance with the E85
equipment certification requirements, and the associated costs.
Nevertheless, we believe that the E85 equipment that is eventually
certified for use will not be substantially different from that on
which our cost estimates are based.\202\
---------------------------------------------------------------------------

    \199\ See Section 4.2 of the DRIA for a discussion of E85
facility costs. These costs include the installation of 2 pumps with
4 E85 refueling positions at 40% of new facilities, and 1 pump with
2 refueling positions at 60% of new facilities. A sensitivity case
was evaluated where it was assumed that all new E85 facilities would
install 3 pumps with 6 refueling positions. The cost per facility
under this sensitivity case is $166,000.
    \200\ 40 CFR 280.32 requires that underground storage tank
systems must be made of or lined with materials that are compatible
with the substance stored in the system.
    \201\ Underwriters Laboratories recently finalized their
requirements for the certification of E85 compatible equipment. No
certifications have been completed to date, because of the time
needed to complete the application for certification including
necessary testing.
    \202\ All retail dispenser components except the hose that
connects the nozzle to the dispenser have been evaluated by UL. Once
such hoses have been evaluated by UL, a certification for the
complete fuel dispenser assembly may be finalized by UL.
---------------------------------------------------------------------------

    Petroleum retailers expressed concerns about their ability to bear
the cost installing the needed E85 refueling equipment. Today's
proposal does not contain a requirement for retailers to carry E85. We
understand that retailers will only install E85 facilities if it is
economically advantageous for them to do so and that they will price
their E85 and E10 in a manner to recover these costs. While the $3
billion total cost for E85 refueling facilities is a substantial sum,
it equates to just 1.5 cents per gallon of E85 throughput.\203\
Therefore, we do not believe that the cost of installing E85 refueling
equipment will represent an undue burden to retailers given the very
large projected consumer demand for E85.
---------------------------------------------------------------------------

    \203\ E85 facility costs were amortized over 15 years at 7% and
the costs spread over the projected volume of E85 dispensed.
---------------------------------------------------------------------------

    Petroleum retailers also expressed concern regarding their ability
to discount the price of E85 sufficiently to persuade flexible fuel
vehicle owners to choose E85 given the lower energy density of ethanol.
This issue is discussed in Section V.D.2.e. of today's preamble.

D. Ethanol Consumption

1. Historic/Current Ethanol Consumption
    Ethanol and ethanol-gasoline blends have a long history as
automotive fuels. However, cheap gasoline/blendstocks kept ethanol from
making a significant presence in the transportation sector until the
end of the 20th century when environmental regulations and tax
incentives helped to stimulate growth.
    In 1978, the U.S. passed the Energy Tax Act which provided an
excise tax exemption for ethanol blended into gasoline that would later
be modified through subsequent regulations.\204\ In the 1980s, EPA
initiated a phase-out of leaded gasoline which created some interest in
ethanol as a gasoline

[[Page 25009]]

oxygenate. Upon passage of the 1990 CAA amendments, states implemented
winter oxygenated fuel (``oxyfuel'') programs to monitor carbon
monoxide emissions. EPA also established the reformulated gasoline
(RFG) program to help reduce emissions of smog-forming and toxic
pollutants. Both the oxyfuel and RFG programs called for oxygenated
gasoline. However, petroleum-derived ethers, namely methyl tertiary
butyl ether (MTBE), dominated oxygenate use until drinking water
contamination concerns prompted a switch to ethanol. Additional support
came in 2004 with the passage of the Volumetric Ethanol Excise Tax
Credit (VEETC). The VEETC provided domestic ethanol blenders with a
$0.51/gal tax credit, replacing the patchwork of existing
subsidies.\205\ The phase-out of MTBE and the introduction of the VEETC
along with state mandates and tax incentives created a growing demand
for ethanol that surpassed the traditional oxyfuel and RFG markets. By
the end of 2004, not only was ethanol the lead oxygenate, it was found
to be blended into a growing number of states' conventional gasoline.\206\
---------------------------------------------------------------------------

    \204\ Gasohol, a fuel containing at least 10% biomass-derived
ethanol, received a partial exemption from the federal gasoline
excise tax. This exemption was implemented in 1979 and a blender's
tax credit and a pure alcohol fuel credit were added to the mix in 1980.
    \205\ The 2008 Farm Bill, discussed in more detail in Section
V.B.2.b, replaces the $0.51/gal ethanol blender credit with a $0.45/
gal corn ethanol blender credit and also introduces a $1.01/gal
cellulosic biofuel producer credit. Both credits are effective January 1, 2009.
    \206\ Based on 2004 Federal Highway Association (FHWA) State
Gasohol Report less estimated RFG and oxyfuel ethanol usage based on
EPA's 2004 RFG Fuel Survey results and knowledge of state oxyfuel
programs and fuel oxygenates. For more on historical ethanol usage
by state and fuel type, refer to Section 1.7.1.1 of the DRIA.
---------------------------------------------------------------------------

    In the years that followed, rising crude oil prices and other
favorable market conditions continued to drive ethanol usage. In May
2007, EPA promulgated a Renewable Fuel Standard (``RFS1'') in response
to EPAct. The RFS1 program set a floor for renewable fuel use reaching
7.5 billion gallons by 2012, the majority of which was ethanol. The
country is currently on track for exceeding the RFS1 requirements and
meeting the introductory years of today's proposed RFS2 program. For a
summary of the growth in U.S. ethanol usage over the past decade, refer
to Table V.D.1.-1.

       Table V.D.1-1--U.S. Ethanol Consumption (Including Imports)
------------------------------------------------------------------------
                                                  Total ethanol use \a\
                                               -------------------------
                     Year                         Trillion
                                                    BTU          Bgal
------------------------------------------------------------------------
1999..........................................          120          1.4
2000..........................................          138          1.6
2001..........................................          144          1.7
2002..........................................          171          2.0
2003..........................................          233          2.8
2004..........................................          292          3.5
2005..........................................          334          4.0
2006..........................................          451          5.3
2007..........................................          566          6.7
2008..........................................          792          9.4
------------------------------------------------------------------------
\a\ EIA Monthly Energy Review March 2009 (Table 10.2).

    Through the years, there have also been several policy initiatives
to increase the number of flexible fuel vehicles (FFVs) capable of
consuming up to 85 volume percent ethanol blends (E85). The Alternative
Motor Vehicle Fuels Act of 1988 provided automakers with Corporate
Average Fuel Economy (CAFE) credits for producing alternative-fuel
vehicles, including FFVs as well as CNG and propane vehicles.
Furthermore, the Energy Policy Act of 1992 required government fleets
to begin purchasing alternative-fuel vehicles, and the majority of
fleets chose FFVs.\207\ As a result of these two policy measures, there
are over 7 million FFVs on the road today.\208\ These vehicles increase
our nation's ethanol consumption potential beyond what is capable with
conventional vehicles. However, most FFVs are currently refueling on
conventional gasoline (E0 or E10) due to limited E85 availability and
the fact that E85 is typically priced 20-30 cents per gallon higher
than gasoline on an energy equivalent basis. As such, we are not
currently tapping into the full ethanol consumption potential of our
FFV fleet. However, we expect refueling patterns to change in the
future under the RFS2 program.
---------------------------------------------------------------------------

    \207\ Source: June 23, 2008 Federal Times, Special Report: Fleet Management.
    \208\ Source: DOE Energy Efficiency and Renewable Energy
(worksheet available at www.eere.energy.gov/afdc/data/index.html.)
---------------------------------------------------------------------------

2. Increased Ethanol Use under RFS2
    To meet the RFS2 standards, ethanol consumption will need to be
much higher than both today's levels and those projected to occur
absent RFS2. The Energy Information Administration (EIA) projected that
under business-as-usual conditions, ethanol usage would grow to just
over 13 billion gallons by 2022.\209\ This represents significant
growth from today's usage, however, this volume of ethanol is capable
of being consumed by today's vehicle fleet albeit with some fuel
infrastructure improvements.\210\ Although EIA projected a small
percentage of ethanol to be blended as E85 in 2022, 13 billion gallons
of ethanol could also be consumed by displacing about 90% of our
country's forecasted gasoline energy demand with E10. The maximum
amount of ethanol our country is capable of consuming as E10 compared
to the projected RFS2 ethanol volumes is shown below in Figure V.D.2-1.\211\
---------------------------------------------------------------------------

    \209\ Source: EIA Annual Energy Outlook 2007, Table 17.
    \210\ For more information on distribution accommodations, refer
to Section V.C.
    \211\ The maximum E10 volumes are a function of the gasoline
energy demand reported in EIA's Annual Energy Outlook 2009, Table 2
adjusted with lower heating values.

---------------------------------------------------------------------------

[[Page 25010]]
[GRAPHIC] [TIFF OMITTED] TP26MY09.006

    As shown in Figure V.D.2-1, under the proposed RFS2 program, we are
projected to hit the E10 ``blend wall'' of about 14.5 billion gallons
of ethanol by 2013. This volume corresponds to 100% E10 nationwide.
However, if gasoline demand falls, or if E10 cannot get distributed
nationwide, the nation could hit the blend wall sooner. Regardless, to
get beyond the blend wall and consume more than 14-15 billion gallons
of ethanol, we are going to need to see significant increases in the
number FFVs on the road, the number of E85 retailers, and the FFV E85
refueling frequency. In the subsections that follow, we will highlight
the variables that impact our nation's ethanol consumption potential
and, more specifically, what measures the market may need to take in
order to consume 34 billion gallons of ethanol by 2022 (assuming the
cellulosic biofuel standard and the majority of the advanced biofuel
standard are met with ethanol).
---------------------------------------------------------------------------

    \212\ Based on the assumption that the cellulosic biofuel
standard and the majority of the advanced biofuel standard would be
met with ethanol.
---------------------------------------------------------------------------

    As explained in Section V.A.2, our primary RFS2 analysis focuses on
ethanol as the main biofuel in the future.\213\ In addition, from an
ethanol consumption standpoint, we have focused on an E10/E85 world.
While E0 is capable of co-existing with E10 and E85 for a while, we
assumed that E10 would replace E0 as expeditiously as possible and that
all subsequent ethanol growth would come from E85. Furthermore, for our
primary analysis, we assumed that no ethanol consumption would come
from the mid-level ethanol blends (i.e., E15 or E20) as they are not
currently approved for use in non-FFVs. However, in Section V.D.3
below, we discuss the potential approval pathways for mid-level ethanol
blends and the volume implications.
---------------------------------------------------------------------------

    \213\ For consideration of other biofuels, refer to Section V.D.3.d.
---------------------------------------------------------------------------

    We acknowledge that, if approved, mid-level ethanol blends could
help the nation meet the proposed RFS2 volume requirements. First, non-
FFVs could consume more ethanol per gallon of ``gasoline''. This could
result in greater ethanol consumption nationwide. In addition, mid-
level blends could allow gasoline retailers to continue to price
ethanol relative to gasoline (as it currently is for E10). For these
reasons, it is possible that mid-level ethanol blends could help the
nation get beyond the E10 blend wall. However, as explained in Section
V.D.3.b, there are numerous actions that would need to be taken to
bring mid-level ethanol blends to market. In addition, mid-level
ethanol blends alone (even if made available nationwide) are not
capable of fulfilling the RFS2 requirements in later years. We would
essentially hit another blend wall 1-6 years later depending on the
intermediate blend, how quickly it could be brought to market, and how
widely mid-level ethanol blends were distributed at retail stations
nationwide. Nevertheless, this time could be very valuable when it
comes to expanding E85/FFV infrastructure and/or commercializing other
non-ethanol cellulosic biofuels.
    Regardless, our primary analysis focuses on an E10/E85 world
because mid-level ethanol blends are not currently approved for use in
conventional gasoline vehicles and nonroad equipment. Before usage
could be legalized, as discussed more in Section V.D.3 below, EPA would
need to grant a waiver declaring that mid-level blends are
substantially similar or ``sub-sim'' to gasoline or perhaps even
reinterpret the meaning of ``sub-sim''. While such a waiver has not yet
been granted, several organizations/agencies are performing vehicle
emission testing and investigating other impacts of mid-

[[Page 25011]]

level blends.\214\ Therefore, as a sensitivity analysis, we have
analyzed what might need to be done to bring mid-level ethanol blends
to market (should a sub-sim waiver be approved) and the extent to which
such blends could help our nation meet the RFS2 ethanol standards, at
least in the near term. Finally we end our ethanol usage discussion by
looking at other strategies for getting beyond the E10 blend wall.
---------------------------------------------------------------------------

    \214\ For more information on mid-level ethanol blends testing,
refer to Section V.D.3.b.
---------------------------------------------------------------------------

a. Projected Gasoline Energy Demand
    The maximum amount of ethanol our country is capable of consuming
in any given year is a function of the total gasoline energy demanded
by the transportation sector. Our nation's gasoline energy demand is
dependent on the number of gasoline-powered vehicles on the road, their
average fuel economy, vehicle miles traveled (VMT), and driving
patterns. For analysis purposes, we relied on the gasoline energy
projections reported by EIA in AEO 2008.\215\ Unlike AEO 2007, AEO 2008
takes the fuel economy improvements set by EISA into consideration and
also assumes a slight dieselization of the vehicle fleet. The result is
a 15% reduction in the projected 2022 gasoline energy demand from AEO
2007 to AEO 2008.\216\ EIA basically has gasoline energy demand
(petroleum-based gasoline plus ethanol) flattening out, and even
slightly decreasing, as we move into the future and implement the EISA
vehicle standards.\217\
---------------------------------------------------------------------------

    \215\ For blend wall discussions, we rely on the most recent AEO
2009 projections. However for our detailed ethanol consumption
analysis presented in this section (and in more detail in Section
1.7.1 of the DRIA), we relied on AEO 2008.
    \216\ EIA Annual Energy Outlook 2007 & 2008, Table 2.
    \217\ For more information on gasoline energy projections, refer
to Section 1.7.1.2.1 of the DRIA.
---------------------------------------------------------------------------

b. Projected Growth in Flexible Fuel Vehicles
    According to DOE's Department of Energy Efficiency and Renewable
Energy, there are currently over 7 million FFVs on the road today
capable of consuming E85.\218\ And that number is growing steadily.
Automakers are incorporating more and more FFVs into their light-duty
production plans. While the FFV system (i.e., fuel tank, sensor,
delivery system, etc.) used to be an option on some vehicles, most FFV
producers are moving in the direction of converting entire product
lines over to E85-capable systems. Still, the number of FFVs that will
be manufactured and purchased in future years is uncertain. For our
cost analysis, we examined several different FFV production scenarios.
But for our ethanol usage analysis, we focused on one primary FFV
scenario, described in more detail below.\219\
---------------------------------------------------------------------------

    \218\ DOE Energy Efficiency and Renewable Energy August 2008
estimate (worksheet available at www.eere.energy.gov/afdc/data/
index.html).
    \219\ For more on the FFV production scenarios we considered,
refer to Section 1.7.1.2.2 of the DRIA.
---------------------------------------------------------------------------

    In response to President Bush's ``20-in-10'' plan of reducing
American gasoline usage by 20% in 10 years, domestic automakers
responded with aggressive FFV production goals. General Motors, Ford
and Chrysler (referred to hereafter as ``The Detroit 3'') announced
plans to produce 50% FFVs by 2012.\220\ And despite the current state
of the economy and the auto industry, it appears U.S. automakers are
still moving forward with their FFV production plans.\221\ Assuming
that The Detroit 3 continue to maintain 50% market share and that total
vehicle sales remain around 16 million per year, at least 4 million
FFVs will be produced by the 2012 model year. Based on 2008 offerings,
we assumed that approximately 80% of The Detroit 3's FFV production
commitment would be met by light-duty trucks and the remaining 20%
would be cars.222 223 We also assumed that all the FFVs in
existence today were produced by The Detroit 3 (and therefore share the
same aforementioned car/truck ratio) and that production would ramp up
linearly beginning in 2008 to reach the 2012 commitment.
---------------------------------------------------------------------------

    \220\ Ethanol Producer Magazine, ``View From the Hill.'' July 2007.
    \221\ Ethanol Producer Magazine, ``Automakers Maintain FFV
Targets in Bailout Plans.'' February 2009.
    \222\ NEVC 2008 Purchasing Guide for Flexible Fuel Vehicles.
    \223\ Several of the FFV assumptions may need to be revised for
the FRM in light of recent events.
---------------------------------------------------------------------------

    Although non-domestic automakers have not made any official FFV
production commitments, Nissan, Mercedes, Izuzu, and Mazda all included
at least one flexible fuel vehicle in their 2008 model year
offerings.\224\ And we anticipate that additional FFVs (or FFV options)
will be added in the future. Ultimately, we predict that non-domestic
FFV production could be as high as 25%, or about 2 million FFVs per
year. While we are not forecasting an official FFV production
commitment from the non-domestic automakers, we believe that this
represents an aggressive, yet reasonable FFV production estimate for
analysis purposes. Furthermore, based on current offerings, we assumed
that the majority of non-domestic FFV production would be trucks. With
respect to timing, we expect that the non-domestic automakers would
ramp up FFV production later than The Detroit 3. For analysis purposes,
we assumed that non-domestic automakers would ramp up FFV production
beginning in 2013, and like The Detroit 3, it would take about five
years for them to reach their FFV production goals (or in this case,
the assumed 25% production level)
---------------------------------------------------------------------------

    \224\ Ibid.
---------------------------------------------------------------------------

    Based on these FFV assumptions and forecasted vehicle phase-out,
VMT, and fuel economy estimates provided by EPA's MOVES Model, we
calculate that the maximum percentage of fuel (gasoline/ethanol mix)
that could feasibly be consumed by FFVs in 2022 would be about 30%. For
more information on our FFV analysis, refer to Section 1.7.1.2.2 of the DRIA.
c. Projected Growth in E85 Access
    According to the National Ethanol Vehicle Coalition (NEVC), there
are currently over 1,900 retailers offering E85 in 45 states plus the
District of Columbia.\225\ While this represents significant industry
growth, it still only translates to about 1% of U.S. retail stations
nationwide carrying the fuel.\226\ As a result, most FFV owners clearly
do not have reasonable access to E85. For our FFV/E85 analysis, we have
defined ``reasonable access'' as one-in-four pumps offering E85 in a
given area.\227\ Accordingly, just over 4% of the nation currently has
reasonable access to E85, up from 3% in 2007 (based on a mid-year NEVC
E85 pump estimate).\228\
---------------------------------------------------------------------------

    \225\ NEVC FYI Newsletter: Volume 15, Issue 5: March 9, 2009.
    \226\ Based on National Petroleum News gasoline station estimate
of 161,768 in 2008.
    \227\ For a more detailed discussion on how we derived our one-
in-four reasonable access assumption, refer to Section 1.6 of the
DRIA. For the distribution cost implications as well as the cost
impacts of assuming reasonable access is greater than one-in-four
pumps, refer to Section 4.2 of the DRIA.
    \228\ Computed as percent of stations with E85 (1,963/161,768 as
of March 2009 or 1,251/164,292 as of July 2007) divided by 25% (one-
in-four stations).
---------------------------------------------------------------------------

    There are a number of states promoting E85 usage by offering FFV/
E85 awareness programs and/or retail pump incentives. A growing number
of states are also offering infrastructure grants to help expand E85
availability. Currently, nine Midwest states have adopted a progressive
Energy Security and Climate Stewardship Platform.\229\

[[Page 25012]]

The platform includes a Regional Biofuels Promotion Plan with a goal of
making E85 available at one third of all stations by 2025. In addition,
on July 31, 2008, Congresswoman Stephanie Herseth Sandlin (D-SD) and
John Shimkus (R-IL) introduced The E85 and Biodiesel Access Act that
would amend IRS tax code and increase the existing federal income tax
credit from $30,000 or 30% of the total cost of improvements to
$100,000 or 50% of the total cost of needed alternative fuel equipment
and dispensing improvements.\230\ While not signed into law, such a tax
credit could provide a significant retail incentive to expand E85 infrastructure.
---------------------------------------------------------------------------

    \229\ The following states have adopted the plan: Indiana,
Kansas, Michigan, Minnesota, Ohio, South Dakota, Wisconsin, Iowa,
and most recently, North Dakota. For more information, visit: 
http://www.midwesterngovernors.org/resolutions/Platform.pdf. 
    \230\ A copy of House Rule 6734 can be accessed at: http://
www.e85fuel.com/news/2008/080108_shimkus_release/shimkus.pdf. 
---------------------------------------------------------------------------

    Given the growing number of state infrastructure incentives and the
proposed Federal alternative fuel infrastructure subsidy, it is clear
that E85 infrastructure will continue to expand in the future. However,
the extent to which nationwide E85 access will grow is difficult to
predict, let alone quantify. For analysis purposes, as a practical
upper bound, we have selected 70% by 2022. This is roughly equivalent
to all urban areas in the United States offering reasonable (one-in-
four-station) access to E85.\231\ We are not concluding that the
percentage of the nation with reasonable access to E85 could not exceed
70% (as a sensitivity, we also modeled the cost impacts of nationwide
access to E85) or that availability would necessarily be concentrated
in urban areas. However, for analysis purposes, we believe that 70% is
a good surrogate for a practical portion of the country that could have
reasonable one-in-four access to E85 by 2022 under the proposed RFS2
program. On average, this translates to about 18% of retail stations
nationwide offering E85. As discussed in Section V.C, we believe this
is feasible based on our assessment of the distribution infrastructure
capabilities. For more information on the projected growth in E85
access, refer to Section 1.7.1.2.3 of the DRIA.
---------------------------------------------------------------------------

    \231\ For this analysis, we've defined ``urban'' as the top 150
metropolitan statistical areas according to the U.S. census and/or
counties with the highest VMT projections according the EPA MOVES
model, all RFG areas, winter oxy-fuel areas, low-RVP areas, and
other relatively populated cities in the Midwest.
---------------------------------------------------------------------------

d. Required Increase in E85 Refueling Rates
    As mentioned above, there were approximately 7 million FFVs on the
road in 2008. If all FFVs refueled on E85 100% of the time, this would
translate to about 6.5 billion gallons of E85 use.\232\ However, E85
usage was only around 12 million gallons in 2008.\233\ This means that,
on average, FFV owners were only tapping into about 0.2% of their
vehicles' E85/ethanol usage potential last year. Assuming that only 4%
of the nation had reasonable one-in-four access to E85 in 2008 (as
discussed above), this equates to an estimated 5% E85 refueling
frequency for those FFVs that had reasonable access to the fuel.
---------------------------------------------------------------------------

    \232\ Based on the assumption that FFV owners travel
approximately 12,000 miles per year and get about 18 miles per
gallon on average under actual in-use driving conditions. For more
information, refer to Section 1.7.1.2.4 of the DRIA.
    \233\ EIA Annual Energy Outlook 2009, Table 17.
---------------------------------------------------------------------------

    There are several reasons for today's low E85 refueling frequency.
For starters, many FFV owners may not know they are driving a vehicle
that is capable of handling E85. As mentioned earlier, more and more
automakers are starting to produce FFVs by engine/product line, e.g.,
all 2008 Chevy Impalas are FFVs.\234\ Consequently, consumers
(especially brand loyal consumers) may inadvertently buy a flexible
fuel vehicle without making a conscious decision to do so. And without
effective consumer awareness programs in place, these FFV owners may
never think to refuel on E85. In addition, FFV owners with reasonable
access to E85 and knowledge of their vehicle's E85 capabilities may
still not choose to refuel on E85. They may feel inconvenienced by the
increased E85 refueling requirements. Based on its lower energy
density, FFV owners will need to stop to refuel 21% more often when
filling up on E85 over E10 (and likewise, 24% more often when refueling
on E85 over conventional gasoline).\235\ In addition, some FFV owners
may be deterred from refueling on E85 out of fear of reduced vehicle
performance or just plain unfamiliarity with the new motor vehicle
fuel. However, as we move into the future, we believe the biggest
determinant will be price--whether E85 is priced competitively with
gasoline based on its reduced energy density and the fact that you need
to stop more often, drive a little further to find an E85 station, and
depending on the retail configuration, wait in longer lines to fill up
on E85.
---------------------------------------------------------------------------

    \234\ NEVC, ``2008 Purchasing Guide for Flexible Fuel
Vehicles.'' Refers to all mass produced 3.5 and 3.9L Impalas.
However, it is our understanding that consumers may still place
special orders for non-FFVs.
    \235\ Based on our assumption that denatured ethanol has an
average lower heating value of 77,930 BTU/gal and conventional
gasoline (E0) has average lower heating value of 115,000 BTU/gal.
For analysis purposes, E10 was assumed to contain 10 vol% ethanol
and 90 vol% gasoline. Based on EIA's AEO 2008 report, E85 was
assumed to contain 74 vol% ethanol and 26 vol% gasoline on average.
---------------------------------------------------------------------------

    To comply with the proposed RFS2 program and consume 34 billion
gallons of ethanol by 2022, not only would we need more FFVs and more
E85 retailers, we would need to see a significant increase in the
current FFV E85 refueling frequency. Based on the FFV and retail
assumptions described above in subsections (b) and (c), our analysis
suggests that FFV owners with reasonable access to E85 in 2022 would
need to fill up on it 74% of the time, a significant increase from
today's estimated 5% refueling frequency. Were there to be fewer FFVs
in the fleet, the E85 refueling frequency would need to be even higher.
Similarly, with more FFVs in the fleet, the E85 refueling frequency
could be lower and still meet the proposed RFS2 requirements. However,
even with an FFV mandate, our analysis suggests that we would need to
see an increase from today's average FFV E85 refueling frequency. In
order for this to be possible, there will need to be an improvement in
the current E85/gasoline price relationship.
e. Market Pricing of E85 Versus Gasoline
    According to a recent online fuel price survey, E85 is currently
priced almost 30 cents per gallon higher than conventional gasoline on
an energy-equivalent basis.\236\ To increase our nation's E85 refueling
frequency to the levels described above, E85 needs to be priced
competitively with (if not lower than) conventional gasoline based on
its reduced energy content, increased time spent at the pump, and
limited availability. Our analysis, described in more detail in Section
1.7.1.2.5 of the DRIA, suggests that E85 would need to be priced about
one-third lower than gasoline at retail (based on 2006 prices) in order
for it to be cost-competitive. As expected, higher crude prices could
make E85 look slightly more attractive while lower crude oil prices
could make E85 look less attractive.
---------------------------------------------------------------------------

    \236\ Based on average E85 and regular unleaded gasoline prices
reported at http://www.fuelgaugereport.com/  on April 23, 2009.
---------------------------------------------------------------------------

    In Brazil, charts are posted at gas stations informing flex-fuel
vehicle owners whether it makes sense to fill up on ``gasoline''
(containing 20-25% denatured anhydrous ethanol) \237\ or ``alcohol''
(100% denatured hydrous ethanol) based on the price and relative energy
density of each. However, in the U.S., FFV owners will likely be on their

[[Page 25013]]

own for figuring out which fuel is more economical.
---------------------------------------------------------------------------

    \237\ The government-mandated gasoline ethanol content was 25%
as of July 2007. Source: F.O. Licht World Ethanol & Biofuels Report
Vol. 5 No. 21 July 9, 2007.
---------------------------------------------------------------------------

    Although in some areas of the country E85 is already priced
significantly lower than gasoline, this is a far cry from a nationwide
trend. And as we move into the future and incorporate cellulosic
ethanol (a fuel that is currently more expensive to produce than corn
ethanol), it may be even more difficult to produce ethanol for a price
that the market would accept. However, a number of measures could be
taken to help encourage FFV E85 refueling.
    The first is increased consumer awareness. To maximize ethanol
usage, it is important that FFV owners are aware of their vehicle's
fueling capabilities, i.e., that their vehicle is capable of refueling
on E85. It is equally important that FFV owners are aware of E85
refueling outlets that may be available to them. Automakers and/or car
dealerships could notify FFV owners of E85 stations in their area.
Together, increased automaker and retail awareness could help increase
our nation's E85 throughput potential. However, in order for consumers
to actually choose E85 over conventional gasoline on a regular basis,
there needs to be a marked price incentive at the pump.
    Current federal and most state tax code does not differentiate
between ethanol sold as E10 and as E85. As of July 2008, state excise
taxes were reported to account for more than $0.18 per gallon of
gasoline (on average).\238\ However, there are a number of states
(e.g., Illinois, Indiana, North Dakota, and South Dakota) that
currently waive or discount excise taxes on E85. This type of fuel tax
structure helps contribute to a retail price relationship that favors
E85 over conventional gasoline.\239\ If states continue to waive/reduce
E85 fuel taxes under RFS2, this could help increase the FFV E85
refueling frequency. As expected, this would have the greatest impact
on ethanol consumption in the areas of the country with the most FFVs.
---------------------------------------------------------------------------

    \238\ Source: The American Petroleum Institute July 2008
Gasoline Tax Report available at: www.api.org/statistics/
fueltaxes/upload/July_2008_gasoline_and_diesel_summary_pages.pdf.
    \239\ Source: DOE Energy Efficiency and Renewable Energy Web
site (http://www.eere.energy.gov/).
---------------------------------------------------------------------------

    The E10/E85 price relationship could also be modified by the
refining industry. Under the proposed program, gasoline refiners (as
well as importers) would be required to purchase RINs to demonstrate
that sufficient volumes of renewable/alternative fuels were used to
meet their volume obligations. This could provide an incentive for
these parties to take the steps necessary to ensure adequate ethanol
use levels to facilitate compliance. One potential action that refiners
might take to ensure a sufficient RIN supply would be to subsidize the
price of the ethanol used to manufacture E85. Such a subsidy might be
financed by an increase in their selling price of gasoline. In
addition, refiners with marketing arms could adjust the retail price
relationship of E10 in E85 in way that encourages E85 throughput while
still maintaining the same average net profit. However, a relatively
small proportion of refiners market their own gasoline and thus have
the ability to make retail price adjustments. Consequently, relying
solely on market mechanisms may create some competitive concerns. We
request comment on viable and cooperative ways refiners and gasoline
retailers could promote E85 throughput to meet the proposed RFS2 requirements.
3. Other Mechanisms for Getting Beyond the E10 Blend Wall
a. Mandate for FFV Production
    One way to increase ethanol usage under RFS2 would be if there were
more FFVs in the fleet. As described above, our primary analysis is
based on the assumption that The Detroit 3 would follow through with
their commitment to produce 50% FFVs by 2012 and the non-domestic
automakers would ramp up FFV production beginning in 2013 and produce
25% FFVs by 2017. Based on the projected number of FFVs in the fleet
(and our E85 infrastructure growth assumptions), FFV owners with
reasonable one-in-four access to E85 would need to refuel on it 74% of
the time. To achieve this optimistic refueling frequency, we believe
there would need to be significant improvements to the E10/E85 price
relationship.
    One way to reduce the required FFV E85 refueling frequency (and in
turn decrease some of the pressure off E85 prices) would be to further
increase the number of FFVs in the fleet. While EPA does not have the
authority to require automakers to produce FFVs, there are a number of
bills in Congress that are set out to do just that. On July 22, 2008
Senator Sam Brownback (R-KS) on behalf of himself and Senators Susan
Collins (R-ME), Joseph Lieberman (I-CT), Ken Salazar (D-CO), and John
Thune (R-SD) introduced the Open Fuel Standard Act of 2008, a bill that
calls for 50% of the U.S. vehicle fleet to be FFVs capable of using
high blends of ethanol or methanol (in addition to gasoline) by 2012.
This number would grow to 80% by 2015.\240\ A similar FFV bill was
introduced by Eliot Engel (D-NY) in the House on July 22, 2008.\241\
---------------------------------------------------------------------------

    \240\ Refer to Senate Bill 3303 which can be found at:
http://thomas.loc.gov/cgi-bin/query/z?c110:S.3303.
    \241\ Refer to House Rule 6559 which can be found at:
http://thomas.loc.gov/cgi-bin/bdquery/z?d110:H.R.6559.
---------------------------------------------------------------------------

    Since a future congressional mandate on FFV production in being
discussed, we have modeled the impact that such a mandate could have on
the RFS2 program. For our sensitivity analysis, we found that if
automakers were required to make all light-duty vehicles E85-capable by
2015 (and our same E85 infrastructure growth assumptions applied), FFV
owners with reasonable one-in-four access to E85 would only need to
refuel on it 33% of the time. This represents a smaller increase from
today's estimated 5% refueling rate. However, implementing such a FFV
mandate would have significant cost implications on the auto industry
and would still not provide certainty that FFV owners would fuel on
E85. For more information on this analysis, as well as other FFV
production scenarios we considered, refer to Section 1.7.1.2.2 of the DRIA.
b. Waiver of Mid-Level Ethanol Blends (E15/E20)
    For our primary ethanol usage analysis, we considered that there
would only be two fuels in the future, E10 and E85. And as explained in
Section V.D.2, we believe it is feasible to consume 34 billion gallons
of ethanol by 2022 given growth in FFV production and E85 availability
and projected improvements in the current E10/E85 price relationship.
    However, several organizations and government entities are
interested in increasing the concentration of ethanol beyond the
current 10% limit in the commercial gasoline pool. Section 211(f)(1) of
the Clean Air Act prohibits the introduction into commerce, or increase
in the concentration in use of, gasoline or gasoline additives for use
in motor vehicles unless they are substantially similar to the gasoline
or gasoline additives used in the certification of new motor vehicles
or motor vehicle engines. EPA may grant a waiver of this prohibition
under Section 211(f)(4) provided that the fuel or fuel additive ``will
not cause or contribute to a failure of any emission control device or
system (over the useful life of the motor vehicle, motor vehicle
engine, nonroad engine or nonroad vehicle in which the device or system
is used) to achieve compliance by the vehicle or engine with the
emission standards to

[[Page 25014]]

which it has been certified.'' The most recent ``substantially
similar'' interpretive rule for unleaded gasoline presently allows
oxygen content up to 2.7% by weight for certain ethers and
alcohols.\242\ E10 contains approximately 3.5% oxygen by weight, which
makes a gasoline-ethanol blend with ten% ethanol not ``substantially
similar'' to certification fuel under the current interpretation.\243\
Since any mid-level blend would have a greater than allowed oxygen
content, any mid-level blend would need to have a waiver under Section
211(f)(4) of the CAA in order to be sold commercially.
---------------------------------------------------------------------------

    \242\ 73 FR 22277 (April 25, 2008).
    \243\ Gas Plus, Inc. submitted an application for a 211(f)(4)
waiver for E10 which was granted, see 44 FR 20777 (April 6, 1979).
---------------------------------------------------------------------------

    Before EPA grants a 211(f)(4) waiver for a new fuel or fuel
additive, an applicant must prove that the new fuel or fuel additive
will meet the waiver requirements outlined in the statute. EPA has
required that applicants provide vehicle/engine testing for tailpipe
emissions, evaporative emissions, materials compatibility, and
driveability. Testing needs to include emissions over the full useful
life of vehicle and equipment. Several interested parties are
investigating the impact that mid-level ethanol blends (e.g., E15 or
E20) may have on these areas among others (i.e. catalyst, engine, and
fuel system durability, and onboard diagnostics). In order to use the
information collected for waiver application purposes, the mid-level
ethanol blend testing will need to consider the different engines and
fuel systems currently in service that could be exposed to mid-level
ethanol blends and the long-term impact of using such blends.\244\
After receiving a waiver application, EPA must give public notice and
comment and has 270 days to grant or deny the waiver request.
---------------------------------------------------------------------------

    \244\ EPA has expressed what such a waiver testing program might
look like, see Karl Simon, ``Mid Level Ethanol Blend Experimental
Framework: Epa Staff Recommendations,'' June 2008, and Ed Nam
``Vehicle Selection & Sample Size Issues for Catalyst and Evap
Durability Testing,'' November 2008, in the docket (EPA-HQ-OAR-2005-0161).
---------------------------------------------------------------------------

    The Department of Energy (DOE) has developed and initiated a
comprehensive testing program to investigate the potential impacts of
mid-level blends of ethanol. Initial testing was conducted on a limited
number of high-volume vehicles and small non-road engines and a
preliminary report was published in October, 2008.\245\ In addition,
DOE is in the process of leveraging existing EPA vehicle and small
engine test programs (originally designed to test up to 10% ethanol) to
add mid-level ethanol blends to the fuel matrix. DOE's comprehensive
test program is intended to evaluate a wide range of emission,
performance, and durability issues associated with mid-level ethanol
blends (additional reports forthcoming).
---------------------------------------------------------------------------

    \245\ Effects of Intermediate Ethanol Blends on Legacy Vehicles
and Small Non-Road Engines, Report 1, Prepared by Oak Ridge National
Laboratory for the Department of Energy, October 2008.
---------------------------------------------------------------------------

    DOE is not alone in pursuing mid-level blends. In 2005, the State
of Minnesota, a large producer of corn ethanol, passed a law requiring
that by 2015, 20% of gasoline (by volume) must be replaced by ethanol.
While this level could be achieved with a high percentage of E85 usage
by FFVs, the state has also expressed an interest in moving to 20%
ethanol blends. Several other states and organizations have also
expressed interest in increasing ethanol use by adopting E15 or E20.
The Renewable Fuels Association (RFA) and the American Coalition for
Ethanol (ACE) have been working with various government entities to
investigate the impact of mid-level blends
    On March 6, 2009, Growth Energy and 54 ethanol manufacturers
submitted an application for a waiver of the prohibition of the
introduction into commerce of certain fuels and fuel additives set
forth in section 211(f) of the Act. This application seeks a waiver for
ethanol-gasoline blends of up to 15 percent by volume ethanol. The
statute directs the Administrator of EPA to grant or deny this
application within 270 days of receipt by EPA, in this instance
December 1, 2009. EPA recently issued a federal register notice
announcing receipt of the Growth Energy waiver application and
soliciting comment on all aspects of it. Refer to 74 FR 18228 (April
21, 2009).
    While the current Growth Energy waiver application is still under
review, as a sensitivity, we considered the implications that adding
E15 or E20 to the marketplace could have on ethanol usage and the
supporting fuel infrastructure should such blends be permitted. For
each case, we assumed that E10 would need to continue to remain in
existence to meet the demand of legacy vehicle and non-road engine
owners. This would also provide consumer choice. Experience in past
fuel programs has shown that many consumers will not be comfortable
refueling on higher ethanol blends and will blame any problems that may
occur on the new fuel (regardless of the actual cause of the vehicle
problems) causing a backlash against the new fuel requirements.
Therefore, we believe it is critical to continue to allow consumers the
choice between mid-level ethanol blends and conventional gasoline
(assumed to be E10 in the future).
    For our optimistic mid-level ethanol blends scenario, we assumed
that E15 or E20 could be available at all retail stations nationwide by
the time the nation hits the E10 blend wall, or around 2013. This
assumes a number of actions are taken to bring mid-level blends to
market (explained in more detail below).\246\ We assumed that E10 would
be marketed as premium-grade gasoline, the mid-level ethanol blend (E15
or E20) would serve as regular, and like today, midgrade would be
blended from the two fuels. Those vehicles and equipment which are
unable to refuel on mid-level ethanol blends (or choose not to) could
continue to fill up on E10. This mid-level ethanol blends scenario,
described in more detail in Section 1.7.1.3 of the DRIA, concluded that
if mid-level ethanol blends were to be distributed at all retail
stations nationwide, they could help increase ethanol usage to over 19
billion gallons (with E15) and 25 billion gallons (with E20).
---------------------------------------------------------------------------

    \246\ Results for other cases are discussed in Section 1.7.1.3 of the DRIA.

---------------------------------------------------------------------------

[[Page 25015]]
[GRAPHIC] [TIFF OMITTED] TP26MY09.007

    As shown in Figure V.D.2-2, in this optimistic phase-in scenario,
adding E15 could postpone the blend wall by about three years to 2016
and adding E20 could postpone it another three years to 2019. Although
mid-level ethanol blends will fall short of meeting the RFS2
requirements, they could provide interim relief while the county ramps
up E85/FFV infrastructure and/or finds other non-ethanol alternatives
(e.g., cellulosic diesel or biobutanol) to reach the RFS2 volumes.
    Our nation's whole system of gasoline fuel regulation, fuel
production, fuel distribution, and fuel use is built around gasoline
with ethanol concentrations limited to E10. As a result, while a waiver
may legalize the use of mid-level ethanol blends under the CAA, there
are a number of other actions that would have to occur to bring mid-
level blends to retail. The time needed to take these actions could
delay the penetration of mid-level ethanol blends into the market. The
CAA only provides a 1 pound RVP waiver for ethanol blends of 10 volume
percent or less. Lacking such an RVP waiver, a special low-RVP gasoline
blendstock would be needed at terminals to allow the formulation of
mid-level ethanol blends that are complaint with EPA RVP requirements.
Providing such a separate gasoline blendstock would present significant
logistical challenges and costs to the fuel distribution system.\247\ A
number of changes would be needed to EPA regulations including those
pertaining to reformulated gasoline, anti-dumping, and gasoline deposit
control additives to accommodate and mid-level ethanol blends. Such
changes would need to be made through the notice and comment process
similar to today's action. In addition, most states require that fuel
comply with the applicable ASTM International (formally known as the
American Standards for Testing and Materials) specification. The
development of an ASTM International specification for mid-level ethanol
blends through an industry consensus process is currently being initiated.
---------------------------------------------------------------------------

    \247\ It may be possible for refiners to formulate a gasoline
blendstock that would be suitable for manufacturing mid-level
ethanol blends and E10 at the terminal. While this would avoid the
logistical problems associated with maintaining separate
blendstocks, there could be significant additional refining costs.
---------------------------------------------------------------------------

    There are a number of requirements regarding the fire and leak
protection safety of retail fuel dispensing and storage equipment. The
Occupational Safety and Health Administration (OSHA) requires that
retail fuel handling equipment be listed with an independent standards
body such as Underwriters Laboratories (UL). No independent standards
body has listed fuel handling equipment for mid-level ethanol blends.
Furthermore, UL has stated that it would not expand listings for in-use
fuel retail equipment originally listed for E10 blends to cover greater
than E10 blends.\248\ EPA's Office of Underground Storage Tanks (OUST)
requires that UST systems must be compatible with the fuel stored in
the system. These requirements pertain to all components of the system
including the storage tank, connecting piping, pumps, seals and leak
detection equipment.
---------------------------------------------------------------------------

    \248\ UL stated that they have data which indicates that the use
of fuel dispensers certified for up to E10 blends to dispense blends
up to a maximum ethanol content of 15 volume percent would not
result in critical safety concerns (http://www.ul.com/newsroom/
newsrel/nr021909.html ). Based on this, UL stated that it would
support authorities having jurisdiction who decide to permit legacy
equipment originally certified for up to E10 blends to be used to
dispense up to 15 volume percent ethanol. The UL announcement did
address the compatibility of underground storage tank systems with
greater than E10 blends.
---------------------------------------------------------------------------

    States typically adopt fire safety codes from either the National
Fire Protection Association (NFPA) or the International Code Council
(ICC). These organizations currently do not have provisions that would
allow the mid-level ethanol blends to be stored/dispensed from existing
equipment at retail. Local safety officials (e.g. fire marshals)
referred to as ``Authorities Having Jurisdiction'' (AHJ's) often
require a UL certification for fuel retail storage/dispensing equipment
although some will accept

[[Page 25016]]

other substantiation of equipment safety such as a manufacture
certification. Fuel retailers must also satisfy the requirements of the
insurance company that they are insured through which may be more
stringent than the legal requirements. Given the liability concerns
associated with leaks from underground storage tanks, these issues have
to be resolved in order to facilitate the widespread use of mid-level
ethanol blends.
    The Department of Energy and EPA are currently working with
industry to evaluate what changes may be necessary to underground
storage tank systems, fuel dispensers, and refueling vapor recovery
equipment at fuel retail facilities to handle a mid-level ethanol
blend. If existing equipment proves tolerant to a mid-level ethanol
blend, this could substantially facilitate its introduction at retail.
If the data supports the suitability of legacy retail equipment to
store/dispense a mid-level blend, then the process of seeking
acceptance by the standard bodies discussed above could commence. The
normal processes used by these standards bodies can be lengthy. For
example, the NFPA has a 3 year cycle for evaluating changes to its
codes with proposals for the current cycle due this June. Thus, apart
from the need to technically evaluate the suitability of legacy retail
equipment to handle a mid-level ethanol blend, the need to secure
recognition from standards bodies could delay the introduction of a
mid-level ethanol blend at retail should a waiver be granted by EPA.
    If some components of the above-ground existing retail hardware are
found to be incompatible with a mid-level ethanol blend, it may be
possible for them to be replaced through normal attrition. For example
the ``hanging hardware'' which includes the nozzle and hose from the
dispenser is typically replaced every 3 to 5 years. It is also possible
that only minor changes might be needed to equipment that has a longer
service life which might be accomplished without too much difficulty/
cost. However, if extensive new equipment is needed and particularly if
this involves the breaking of concrete, we believe that it is unlikely
that fuel retailer would opt to install equipment specifically for a
mid-level ethanol blend given the projected future need for retail
equipment capable of handling E85.\249\
---------------------------------------------------------------------------

    \249\ As discussed previously, significant penetration of E85 is
projected to be needed to facilitate the use of the volumes of
ethanol we project would be needed to satisfy the requirements of the EISA.
---------------------------------------------------------------------------

    Finally, all vehicles and nonroad equipment currently in use are
only warranted for ethanol levels not exceeding E10 (except for FFVs),
and the owner's manuals are written to reflect this. Before widespread
acceptance of mid-level ethanol blends by consumers can occur, these
warranty issues would need to be addressed.
c. Partial Waiver for Mid-Level Blends
    CAA section 211(f)(4), the waiver provision, states that the
Administrator may grant a fuel waiver if a fuel manufacturer can
demonstrate that the fuel ``will not cause or contribute to a failure
of any emission control device or system (over the useful life of the
motor vehicle, motor vehicle engine, nonroad engine or nonroad vehicle
in which such device or system is used) to achieve compliance by the
vehicle or engine with the emission standards with respect to which it
has been certified.'' For reasons discussed below, it may be possible
that these criteria for a mid-level blend waiver may be met for a
subset of gasoline vehicles or engines but not for all gasoline
vehicles or engines. The waiver criteria are applied over the useful
life of ``the motor vehicle, motor vehicle engine, nonroad engine or
nonroad vehicle in which such device or system is used.'' Assuming the
criteria is met for a certain subset of vehicles, and that adequate
measures could be put in place to ensure that a waiver fuel were only
used in that subset of vehicles or engines, one interpretation of this
provision is that the waiver could apply only to that subset of
vehicles or engines.
    One potential outcome from a review of the entire body of
scientific and technical information available may be an indication
that mid-level ethanol blends could meet the criteria of a section
211(f)(4) waiver for some vehicles and engines but not for others. It
may be that certain vehicles and engines operate as intended using mid-
level blends but others may be more susceptible to emissions increases
or durability problems. For example, vehicles or engines without newer
technology that do not readily adjust for the higher oxygen level in
the fuel may experience problems, while newer technology vehicles such
as those meeting our Tier 2 standards may be able to adjust for such
changes as a result of more advanced emissions and fuel control
equipment. Nonroad engines, which are typically small, are likely to be
most susceptible given the less sophisticated technology associated
with such engines. Given this potential outcome, EPA requests comment
on all aspects, both legal and technical, as to the possibility that a
section 211(f)(4) waiver might be granted, in a partial way with
conditions, such that the use of mid-level blends would be restricted
to a subset of the gasoline vehicles or engines covered by the waiver
provision, while those nonroad engines and vehicles not covered by the
waiver would continue using fuels with blends no greater than E10.
    Any waiver approval, either fully or partially, is likely to elicit
a market response to add E15 blends to E10 and E0 blends in the
marketplace, rather than replace them. Thus consumers would merely have
an additional choice of fuel.
    Experience in past fuel programs has shown that even with consumer
education and fuel implementation efforts, there sometimes continues to
be public concern for new fuel requirements. Several examples include
the phasedown of the amount of lead allowed in gasoline in the 1980s
and the introduction of reformulated gasoline (RFG) in 1995. Some
segments of the public were convinced that the new fuels caused vehicle
problems or decreases in fuel economy. Although substantial test data
proved otherwise, these concerns lingered in some cases for several
years. As a direct result of these experiences, EPA wants to be assured
that prior to potentially granting a waiver for mid-level blends,
sufficient testing has been conducted to demonstrate the compatibility
of a waiver fuel with engine, fuel and emission control system components.
    EPA has previously granted waivers with certain restrictions or
conditions. Among other things, these restrictions have included
requiring fuels to meet certain voluntary consensus-based gasoline
standards such as those developed by the American Society of Testing
and Materials (ASTM standards), requirements that precautions be taken
to prevent using the waiver fuel as a base fuel for adding oxygenates,
and that certain corrosion inhibitors be utilized when producing the
waived fuel.\250\ However, in those waivers, the conditions placed upon
the fuel manufacturer were directly related to manufacturing the fuel
itself. Here, the conditions placed upon the fuel manufacturer would be
on the use of the fuel in certain vehicles or engines. In other words,
the fuel manufacturer would have to ensure that the mid-level blend was
only used in that particular subset of vehicles or engines to be able
to legally manufacture and sell the fuel

[[Page 25017]]

under the terms of the waiver. Since it would become the fuel
manufacturer's responsibility to prevent misfueling, the following
discussion highlights some of the ideas that the fuel manufacturer
could implement, based on particular subsets of vehicles,\251\ to
prevent misfueling.
---------------------------------------------------------------------------

    \250\ See, for example, 53 FR 3636, February 8, 1988, and 53 FR
33846, September 1, 1988.
    \251\ Although it is not possible at this time to know the
contours of a partial waiver with conditions, or even if one might
be appropriate, the remainder of this discussion will refer only to
vehicles covered by the waiver (and not engines) since newer
vehicles are more likely to have more sophisticated emissions and
fuel control equipment, while certain engines might be more affected
for the reasons stated above.
---------------------------------------------------------------------------

    If a partial waiver covered only newly manufactured vehicles,
methods focused on the manufacturing of the vehicle could be utilized
to inform the buyer that the vehicle was capable of operating on the
waiver fuel. In this case, approaches such as the use of vehicle
fueling inlet labels and owner's manuals could be utilized in tandem
with retail station fuel dispenser labels. Such an approach depends on
the attention of the vehicle operator to ensure compliance with the
waiver. Additionally, retail station attendants could be trained to
provide guidance to operators on which vehicles are covered under the waiver.
    If only vehicles of certain model years were covered, owners would
know if they could utilize the mid-level blends simply by knowing the
model year (again, in tandem with pump labeling). Alternatively, if
some portion of the existing fleet, not based upon model-year (such as
vehicles meeting EPA Tier 2 emission standards), would also be covered,
the approach would have to include some means by which the operator of
such a vehicle would be made aware that the vehicle being fueled was
covered or not covered by the waiver. Such an approach would likely
involve notification of owners of covered vehicles, through direct
contact or education campaigns, and would likely require the assistance
of the vehicle manufacturers. This approach, as with other approaches,
would require pump labeling.
    Other approaches may bring about tighter control of misfueling
situations but may present additional challenges. For example, one
approach might be to provide owners of covered vehicles with a
transaction card similar to a credit card that could be swiped at the
dispenser to allow for the dispensing of a waived mid-level blend.
Presumably, software and/or hardware at dispensing pumps may be able to
be adjusted to accommodate such an approach. Some retail station chains
have already utilized transponder mechanisms to record sales. Similar
transponder systems could be utilized in place of transaction cards.
    The above discussion is not meant to be an exhaustive list of
possible approaches for ensuring compliance with a partial waiver, nor
does it explore all the facets of any single approach. EPA recognizes
that there may be legal and practical limitations on what a fuel
manufacturer may be able to do to ensure compliance with the conditions
of the partial waiver. EPA has not previously imposed this type of
``downstream'' condition on the fuel manufacturer as part of a section
211(f)(4) waiver. EPA does, however, have experience with compliance
problems occurring when two types of gasoline have been available at
service stations. Beginning in the mid-1970s with the introduction of
unleaded gasoline and continuing into the 1980s as leaded gasoline was
phased out, there was significant intentional misfueling by consumers.
At the time most service stations had pumps dispensing both leaded and
unleaded gasoline and a price differential as small as a few cents per
gallon was enough to cause some consumers to misfuel. Higher price
differentials could occur if, as expected, mid-level ethanol blends
were to be marketed as the regular grade and E0 or E10 as the premium
grade. The Agency seeks comment regarding whether this is a reasonable
or practical condition for this type of waiver. EPA acknowledges that
the issue of misfueling would be challenging in a situation where a
partial waiver is granted. Therefore, EPA solicits comments on what
measures a fuel manufacturer, EPA or others in the gasoline distribution
network could take for ensuring compliance with a partial waiver.
    While EPA has not analyzed the specific cost of a conditional
waiver, such a waiver would likely carry a cost similar to the costs
described above in Section V.D.3.b. Because existing equipment in
retail stations is certified by Underwriters Laboratories only up to
ten percent ethanol, existing equipment would need to be evaluated for
its acceptability for use with mid-level blends (and deemed to be
acceptable if possible) or it would have to be modified/replaced before
any ethanol blend greater than ten percent could be effectuated in the
marketplace.\252\ If existing retail equipment is found not to be
acceptable for storing/dispensing mid-level blends, the aforementioned
infrastructure challenges would be present and additional costs would
be associated with measures adopted for the prevention of releases due
to material incompatibility, as well as those associated with
misfueling. EPA therefore seeks comment on the compatibility of the
existing retail fuel storage/dispensing equipment with mid-level
ethanol blends. Further, adoption of such a waiver would mean that
fewer vehicles/engines would be able to utilize mid-level blends and,
therefore, the full impact of mid-level blends on the E10 blend wall
under such a scenario would not be as significant as full unrestricted
utilization of such blends.
---------------------------------------------------------------------------

    \252\ See previous discussion in Section V.D.3.b of this
preamble regarding the issues that would need to be addressed to
facilitate the introduction of mid-level ethanol blends at retail.
---------------------------------------------------------------------------

d. Non-Ethanol Cellulosic Biofuel Production
    While our analysis describes possible pathways by which the market
could meet the RFS2 requirements with 34 billion gallons of ethanol as
E10 and E85, our analysis of the required FFV and E85 infrastructure
growth as well as the required changes to the E10/E85 price
relationship suggests some inherent challenges. Furthermore, we
conclude that the introduction of mid-level ethanol blends (contingent
upon waiver approval) would by itself not allow the country to achieve
the RFS2 standards. Another means of achieving the RFS2 volume
requirements would be through the introduction of non-ethanol
cellulosic biofuels. The growing spread in gasoline and diesel pricing
implies that we are currently moving in the direction of being
oversupplied with gasoline and undersupplied with diesel.\253\ As such,
it makes sense that the market might preferentially investigate diesel
fuel replacements, e.g., cellulosic diesel via Fischer-Tropsch
synthesis, pyrolysis, or catalytic depolymerization. These fuels would
meet the definition of cellulosic biofuel (as well as advanced biofuel)
under the proposed RFS2 program and help reduce the ethanol blend wall
impacts associated with this rule. Although for our analysis we assumed
that the cellulosic biofuel standard would be met with ethanol, the
market could choose a significant volume of other non-ethanol renewable
fuels. DOE and other agencies are currently providing grants to support critical

[[Page 25018]]

research into these second-generation cellulosic feedstock conversion
technologies. DOE is also providing loan guarantees to help with the
commercialization of such technologies. For more information on non-ethanol
cellulosic biofuels, refer to Section V.A. or Section 1.4.3 of the DRIA.
---------------------------------------------------------------------------

    \253\ According to EIA, gasoline and diesel prices were pretty
similar on average for a decade from 1995-2004. However, over the
past four years, diesel prices have begun to track consistently
higher than gasoline prices. To date in 2008, diesel has been priced
more than $0.50/gallon higher than gasoline on average. Source:
http://tonto.eia.doe.gov/oog/info/gdu/gasdiesel.asp.
---------------------------------------------------------------------------

e. Measurement Tolerance For E10
    Some stakeholders have suggested that the implementation of a
tolerance in the measurement of the ethanol content of gasoline could
allow more ethanol to be used in existing vehicles without the need for
a formal waiver and without the need for more FFVs. Such a tolerance
could allow ethanol contents slightly higher than 10 volume percent
while still treating such blends as meeting the 10 volume percent
limitation on the ethanol content of gasoline.
    Although there is no explicit written precedent for permitting
ethanol contents higher than 10 vol%, some have speculated that current
vehicles would not exhibit any noticeable change in performance,
durability, or emissions if a small measurement tolerance for ethanol
content of gasoline were allowed. The current specified test method for
oxygen content ASTM D-5599-00 includes estimates of the measurement
reproducibility that could be used to inform the determination of an
appropriate tolerance for ethanol content in gasoline. For instance,
based on the provided reproducibility, a measurement as high as 11 vol%
ethanol in gasoline might be possible for gasoline that was blended to
meet a 10 vol% ethanol requirement. Historically, however, EPA has
always enforced the 10 vol% waiver at the 10 vol% level without any
tolerance.
    The 1978 gasohol waiver application requested a blend of 90%
unleaded gasoline and 10% anhydrous ethanol. Although not specified in
the application, the convention and the practical approach for blending
ethanol into gasoline in 1978 was by volume, and it has continued to be
by volume. Thus, the limit on ethanol in gasoline under the waiver is
10% by volume. This is approximately 3.5% oxygen by weight. The waiver
request did not apply to a level of ethanol in gasoline beyond 10%, and
since the application was approved by default after 180 days due to the
fact that the Administrator did not make an explicit decision in this
timeframe, there is no formal approval that could have indicated what
measurement tolerances might have been acceptable. Thus it has
historically been enforced at the 10 vol% limit without any enforcement
tolerance. However, parties who have raised this option have suggested
that the Agency's previous treatment of the oxygenate content of
gasoline may provide a precedent that would allow for a higher
measurement tolerance for ethanol content.
    Prior to and after 1981, several waivers issued by the Agency
allowed the use of various alcohols and ethers in unleaded gasoline. In
1981, the ``substantially similar'' interpretive rule for unleaded
gasoline allowed certain alcohols and ethers at up to 2.0% oxygen by
weight. In 1991 the limit was increased to 2.7% oxygen by weight. For
each of these waivers, the unleaded gasoline base to which the
oxygenate was to be added was to be initially free of oxygenate. With
the exception of ethanol, oxygenates, mostly MTBE, were blended at the
refinery, with the refiner in control of the gasoline used for
blending. This enabled the refiner to ensure that it was free of
oxygenate prior to blending. Ethanol was primarily blended at
terminals. In order to ensure that gasoline blended with ethanol at the
terminal was free of other oxygenates, the ethanol blender first had to
check for the presence of other oxygenates in the base gasoline. In the
mid-1980's ethanol blenders informed EPA that they were having
difficulty finding oxygenate-free gasoline. Much of gasoline had at
least trace amounts of MTBE due to commingling of gasolines with
different oxygenates in the fungible pipeline system. In order to
continue to allow the blending of ethanol up to the 10 vol% limit, EPA
issued a letter stating that it would not consider it to be a violation
of the ethanol sub-sim waiver if up to 10% by volume ethanol were added
to unleaded gasoline containing no more than 2% by volume MTBE.
However, the MTBE must have been present only as a result of
commingling during storage or transport and not purposefully added as
an additional component to the ethanol blend.
    Subsequently, two other statements by EPA provided guidance on the
allowable oxygen content of oxygenated fuels. For instance, in a
memorandum dated October 5, 1992, EPA provided interim guidance for
states that allowed averaging programs.\254\ This guidance allowed the
oxygen content of ethanol to be as high as 3.8% by weight, but did not
indicate that the ethanol concentration could be higher than 10 vol%.
Also, in a 1995 RFG/Anti-dumping Q&A it was noted that the maximum
oxygen range for the simple and complex models was 4.0% by weight. This
range was implemented to once again continue to allow the blending of
ethanol up to the 10 vol% limit in cases where an extremely low
gasoline density might increase the calculated weight percent oxygen
content for E10 above the more typical 3.5-3.7 wt% range.
---------------------------------------------------------------------------

    \254\ Memorandum from Mary T. Smith, Director of the Field
Operations and Support Division, to State/Local Oxygenated Fuels
Contacts, October 5, 1992. Subject: ``Testing Tolerance''.
---------------------------------------------------------------------------

    Although we acknowledge that the currently specified test method
ASTM D-5599-00 includes some variability, ethanol is different than
many other fuel properties and components that are controlled in other
fuel programs in one important respect. Fuel properties such as RVP,
and components such as sulfur and benzene, are natural characteristics
of gasoline as a result of the chemical nature of crude oil and the
refining process. Their level or concentration in gasoline is unknown
until measured, and then is dependent upon accuracy of the test method.
In contrast, ethanol is intentionally added in known amounts using
equipment designed to ensure a specific concentration within a small
fraction of one percent. Parties that blend ethanol into gasoline
therefore have precise control over the final concentration. Thus, a
measurement tolerance for ethanol would be less appropriate than
measurement tolerances for other fuel properties and components.
    We request comment on whether a measurement tolerance should be
allowed for the ethanol content of gasoline, the basis for such a
tolerance, and what tolerance if any would be appropriate. We also
request comment on whether such a tolerance would fit within the
existing Underwriters Laboratories, Inc. (UL) approval for the safety
of equipment at refueling stations, including underground storage
tanks, pumps, piping, seals, etc.
f. Redefining ``Substantially Similar'' to Allow Mid-Level Ethanol Blends
    Section 211(f)(1) prohibits the introduction into commerce, or
increase in the concentration in use of, gasoline or gasoline additives
for use in motor vehicles unless they are substantially similar to the
gasoline or gasoline additives used in the certification of new motor
vehicles or motor vehicle engines. EPA may grant a waiver of this
prohibition under section 211(f)(4) of the Clean Air Act provided that
the fuel or fuel additive ``will not cause or contribute to a failure
of any emission control device or system (over the useful life of the
motor vehicle, motor vehicle engine, nonroad engine or nonroad vehicle
in which the device or system

[[Page 25019]]

is used) to achieve compliance by the vehicle or engine with the
emission standards to which it has been certified.''
    EPA first interpreted the term ``substantially similar'' for
unleaded gasoline and its additives in 1978.\255\ Recognizing that this
interpretation was too limited, EPA updated it in 1980, and again in
1981.\256\ EPA set the limits contained in the interpretation based on
the physical and chemical similarities of the fuel or fuel additives to
those used in the motor vehicle certification process. EPA also
considered information available regarding the emission effects that
such fuels and additives would exhibit relative to the emissions
performance of the certification fuels and fuel additives. The 1981
interpretative rule identified the characteristics and specifications
that EPA determined would make a fuel or fuel additive ``substantially
similar'' to those used in certification. Under this rule, a fuel or
fuel additive would be considered substantially similar if it satisfied
certain limits on fuel and fuel additive composition, did not exceed a
maximum allowable oxygen content of fuel at 2.0% by weight, and met
certain ASTM specifications. Comments on this interpretative rule
requested that EPA increase the maximum oxygen concentration up to 3.5%
oxygen by weight, but EPA rejected this recommendation, stating that it
would keep the limit at 2.0% because of concerns over emissions,
material compatibility, and drivability from use of various alcohols at
higher oxygen contents.
---------------------------------------------------------------------------

    \255\ 43 FR 11258 (March 17, 1978), 43 FR 24131 (June 2, 1978).
    \256\ 45 FR 67443 (October 10, 1980), 46 FR 38582 (July 28, 1981).
---------------------------------------------------------------------------

    In 1991, EPA amended the interpretive rule by revising the oxygen
content criteria to allow fuels containing aliphatic ethers and/or
alcohols (excluding methanol) to contain up to 2.7% by weight
oxygen.\257\ EPA based this increase in the oxygen content on its
review of information on a wide variety of alcohol and ether blends,
leading it to determine that ``unleaded gasolines with such oxygen
content are chemically and physically substantially similar to, and
have been shown to have emissions properties substantially similar to,
unleaded gasolines used in light-duty vehicle certification.'' \258\
Finally, in 2008, EPA amended the interpretive rule to allow
flexibility for the vapor/liquid ratio specification for fuel
introduced into commerce in the state of Alaska to improve cold
starting for vehicles during the winter months in Alaska.\259\ Thus the
``substantially similar'' interpretive rule for unleaded gasoline
presently allows oxygen content up to 2.7% by weight for certain ethers
and alcohols.
---------------------------------------------------------------------------

    \257\ 56 FR 5352 (February 11, 1991).
    \258\ 56 FR at 5353.
    \259\ 73 FR 22277 (April 25, 2008).
---------------------------------------------------------------------------

    A waiver of the substantially similar prohibition was provided by
operation of law in 1979 under CAA section 211(f)(4), allowing a
gasoline-alcohol fuel blend with up to 10% ethanol by volume (E10)
(``E10 Waiver''). E10 has an oxygen content which typically ranges
between 3.5 and 3.7% by weight, depending on the specific gravity of
the gasoline. Any ethanol blends with greater than 10% ethanol by
volume would have an oxygen content which exceeds the 2.7% by weight
allowed under the current interpretation of ``substantially similar.''
Therefore, under the 1991 interpretive rule, mid-level ethanol blends
would not be considered substantially similar and would require a CAA
section 211(f)(4) waiver.
    It has been suggested to EPA that we should update the interpretive
rule such that mid-level ethanol blends would be considered
substantially similar. As in the past, this would involve consideration
of the physical and chemical similarities of such mid-level blends to
fuels used in the certification process, as well as information about
the expected emissions effects of such mid-level blends.\260\ EPA
invites comment on whether mid-level blends of ethanol are physically
and chemically similar enough to the fuels used in the motor vehicle
certification process such that they could be considered
``substantially similar'' to the certification fuels used by EPA. With
respect to the emissions effects of mid-level blends on emissions
performance, EPA recognizes that there may be different impacts
depending on the kind of motor vehicle involved. For example, it has
been suggested that older technology motor vehicles and engines may
have emissions and durability impacts from ethanol blends higher than
10 percent, while Tier 2 and later technology vehicles--2004 and later
model year vehicles--may have fewer such impacts.\261\ These more
recent technology vehicles represent an ever growing proportion of the
in-use fleet. DOE is currently conducting various test programs to
ascertain the impacts of higher level ethanol blends on vehicles and equipment.
---------------------------------------------------------------------------

    \260\ One point to be clear on is that the substantially similar
provision relates to fuels used in certification. It is not an issue
of whether mid-level blends are substantially similar to a fuel that
has received a waiver of this prohibition. See 46 FR 38582, 38583
(July 28, 1981). The fuels used in certification include the test
fuels used for exhaust testing, test fuels for evaporative emissions
testing, and the fuels used in the durability process.
    \261\ It has also been suggested that nonroad engines and
equipment may experience greater emissions effects and durability
problems when using mid-level blends.
---------------------------------------------------------------------------

    EPA seeks comment on all of the issues involved with reconsidering
its interpretation of the term ``substantially similar'' to include
gasoline blended with ethanol to contain up to 4.5% oxygen by weight.
If EPA revised the substantially similar interpretation in this manner,
gasoline blended with up to 12% ethanol by volume (E12) would be
considered ``substantially similar.'' \262\ Given the possibility,
based upon engineering judgment, of a varying impact of a mid-level
ethanol blends on different technology vehicles, EPA invites comment on
limiting such an interpretation to gasoline intended for use in Tier 2
and later motor vehicles. We estimate that defining E12 as
``substantially similar'' for Tier 2 and later motor vehicles could
delay the saturation of the gasoline market with ethanol for up to a
year, allowing for more comprehensive testing on higher blend levels to
be carried out. However, before EPA could determine whether it was
appropriate to revise the interpretation of ``substantially similar''
for gasoline to include gasoline-alcohol fuels blended with up to 12%
ethanol, information would need to be provided to EPA that would allow
for a robust assessment of the impact of E12 over the full useful life
of Tier 2 and later motor vehicles addressing emissions (both tailpipe
and evaporative emissions), materials compatibility, and drivability.
Furthermore, E12 would still need to fulfill registration requirements
(i.e. speciation and health effects testing found at 40 CFR 79.52 and
40 CFR 79.53).
---------------------------------------------------------------------------

    \262\ As mentioned earlier, EPA has typically used the oxygen
weight percent convention when interpreting the ``substantially
similar'' provision. A change in the ``substantially similar''
interpretation to allow for up to 4.5% oxygen by weight in the form
of ethanol would essentially accommodate ethanol blends up to 12% by
volume since the vast majority of gasolines blended at 12% by volume
ethanol would not exceed this oxygen weight percent limit.
---------------------------------------------------------------------------

    EPA also seeks comments on additional regulatory and implementation
issues that would arise as a result of changing the ``substantially
similar'' definition to allow for E12. These issues as identified for
mid-level blends in the discussion in Section V.D.3.b include, but are
not necessarily limited to, the applicability of the 1.0 psi RVP waiver
with regard to 10% ethanol blends found at 40 CFR

[[Page 25020]]

80.27(d), Clean Air Act section 211(h); the accommodation of ethanol
blends in making calculations utilizing the complex model for
reformulated and conventional gasoline at 40 CFR 80.45; and detergent
certification requirements found at 40 CFR 80 (Subpart G). Emissions
speciation and health effects testing is required for oxygenate-
specific blends under 40 CFR 79 (Subpart F). Such testing is currently
underway for 10% ethanol blends but not for ethanol levels higher than
10 percent. Additionally, if E12 was allowed under the ``substantially
similar'' definition, presumably such a blend would have to meet one of
the volatility classes of ASTM D4814-88, which is not now the case with
some blends of 10% ethanol blended under the E10 Waiver. Any change in
the allowable maximum ethanol level in motor fuels will impact these
and, potentially, other motor fuel regulations.
    Furthermore, there are also implications beyond EPA's motor fuel
regulations. Existing equipment in retail stations is certified by
Underwriters Laboratories only up to 10% ethanol. Thus, either existing
equipment would need to be recertified for E12 (if possible) or it
would have to be replaced before E12 could be effectuated in the
marketplace. In addition, the substantially similar prohibition applies
to the fuel manufacturer, and if the reinterpretation only applied to
gasoline used with Tier 2 and later motor vehicles, then the
manufacturer of a mid-level blend could not introduce it into commerce
for use with any other motor vehicles. This means that the fuel
distribution system would need to be structured in such a way that the
fuel manufacturer could appropriately ensure that the fuel was only
used in Tier 2 or later motor vehicles. Preventing the misfueling of
mid-level blends into vehicles and engines not specified in the
interpretive rule, and ensuring the availability of fuels for other
vehicles and engines, poses a major problem with reinterpreting
``substantially similar'' to include mid-level blends with a
restriction for use in Tier 2 and later motor vehicles. (For a more
detailed discussion on this issue, see Section V.D.3.c above). We seek
comment on these logistical and regulatory concerns as well.

VI. Impacts of the Program on Greenhouse Gas Emissions

A. Introduction

    Lifecycle modeling, often referred to as fuel cycle or well-to-
wheel analysis, assesses the net impacts of a fuel throughout each
stage of its production and use including production/extraction of the
feedstock, feedstock transportation, fuel production, fuel
transportation and distribution, and tailpipe emissions.\263\ This
section describes and seeks comment on the methodology developed by EPA
to determine the lifecycle greenhouse gas (GHG) emissions of biofuels
fuels as required by EISA as well as the petroleum-based transportation
fuels being replaced. While much of the discussion below focuses on
those portions of lifecycle assessment particularly important to
biofuel production, the basic methodology was the same for analyzing
both petroleum-based fuels and biofuels. This methodology was utilized
to determine which biofuels (both domestic and imported) qualify for
the four different GHG reduction thresholds established in EISA. This
threshold assessment compares the lifecycle emissions of a particular
biofuel including its production pathway against the lifecycle
emissions of the petroleum-based fuel it is replacing (e.g., ethanol
replacing gasoline or biodiesel replacing diesel). This section also
seeks comment on the Agency's proposal to utilize the discretion
provided in EISA to adjust these thresholds downward should certain
conditions be met. We also explain how feedstocks and fuel types not
included in our analysis will be addressed and incorporated in the
future. The overall GHG benefits of the RFS program, which are based on
the same methodology presented here, are provided in Section VI.F.
---------------------------------------------------------------------------

    \263\ In this preamble, we are considering ``lifecycle
analysis'' in the context of estimating GHG emissions, as required
by EISA. More generally, the term ``lifecycle analysis'' or
``assessment'' has been defined as an evaluation of all the
environmental impacts across the range of media/exposure pathways
that are associated with a ``cradle to grave'' view of a product or
set of policies. For more information on this broader context,
please see the 2006 EPA publication ``Life Cycle Assessment:
Principles and Practice (EPA/600/R-06/060).
---------------------------------------------------------------------------

    As described in detail below, EPA has analyzed the lifecycle GHG
impacts of the range of biofuels currently expected to contribute
significantly to meeting the volume mandates of EISA through 2022. In
these analyses we have used the best science available. Our analysis
relies on peer reviewed models and the best estimate of important
trends in agricultural practices and fuel production technologies as
these may impact our prediction of individual biofuel GHG performance
through 2022. We have identified and highlighted assumptions and model
inputs that particularly influence our assessment and seek comment on
these assumptions, the models we have used and our overall methodology
so as to assure the most robust assessment of lifecycle GHG performance
for the final rule.
    EPA believes that compliance with the EISA mandate--determining the
aggregate GHG emissions related to the full fuel lifecycle, including
both direct emissions and significant indirect emissions such as land
use changes--makes it necessary to assess those direct and indirect
impacts that occur not just within the United States and also those
that occur in other countries. This applies to determining the
lifecycle emissions for petroleum-based fuels, to determine the
baseline, as well as the lifecycle emissions for biofuels. For
biofuels, this includes evaluating significant emissions from indirect
land use changes that occur in other countries as a result of the
increased production and importation of biofuels in the U.S. As
detailed below, we have included the GHG emission impacts of
international indirect land use changes. We recognize the significance
of including international land use emissions impact and in our
analysis presentation we have been transparent in breaking out the
various sources of GHG emissions so that the reader can readily see the
impact of including international land use impacts.
    In addition to the many technical issues addressed in this
proposal, this section also discusses the emissions decreases and
increases associated with the different parts of the lifecycle
emissions of various biofuels, and the timeframes in which these
emissions changes occur. Determining a single lifecycle value that best
represents this combination of emissions increases and decreases
occurring over time led EPA to consider various alternative ways to
analyze the timeframe of emissions related to biofuel production and
use as well as options for adjusting or discounting these emissions to
determine their net present value. Several variations of time period
and discount rate are discussed. The analytical time horizon and the
choice whether to discount GHG emissions and, if so, at what
appropriate rate can have a significant impact on the final assessment
of the lifecycle GHG emissions impacts of individual biofuels as well
as the overall GHG impacts of these EISA provisions and this rule.
    We believe that our lifecycle analysis is based on the best
available science, and recognize that in some aspects it represents a
cutting edge approach to addressing lifecycle GHG emissions. Because of
this, varying degrees of uncertainty are in our analysis. For this
proposal, we conducted a number of

[[Page 25021]]

sensitivity analyses which focus on key parameters and demonstrate how
our assessments might change under alternative assumptions. By focusing
attention on these key parameters, the comments we receive as well as
additional investigation and analysis by EPA will allow narrowing of
uncertainty concerns for the final rule. In addition to this
sensitivity analysis approach, we will also explore options for more
formal uncertainty analyses for the final rule to the extent possible.
    Because lifecycle analysis is a new part of the RFS program, in
addition to the formal comment period on the proposed rule, EPA is
making multiple efforts to solicit public and expert feedback on our
proposed approach. As discussed in Section XI, EPA plans to hold a
public workshop during the comment period focused specifically on our
lifecycle analysis to help ensure full understanding of the analyses
conducted, the issues addressed and options that should be considered.
We expect that this workshop will help ensure that we receive the most
thoughtful and useful comments to this proposal and that the best
methodology and assumptions are used for calculating GHG emissions
impacts of fuels for the final rule. Additionally we will conduct peer-
reviews of key components of our analysis. As explained in more detail
in the following sections, EPA is specifically seeking peer review of:
Our use of satellite data to project future land use changes; the land
conversion GHG emissions factors estimated by Winrock; our estimates of
GHG emissions from foreign crop production; methods to account for the
variable timing of GHG emissions; and how models are used together to
provide overall lifecycle GHG estimates.
    The regulatory purpose of the lifecycle greenhouse gas emissions
analysis is to determine whether renewable fuels meet the GHG
thresholds for the different categories of renewable fuel.
1. Definition of Lifecycle GHG Emissions
    The GHG provisions in EISA are notable for the GHG thresholds
mandated for each category of renewable fuel and also the mandated
lifecycle approach to those thresholds. Renewable fuel must, unless
``grandfathered'' as discussed in Section II.B.3., achieve at least 20%
reduction in lifecycle greenhouse gas emissions compared to the average
lifecycle greenhouse gas emissions for gasoline or diesel sold or
distributed as transportation fuel in 2005. Similarly, biomass-based
diesel and advanced biofuels must achieve a 50% reduction, and
cellulosic biofuels a 60% reduction, unless these thresholds are
adjusted according to the provisions in EISA. To EPA's knowledge, the
GHG reduction thresholds presented in EISA are the first lifecycle GHG
performance requirements included in federal law. These thresholds, in
combination with the renewable fuel volume mandates, are designed to
ensure significant GHG emission reductions from the use of renewable
fuels and encourage the use of GHG-reducing renewable fuels.
    The definition of lifecycle greenhouse gas emissions established by
Congress is also critical. Congress specified that:

    The term `lifecycle greenhouse gas emissions' means the
aggregate quantity of greenhouse gas emissions (including direct
emissions and significant indirect emissions such as significant
emissions from land use changes), as determined by the
Administrator, related to the full fuel lifecycle, including all
stages of fuel and feedstock production and distribution, from
feedstock generation or extraction through the distribution and
delivery and use of the finished fuel to the ultimate consumer,
where the mass values for all greenhouse gases are adjusted to
account for their relative global warming potential.\264\
---------------------------------------------------------------------------

    \264\ Clean Air Act Section 211(o)(1).

    This definition requires EPA to look broadly at lifecycle analyses
and to develop a methodology that accounts for all the important
factors that may significantly influence this assessment, including the
secondary or indirect impacts of expanded biofuels use. EPA's analysis
described below indicates that the assessment of lifecycle GHG
emissions for biofuels is significantly affected by the secondary
agricultural sector GHG impacts from increased biofuel feedstock
production (e.g., changes in livestock emissions due to changes in
agricultural commodity prices) and also by the international impact of
land use change from increased biofuel feedstock production. Thus,
these factors must be appropriately incorporated into EPA's lifecycle
methodology to properly assess full lifecycle GHG performance of
biofuels in accordance with the EISA definition.
2. History and Evolution of GHG Lifecycle Analysis
    Traditionally, the GHG lifecycle analysis of fuels has involved
calculating the emissions associated with each individual stage in the
production and use of the fuel (e.g., growing or extracting the
feedstock, moving the feedstock to the processing plant, processing the
feedstock into fuel, moving the fuel to market, and combusting the
fuel.) EPA used this approach for the lifecycle modeling conducted for
the RFS1 program in 2005. However, it has become increasingly apparent
that this type of first order or attributional lifecycle modeling has
notable shortcomings, especially when evaluating the implications of
biofuel policies.\265\ In fact, the main criticism EPA received in
reaction to our previous RFS1 lifecycle analysis was that we did not
include important secondary, indirect, or consequential impacts of
biofuel production and use.
---------------------------------------------------------------------------

    \265\ See also, Conceptual and Methodological Issues in
Lifecycle Analysis of Transportation Fuels, Mark A. Delucchi,
Institute of Transportation Studies, University of California,
Davis, 2004, UCD-ITS-RR-04-45 for a description of issues with
traditional lifecycle analysis used to model GHG impacts of biofuels
and biofuel policies.
---------------------------------------------------------------------------

    Several studies and analyses conducted since the completion of RFS1
have contributed to our understanding of the lifecycle GHG emissions of
biofuel production. These studies, and others, have highlighted the
potential impacts of biofuel production on the agricultural sector and
have specifically identified land use change impacts as an important
consideration when determining GHG impacts of
biofuels.266 267 In the meantime, the dramatic increase in
U.S. production of biofuels has heightened the concern about the
impacts biofuels might have on land use and has increased the
importance of considering these indirect impacts in lifecycle analysis.
---------------------------------------------------------------------------

    \266\ Fargione, J., J. Hill, D. Tilman, S. Polasky, and P.
Hawthorne. 2008. Land clearing and the biofuel carbon debt. Science
319:1235-1238. See http://www.sciencemag.org/cgi/reprint/319/5867/
1235.pdf. 
    \267\ Searchinger, T., R. Heimlich, R.A. Houghton, F. Dong, A.
Elobeid, J. Fabiosa, S. Tokgoz, D. Hayes, and T.-H. Yu. 2008. Use of
U.S. croplands for biofuels increases greenhouse gases through
emissions from land-use change. Science 319:1238-1240. See 
http://www.sciencemag.org/cgi/reprint/319/5867/1238.pdf. 
---------------------------------------------------------------------------

    Based on the evolution of lifecycle analysis and the new
requirements of EISA, we have developed a comprehensive methodology for
estimating the lifecycle GHG emissions associated with renewable fuels.
Through dozens of meetings with a wide range of experts and
stakeholders, EPA has shared and sought input on this methodology. We
also have relied on the expertise of the U.S. Department of Agriculture
(USDA) and the Department of Energy (DOE) to help inform many of the
key assumptions and modeling inputs for this analysis. Dialogue with
the State of California and the European Union on their parallel, on-
going efforts in GHG

[[Page 25022]]

lifecycle analysis has also helped inform EPA's methodology. As part of
this discussion, we have identified several of the key drivers
associated with these lifecycle GHG emissions estimates, including
assumptions about international land use change and the timing of GHG
emissions over time. The inputs we have received through these
interactions are reflected throughout this section.
    Specifically EPA has worked closely with the California Air
Resources Board (CARB) regarding their development of transportation
fuels lifecycle GHG impacts. California Executive Order S-1-07, the Low
Carbon Fuel Standard (LCFS) (issued on January 18, 2007), calls for a
reduction of at least 10 percent in the carbon intensity of
California's transportation fuels by 2020. CARB has worked to develop
lifecycle GHG impacts of different fuels for this Executive Order
rulemaking. More information about this rulemaking and the lifecycle
analysis conducted by California can be found at http://www.arb.ca.gov/
fuels/lcfs/lcfs.htm. EPA will continue to coordinate with California on
this rulemaking and the biofuels lifecycle GHG analysis work.
    Because this lifecycle GHG emissions analysis is complex and
requires the use of sophisticated computer models, we have taken
several steps to increase the transparency associated with our
analysis. For example, we have updated the model documentation for the
Forest and Agricultural Sector Optimization Model (FASOM), which is
included in the docket. In addition, we have highlighted key
assumptions in FASOM and the Food and Agricultural Policy Research
Institute (FAPRI) models that impact the results of our analysis.
Finally, this NPRM provides an important opportunity for the Agency to
present our work and to receive input from stakeholders and experts in
this field. We will also continue to refine our analysis between the
proposed and final rules, and we will add or update information to the
docket as it becomes available.

B. Methodology

    This section describes EPA's methodology for assessing the
lifecycle GHG emissions associated with each biofuel evaluated as well
as the petroleum-based gasoline and diesel fuel these biofuels would
replace. Whereas lifecycle GHG emission methodologies have been well
studied and established for petroleum-based gasoline and diesel fuel,
much of EPA's work has focused on newly developing lifecycle
methodologies for biofuels. Therefore, much of the following section
describes the biofuels-related methodologies and identifies important
issues for comment. Assessing the complete lifecycle GHG impact for
each individual biofuel mandated by EISA requires that a number of key
methodological issues be addressed--from the choice of a baseline to
the selection of the most credible technique for predicting
international land use conversion due to the increase in U.S. renewable
fuels demand, to accounting for the time dimension of changes in GHG
emissions. In this section, we first describe the scenarios we have
analyzed for this proposal. Second, we discuss the scope of our
analysis and what is included in our estimates. Third, we provide
details on the tools and models we used to quantify the GHG emissions
associated with the different fuels. Fourth, we discuss the
uncertainties associated with lifecycle analysis and how we have
addressed them. Fifth, we describe the different components of the
lifecycle that we have analyzed and the key questions we have addressed
in this analysis.
1. Scenario Description
    To quantify the lifecycle GHG emissions associated with the
increase in renewable fuel mandated by EISA, we compared the
differences in total GHG emissions between two future scenarios. The
first assumed a ``business as usual'' volume of a particular renewable
fuel based on what would likely be in the fuel pool in 2022 without
EISA, as predicted by the Energy Information Agency's Annual Energy
Outlook (AEO) for 2007 (which took into account the economic and policy
factors in existence in 2007 before EISA). The second assumed the
higher volume of renewable fuels as mandated by EISA for 2022. For each
individual biofuel, we analyzed the incremental GHG emission impacts of
increasing the volume of that fuel to the total mix of biofuels needed
to meet the EISA requirements. Rather than focus on the impacts
associated with a specific gallon of fuel and tracking inputs and
outputs across different lifecycle stages, we determined the overall
aggregate impacts across sections of the economy in response to a given
volume change in the amount of biofuel produced.\268\
---------------------------------------------------------------------------

    \268\ We then normalize those impacts for each gallon of fuel
(or Btu) by dividing total impacts over the given volume change.
---------------------------------------------------------------------------

    This analysis is not a comparison of biofuel produced today versus
biofuel produced in the future. Instead, it is a comparison of two
future scenarios. Any projected changes in factors such as crop yields,
energy costs, or production plant efficiencies, both domestically and
internationally, are reflected in both scenarios. We focused our
analyses on 2022 results for three reasons. First, it would require an
extremely complex assessment and administratively difficult
implementation program to track how biofuel production might
continuously change from month to month or year to year. Instead, it
seems appropriate that each biofuel be assessed a level of GHG
performance that is constant over the implementation of this rule,
allowing fuel providers to anticipate how these GHG performance
assessments should affect their production plans. Second, it is
appropriate to focus on 2022, the final year of ramp up in the required
volumes of renewable fuel as this year. Assessment in this year allows
the complete fuel volumes specified in EISA to be incorporated. Third,
since the GHG assessment compares performance between a business as
usual case and the mandated volumes case, many of the factors that
change over time such as crop yield per acre are reflected in both
cases. Therefore the differences in these parallel assessments are
unlikely to vary significantly over time.
    EPA requests comment on its proposal to adopt fixed assessments of
fuels meeting the GHG thresholds based on a 2022 performance
assessment. Additional information on the scenarios modeled and the
supplemental analyses that will be conducted for the final rule is
included in Chapter 2 of the DRIA.
    In the existing Renewable Fuel Standard rules adopted in response
to the Energy Policy Act of 2005, biofuels and RINs associated with
them are not based on regional differences of where the feedstock was
grown or the biofuel was produced. In effect, the RINs apply to a
national average of the fuel type. Similarly, this proposal does not
distinguish biofuel on the basis of where within the country the
biofuel feedstock was grown or the biofuel produced. Thus, for example,
ethanol produced from corn starch using the same production technology
will receive the same GHG lifecycle assessment regardless of where the
corn was grown or at what facility the biofuel was produced. There are
regional differences in soil types, weather conditions, and other
factors which could affect, for example, the amount of fertilizer
applied and thus the GHG impact of corn production. Such factors could
vary somewhat across a region, within a state and even within a county.
The agricultural models used to conduct this analysis do distinguish
crop production

[[Page 25023]]

by region domestically and by country internationally. However, biofuel
feedstocks such as corn or soybean oil are well traded commodities
including internationally. So, for example, if corn in a certain
location in Iowa is used to produce ethanol, corn from all other
regions will be used to replace that corn for all its other potential
uses. Therefore, it is not appropriate to ascribe the indirect affects,
both domestically and internationally, to corn grown in one area
differently to corn (or other biofuel feedstock) grown in another area.
Our national treatment of biofuel feedstock also pertains to fuels
produced in other countries. Thus for example, sugarcane-based ethanol
produced in Brazil is all treated the same regardless of where the
sugarcane was grown in Brazil. Nevertheless, comments are invited on
the option of differentiating biofuels in the future based on the
location of their feedstock production within a country.
2. Scope of the Analysis
a. Legal Interpretation of Lifecycle Greenhouse Gas Emissions
    As described in VI.A.1, the definition of lifecycle greenhouse gas
emissions refers to the ``aggregate quantity of GHG emissions'' that
are ``related to the full fuel lifecycle.'' The fuel lifecycle includes
``all stages of fuel and feedstock production and distribution, from
feedstock generation or extraction through * * * use of the finished
fuel to the ultimate consumer.'' The aggregate quantity of GHG
emissions includes ``direct emissions'' and ``significant indirect
emissions such as significant emission from land use changes.'' This
provision is written in generally broad and expansive terms, such as
``aggregate quantity'', ``related to'', ``full fuel lifecycle'', and
``all stages'' of production and distribution. At the same time, these
and other terms are not themselves defined and provide discretion to
the Administrator in implementing this definition. For example, the
word ``significant,'' which is used to modify ``indirect emissions,''
is not defined.
    The definition includes both ``direct'' and ``significant
indirect'' emissions related to the full fuel lifecycle. We consider
direct emissions as those that are emitted from each stage of the full
fuel lifecycle, and indirect emissions as those from second order
effects that occur as a consequence of the full fuel lifecycle. For
example, direct emissions for a renewable fuel would include those from
the growing of renewable fuel feedstock, the distribution of the
feedstock to the renewable fuel producer, the production of renewable
fuel, the distribution of the finished fuel to the consumer, and the
use of the fuel by the consumer as transportation fuel. Similarly,
direct emissions associated with the baseline fuel would include
extraction of the crude oil, distribution of the crude oil to the
refinery, the production of gasoline and diesel from the crude oil, the
distribution of the finished fuel to the consumer, and the use of the
fuel by the consumer. Indirect emissions would include other emissions
impacts that result from fuel production or use, such as changes in
livestock emissions resulting from changes in livestock numbers, or
shifts in acreage between different crop types. The definition of
indirect emissions specifically includes ``land use changes'' which
would include changes in the kind of usage that land is put to such as
changes in forest, pasture, savannah, and crop use.\269\
---------------------------------------------------------------------------

    \269\ Arguably shifts in acreage between different crops also
could be considered a land use change, but we believe there will be
less confusion if the term land use change is used with respect to
changes in land such as changing from savannah or forest to
cropland. There is no difference in result, as in both cases the
emissions need to be significant.
---------------------------------------------------------------------------

    In considering how to address land use changes in our lifecycle
analysis, two distinct questions have been raised--whether to account
for emissions that occur outside of the U.S., and under what
circumstances land use change should properly be included in the
lifecycle analysis.
    On the question of considering GHG emissions that occur outside of
the U.S., it is important to be clear that including such emissions in
the lifecycle analysis does not exercise regulatory authority over
activities that occur solely outside the U.S., and does not raise
questions of extra-territorial jurisdiction. EPA's regulatory action
involves classification of products either produced in the U.S. or
imported into the U.S. EPA is simply assessing whether the use of these
products in the U.S. satisfies requirements under the Clean Air Act for
the use of designated volumes of renewable fuel, cellulosic biofuel,
biomass-based diesel and advanced biofuel, as those terms are defined
in the Act. Considering international emissions in determining the
lifecycle GHG emissions of the domestically produced or imported fuel
does not change the fact that the actual regulation of the product
involves its use solely inside the U.S.
    When looking at the issue of international versus domestic
emissions, it is important to recognize that a large variety of
different activities outside the U.S. play a major part of the full
fuel lifecycle of baseline and renewable fuels. For example, for
baseline fuels (i.e., gasoline and diesel fuels used as transportation
fuel in 2005), GHG emissions associated with extraction and delivery of
crude oil imported to the U.S. all have occurred overseas. In addition,
for imported gasoline or diesel, all of the crude extraction and
delivery emissions, as well as the emissions associated with refining
and distribution of the finished product to the U.S., would have
occurred overseas. For imported renewable fuel all of the emissions
associated with feedstock production and distribution, processing of
the feedstock into renewable fuel, and delivery of the finished
renewable fuel to the U.S. would have occurred overseas. The definition
of lifecycle greenhouse gas emissions makes it clear that EPA is to
determine the aggregate emissions related to the ``full'' fuel
lifecycle, including ``all stages of fuel and feedstock production and
distribution.'' Thus, EPA could not, as a legal matter, ignore those
parts of a fuel lifecycle that occur overseas.
    Drawing a distinction between GHG emissions that occur inside the
U.S. as compared to emissions that occur outside the U.S. would
dramatically alter the lifecycle analysis in a way that bears no
apparent relationship to the purpose of this provision. The purpose of
including lifecycle GHG thresholds in this statutory provision is to
require the use of renewable fuels that achieve reductions in GHG
emissions compared to the baseline. Drawing a distinction between
domestic and international emissions would ignore a large part of the
GHG emission associated with the different fuels, and would result in a
GHG analysis of baseline renewable fuels that bears no relationship to
the real world emissions impact of the fuels. The baseline would be
significantly understated, given the large amount of imported crude
used to produce gasoline and diesel, and the importation of finished
gasoline and diesel, in 2005. Likewise, the emissions associated with
imported renewable fuel would be understated, as it would only consider
the emissions from distribution of the fuel to the consumer and the use
of the fuel by the consumer, and would ignore both the emissions that
occurred overseas as well as the emissions reductions from the intake
of CO₂ from growing of the feedstock. While large
percentages of GHG emissions would be ignored, this would take place in
a context where the global warming impact of emissions is irrespective of

[[Page 25024]]

where the emissions occur. Thus taking such an approach would
essentially undermine the provision, and would be an arbitrary
interpretation of the broadly phrased text used by Congress.
    While the emissions discussed above would more typically be
considered direct emissions related to the full fuel lifecycle, there
would also be no basis to cover just foreign direct emissions while
excluding foreign indirect emissions. The text of the statute draws no
such distinction, nor is there a distinction in achieving the purposes
of the provision. GHG emissions impact global warming wherever they
occur, and if the purpose is to achieve some reduction in GHG emissions
in order to help address global warming, then ignoring GHG emissions
because they are emitted outside our borders versus inside our borders
interferes with the ability to achieve this objective.
    For example, domestic production of a renewable fuel could lead to
indirect emissions, whether from land use changes or otherwise, some
occurring within the U.S. and some occurring in other countries.
Similarly, imported renewable fuel could have resulted in the same
indirect emissions whether occurring in the country that produced the
biofuel or in other countries. It would be arbitrary to assign the
indirect emissions to the domestic renewable fuel but not to assign the
identical indirect emissions that occur overseas to an imported product.
    Based on the above, EPA believes that the definition of lifecycle
greenhouse gas emissions is properly interpreted as including all
direct and significant indirect GHG emissions related to the full fuel
lifecycle, whether or not they occur in the U.S. This applies to both
the baseline lifecycle greenhouse emissions as well as the lifecycle
greenhouse gas emissions for various renewable fuels.
    EPA recognizes, as discussed later, our estimates of domestic
indirect emissions are more certain than our estimate of international
indirect emissions. The issue of how to evaluate and weigh the various
elements of the lifecycle analysis, and properly account for
uncertainty in our estimates, is a different issue, however. The issue
here is whether the definition of lifecycle greenhouse gas emissions is
properly interpreted as including direct and significant indirect emissions
that occur outside the U.S. as well as those that occur inside the U.S.
    As to the question of which land use changes should be included in
our lifecycle analyses, a central element to focus on is the
requirement that such indirect emissions be related to the full fuel
lifecycle. The term ``related to'' is generally interpreted as
providing a broad and expansive scope for a provision. It has routinely
been interpreted as meaning to have a connection to or refer to a
matter. To determine whether an indirect emission has the appropriate
connection to the full fuel lifecycle, we must look at both the
objectives of this provision as well as the nature of the relationship.
    In this case, EPA has used a global model that projects a variety
of agricultural impacts that stem from the use of feedstocks to produce
renewable fuel. We have estimated shifts in types of crops planted and
increases in crop acres planted. There is a direct relationship between
these shifts in the agricultural market as a consequence of the
increased demand for biofuels in the U.S. Increased U.S. demand for
biofuel feedstocks diverts these feedstocks from other competing uses,
and also increases the price of the feedstock, thus spurring
production. To the extent feedstocks like corn and soybeans are traded
internationally, this combined impact of lower supply from the U.S. and
higher commodity prices encourages international production to fill the
gap. Our analysis uses country specific information to determine the
amount, location, and type of land use change that would occur to meet
this change in production patterns. The linkages are generally close,
and are not extended or overly complex. While there is clearly
significant uncertainty in determining the specific degree of land use
change and the specific impact of those changes, there is considerable
overall certainty as to the existence of the land use changes in
general, the fact that GHG emissions will result, and the cause and
effect linkage of these emissions impacts to the increased use of
feedstock for production of renewable fuels.
    Overall, EPA is confident that it is appropriate to consider the
estimated emissions from land use changes as well as the other indirect
emissions as ``related to'' the full fuel lifecycle, based on the
reasonable technical basis provided by the modeling for the connection
between the full fuel lifecycle and the indirect emissions, as well as
for the determination that the emissions are significant. EPA believes
uncertainty in the resulting aggregate GHG estimates should be taken
into consideration, but that it would be inappropriate to exclude
indirect emissions estimates from this analysis. Developing a
reasonable estimate of these kinds of indirect emissions will allow for
a reasoned evaluation of total GHG impacts, which is needed to promote
the objectives of this provision, as compared to ignoring or not
accounting for these indirect emissions.
b. System Boundaries
    It is important to establish clear system boundaries in this
analysis. By determining a common set of system boundaries, different
fuel types can then be validly compared. As described in the previous
section, we have assessed the direct and indirect GHG impacts in each
stage of the full fuel lifecycle for biofuels and petroleum fuels.
    To capture the direct emissions impacts of feedstock production in
our analysis, we included the agricultural inputs (e.g., the fuel used
in the tractor, the energy used to produce and transport fertilizer to
the field) needed to grow crops directly used in biofuel production. We
also included the N₂O emissions associated with agricultural
sector practices used in biofuel production (including direct and
indirect N₂O emissions from synthetic fertilizer
application, N fixing crops, crop residue, and manure management), as
well as the land use change associated with converting land to grow
crops directly used in biofuel production. To capture the indirect, or
secondary, GHG emissions that result from biofuel feedstock production,
we relied on the internationally accepted lifecycle assessment
standards developed by the International Organization for
Standardization (ISO). Examples of significant secondary impacts
include the agricultural inputs associated with crops indirectly
impacted by the use of feedstock for biofuel production (domestically
and internationally), the emissions associated with land use change
that are indirectly impacted by using feedstocks for biofuel production
(e.g., to make up for lost U.S. exports), changes in livestock herd
numbers that result from higher feed costs, and changes in rice methane
emissions indirectly impacted by shifts in acres to produce feedstocks
for biofuel production. These indirect or secondary impacts would not
have occurred if it were not for the use of biomass to produce a biofuel.
    We did not include the infrastructure related GHG emissions (e.g.,
the energy needed to manufacture the tractor used on the farm) or the
facility construction-related emissions (e.g., steel or concrete needed
to construct a refinery). As part of the GHG analysis performed for
RFS1, we performed a sensitivity analysis on expanding the corn
production system to include farm equipment production to determine the
impact it has on the overall results of our analysis. We found that including

[[Page 25025]]

farm equipment production energy use and emissions increases corn
ethanol lifecycle energy use and GHG emissions and decreases the corn
ethanol lifecycle GHG benefit as compared to petroleum gasoline by
approximately 1%. Furthermore, to be consistent in the modeling if
system boundaries are expanded to include production of farming
equipment they should also be expanded to include producing other
material inputs to both the ethanol and petroleum lifecycles. The net
effect of this would be a slight increase in both the ethanol and
petroleum fuel lifecycle results and a smaller or negligible effect on
the comparison of the two.
    For this proposal, we have not yet incorporated secondary energy
sector impacts, however we plan to have this analysis complete for the
final rule. Additional details on the system boundaries are included in
the DRIA Chapter 2.
3. Modeling Framework
    Currently, no single model can capture all of the complex
interactions associated with estimating lifecycle GHG emissions for
biofuels, taking into account the ``significant indirect emissions such
as significant emissions from land use change'' required by EISA. For
example, some analysis tools used in the past focus on process
modeling--the energy and resultant emissions associated with the direct
production of a fuel at a petroleum refinery or biofuel production
facility. But this is only one component in the production of the fuel.
Clearly in the case of biofuels, impacts from and on the agricultural
sector are important, because this sector produces feedstock for
biofuel production. Commercial agricultural operations make many of
their decisions based on an economic assessment of profit maximization.
Assessment of the interactions throughout the agricultural sector
requires an analysis of the commodity markets using economic models.
However, existing economy wide general equilibrium economic models are
not detailed enough to capture the specific agricultural sector
interactions critical to our analysis (e.g., changes in acres by crop
type) and would not provide the types of outputs needed for a thorough
GHG analysis. As a result, EPA has used different tools that have
different strengths for each specific component of the analysis to
create a more comprehensive estimate of GHG emissions. Where no direct
links between the different models exist, specific components and
outputs of each are used and combined to provide an analytical
framework and the composite lifecycle assessment results. As this is a
new application of these modeling tools, EPA plans to organize peer
review of our modeling approach. The individual models are described in
the following sections and in more detail in Chapter 2 of the DRIA.
    To quantify the emissions factors associated with different steps
of the production and use of various fuels (e.g., extraction of
petroleum products, transport of feedstocks), we used the spreadsheet
analysis tool developed by Argonne National Laboratories, the
Greenhouse gases, Regulated Emissions, and Energy use in Transportation
(GREET) model. This analysis tool includes the GHG emissions associated
with the production and combustion of fossil fuels (diesel fuel,
gasoline, natural gas, coal, etc.). These fossil fuels are used both in
the production of biofuels, (e.g., diesel fuel used in farm tractors
and natural gas used at ethanol plants) and could also be displaced by
renewable fuel use in the transportation sector. GREET also estimates
the GHG emissions estimates associated with electricity production
required for biofuel and petroleum fuel production. For the
agricultural sector, we also relied upon GREET to provide GHG emissions
associated with the production and transport of agricultural inputs
such as fertilizer, herbicides, pesticides, etc. While GREET provides
direct GHG emissions estimates associated with the extraction-through-
combustion phases of fuel use, it does not capture some of the
secondary impacts associated with the fuel, such as changes in the
composition of feed used for animal production, which would be expected
due to changes in cost. EPA addresses these secondary impacts through
other models described later in this section. GREET has been under
development for several years and has undergone extensive peer review
through multiple updates. Of the available sources of information on
lifecycle GHG emissions of fossil energy consumed, we believe that
GREET offers the most comprehensive treatment of emissions from the
covered sources.
    For some steps in the production of biofuels, we used more detailed
models to capture some of the dynamic market interactions that result
from various policies. Here, we briefly describe the different models
incorporated into our analysis to provide specific details for various
lifecycle components.
    To estimate the changes in the domestic agricultural sector (e.g.,
changes in crop acres resulting from increased demand for biofuel
feedstock or changes in the number of livestock due to higher corn
prices) and their associated emissions, we used the FASOM model,
developed by Texas A&M University and others. FASOM is a partial
equilibrium economic model of the U.S. forest and agricultural sectors.
EPA selected the FASOM model for this analysis for several reasons.
FASOM is a comprehensive forestry and agricultural sector model that
tracks over 2,000 production possibilities for field crops, livestock,
and biofuels for private lands in the contiguous United States. It
accounts for changes in CO₂, methane, and N₂O
from most agricultural activities and tracks carbon sequestration and
carbon losses over time. Another advantage of FASOM is that it captures
the impacts of all crop production, not just biofuel feedstock. Thus,
as compared to some earlier assessments of lifecycle emissions, using
FASOM allows us to determine secondary agricultural sector impacts,
such as crop shifting and reduced demand due to higher prices. It also
captures changes in the livestock market (e.g., smaller herd sizes that
result from higher feed costs) and U.S. export changes. FASOM also has
been used by EPA to consider U.S. forest and agricultural sector GHG
mitigation options.\270\
---------------------------------------------------------------------------

    \270\ Greenhouse Gas Mitigation Potential in U.S. Forestry and
Agriculture, EPA Document 430-R-05-006. See http://www.epa.gov/
sequestration/greenhouse_gas.html.
---------------------------------------------------------------------------

    To estimate the impacts of biofuels feedstock production on
international agricultural and livestock production, we used the
integrated FAPRI international models, developed by Iowa State
University and the University of Missouri. These models capture the
biological, technical, and economic relationships among key variables
within a particular commodity and across commodities. FAPRI is a
worldwide agricultural sector economic model that was run by the Center
for Agricultural and Rural Development (CARD) at Iowa State University
on behalf of EPA. The FAPRI models have been previously employed to
examine the impacts of World Trade Organization proposals and changes
in the European Union's Common Agricultural Policy, to analyze farm
bill proposals since 1984, and to evaluate the impact of biofuel
development in the United States. In addition, the FAPRI models have
been used by the USDA Office of Chief Economist, Congress, and the
World Bank to examine agricultural impacts from government policy
changes, market developments, and land use shifts.
    Although FASOM predicts land use and export changes in the U.S. due to

[[Page 25026]]

greater demand for domestic biofuel feedstock, it does not assess how
international agricultural production might respond to these changes in
commodity prices and U.S. exports. The FAPRI model does predict how
much crop land will change in other countries but does not predict what
type of land such as forest or pasture will be affected. We used data
analyses provided by Winrock International to estimate what land types
will be converted into crop land in each country and the GHG emissions
associated with the land conversions. Winrock has used 2001-2004
satellite data to analyze recent land use changes around the world that
have resulted from the social, economic, and political forces that
drive land use. Winrock has then combined the recent land use change
patterns with various estimates of carbon stocks associated with
different types of land at the state level. This international land use
assessment is an important consideration in our lifecycle GHG
assessment and is explained in more detail later in this section.
    To test the robustness of the FASOM, FAPRI and Winrock results, we
are also evaluating the Global Trade Analysis Project (GTAP) model, a
multi-region, multi-sector, computable general equilibrium model that
estimates changes in world agricultural production. Maintained through
Purdue University, GTAP projects international land use change based on
the economics of land conversion, rather than using the historical data
approach applied by FAPRI/Winrock. GTAP is designed to project changes
in international land use as a result of the change in U.S. biofuel
policies, based on the relative land use values of cropland, forest,
and pastureland. The GTAP design has the advantage of explicitly
modeling the competition between different land types due to a change
in policy. As further discussed in Section VI.B.5.iv, GTAP has several
disadvantages, some of which prevented its use for the proposal. We
expect to correct several of these shortcomings between the proposed
and final rules and therefore continue to evaluate how the GTAP model
could be used as part of the final rule.
    The assessments provided in this proposal use the values provided
by the Intergovernmental Panel on Climate Change (IPCC) to estimate the
impacts of N₂O emissions from fertilizer application.
However, due to concern that this may underestimate N₂O
emissions from fertilizer application, \271\ we are working with the
CENTURY and DAYCENT models, developed by Colorado State University, to
update our assessments. The DAYCENT model simulates plant-soil systems
and is capable of simulating detailed daily soil water and temperature
dynamics and trace gas fluxes (CH₄, N₂O,
NO_X and N₂). The CENTURY model is a generalized
plant-soil ecosystem model that simulates plant production, soil carbon
dynamics, soil nutrient dynamics, and soil water and temperature. We
anticipate the results of this new modeling work will be reflected in
our assessments for the final rule. More description of this ongoing
work is included in the Chapter 2 of the DRIA.
---------------------------------------------------------------------------

    \271\ Crutzen, P. J., Mosier, A. R., Smith, K. A., and
Winiwarter, W.: N₂O release from agro-biofuel production
negates global warming reduction by replacing fossil fuels, Atmos.
Chem. Phys., 8, 389-395, 2008. See http://www.atmos-chem-phys.net/8/
389/2008/acp-8-389-2008.pdf. 
---------------------------------------------------------------------------

    To estimate the GHG emissions associated with renewable fuel
production, we used detailed ASPEN-based process models developed by
USDA and DOE's National Renewable Energy Laboratory (NREL). While GREET
contains estimates for renewable fuel production, these estimates are
based on existing technology. We expect biofuel production technology
to improve over time, and we projected improvements in process
technology over time based on available information. These projections
are discussed in DRIA Chapter 4. We then utilized the ASPEN-based
process models to assess the impacts of these improvements. We also
cross-checked the ASPEN-based process model predictions by comparing
them to a number of industry sources and other modeling efforts that
estimate potential improvements in ethanol production over time,
including the Biofuel Energy Systems Simulator (BESS) model. BESS is a
software tool developed by the University of Nebraska that calculates
the energy efficiency, greenhouse gas (GHG) emissions, and natural
resource requirements of corn-to-ethanol biofuel production systems. We
used the GREET model to estimate the GHG emissions associated with
current technology as used by petroleum refineries, because we do not
expect significant changes in petroleum refinery technology.
    We used the EPA-developed Motor Vehicle Emission Simulator (MOVES)
to estimate vehicle tailpipe GHG emissions. The MOVES modeling system
estimates emissions for on-road and nonroad sources, covers a broad
range of pollutants, and allows multiple scale analysis, from fine-
scale analysis to national inventory estimation.
    Finally, for the FRM we intend to use an EPA version of the Energy
Information Administration's National Energy Modeling System (NEMS) to
estimate the secondary impacts on the energy market associated with
increased renewable fuel production. NEMS is a modeling system that
simulates the behavior of energy markets and their interactions with
the U.S. economy by explicitly representing the economic decision-
making involved in the production, conversion, and consumption of
energy products. NEMS can reflect the secondary impacts that greater
renewable fuel use may have on the prices and quantities of other
sources of energy, and the greenhouse gas emissions associated with
these changes in the energy sector. It was not possible to complete
this analysis in time for the NPRM
    While EPA is using state-of-the-art tools available today for each
of the lifecycle components considered, using multiple models
necessitates integrating these models and, where possible, applying a
common set of assumptions. As discussed later in this section, this is
particularly important for the two agricultural sector models, FASOM
and FAPRI, which are being used in combination to describe the
agricultural sector impacts domestically and internationally. As
described in more detail in the DRIA Chapter 5, we have worked with the
FAPRI and FASOM models to align key assumptions. As a result, the
projected agricultural impacts described in Section IX are relatively
consistent across both models. One outstanding issue is the differences
between the modeling results associated with increased soybean-based
biodiesel production. We intend to further refine the soybean biodiesel
scenarios for the final rule. Additional details on all of the models
used can be found in DRIA Chapter 2. Finally, as noted earlier, we are
planning to have a number of aspects of our modeling framework peer
reviewed before finalizing these regulations. In the sections below, we
have identified specific peer review plans.
4. Treatment of Uncertainty
    While EPA believes the methodology presented here represents a
robust and scientifically credible approach, we recognize that some
calculations of GHG emissions are relatively straight-forward, while
others are not. The direct, domestic emissions are relatively well
known. These estimates are based on well-established process models
that can relatively accurately capture

[[Page 25027]]

emissions impacts. For example, the energy and GHG emissions used by a
natural gas-fired ethanol plant to produce one gallon of ethanol can be
calculated through direct observations, though this will vary somewhat
between individual facilities. The indirect domestic emissions are also
fairly well understood; however, these results are sensitive to a
number of key assumptions (e.g., current and future corn yields). We
address uncertainty in this area by testing the impact of changing
these assumptions on our results. Finally, the indirect, international
emissions are the component of our analysis with the highest level of
uncertainty. For example, identifying what type of land is converted
internationally and the emissions associated with this land conversion
are critical issues that have a large impact on the GHG emissions
estimates. We address this uncertainty by using sensitivity analyses to
test the robustness of the results based on different assumptions. We
also identify areas of additional work that will be completed prior to
the final rulemaking. For example, while we utilized an approach using
comprehensive agricultural sector models and recent satellite data to
determine the emissions resulting from international land use impacts,
we are also considering an alternative methodology (the analyses using
GTAP) that estimates changes in land use based on the relative land use
values of cropland, forest, and pastureland. Additionally, we are
considering country-specific information which may allow us to better
predict specific trends in land use such as the degree to which
marginal or abandoned pasture land will need to be replaced if used
instead for crop production. In addition to the sensitivity analysis
approach, we will also explore options for more formal uncertainty
analyses for the final rule to the extent possible. However, formal
uncertainty analyses generally include an assumption of a statistically
based distribution of likely outcomes. In the time available for
developing this proposal, we have not developed an analytical technique
which allows us to determine the likelihood of a range of possible
outcome across the wide range of critical factors affecting lifecycle
GHG assessment. We specifically ask for recommendations on how best to
conduct a sound, statistically based uncertainty analysis for the final rule.
    Despite the uncertainty associated with international land use
change, we would expect at least some international land use change to
occur as demand for crop land increases as a result of this rule.
Furthermore, the conversion of crop land will lead to GHG emission from
land conversion that must be accounted for in the calculation of
lifecycle GHG emissions. As discussed above, we believe that
uncertainty in the effects and extent of land use changes is not a
sufficient reason for ignoring land use change emissions. Although
uncertainties are associated with these estimates, it would be far less
scientifically credible to ignore the potentially significant effects
of land use change altogether than it is to use the best approach
available to assess these known emissions. We anticipate that comment
and information received in response to this proposal as well as
additional analyses will improve our assessment of land use impacts for
the final rule. Finally, we note that further research on key variables
will result in a more robust assessment of these impacts in the future.
5. Components of the Lifecycle GHG Emissions Analysis
    As described previously, GHG emissions from many stages of the full
fuel lifecycle are included within the system boundaries of this
analysis. Details on how these emissions were calculated are included
in the DRIA Section 2. This section highlights the key questions that
we have attempted to address in our analysis. In addition, this section
identifies some of the key assumptions that influence the GHG emissions
estimates in the following section.
a. Feedstock Production
    Our analysis addresses the lifecycle GHG emissions from feedstock
production by capturing both the direct and indirect impacts of growing
corn, soybeans, and other renewable fuel feedstocks. For both domestic
and international agricultural feedstock production, we analyzed four
main sources of GHG emissions: agricultural inputs (e.g., fertilizer
and energy use), fertilizer N₂O, livestock, and rice
methane. (Emissions related to land use change are discussed in the
next section).
    As described in Section IX.A, EPA uses FASOM to model domestic
agricultural sector impacts and uses FAPRI to model international
agricultural sector impacts. However, we also recognize that these
emission estimates rely on a number of key assumptions, including crop
yields, fertilizer application rates, use of distiller grains and other
co-products, and fertilizer N₂O emission rates. As described
in the following sections, we have used sensitivity analyses to test
the impact of changing these assumptions on our results.

i. Domestic Agricultural Sector Impacts

    Agricultural Sector Inputs: GHG emissions from agricultural sector
inputs (chemical and energy) are determined based on output from FASOM
combined with default factors for GHG emissions from GREET. Fuel use
emissions from GREET include both the upstream emissions associated
with production of the fuel and downstream combustion emissions. Inputs
are based on historic rates by region and include projected increases
to account for yield improvements over time. This yield increase does
not capture changes due to cropping practices such as shifts to corn-
after-corn rotations.
    N₂O Emissions: FASOM estimates N₂O emissions
from fertilizer application and nitrogen fixing crops based on the
amount of fertilizer used and different regional factors to represent
the percent of nitrogen (N) fertilizer applied that result in
N₂O emissions. This approach is consistent with IPCC
guidelines for calculating N₂O emissions from the
agricultural sector.\272\ A recent paper \273\ raised the question of
whether N₂O emissions are significantly higher than
previously estimated. To better understand this issue, we are working
with Colorado State University to analyze N₂O emissions.
Specifically, Colorado State University will provide several key
refinements for a re-analysis of land use and cropping trends and GHG
emissions in the FASOM assessment, including:
---------------------------------------------------------------------------

    \272\ 2006 Intergovernmental Panel on Climate Change (IPCC)
Guidelines for National Greenhouse Gas Inventories, Volume 4,
Chapter 11, N₂O emissions from Managed Soils, and
C_O2 Emissions from lime and Urea Application. See 
http://www.ipcc-nggip.iges.or.jp/public/2006gl/vol4.html. 
    \273\ Crutzen et al., 2008.
---------------------------------------------------------------------------

    • Direct N₂O emissions based on DAYCENT
simulations with an accounting of all N inputs to agricultural soils,
including mineral N fertilizer, organic amendments, symbiotic N
fixation, asymbiotic N fixation, crop residue N, and mineralization of
soil organic matter. Colorado State University will provide (1) the
total emission rate on an acre basis for each simulated bioenergy crop
in the 63 FASOM regions and (2) a total emissions for each N source.
    • Indirect N₂O emissions on a per acre basis
using results from DAYCENT simulations of volatilization, leaching and
runoff of N from each bioenergy crop included in the analysis for the
63 FASOM regions, combined with IPCC

[[Page 25028]]

factors for the N₂O emission associated with the simulated N losses.
    The analyses with updated N₂O estimates are not yet
complete and are not included in this proposal. We expect to complete
these analyses for the final rule.
    Livestock Emissions: GHG emissions from livestock have two main
sources: enteric fermentation and manure management. Enteric
fermentation produces methane emissions as part of the normal digestive
processes in animals. The FASOM modeling reflects changes in livestock
enteric fermentation emissions due to changes in livestock herds. As
more corn is used in producing ethanol the price of corn increases,
driving changes in livestock production costs and demand. The FASOM
model predicts reductions in livestock herds. IPCC factors for
different livestock types are applied to herd values to get GHG
emissions. The management of livestock manure can produce methane and
N₂O emissions. Methane is produced by the anaerobic
decomposition of manure. N₂O is produced as part of the
nitrogen cycle through the nitrification and denitrification of the
organic nitrogen in livestock manure and urine. FASOM calculates these
manure management emissions based on IPCC default factors for emissions
factors from the different types of livestock and management methods.
Manure management emissions are projected to be reduced as a result of
lower livestock animal numbers. Use of distiller grains (DGs), as
discussed in Section VI.B.5.b, has been shown to decrease methane
produced from enteric fermentation if replacing corn as animal
feed.\274\ This effect is not currently captured in the models but will
be considered for the final rule.
---------------------------------------------------------------------------

    \274\ Salil Arora, May Wu, and Michael Wang, ``Update of
Distillers Grains Displacement Ratios for Corn Ethanol Life-Cycle
Analysis,'' September 2008. See http://www.transportation.anl.gov/
pdfs/AF/527.pdf.
---------------------------------------------------------------------------

    Methane from Rice: Most of the world's rice, and all rice in the
United States, is grown in flooded fields. When fields are flooded,
aerobic decomposition of organic material gradually depletes most of
the oxygen present in the soil, causing anaerobic soil conditions. Once
the environment becomes anaerobic, methane is produced through
anaerobic decomposition of soil organic matter by methanogenic
bacteria. FASOM predicts changes in rice acres resulting from the RFS2
program and calculates changes in methane emissions using IPCC factors.

ii. International Agricultural Sector GHG Impacts

    Agricultural Sector Inputs: The FAPRI model does not directly
provide an assessment of the GHG impacts of changes in international
agricultural practices (e.g., changes in fertilizer load and fuels
usage), however it does predict changes in the land area and production
by crop type and by country. We therefore determined international
fertilizer and energy use based on international data collected by the
Food and Agriculture Organization (FAO) of the United Nations and the
International Energy Agency (IEA). We used the historical trends based
on these FAO and IEA data to project chemical and energy use in 2022.
Additional details on the data used are included in DRIA Chapter 2. We
intend to review input changes required to increase yields for the
final rule and request comment on the extent to which historic trends
adequately project what could occur in 2022 or what alternative
assumptions should be made and the bases for these assumptions. For
example, will changes in farming practices or seed varieties likely
result in significantly different impacts on fertilizer use
internationally than suggested by recent trends? Additionally, we
intend to have the selection and application of this data peer reviewed
before the final rule.
    N₂O Emissions: For international N₂O
emissions from crops, we apply the IPCC emissions factors based on
total amount of fertilizer applied and N₂O impacts of crop
residue by type of crop produced. As noted above, we are also working
with Colorado State University to update these factors as part of the
final rule analysis. Additional details on the factors used are
included in DRIA Chapter 2.
    Livestock Emissions: Similar to domestic livestock impacts
associated with an increase in biofuel production, FAPRI model predicts
international changes in livestock production due to changes in
commodity prices. The GHG impacts of these livestock changes, including
enteric fermentation and manure management GHG emissions, were included
in our analysis. Unlike FASOM, the FAPRI model does not have GHG
emissions built in and therefore livestock GHG impacts were based on
activity data provided by the FAPRI model (e.g., number and type of
livestock by country) multiplied by IPCC default factors for GHG
emissions. We seek comments on the extent to which the use of this
methodology is appropriate.
    Rice Emissions: To estimate rice emission impacts internationally,
we used the FAPRI model to predict changes in international rice
production as a result of the increase in biofuels demand in the U.S.
Since FAPRI does not have GHG emissions factors built into the model,
we applied IPCC default factors by country based on predicted changes
in rice acres. We seek comments on this methodology.
b. Land Use Change
    We are also addressing GHG emissions associated with land use
changes that occur domestically and internationally as a result of the
increase in renewable fuels demand in the U.S. Key questions we address
in this analysis include the land area converted to crop production,
where those acreage changes occur, lands types converted, and the GHG
emission impacts associated with different types of land conversion.
    EPA recognizes that analyzing international impacts of land use
change can introduce additional uncertainty to the GHG emissions
estimates. At this time, we do not have the same quality of data for
international crop production and projected future trends as we do for
the United States. For example, prediction of the economic and
geographic development of developing country agricultural systems is
far more difficult than prediction of future U.S. agricultural
development. The U.S. has a very mature agriculture system in which the
high quality agricultural lands are already under production and the
infrastructure to move crops to market is already in place. This is not
necessarily the case in other countries. Some very large countries
expected to play a significant role in future agricultural production
are still developing their agricultural system. Brazil, for example,
has vast areas of land that may be suitable for commercial agricultural
production that would allow for significant expansion in crop lands.
One of the restraints on expansion is the relative lack of
infrastructure (e.g., road and rail systems) that would allow shipment
of expanded crop production to market. Identifying what type of land is
converted internationally and the emissions associated with this land
conversion can significantly affect our assessment of GHG impacts. We
present a range of results for differences in these and other
assumptions in Section VI.C.2, and we seek comment on our approach so
that the final rule will use the best science to provide credible
estimates of lifecycle GHG emissions for each biofuel.

[[Page 25029]]

i. Amount of Land Converted

    The main question regarding the amount of new land needed to meet
an increasing demand for biofuels hinges on assumptions about the
intensification of existing production versus expansion of production
to other lands. This interaction is driven by the relative costs and
returns associated with each option, but there are other factors as
described below.
    Co-Products: One factor determining the amount of new crop acres
required under an increased biofuel scenario is the treatment of co-
products. For example, distillers grains (DGs) are the major co-product
of dry mill ethanol production that is also used as animal feed.
Therefore, using the DGs as an animal feed to replace the use of corn
tends to offset the loss of corn to ethanol production, and reduces the
need to grow additional corn to feed animals. As the renewable fuels
industry expands, the handling and use of co-products is also
expanding. Some uncertainty is associated with how these co-products
will be used in the future (e.g., whether it can be reformulated for
higher incorporation into pork and poultry diets, whether it will be
dried and shipped long distances, whether fractionation will become widespread).
    Both our FASOM and FAPRI models account for the use of DGs in the
agricultural sector. The FASOM and FAPRI models both assume that a
pound of co-product would displace roughly a pound of feed. However, a
recent paper by Argonne National Laboratory \275\ estimates that 1
pound of DGs can displace more than a pound of feed due to the higher
nutritional value of DGs compared to corn.
---------------------------------------------------------------------------

    \275\ Salil et al., 2008.
---------------------------------------------------------------------------

    The Argonne replacement ratios do not take into account the dynamic
least cost feed decisions faced by livestock producers. The actual use
of DGs will depend on the maximum inclusion rates for each type of
animal (based on the digestibility of DGs), the displacement ratio for
each type of animal (based on DGs energy and protein content), and the
adoption rate (based on the feed value relative to price). Furthermore,
as world vegetable oil prices increase, dry mill ethanol producers will
have an incentive to extract the corn oil from the DGs. This step
changes the nutritional content of the DGs, which results in different
replacement rates than the ones currently used in FASOM or described by
Argonne. As we plan to evaluate and incorporate a least cost feed
rationing approach for the final rule, we invite comment on the
expected future uses of DGs and their displacement ratios.
    Crop Yields: Assumptions about yields and how they may change over
time can also influence land use change predictions. Domestic yields
were based on USDA projections, extrapolated out to 2022. In 2022, we
estimate that the U.S. average corn yield will be approximately 180
bushels/acre (a 1.6% annual increase consistent with recent trends) and
average U.S. soybean yields will be approximately 50 bushels per acre
(a 0.4% annual increase).\276\ Using the FASOM model, we conducted a
sensitivity analysis on the impact of higher and lower yields in the
U.S. Details on this scenario are included in DRIA Chapter 5.1.
International yields changes are also based on the historic trends. The
FAPRI model contains projected yields and yield growths that are
generally lower in other countries compared to the U.S. We request
comment on the projected increase in crop yields in the U.S (including
consideration of how emerging seed types might be expected to increase
average crop yields). We also request comment on the use of historical
trends to predict future agricultural production in other countries and
request information on alternative methodologies and supporting data
that would allow us to base our predictions on alternative assumptions.
---------------------------------------------------------------------------

    \276\ Note that these same assumptions apply in both the
reference case and the control cases.
---------------------------------------------------------------------------

    The FASOM and FAPRI models currently do not take into account
changes in productivity as crop production shifts to marginal acres or
the impact of price induced yield changes on land use change. We would
expect these two factors could work in opposite directions and
therefore could tend to offset each other's impacts. Marginal acres in
fully developed agricultural systems are expected to have lower yields,
because most productive acres are already under cultivation. This may
not be the case in developing systems where prime agricultural lands
are not currently in full production due to, for example, lack of
supporting infrastructure. Changes in agricultural inputs (e.g.,
fertilizer, pesticides) can also change crop yield per acre. Higher
commodity prices might provide an incentive to increase production in
existing acres. If the costs of increasing productivity on existing
land were minimal relative to the value of the increased production,
then agricultural landowners would presumably adopt these productivity-
enhancing actions under the reference case. Although it is reasonable
to assume a trend wherein some productivity-enhancing practices may
become profitable if commodity prices are high enough such as might
occur as the result of increased biofuel production, it is not clear
that farmers would find significant increases in production per acre
profitable. If crop yields either domestically or international are
significantly impacted by higher commodity prices driven by general
increase in worldwide demand, this could affect our assessment of land
use impacts and the resulting GHG emissions due to increased biofuel
demand in the U.S. However, as described in Section IX, the change in
commodity prices associated with the increase in U.S. biofuel as a
result of the EISA mandates are very small and perhaps not large enough
to induce significant increased yield changes. We invite comment on
projected yields and the potential impact of increased use of marginal
lands and price induced yield changes. For the final rule we plan to
explicitly model the impact of price induced yield changes.
    Land Conversion Costs: The assumed cost associated with different
types of land conversion can also play a key role in determining how
much land will be converted. In FASOM, the decision to convert land
from pasture or forest to cropland is based on whether the landowner
can increase the net present value of expected returns through
conversion (including any costs of conversion). Among other things, the
decision to convert land depends on regional yields, costs, and other
factors affecting profitability and on the returns to alternative land
uses. In other words, FASOM assumes that land conversion is based on
maximizing profits rather than minimizing costs. Additional details on
land conversions costs incorporated in FASOM are included in DRIA Chapter 2.
    FAPRI does not explicitly model land conversion costs, however the
international production supply curves used by the FAPRI model
implicitly take into account conversion costs. FAPRI's supply curves
are based on historical responses to price changes, which take into
account the conversion costs of land, based on expected future returns
associated with land conversion. Thus, we believe that our assessments
of international land use changes are based on economic land-use decisions.
ii. Where Land is Converted
    The first step in determining what domestic and international land
will be converted due to biofuels production is to estimate the extent
to which the increased demand for biofuel feedstock

[[Page 25030]]

will be met through increased U.S. agriculture production or reductions
in U.S. exports.
    This question has several implications. For example, U.S.
agriculture production is typically more energy and input intensive but
has higher yields than agricultural production in other parts of the
world. This implies that increased production in the U.S. has higher
input GHG emission impacts but lower land use change impacts compared
to overseas production. In addition, the types of land where
agriculture would expand would be different in the U.S. vs. other parts
of the world.
    EPA's analysis relies on FASOM predictions to represent changes in
the U.S. agricultural sector, including land use, and on FAPRI to
predict the resulting international agricultural sector impacts
including the amount of additional cropland needed under different
scenarios. The impact on the international agricultural sector is
highly dependent on the U.S. export assumptions. As the FASOM model was
used to represent domestic agricultural sector impacts with an assumed
export picture, the international agricultural sector impacts from
FAPRI needed to be based on a consistent set of export assumptions. We
worked with FASOM and FAPRI modelers to ensure this consistency. This
involved coordinating crop yields, ethanol yields and co-product use,
assumptions regarding CRP acres, a consistent export response, and a
consistent livestock demand and feed use in both models.
    As shown in Chapter 2 of the DRIA, coordination of assumptions has
generated a consistent export picture response from both the FASOM and
FAPRI model for the majority of biofuel and feedstock scenarios
considered. Differences in responses in the biodiesel scenario remain
between the two models. FASOM assumes more biodiesel will come from new
soybean acres (but total domestic acres are relatively constant as
reductions in other crops offset the increase in soybean acres). In
comparison, FAPRI contains more types of oil seed crops and has a more
elastic demand in the soybean oil market. The FAPRI model also allows
for some corn oil fractionation from DGs, which can be used as a
substitute for soybean oil. The FASOM model predicts a larger change in
net exports than the FAPRI model. Since we are using the FAPRI model as
the basis for estimating international land use changes, we may be
underestimating the international land use change emissions associated
with soybean based biodiesel. For the final rule, EPA will work, in
particular, to resolve the differences in soybean production impact
between the models. This, too, may modify our assessment of the
biodiesel lifecycle GHG emissions.
    Due to the wide range of carbon and biomass properties associated
with land in different parts of the world, the location of crop
conversion is also important to our lifecycle analysis. For example, an
average acre of forest in Southeast Asia stores a much larger quantity
of carbon than a typical acre of forest in Northern Europe. The FAPRI
model provides estimates of the acreage change by country and crop that
result from a decrease in U.S. exports due to the increase in U.S.
biofuel demand. These estimates are based on historic responsiveness to
changes in prices in other countries. Implicit in these supply curves
are the costs associated with converting new land to crop production
and the relative competitiveness of each country to increase production
based on production costs, yields, transportation costs, and currency
fluctuations. FAPRI also includes in its baseline projections of future
population growth, GDP growth, and other macroeconomic changes. FAPRI
also takes into account the fact that not all U.S. exports will need to
be made up in international production, as there is a small decreases
in demand due to shifts in crop production and higher prices.

iii. What Type of Land is Converted

    In the same way that the location of land conversion is important,
the type of land that is converted is critical to the magnitude of
impact on the GHG emissions associated with biofuel production. For
example, the conversion of rainforest results in a much larger increase
in GHG emissions than the conversion of grassland. There are several
options for determining what type of land will be converted to crop
acreage. One option is to model land rental rates for different types
of land (e.g., forest, pasture, and crop production), and allow the
model to choose the type of land that is expected to have the highest
net returns. This approach is used by FASOM on the domestic side.
Another option is to use historical land conversion trends in a given
country or region. The FAPRI/Winrock approach uses this approach for
international land use conversion.
    Domestic: The FASOM model includes competition between land types,
agriculture, pasture, and forest land. The interaction is based on
providing the highest rate of return across the different land types.
Therefore domestically we have the ability to explicitly model what
types of land would be converted to increased agriculture based on the
rates of return for different land types for the 63 regions in FASOM.
For this draft proposal we incorporated the agricultural component
(which includes both existing cropland and pasture) of the FASOM model,
but not the forestry component (see Section IX.A for explanation).
Therefore, this analysis assumes that all additional cropland predicted
by FASOM comes from pasture. As we incorporate the forestry component
for the final rule analysis we would expect to see more interaction
between the forestry and agriculture sector such that there may be
conversion of forest to agriculture based on additional cropland
needed. While we do not know if forest will be converted to cropland or
the extent that this might occur, if domestic forests were converted to
cropland, we would expect domestic GHG emissions would increase. This
work will be incorporated for our final rule.
    International: Basing land use change on the economics and rates of
return of different land uses offers advantages for capturing potential
future land use changes. However, the only model potentially capable of
fully incorporating this calculation internationally, GTAP, is still in
the process of being updated and modified for this purpose. Thus, EPA
has chosen to use historical patterns as identified by satellite images
to estimate future land conversion. This approach is referred to here
as the FAPRI/Winrock approach because it relies on the integration of
each of these tools.
    EPA believes that FAPRI/Winrock is a scientifically credible
modeling approach at this time. However, we will continue to work with
the GTAP model to help test the results generated by our primary approach.

FAPRI/Winrock

    Since FAPRI does not contain information on what type of land is
being converted into cropland, we worked with Winrock International, a
global nonprofit organization, to address this question. A key
advantage of Winrock is that they can accurately measure and monitor
trends in forest and land use change, forest carbon content,
biodiversity, and the impact of infrastructure development.
Furthermore, several of the Winrock staff were involved in the
development of the IPCC land use change good practice guidance and are
widely recognized as the leaders in this field.
    Using satellite data from 2001-2004, Winrock provided a breakdown
of the types of land that have been converted

[[Page 25031]]

into cropland for a number of key agriculturally producing countries
based on the International Geosphere-Biosphere Programme (IGBP).\277\
The IGBP land cover list includes eleven classes of natural vegetation,
three classes of developed and mosaic lands, and three classes of non-
vegetated lands. The natural vegetation units distinguish evergreen and
deciduous, broadleaf and needle-leaf forests, mixed forests, where
mixtures occur; closed shrublands and open shrublands; savannas and
woody savannas; grasslands; and permanent wetlands of large areal
extent. The three classes of developed and mosaic lands distinguish
among croplands, urban and built-up lands, and cropland/natural
vegetation mosaics. Classes of non-vegetated land cover units include
snow and ice; barren land; and water bodies. Winrock aggregated these
categories into five similar classes: five classes of forest were
combined into one, two classes of savanna were combined into one, and
two classes of shrubland were combined into one. The final land cover
categories for this analysis are forest, cropland, grassland, savanna,
and shrubland. The rest of the IGBP categories not of interest to this
analysis were reclassified into the background. The satellite data
represents different types of land cover, which we are using as a proxy
for land use.
---------------------------------------------------------------------------

    \277\ U.S. Geological Survey MODIS Data Set Documentation. See
http://edcdaac.usgs.gov/modis/mod12q1v4.asp.
---------------------------------------------------------------------------

    A key strength of this approach is that satellite information is
based on empirical data instead of modeled predictions. Furthermore, it
is reasonable to assume that recent land use changes have been driven
largely by economics and recent historical patterns will continue in
the future absent major economic or land use regime shifts caused, for
example, by changes in government policies. We are using the FAPRI
model to predict where in the world, based on economic conditions, the
most likely increase in agriculture production will occur as a result
of the EISA mandates. We are then using the historical satellite data
to address the key question: If additional land is needed for crop
production in a particular country, what type of land will be used? The
Winrock analysis addresses this question by calculating the weighted
average type of land that was converted to cropland between 2001 and
2004. Essentially, we are using the Winrock data to determine the type
of land that is most likely to be converted to cropland, should
additional acres be needed as predicted by FAPRI.
    Table VI.B.5-1 shows the percentage of land converted to cropland
between 2001 and 2004 according to the Winrock satellite data analysis
for the countries currently available. We use these percentages to
calculate a weighted average of the types of land converted into
cropland. For example, if FAPRI predicts that one additional acre of
cropland will be brought into production in Argentina, we used the
Winrock data to estimate that 8% on average of that acre will come from
forest, 40% of that acre will come from grassland, 45% of that land
will come from savanna, and 8% of that land will come from shrubland.
Using GTAP might result in different percentage weights.

                         Table VI.B.5-1--Types of Land Converted to Cropland by Country
                                                  [In percent]
----------------------------------------------------------------------------------------------------------------
                     Country                          Forest         Grassland        Savanna          Shrub
----------------------------------------------------------------------------------------------------------------
Argentina.......................................               8              40              45               8
Brazil..........................................               4              18              74               4
China...........................................              17              38              23              21
EU..............................................              27              16              36              21
India...........................................               7               7              33              53
Indonesia.......................................              34               5              58               4
Malaysia........................................              74               3              19               3
Nigeria.........................................               4              56              36               4
Philippines.....................................              49               5              44               3
South Africa....................................              10              22              53              15
----------------------------------------------------------------------------------------------------------------
Source: Winrock Satellite Data (2001-2004).

    We are assuming that the weighted average, resulting from
agriculture demand as well as other possible drivers, is a reasonable
estimate of the land use change attributable to increased agricultural
demand. A shortcoming of this approach is that it assumes that when new
crop acres are needed to meet increased agricultural demand these crop
acres will follow the average pattern of recent historical land
conversion, recognizing that this pattern is based on a variety of
drivers of land use change, not all of which are associated with
agricultural demand. This approach is not able to isolate from the
historical pattern the land use changes stemming just from increased
agricultural demand. For example, it is likely that in some cases trees
are being removed from forests for the value of the wood. However,
having removed valuable wood, additional clearing may occur to allow
the land to be used for pasture or cropland. In that case the GHG
emissions associated with the removal of the trees would not occur as a
consequence of increased agricultural demand, but as a consequence of
increased demand for the wood, while the GHG emissions associated with
the additional clearing would occur as a consequence of the
agricultural demand.
    As a result, the Winrock data also does not distinguish between the
land-use impacts associated with one crop versus another. Indeed, at
least in the case of sugarcane production in Brazil, a number of
researchers argue that expanded sugarcane production is likely to occur
in significant part through the use of degraded or abandoned pasture
land without additional land use impact.\278\ These research reports
suggest that general historical trends in land use change to grow crops
in Brazil may not pertain to expected growth in sugarcane production.
Ideally, an analysis of a U.S. biofuels policy's influence on land use
change would

[[Page 25032]]

model the marginal impact that U.S. biofuel would have on land use and
land use change in addition to baseline land use change. Use of
historic land use change data is capturing some of this baseline land
use change. Comments are requested on our approach of assuming
historical land use changes will continue to be followed in response to
increased agricultural demand associated with our biofuel policy. We
also invite comment on what alternative methodologies and data are
available, if any, to better link the impacts of biofuels to land use
change. To the extent additional information or data may be available
for certain countries such as the example of Brazil, we also ask how
this country-specific data and similar information might best be
integrated with the modeling results otherwise available. Furthermore,
to the extent different government policies can shift land use patterns
(e.g., through regulations, financial supports), these weighted
averages could change in the future. We request comment on whether
these government policies and regulations should be incorporated into
the future land use change calculations and the best methodology for
taking into account these changes.
---------------------------------------------------------------------------

    \278\ See for example ``Mitigation of GHG emissions using
sugarcane bio-ethanol--Working Paper'' by Isaias C. Macedo and
Joaquim E. A. Seabra, and ``Prospects of the Sugarcane Expansion in
Brazil: Impacts on Direct and Indirect Land Use Changes--Working
Paper'' by Andre Nassar et al., both received by EPA October 13, 2008.
---------------------------------------------------------------------------

    The Winrock data and analyses present an aggregate picture of land
use changes; they cannot predict the nature of the land use change that
will result due to an additional acre of corn planted in a country
versus an additional acre of sugarcane or soybeans. In reality,
sugarcane may be more suitable for planting in different regions with
different soil types compared to corn or soybeans. However, because we
are using weighted averages to characterize the type of land that is
converted to crop acres, all additional crop acres in a particular
country are treated identically.
    Winrock also provides information on land conversions between other
categories (e.g., forest to savanna). For one set of GHG analyses, we
assumed that land taken out of actively managed use \279\ (e.g.,
pasture used for livestock production) would have to be replaced with
new pasture acreage, thereby capturing some of the domino effect
associated with converting previously productive land into cropland.
Therefore, in addition to land conversion shown in Table VI.B.5-1, we
also include land conversion to replace some of the grassland and
savanna that is used as pasture. An alternative approach would be to
assume that no additional land is necessary, since there is a
significant amount of pastureland that could be used more intensively
for grazing purposes. For example, as noted above, in Brazil almost all
of the direct land conversion associated with expanding sugarcane
production is coming out of existing pasture land, in some cases,
depleted, low value pasture land, that may have relatively low levels
of stored carbon compared to other land. Also in Brazil there is a
trend toward more intensive use of existing pasture land by grazing
higher numbers of cattle per unit of pasture, decreasing the need to
replace pasture converted to cropland. This more intensive use of
pasture is encouraged by two factors: improved grasses which can
sustain more intensive grazing and lack of transportation
infrastructure which tends to constrain geographic expansion of pasture
lands. However, we also note that depleted cropland in Brazil might
also be suitable for other crop production. To extend sugarcane limits
to production of these other crops on this land, the indirect effect
could be that these crops move into other areas of Brazil and resulting
in increased emissions due to land use change. We invite comment on
alternative methodologies for predicting land use changes in particular
in other countries. Some alternative methodologies are described in
more detail in Chapter 2 of the DRIA.
---------------------------------------------------------------------------

    \279\ GTAP Land Cover Data (2000-2001).
---------------------------------------------------------------------------

    The FAPRI model results have been used in peer reviewed literature
in conjunction with satellite data to assess land use changes \280\ and
we also believe it is an appropriate method for projecting biofuel
induced land use changes. However, we recognize the uncertainty
associated with this approach and, in addition to seeking public
comment, we plan to conduct an expert peer review of the data and
methods used, including the appropriateness of using historic satellite
data to project future land use changes.
---------------------------------------------------------------------------

    \280\ Searchinger et al., 2008.
---------------------------------------------------------------------------

iv. What Are the GHG Emissions Associated With Different Types of Land
Conversion?
    Our estimates of domestic land use change GHG emissions are based
on outputs of the FASOM model. As we are just using the agricultural
portion of the FASOM model for this analysis the land use change GHG
emissions are limited to changes in land use for existing crop and
pasture land. Some of that crop land could currently be fallow and some
of the pasture land could currently be unused. However, no new crop or
pasture land (beyond some Conservation Reserve Program (CRP) land due
to legislative changes in the program) is added compared to current
levels. Thus FASOM only models shifts in the use of this land.
    Changes in the agricultural sector due to increased corn used for
ethanol have impacts on land use change in a number of ways. FASOM
explicitly models change in soil carbon from increased crop production
acres and from different types of crop production. FASOM also models
changes in soil carbon from converting non crop land into crop
production. Land converted to crop land could include pasture land. As
recommended by USDA, we are assuming that 32 million acres of CRP land
will remain in that program even if crop prices increase and thus
increase land values. This assumption is consistent with the 2008 Farm
Bill, which limits CRP acres to 32 million. A sensitivity analysis on
this assumption is included in Chapter 5 of the DRIA.
    For the international impacts, we used the 2006 IPCC Agriculture,
Forestry, and Other Land Use (AFOLU) Guidelines \281\ and the Winrock
provided GHG emissions factors for each country based on the weighted
average type of land converted. GHG emissions estimates were based on
immediate releases (e.g., changes in biomass carbon stocks, soil carbon
stocks, and non-CO₂ emissions assuming the land is cleared
with fire) and foregone forest sequestration (the future growth in
vegetation and soil carbon). Additional details on these calculations
are included in Chapter 2 of the DRIA. For the emissions factors
presented, we assume forests cleared would have continued to sequester
carbon for another 80 years, based on the amount of time it takes for
forests to reach a general equilibrium stage. We request comment on
whether it is appropriate to include foregone sequestration in the GHG
emissions estimates. Carbon soil calculations \282\ take into account
the annual changes in carbon content in the top 30 centimeters of soil
over the first 20 years, based on IPCC recommendations.\283\ We also
request comment on whether soil carbon calculations should be based on
the top 30 centimeters of soil. These emission factors do not include
credits for harvested wood products, based on the expectation that they
would have a

[[Page 25033]]

very small impact on our estimates of land use change emissions.
However, we intend to analyze the impact of wood product credits for
the final rule. We invite comment on whether it is appropriate to
include wood product credits in the GHG emissions estimates.
---------------------------------------------------------------------------

    \281\ 2006 IPCC Guidelines for National Greenhouse Gas
Inventories, Volume 4, Agriculture, Forestry and Other Land Use
(AFOLU). See http://www.ipcc-nggip.iges.or.jp/public/2006gl/
vol4.html. 
    \282\ See ftp://www.daac.ornl.gov/data/global_soil/
IsricWiseGrids.
    \283\ 2006 IPCC Guidelines for National Greenhouse Gas
Inventories, Volume 4, Section 5.3.3.4.
---------------------------------------------------------------------------

    GHG emissions associated with land use changes vary significantly
based on the type of land and the geographic region. For example, the
GHG emissions associated with converting an acre of grassland to
cropland in China are lower than the emissions associated with
converting an acre of shrubland to cropland in China. Similarly, the
GHG emissions associated with converting an acre of forest to cropland
in Malaysia are larger than the emissions associated with converting an
acre of forest in Nigeria to cropland. Where country specific emission
factors were not available in time for the proposal, we used world
average. For the proposal, we focused on the countries with the largest
projected changes in crop acreage. The Winrock data currently covers
63% of total land use change acres associated with corn ethanol, 53% of
the acres associated with biodiesel, 57% of the acres associated with
switchgrass, and 87% of the acres associated sugarcane ethanol. We will
continue to add additional countries for our analysis for the final
rule. Two changes that may impact these results for the final rule
include the addition of perennial crops and the conversion on land with
peat soils. We request comment on our calculation of emission factors
due to land use change; improved data and assumptions are specifically
requested. Additionally, we plan to have the calculation of these
emissions factors reviewed by experts in this field. Details on the
Winrock estimates are included in the DRIA Chapter 2.

GTAP Approach:

    GTAP is an economy-wide general equilibrium model that was
originally developed for addressing agricultural trade issues among
countries. The databases and versions of the model are widely used
internationally.\284\ Since its inception in 1993, GTAP has rapidly
become a common ``language'' for many of those conducting global
economic analysis. For example, the WTO and the World Bank co-sponsored
two conferences on the so-called Millennium Round of Multilateral Trade
talks in Geneva. Here, virtually all of the quantitative, global
economic analyses were based on the GTAP framework. Over the past few
years, a version of the model was developed to explicitly model global
competition among different land types (e.g., forest, agricultural
land, pasture) and different qualities of land based on the relative
value of the alternative land-uses. More recently, it was modified to
include biofuel substitutes for gasoline and diesel. In simulating land
use changes due to biofuels production, GTAP explicitly models land-use
conversion decisions, as well as land management intensification. For
example, it allows for price-induced yield changes (e.g., farmers can
reallocate inputs to increase yields when commodity prices are high)
and considers the marginal productivity of additional land (e.g.,
expansion of crop land onto lower quality land as a result of the
increased use of biofuels). Most importantly, in contrast to other
models, GTAP is designed with the framework of predicting the amount
and types of land needed in a region to meet demands for both food and
fuel production. The GTAP framework also allows predictions to be made
about the types of land available in the region to meet the needed
demands, since it explicitly represents different land types within the model.
---------------------------------------------------------------------------

    \284\ https://www.gtap.agecon.purdue.edu. 
---------------------------------------------------------------------------

    The global modeling of land-use competition and land management
decisions is relatively new, and evolving.\285\ GTAP does not yet
contain cellulosic feedstocks in the model. In addition, GTAP does not
currently contain unmanaged land, which could be a major factor driving
current GTAP land use projections and is a significant potential source
of GHG emissions. We expect to update GTAP with cellulosic feedstocks
and unmanaged land in time for the final rule.
---------------------------------------------------------------------------

    \285\ See Hertel, Thomas, Steven Rose, Richard Tol (eds.), (in
press). Economic Analysis of Land Use in Global Climate Change
Policy, Routledge Publishing.
---------------------------------------------------------------------------

    Our proposal is therefore based on the FAPRI/Winrock estimates.
There are advantages and disadvantages associated with any model choice
and we have chosen the FAPRI/Winrock combination as the best approach
available for preparing the proposal. Although we have not relied on
the current version of GTAP for the principal analyses in this
proposal, others have used versions of the current model to assess land
use changes which could result from expanded biofuel demand. The
California Air Resources Board as part of the analysis for their low
carbon fuel standard used GTAP to model indirect land use change for
biofuels. More information on their program and GTAP analysis can be
found at http://www.arb.ca.gov/fuels/lcfs/lcfs.htm. Furthermore,
researchers from Purdue University have released a report on work using
GTAP to look at land use change associated with corn ethanol production
scenarios.\286\ This work was partially funded by Argonne National Lab
for possible inclusion in the GREET model. We anticipate additional
refinements will be made to the GTAP model between the proposal and
final rule and we will include this information and results in the
docket as they become available. We invite comments in this NPRM on the
use of the GTAP model in helping to establish the GHG emissions
estimates for the final rule.
---------------------------------------------------------------------------

    \286\ Land Use Change Carbon Emissions due to US Ethanol
Production, Wallace E. Tyner, Farzad Taheripour, Uris Baldos,
January 2009. Available at http://www.agecon.purdue.edu/papers/
biofuels/Argonne-GTAP_Revision%204a.pdf. 
---------------------------------------------------------------------------

v. Assessing GHG Emissions Impacts Over Time and Potential Application
of a GHG Discount Rate
    When comparing the lifecycle GHG emissions associated with biofuels
to those associated with gasoline or diesel emissions, it is critical
to take into consideration the time profile associated with each fuel's
GHG emissions stream. With gasoline, a majority of the lifecycle GHG
emissions associated with extraction, conversion, and combustion are
likely to be released over a short period of time (i.e., annually) as
crude oil is converted into gasoline or diesel fuel which quickly pass
to market. This means that the lifecycle GHG emissions of a gallon of
gasoline produced one year are unlikely to vary much from the lifecycle
GHG emissions of a similar gallon of gasoline produced in a subsequent year.
    In contrast, the lifecycle GHG emissions from the production of a
typical biofuel may continue to occur over a long period of time. As
with petroleum based fuels, renewable fuel lifecycle GHG emissions are
associated with the conversion and combustion of biofuels in every year
they are produced. In addition, GHG emissions could be released through
time if new acres are needed to produce corn, soybeans or other crops
as a replacement for crops that are directly used for biofuel
production or displaced due to biofuels production. The GHG emissions
associated with converting land into crop production would accumulate
over time with the largest release occurring in the first few years due
to clearing with fire or biomass decay. After the land is converted,
moderate amounts of soil carbon would continue to be released for

[[Page 25034]]

approximately 20 years.\287\ Furthermore, there would be foregone
sequestration associated with forest clearing as this forest would have
continued to sequester carbon had it not been cleared for approximately
80 years.
---------------------------------------------------------------------------

    \287\ Following Section 5.3.3.4 of the IPCC AFOLU guidelines,
the total difference in soil carbon stocks before and after
conversion was averaged over 20 years.
---------------------------------------------------------------------------

    Therefore, we have included an analysis which considers GHG
emissions from land use change that may continue for up to 80 years,
based on our estimate of the average length of foregone sequestration
after a forest is cleared. Annual foregone sequestration rates were
estimated by ecological region using growth rates for forests greater
then 20 years old from the 2006 IPCC guidelines for Agriculture,
Forestry and Other Land Use.\288\ Studies have estimated that new
forests grow for 90 years to over 120 years.\289\ More recent estimates
suggest that old growth forests accumulate carbon for up to 800
years.\290\ The foregone sequestration methods used in this proposal
are within the range supported by the scientific literature and the
2006 IPCC guidelines. Details of the foregone sequestration estimates
are included in DRIA Chapter 2. We seek comment on our estimate of the
average length of annual foregone forest sequestration for
consideration in biofuel lifecycle GHG analysis.
---------------------------------------------------------------------------

    \288\ Table 4.9 from the 2006 GL AFOLU was used to estimate the
lost C sequestration of forests that were converted to another land use.
    \289\ See Greenhouse Gas Mitigation Potential in U.S. Forestry
and Agriculture, EPA Document 430-R-05-006 for a discussion of the
time required for forests to reach carbon saturation.
    \290\ Luyassert, S et al., 2008. Old-growth forests as global
carbon sinks. Nature 455: 213-215. Link: http://www.nature.com/
nature/journal/v455/n7210/abs/nature07276.html. 
---------------------------------------------------------------------------

    Figure VI.B.5-1 shows how lifecycle GHG emissions vary over time
for a natural gas fired dry mill corn ethanol plant assuming that all
land use change occurs in 2022. While biomass feedstocks grown each
year on new cropland can be converted to biofuels that offer an annual
GHG benefit relative to the petroleum product they replace, these
benefits may be small compared to the upfront release of GHG emissions
from land use change. Depending on the specific biofuel in question, it
can take many years for the benefits of the biofuel to make up for the
large initial releases of carbon that result from land conversion
(e.g., the payback period). As shown in Figure VI.B.5-1, the payback
period for a natural gas-fired dry mill corn ethanol plant which begins
operation in 2022 would be approximately 33 years. We present a similar
payback period calculation for the full range of biofuels analyzed in
Section VI.C.
[GRAPHIC] [TIFF OMITTED] TP26MY09.008

    As required by EISA, our analysis must demonstrate whether biofuels
reduce GHG emissions by the required percentage relative to the 2005
petroleum baseline. A payback period alone cannot answer that question.
Since the payback period alone is not sufficient for our analysis, we
have considered accounting methods for capturing the full stream of
emissions and benefits over time. There are at least two necessary
criteria for the accounting methods we have considered. First, they
must provide an estimate of renewable fuel lifecycle GHG emissions that
is consistent over time. Otherwise, for example, all of the upfront
emissions due to land clearing would be assigned to corn ethanol
produced in the first year, and none of those emissions to corn ethanol
produced the following years even though this land use change is
central to the production over these following years. Second, the
accounting method must also provide a common metric that allows for a
direct comparison of biofuels to gasoline or

[[Page 25035]]

diesel. When accounting for the time profile of lifecycle GHG
emissions, the two most important assumptions in the determination of
whether a biofuel meets the specified emissions reduction thresholds
include: (1) The time period considered and (2) the discount rate
(which could be zero) applied to future emissions streams.

Time Periods Considered

    The illustration of the payback period in Figure VI.B.5-1
demonstrates the importance of the time period over which to consider
both the lifecycle GHG emissions increases associated with the
production of a biofuel as well as the benefits from using the biofuel.
As mentioned above, based on our lifecycle GHG analysis for this
proposed rule we estimate that the payback period for corn ethanol
produced in a natural gas-fired dry mill is approximately 33 years. In
this case, if we measure GHG impacts over a time period of less than 33
years we will determine that the total corn ethanol produced over this
time period increases lifecycle GHG emissions. Conversely, total corn
ethanol production will reduce net lifecycle GHG emissions if we look
beyond 33 years, with net emissions reductions increasing the further
into the future we extend our analysis. To inform our decision of which
time period for analysis is most appropriate, we must consider a number
of factors including but not limited to the length of time over which
we expect a particular biofuel to be produced, the time over which
biofuel production continues to impact GHG emissions into the future,
the importance of achieving near-term GHG emissions reductions, and the
increasing uncertainty of projecting GHG emissions impacts into the
future. Based on these considerations, our discussion of lifecycle
analyses prepared for this proposed rule focuses on time periods of 100
years and 30 years.
    There are advantages and disadvantages to using the 100 and 30 year
time frames to represent both emissions impacts as well as emissions
benefits of use of biofuels over time. There are several principal
reasons for using the 100 year time frame. First, greenhouse gases are
chemically stable compounds and persist in the atmosphere over long
time scales that span two or more generations. Second, the 100 year
time frame captures the emissions associated with land use change that
may continue for a long period of time after biofuel-induced land
conversion first takes place.\291\ For example, physical changes in
carbon stocks on unmanaged lands may not slow until after 100 years,
and optimal forest rotation ages can influence greenhouse gas emissions
for 100 years on managed lands. Similarly, a 100 year time frame would
allow estimating the future changes in the land should the need for
these changes due to biofuel production cease. For example, as
discussed in more detail below, if production of a biofuel ended, then
the land use impacts associated with that biofuel would also tend to go
away in a process known as land use reversion. A longer time frame
would allow assessment of the impacts of that land use reversion.
---------------------------------------------------------------------------

    \291\ Luyassert, S et al., 2008. Old-growth forests as global
carbon sinks. Nature 455: 213-215. Link: http://www.nature.com/
nature/journal/v455/n7210/abs/nature07276.html. 
---------------------------------------------------------------------------

    For a number of reasons we believe that biofuel production could
continue for a long time into the future. As biofuel technologies
advance and production costs are decreased, it is likely that renewable
fuels will become increasingly competitive with petroleum-based fuels.
Another reason for expecting long term biofuel production is that,
unlike a specific facility that has an expected lifetime, the RFS
program does not have a specified expiration date. The expectation that
renewable fuel production will continue for a long time provides
justification for using a longer time frame for analysis, such as 100
years. Another reason for considering an inter-generational time period
such as 100 years for lifecycle GHG analysis is that climate change is
a long-term environmental problem that may require GHG emissions
reductions for many decades.
    The 100 year time frame also has drawbacks. A general concern with
projecting impacts over a very long time period is that uncertainty
increases the further the analysis is extended into the future. For
example, a 100 year analysis presumes that production of a particular
biofuel will continue for at least 100 years. Although we expect
renewable fuel production as a whole to continue for a long time, it is
possible that due to changing market conditions or other factors, the
use of first generation biofuels (e.g., corn ethanol) could see a
decline in use over a shorter period of time.
    For this proposal, we are also showing the results of analyzing
both GHG emissions impacts of producing a biofuel as well as benefits
from using the biofuel over 30 years, a time frame which has been used
in the literature to estimate the greenhouse gas impacts of
biofuels.292 293 Since a time period such as 30 years would
truncate the potential GHG benefits that accumulate over time, this
second option would reduce the GHG benefits of biofuels relative to
gasoline or diesel compared to assuming a longer time frame for biofuel
production such as 100 years.
---------------------------------------------------------------------------

    \292\ Searchinger et al., 2008.
    \293\ M. Delucchi, ``A multi-country analysis of lifecycle
emissions from transportation fuels and motor vehicles'' (UCD-ITS-
RR-05-10, University of California at Davis, Davis, CA 2005). See
also http://www.its.ucdavis.edu/people/faculty/delucchi/. 
---------------------------------------------------------------------------

    One advantage of using a shorter time period is that it is more
``conservative'' from a climate change policy perspective. In general,
the further out into the future an analysis projects, the more
uncertainty is introduced into the results. For example, with a longer
time period for analysis, it is more likely that significant changes in
market factors or policies will change the incentives for producing
biofuels. If a biofuel only has greenhouse benefits when considered in
an extended future time frame, it is not clear that these benefits will
be realized due to the inherent uncertainty of the future. Also,
potential irreversible climate change impacts or future actions in
other sectors of the economy, such as reductions from stationary
sources, could influence the relative importance of renewable fuel GHG
impacts. The timing and severity of these potential irreversible
climate change impacts are clearly uncertain as is the degree to which
near-term lifecycle emissions related to biofuel production influences
these climate change impacts. Given these uncertainties, it may be
appropriate to limit our analysis horizon to a much shorter time period
such as 30 years.
    Several disadvantages are also associated with choosing the 30 year
time frame to represent both emissions impacts as well as emissions
benefits. One key disadvantage is that it ignores the potential sources
of GHG emissions impacts of producing biofuel after 30 years such as
foregone sequestration from forests that may have been removed which
could have continuing impacts even after production of a biofuel has
ended. Thus, it doesn't account for the full land use emissions
``signature'' of biofuels. In addition, even if second generation fuels
start to dominate new construction, building a first generation fuel
production facility such as a corn ethanol refinery represents a
significant capital investment. Once the facility is built and
financed, it may continue

[[Page 25036]]

producing biofuel as long as it is covering its operating costs. This
suggests that, once a plant is built, if the variable cost of corn
ethanol production is less than the cost to produce gasoline, then corn
ethanol production at that facility may continue. This economic
advantage may contribute to the longevity of first generation biofuel
production and usage far into the future.
    An appropriate time frame for analysis could also be different for
different biofuels. While we could assume that corn ethanol would be
phased out after a shorter time period such as 30 years, it might be
more appropriate to use a longer time period over which to analyze the
benefits of other advanced biofuels such as cellulosic biofuels. It
could be reasonably assumed that cellulosic biofuels will be produced
for more than 30 years, perhaps for 100 years or longer. However, even
if cellulosic biofuels are expected to be produced for 100 years or
longer, a shorter time period, such as 30 years, may still be the most
relevant period over which to assess GHG emissions, given the
importance of near-term emissions reductions and the increasing
uncertainty of future events. We specifically seek comments on the 100
year and 30 year time frames discussed in this proposal. We also seek
general comments on the most appropriate time periods for analysis of
biofuels, and whether we should use different time periods for
different types of renewable fuels.
    Another way to look at the time period issue, which we have not
specifically analyzed for this proposed rule, would separate the time
period into two parts. The first part would consider how long we expect
production of a particular biofuel to continue into the future. We
refer to this concept, which is similar to the project lifetime often
considered in traditional cost benefit analysis, as the ``project''
time horizon. The second part would address the length over which to
account for the changes in GHG emissions due to land use changes which
result from biofuel production. We call this the ``impact'' time horizon.
    Our analysis for this proposed rule has not considered a scenario
where the project time horizon is shorter than impact time horizon.
However, we are considering this option for the final rule. For
example, we could look at a scenario where corn ethanol production
continues for 30 years and land use related GHG emissions are estimated
for 100 years. Specifically, we are considering whether to use 30 years
after 2015 (as an approximation of when ethanol production from corn
starch reaches 15 billion gallons) as a reasonable estimate of when
corn will no longer be used for ethanol production due to advances in
other biofuels and the competing demand to use corn for food rather
than biofuel feedstock. We specifically ask whether a 30 year estimate
of continued corn starch ethanol production (i.e., through 2045) is a
reasonable estimate for assessing the stream of GHG benefits from corn
ethanol use while 100 years would be appropriate for assessing impacts
of the land use change. Under such an assumption a 100 time horizon
would capture the longer term emission impacts of corn ethanol
production (including indirect land use impacts) while the benefits
from 31 through 100 years would be zero since corn ethanol would be
modeled as no longer in use.
    In that scenario, we would have to consider the lifecycle GHG
impacts after the production of corn ethanol ends. This would include
the issue of land reversion, or what happens to the land used to
produce a biofuel feedstock after its use for biofuel production has
ceased. A full accounting of land reversion would involve economic
modeling to determine how long we expect production of a particular
biofuel to last, and to determine the land use changes after that
biofuel production ends. Ideally this modeling would extend well beyond
2022 to the point where land reversion is complete, and it would
include projections for global crop yield improvements, population
trends, food demand, and other key factors. For this proposal, we have
not projected the GHG emissions associated with land reversion, but we
plan to consider land reversion in our final rule analysis and we seek
comments on methodologies and approaches for doing this. We also seek
comment on the related issue of how best to estimate how long each type
of biofuel is most likely to continue to be produced, and whether
production of these biofuels is likely to end abruptly or phase out gradually.
    Agricultural and economic models that look beyond 2022 would not
only help to estimate the impacts of land reversion after biofuel
production ends, they would also help to project how evolving
agricultural conditions could influence the lifecycle GHG emissions of
biofuels beyond 2022. For example, corn yields per acre are expected to
continue increasing after 2022; this increase in yields per acre will
decrease the amount of land required to produce a bushel of corn. At
higher yields, fewer acres are required to grow the corn used for the
15 billion gallons of corn starch ethanol modeled for the rule. The
indirect impacts of maintaining 15 billion gallons of corn ethanol
production would similarly be reduced. EPA intends to more carefully
model these transitions in particular to better account for future land
use impacts and we invite comments on methodology, sources of data,
factors that should be considered in assessing whether and when a
particular biofuel such as ethanol from corn starch, for example, will
no longer be produced and recommendations on how to improve on our
assessment of the likely stream of GHG emissions after 2022 that will
result from the EISA mandates.
    A complicating consideration in this analysis arises in determining
future use of the land (post-biofuel production) is the fact that
perhaps significant land use change occurred as a result of biofuel
production and that land is now more easily suited for alternative uses
compared to its pre-biofuel state. For example, the demand created by
biofuel production may have justified clearing forested lands and
turning them into productive cropland. Even if the need for the land to
produce crops in response to biofuel demand ceases when the biofuel
production ends, the land will still be in an altered form making it,
for example, more economically available for other crop production than
when it had been forested. How this land is subsequently used can
affect its impact on GHG emissions. If it is used for intensive crop
production, the land will have a much different carbon sequestration
profile, for example, than if it returned to its pre-biofuel forested
state. EPA asks for suggestions on how to best treat these lingering
effects of land use change when attributing the effects of biofuel
demand to uses of land even after biofuel production ends.
    For the determination of whether biofuels meet the GHG emissions
reduction required by EISA, we present the results for a range of time
periods, including both 100 years and 30 years in Section VI.C and
specifically invite comment on whether use of a 100 year time frame, a
30 year time frame, or some other time frame, would be most appropriate.
    In addition to this general issue of the appropriate time frames
for analysis, several more specific issues exist. Since it would be
likely that corn starch ethanol production will phase out gradually
rather than stopping all of a sudden in 2045, we also are evaluating
options for estimating the phase out of corn starch ethanol production.
One simplifying assumption would have corn ethanol production phase out

[[Page 25037]]

linearly between 2022 and 2045 as production of other biofuels such as
cellulosic biofuels continue to expand. Comments are requested on the
option of linearly phasing out corn ethanol production from 2022
through 2045 and other approaches for estimating this transition in
corn ethanol production. Finally, its not only corn starch ethanol that
might be replaced in future years. For example, the use of soy oil for
biodiesel fuel production might be replaced by other non-food oils such
as oil from algae. Comments are requested on whether other biofuels
will similarly phase out of use and therefore the land use change
impacts need to be similarly considered.
    In addition to seeking comments on all of the issues related to the
time periods for lifecycle analysis, EPA plans to convene a peer review
of the range of time periods considered in this proposed rule. This
peer review will also seek expert feedback on all of the issues raised
above in this section, including how to determine the most appropriate
time periods for consideration in the final rule.

Discounting of Lifecycle GHG Emissions

    Economic theory suggests that in general consumers have a time
preference for benefits received today versus receiving them in the
future. Therefore, future benefits are often valued at a discounted
rate. Although discount rates are most often applied to the monetary
valuation of future versus present benefits, a discounting approach can
also be used to compare physical quantities (i.e., total GHG emissions
per gallon of fuel used).
    The concept of weighting physical units accruing at different times
has been used in the environmental and resource economics
literature,\294\ and is analogous to valuing the monetary cost and
benefits of a policy, only that in this case the metric that we `value'
is the reduction in GHG emissions. \295\ An important part of the
economic theory of time is the idea that benefits expected to accrue in
the long term are less certain than benefits in the near term. This is
true in the case of GHG emissions changes from biofuel production which
are dependent upon how long biofuel production will continue, how
technologies will develop over time, and other factors. Another reason
to give more weight to near-term emissions changes is that the risks
associated with climate change in the future include the possibility of
extreme climate change and threshold impacts (e.g., species and
ecosystem thresholds, catastrophic events). Increased GHG emissions in
the near-term may be more important in terms of physical damage to the
world's environment. Some scientists, for example, believe that effects
on factors such as arctic summer ice, Himalayan-Tibetan Glaciers, and
the Greenland ice sheet are more likely to be effected by near-term GHG
emissions, causing non-linearities in the effects attributable to GHG
emissions.\296\ Long-term GHG reductions may be too late to mitigate
these irreversible impacts, providing further justification for
discounting GHG emissions changes that are expected in the distant
future. Under this perspective, it would be appropriate to discount the
physical quantities of future emissions, and especially in a long term
analysis of lifecycle GHG emissions. Thus in our analysis with a 100
year time frame, or impact horizon, we discount the value of future GHG
emissions changes.
---------------------------------------------------------------------------

    \294\ Herzog et al. 2003 (See http://sequestration.mit.edu/pdf/
climatic_change.pdf ), Richards 1997, Stavins and Richards 2005 (See
http://www.pewclimate.org/docUploads/Sequest_Final.pdf ).
    \295\ Sunstein and Rowell, 2007, On Discounting Regulatory
Benefits: Risk, Money, and Intergenerational Equity, Chicago Law Review.
    \296\ Ramanathan and Feng, 2008. On avoiding dangerous
anthropogenic interference with the climate system: Formidable
challenges ahead. Proceedings of the National Academy of Sciences
105:143245-14250.
---------------------------------------------------------------------------

    Despite the rationale for discounting future GHG emissions changes
discussed above, there are reasons to be cautious about the application
of discounting in lifecycle GHG analysis. One argument is that it may
only be appropriate to discount monetized values. Our lifecycle
analysis estimates GHG emission impacts, not their monetary value, and
under this argument emissions should not be directly discounted.
Rather, the physical GHG emissions should be converted into monetary
impacts, where these monetary impacts are also a function of climate
science. The resulting climate impacts would then have to be translated
into monetary values. This presents significant challenges for
lifecycle GHG analysis because it is difficult to translate dynamic GHG
emissions into a single estimate of physical impacts, much less a
single estimate of monetized impacts. This is the case for a number of
reasons, including the complex physical systems associated with climate
change (e.g., the relationship between atmospheric degradation rates
with atmospheric carbon stocks) that may create non-constant marginal
damages from GHG emissions over time. Furthermore, converting lifecycle
GHG emissions into monetized impacts may be inconsistent with the EISA
definition of lifecycle GHG emissions provided above in Section VI.A.1,
which stipulates that lifecycle GHG emissions are the ``aggregate
quantity of greenhouse gas emissions * * * where the mass values for
all greenhouse gases are adjusted to account for their relative global
warming potential.''
    Another argument against discounting GHG emissions changes is the
concept of inter-generational equity, which argues that benefits or
damages affecting future generations merit just as much weight as
impacts felt by current generations. It is argued that this would
invalidate the practice of discounting emissions impacts that could
affect future generations.
    Finally, earlier in this section we discussed the possible ranges
of time frames for analyzing the GHG emissions impacts. For shorter
time frames such as 30 years, there would be less uncertainty in the
emissions stream so the benefit of discounting to address uncertainty
is also lessoned.
    Comments are requested on the concept of discounting a stream of
GHG emissions for the purpose of estimating lifecycle GHG emissions
from transportation fuels as specified in EISA.

Appropriate Level of Discount Rate

    As described in more detail in Section IX on GHG emission reduction
benefits, GHG emissions have primarily consumption effects and inter-
generational impacts, as changes in GHG emissions today will continue
to have impacts on climate change for decades to centuries. If a
discount rate is applied to future GHG emissions, an appropriate
discount rate should be based on a consumption-based discount rate
given that monetized climate change impacts are primarily consumption
effects (i.e., impacts on household purchases of goods and services). A
consumption-based discount rate reflects the implied tradeoffs between
consumption today and in the future. Discount rates of 3% or less are
considered appropriate for discounting climate change impacts, since
they reflect the long run uncertainty in economic growth and interest
rates and the risk of high impact climate damages that could reduce
economic growth.\297\
---------------------------------------------------------------------------

    \297\ Technical Support Document on Benefits of Reducing GHG
Emissions, U.S. Environmental Protection Agency, June 12, 2008,
www.regulations.gov (search phrase ``Technical Support Document on
Benefits of Reducing GHG Emissions'').

---------------------------------------------------------------------------

[[Page 25038]]

    When analyzing the GHG emissions associated with a 100 year time
period, we examined a variety of alternative discount rates (e.g., 0,
2, 3, 7 percent) to show the sensitivity of greenhouse gas emissions
estimates to the choice of the discount rate. A zero discount rate
estimates GHG emission impacts as if each ton of GHG emissions is
treated equally through time. Previous methodologies of lifecycle GHG
benefits have presented results using a zero discount rate.\298\
However, some of the climate change literature supports using a higher
discount rate, as described in Section IX.C. We show the 7% discount
rate for illustrative purposes; however climate change benefit analyses
from global long-run growth models typically use discount rates well
under 7% for standard analysis.\299\ High discount rates imply very low
values for the future GHG emission impacts resulting from today's
actions on the welfare of future generations. Therefore, lower discount
rates such as 2-3% are considered more appropriate for discounting long
term climate change impacts.\300\
---------------------------------------------------------------------------

    \298\ Searchinger et al., 2008.
    \299\ Tol, 2005.
    \300\ Newell and Pizer, 2003.
---------------------------------------------------------------------------

    In the analysis for this proposal we use a 2% discount rate to
assess the present value of GHG emissions changes which occur over a
100 year time frame. This discount rate is consistent with the Office
of Management and Budget (OMB) \301\ and EPA \302\ guidance and is one
of the discount rates that has been used in the literature to monetize
the impacts of climate change.\303\ EPA has considered this issue
previously, and after weighing the pros and cons of different values,
set forth a guidance document recommending using a range of consumption
based discount rates of 0.5-3%. OMB and EPA guidance on inter-
generational discounting suggests using a low but positive discount
rate if there are important inter-generational benefits and costs. In
selecting a 2% discount rate coupled with a 100 year emission stream
estimate, EPA would be recognizing the long term nature of the emission
impacts of biofuel production, the uncertainty in estimating these
emission impacts and their consequences plus the significance of nearer
term emission changes in avoiding future consequences. Other options
for intergenerational discounting have been discussed in the economic
literature, such as dealing with uncertainty by using a non-constant,
declining, or negative discount rate.\304\ Comments could consider how
discounting appropriately reflects the uneven emission of greenhouse
gases from biofuels over time, the uncertainty in predicting emissions
in more distant futures and the impacts these emissions could have on
climate change. Alternative approaches for inter-generational
discounting are described in Chapter 5.3 of the DRIA.
---------------------------------------------------------------------------

    \301\ OMB Circular A-4, 2003 provides a range of 1-3% for
consumption based discount rates.
    \302\ EPA Guidelines for Preparing Economic Analyses, 2000.
    \303\ Tol (2005, 2007).
    \304\ Newell and Pizer, 2003, Weitzman (1999, 2001), Nordhaus
(2008), Guo et al., (2006), Saez, C.A. and J.C. Requena,
``Reconciling sustainability and discounting in Cost-Benefit
Analysis: A methodological proposal'', Ecological Economics, 2007,
vol. 60, issue 4, pages 712-725.
---------------------------------------------------------------------------

    Because we are considering not discounting GHG emissions and in
particular since the justifications for discounting physical emissions
are not as strong for shorter time periods, in Section VI.C.2, we also
present the GHG emissions reductions associated with biofuels using a
30 year time period and no discount rate. Using a zero percent or no
discount rate implies that all emission releases and uptakes during
this time period are valued equally. For a shorter time period such as
thirty years, we are relatively certain of the emission trends.
Furthermore, all of these emissions occur in a relatively short period
of time so their impact on climate change and the consequences of that
climate change could all be considered the same regardless of whether
those emissions occurred early or late in this 30-year time period.
    We specifically invite comment on our use of a 2% discount rate
with a 100 year time period for analysis of lifecycle GHG emissions,
and our use of no discount rate in our analysis of GHG emissions over
30 years. We also invite comments on whether using physical science
metrics such as the actual quantities of climate forcing gasses in the
atmosphere, actual quantities of climate forcing gasses in the
atmosphere weighted by global warming potential (GWP), or cumulative
radiative forcing should be used to evaluate emissions over time.
Specifically, we seek comment on an approach for comparing GHG
emissions based on the time profile of the greenhouse gas emissions in
the atmosphere, and whether this approach would be consistent with the
legal definition of lifecycle GHG emissions in EISA. One such method is
the Fuel Warming Potential as outlined in a memo to the EPA from the
Union of Concerned Scientists which is available on the public docket
for this rulemaking.\305\ This approach is based on the ratio of the
cumulative radiative forcing between the biofuel and the gasoline case
over a specified time frame.
---------------------------------------------------------------------------

    \305\ See Memo to EPA, Office of Transportation and Air Quality
from Union of Concerned Scientists, Re: Treatment of Time in Life
Cycle Accounting, February 18, 2009.
---------------------------------------------------------------------------

    The EISA definition of lifecycle GHG emissions stipulates that the
mass values for all greenhouse gas emissions shall be adjusted to
account for their relative GWP. We are proposing to use the standard
100-year GWP's published in the IPCC Second Assessment
Report.306 307 We invite comment on whether it is
appropriate to discount GWP-weighted emissions and how such discounting
might appropriately apply across the several greenhouse gases.
---------------------------------------------------------------------------

    \306\ See http://www.ipcc.ch/ipccreports/assessments-
reports.htm. 
    \307\ O'Hare, Plevin, Martin, Jones, Kendal and Hopson; ``Proper
accounting for time increases crop-based biofuel's greenhouse gas
deficit versus petroleum''; Environmental Research Letters, 4 (2009)
024001.
---------------------------------------------------------------------------

    Furthermore, if alternative time periods for the production of
biofuels and the GHG impacts of biofuel production are considered as
discussed above, and the choice is made to discount GHG emissions, the
question that arises is: What discount rate or combination of discount
rates should be considered? For example, if ethanol production is
assumed to occur for 30 years and the GHG impacts are assumed to span
across 80-100 years, should a single discount rate be applied to the
emissions stream or alternative discount rates based upon the different
time frames? EPA is taking comment on whether and how to apply
discounting when different time frames between the production and long-
run GHG impacts are utilized to analysis biofuels. Specifically, EPA is
considering and requests comment on the option of using either no
discount rate or a 3% discount rate to assess those emissions that
occur during the relatively shorter time frame for biofuel use which
could phase out within 30 years as in our corn ethanol example and a 2%
discount rate over the reminder of the 100 years in assessing the
longer term GHG emissions impacts resulting from land use changes
related to biofuel production (including land reversion considerations).
    EPA is considering a range of discount rates including zero or no
discounting for reasons as described above and requests comments on the
appropriate discount rate to use when assessing the stream of GHG
emission changes that are likely to result from biofuel production and
use. Other

[[Page 25039]]

options for intergenerational discounting have been discussed in the
economic literature, such as dealing with uncertainty by using a non-
constant, declining, or negative discount rate.\308\ Comments could
consider how discounting appropriately reflects the uneven release of
greenhouse gases from biofuels over time, the uncertainty in predicting
emissions in more distant futures and the impacts these emissions could
have on climate change. Alternative approaches for inter-generational
discounting are described in Chapter 5.3 of the DRIA.
---------------------------------------------------------------------------

    \308\ Newell and Pizer, 2003, Weitzman (1999, 2001), Nordhaus
(2008), Guo et al., (2006), Saez, C.A. and J.C. Requena,
``Reconciling sustainability and discounting in Cost-Benefit
Analysis: A methodological proposal'', Ecological Economics, 2007,
vol. 60, issue 4, pages 712-725.
---------------------------------------------------------------------------

    EPA recognizes that the time horizon for analysis and the treatment
of future emissions including the appropriateness of applying discount
factors are key factors in determining biofuel lifecycle GHG impacts;
therefore, we plan to organize an expert peer review of these issues
before the final rule.
c. Feedstock Transport
    The GHG impacts of transporting corn from the field to the ethanol
facility and transporting the co-product DGs from the ethanol facility
to the point of use were included in this analysis. The GREET default
of truck transportation of 50 miles was used to represent corn
transportation from farm to plant. Transportation assumptions for DGs
transport were 14% shipped by rail 800 miles, 2% shipped by barge 520
miles, and 86% shipped by truck 50 miles. The percent shipped by mode
was from data provided by USDA and based on Association of American
Railroads, Army Corps of Engineers, Commodity Freight Statistics, and
industry estimates. The distances DGs were shipped were based on GREET
defaults for other commodities shipped by those transportation modes.
The GHG emissions from transport of corn and DGs are based on GREET
default emission factors for each type of vehicle including capacity,
fuel economy, and type of fuel used. Similar detailed analyses were
conducted for the transport of cellulosic biofuel feedstock and
biomass-based diesel feedstock.
    As part of this rulemaking analysis we have conducted a more
detailed analysis of biofuel production locations and transportation
distances and modes of transport used in the criteria pollutant
emissions inventory calculations described in DRIA Chapter 1.6 and for
the cost analysis of this rule described in DRIA Chapter 4.2. Given the
timing of when the current analysis was completed we were not able to
incorporate this updated transportation information into our lifecycle
analysis but plan to do that for the final rule.
    Furthermore, the transportation modes and distances assumed for
corn and DGs do not account for the secondary or indirect
transportation impacts. For example, decreases in exports might reduce
overall domestic agricultural commodity transport and emissions but
might increase transportation of commodities internationally. We plan
to consider these secondary transportation impacts in our final rule analysis.
d. Processing
    The GHG emissions estimates associated with the processing of
renewable fuels is dependent on a number of assumptions and varies
significantly based on a number of key variables. The ethanol yield
impacts the total amount of corn used and associated agricultural
sector GHG emissions. The amount of DGs and other co-products produced
will also impact the agricultural sector emissions in terms of being
used as a feed replacement. Finally the energy used by the ethanol
plant will result in GHG emissions, both from producing the fuel used
and through direct combustion emissions.
    As mentioned above, in traditional lifecycle analyses, the energy
consumed and emissions generated by a renewable fuel plant must be
allocated not only to the renewable fuel, but also to each of the by-
products. For corn ethanol production, our analysis avoids the need to
allocate by accounting for the DGs and other co-products directly in
the FASOM and FAPRI agricultural sector modeling described above. DGs
are considered a partial replacement for corn and other animal feed and
thus reduce the need to make up for the corn production that went into
ethanol production. Since FASOM takes the benefits from the production
and use of DGs into account (e.g., displacing the need to grow
additional crops for feed and therefore reducing GHG emissions in the
agricultural sector), no further allocation was needed at the ethanol
plant and all plant emissions are accounted for here.
    In terms of the energy used at renewable fuel facilities, there is
a lot of variation between plants based on the process type (e.g., wet
vs. dry milling) and the type of fuel used (e.g., coal vs. natural
gas). There can also be variation between the same type of plants using
the same fuel source based on the age of the plant and types of
processes included, etc. For our analysis we considered different
pathways for corn ethanol production. Our focus was to differentiate
between facilities based on the key differences between plants, namely
the type of plant and the type of fuel used. One other key difference
we modeled between plants was the treatment of the co-products DGs. One
of the main energy drivers of ethanol production is drying of the DGs.
Plants that are co-located with feedlots have the ability to provide
the co-product without drying. This has a big enough impact on overall
results that we defined a specific category for wet vs. dry co-product.
One additional factor that appears to have a significant impact on GHG
emissions is corn oil fractionation from DGs. Therefore, this category
is also broken out as a separate category in the following section. See
DRIA Chapter 1.4 for a discussion of corn oil fractionation.
    Furthermore, as our analysis was based on a future timeframe, we
modeled future plant energy use to represent plants that would be built
to meet requirements of increased ethanol production, as opposed to
current or historic data on energy used in ethanol production. The
energy use at dry mill plants was based on ASPEN models developed by
USDA and updated to reflect changes in technology out to 2022 as
described in DRIA Chapter 4.1.
    The GHG emissions from renewable fuel production are calculated by
multiplying the Btus of the different types of energy inputs by
emissions factors for combustion of those fuel sources. The emission
factors for the different fuel types are from GREET and are based
primarily on assumed carbon contents of the different process fuels.
The emissions from producing electricity are also taken from GREET and
represent average U.S. grid electricity production emissions. The
emissions from combustion of biomass fuel source are not assumed to
increase net atmospheric CO₂ levels the CO₂
emitted from biomass-based fuels combustion does not increase
atmospheric CO₂ concentrations, assuming the biogenic carbon
emitted is offset by the uptake of CO₂ resulting from the
growth of new biomass. Therefore, CO₂ emissions from biomass
combustion as a process fuel source are not included in the lifecycle
GHG inventory of the ethanol plant.
e. Fuel Transport
    Transportation and distribution of ethanol, biomass-based diesel,
petroleum diesel and gasoline were also included in this analysis based
on GREET defaults. The GREET defaults for

[[Page 25040]]

both ethanol and gasoline transport from plant/refinery to bulk
terminals were used. The GREET defaults for both ethanol and gasoline
distribution from the bulk terminal to the service station were also used.
    As with feedstock transport we have conducted a more detailed
analysis of fuel transport and distribution impacts for use in criteria
pollutant inventories (see DRIA Chapter 1.6) and for our cost analysis
described in DRIA Chapter 4.2. Due to the timing of this analysis we
were not able to incorporate the results in our proposed lifecycle
calculation but plan to do that for the final rule.
f. Tailpipe Combustion
    Combustion CO₂ emissions for ethanol, biomass-based
diesel, petroleum diesel and gasoline were based on the carbon content
of the fuel. However, over the full lifecycle of the fuel, the
CO₂ emitted from biomass-based fuels combustion does not
increase atmospheric CO₂ concentrations, assuming the
biogenic carbon emitted is offset by the uptake of CO₂
resulting from the growth of new biomass. As a result, CO₂
emissions from biomass-based fuels combustion are not included in their
lifecycle emissions results. Net carbon fluxes from changes in biogenic
carbon reservoirs in wooded or crop lands are accounted for separately
in the land use change analysis as outlined in the agricultural sector
modeling above.
    When calculating combustion GHG emissions, however, the methane and
N₂O emitted during biomass-based fuels combustion are
included in the analysis. Unlike CO₂ emissions, the
combustion of biomass-based fuels does result in net additions of
methane and N₂O to the atmosphere. Therefore, combustion
methane and N₂O emissions are included in the lifecycle GHG
emissions results for biomass-based fuels.
    Combustion related methane and N₂O emissions for both
biomass-based fuels and petroleum-based fuels are based on EPA MOVES
model results.
6. Petroleum Baseline
    To establish the lifecycle greenhouse gas emissions associated with
the petroleum baseline against which the renewable fuels were compared,
we used an updated version of the GREET model. Lifecycle energy use and
associated emissions for petroleum-based fuels in GREET is calculated
based on an energy efficiency metric for the different processes
involved with petroleum-based fuels production. The energy efficiency
metric is a measure of how many Btus of input energy are needed to make
a Btu of product. GREET has assumptions on energy efficiency for
different finished petroleum products as well as for different types of
crude oil.
    We are using the latest version of the GREET model for this
analysis (Version 1.8b) which includes recent updates to the energy
efficiencies of petroleum refining. To represent baseline petroleum
fuels we have used the 2005 estimates of actual gasoline and diesel
fuel used. For 2005, 86% of gasoline and 92% of diesel fuel was
produced domestically with the rest imported finished product. To
represent international production we assume the same GHG refinery
emissions from GREET as used domestically. We did not include indirect
land use impacts in assessing the lifecycle GHG performance of the 2005
baseline fuel pool as we believe these would insignificantly impact the
average performance assessment of the baseline. Additionally,
consistent with our assessment of energy security impacts, we did not
include as an indirect GHG impact the potential impact of maintaining a
military presence.
    GREET also has assumptions on the mix of energy sources used to
provide the energy input, which determine GHG emissions. For example if
coal, natural gas, or purchased electricity is used as an energy
source. The GHG emissions associated with petroleum fuel production are
based on the emissions from producing and combusting the input energy
sources needed, like GHG emissions from using natural gas at the
petroleum refinery. Non-combustion GHG sources like fugitive methane
emissions are added in where applicable.
    Based on the EISA requirements, we used the 2005 mix of crude as
the petroleum baseline. We developed emissions factors for those crude
types since they are not currently included in GREET. In 2005, 5% of
crude was Canadian tar sand, 1% was Venezuela extra heavy, and 23% was
heavy crude.
    For this proposal, we are using the average GHG emissions
associated with the 2005 petroleum baseline, as required by EISA.
However, we recognize that an additional gallon of renewable fuel
replaces the marginal gallon of petroleum fuel. To the extent that the
marginal gallon is from oil sands or other types of crude oil that are
associated with higher than average GHG emissions, replacing these
fuels could have a larger GHG benefit. Conversely to the extent the
marginal gallon displaced is from imported gasoline produced from light
crude, replacing these fuels would have a smaller GHG benefit. We
solicit comment on whether--strictly for purposes of assessing the
benefits of the rule (and not for purposes of determining whether
certain renewable fuel pathways meet the GHG reduction thresholds set
forth in EISA), we should assess benefits based on a marginal
displacement approach and, if so, what assumptions we should use for
the marginal displacements.
    In December 2008, the U.S. Department of Energy's National Energy
Technology Laboratory (NETL) released a report that estimates the
average lifecycle GHG emissions from petroleum-based fuels sold or
distributed in 2005.\309\ The estimates in the report are based on a
slightly different methodology than EPA's analysis of lifecycle GHG
emissions for the petroleum baseline. The NETL report is available on
the docket for this rulemaking. We invite comments on whether NETL's
analysis has significant implications for how EPA is estimating
petroleum baseline lifecycle GHG emissions.
---------------------------------------------------------------------------

    \309\ DOE/NETL. 2008. Development of Baseline Data and Analysis
of Life Cycle Greenhouse Gas Emissions of Petroleum-Based Fuels.
DOE/NETL-2009/1346.
---------------------------------------------------------------------------

7. Energy Sector Indirect Impacts
    Increased demand for natural gas to power corn ethanol plants could
have additional impacts on the U.S. energy sector. As demand for
natural gas increases, the use of natural gas in other sectors (e.g.,
electric generation) could decrease. For this analysis, we are using
the NEMS model to project the secondary or indirect impacts on the
energy sector. However, we were not able to include this analysis in
the GHG emissions estimates presented in this proposal. We hope to have
this analysis for the final rule. Additional details on the methodology
are included in the DRIA Chapter 2, and we invite comments on this approach.
    We are assuming, for the proposal, that a gallon of renewable fuel
replaces an energy equivalent gallon of petroleum fuel. This analysis
presumes that petroleum-based fuels as they are currently produced will
continue to be used for transportation fuels and will be replaced on a
Btu for Btu basis. Many factors could affect this assumption including
advances in petroleum fuel technology, availability of other fossil
fuels for transportation use, and of course the supply and cost of
petroleum. We have not tried to analyze these potential impacts in this
rule. However we invite comment on such an approach.
    We have also not assessed whether expanded use of biofuels in the U.S.

[[Page 25041]]

will impact the energy markets in other countries. For example,
reducing demand for petroleum-based fuel in the U.S. may reduce
worldwide petroleum prices and impact the use of petroleum in other
countries. We invite comment on how best to assess these potential
impacts and will attempt to do so for the final rule.

C. Fuel Specific GHG Emissions Estimates

    While the results presented in this section represent the most up-
to-date information currently available, this analysis is part of an
ongoing process. Because lifecycle analysis is a new part of the RFS
program, in addition to the formal comment period on the proposed rule,
EPA is making multiple efforts to solicit public and expert feedback on
our proposed approach. As discussed in Section XI, EPA plans to hold a
public workshop focused specifically on lifecycle analysis during the
comment period to assure full understanding of the analyses conducted,
the issues addressed and options that should be considered. We expect
that this workshop will allow the most thoughtful and useful comments
to this proposal and assure the best methodology and assumptions are
used for calculating GHG emissions impacts of fuels for the final rule.
Additionally we will conduct peer-reviews of key components of our
analysis. As part of ongoing analysis for the final rule, EPA will seek
peer review of: Our use of satellite data to project future land use
changes; the land conversion GHG emissions factors estimated by
Winrock; our estimates of GHG emissions from foreign crop production;
methods to account for the variable timing of GHG emissions; and how
models are used together to provide overall lifecycle GHG estimates.
    In addition to the refinements to the methodology that we plan to
undertake for the final rule, we also intend to update our results
periodically. EPA recognizes that the state of the science for
lifecycle GHG analysis will continue to evolve over time as new data
and modeling techniques become available and as there are improvements
in agricultural and renewable fuel production practices as well as new
feedstocks. We invite comments on the appropriate amount of time for
periodic review of the lifecycle assessment methodology, but we propose
that performing an update of the methodology every 3-5 years would be
appropriate. We would expect the first update to this analysis would
occur closer to 3 years. This timeframe would allow us to undergo a
formal review process after the final rule to ensure that this
methodology takes into account the most state-of-the-art science and
reflects the input of appropriate experts in this field. However, any
change in lifecycle methodology as contemplated here would not affect
the eligibility of biofuels produced at facilities covered by the
grandfathering provisions of EISA at section 211(o)(4)(g).
1. Greenhouse Gas Emissions Reductions Relative to the 2005 Petroleum Baseline
    In this section we present detailed lifecycle GHG results for
several specific biofuels representing a range biofuel pathways. This
section also includes the results of sensitivity analysis for key
variables. The sensitivity of the time period and discount rate are
discussed below. In the rest of this section we focus on two sets of
lifecycle GHG results. One set of results that uses a 100 year time
period and 2% discount rate and a parallel set of results using a 30
year time period and a 0% discount rate. In Section IV.C.2 which
follows, we also present the results for some additional combinations
of time horizon for assessing GHG emission changes as well as assuming
other discount rates. Additional pathways, not included in the results
presented in this section, distinguishing other combinations of
feedstock and processing technologies have been evaluated. These
additional pathways are described in detail in the DRIA and are
included in these proposed regulations.
a. Corn Ethanol
    Table VI.C.1-1 presents the breakout of the net present value of
lifecycle GHG emissions per million British thermal unit (mmbtu) of
corn ethanol and gasoline. The results are broken out by lifecycle
stage. Values are shown for a standard dry mill corn ethanol plant in
2022 using natural gas for process energy and drying the co-product of
distillers grains (DGs). Results indicate where the major contributions
of GHG emissions are across the fuel lifecycle. Fuel processing and
indirect land use change are the main contributors to corn ethanol
lifecycle GHG emissions. Net domestic and international agricultural
impacts (w/o land use change) include direct and indirect impacts, such
as reductions in livestock enteric fermentation.
---------------------------------------------------------------------------

    \310\ For this proposal, our preliminary analysis suggests land
use impacts of petroleum production for the fuels used in the U.S.
in 2005 would not have an appreciable impact on the 2005 baseline
GHG emissions assessment. However, we expect to more carefully
consider potential land use impacts of petroleum-based fuel
production for the final rule and invite comment and information
that would support such an analysis.
    \311\ 2005 petroleum baseline fuel production includes crude oil
extraction, transportation, refining, and transport of finished product.
    \312\ Ethanol tailpipe emissions include CH₄ and
N₂O emissions but not CO₂ emissions as these
are assumed to be offset by feedstock carbon uptake.

        Table VI.C.1-1--Absolute Lifecycle GHG Emissions for Corn Ethanol and the 2005 Petroleum Baseline
                                                 [CO2-eq/mmBtu]
----------------------------------------------------------------------------------------------------------------
                                                                    Natural gas                     Natural gas
                                                   2005 Gasoline   dry mill with   2005 Gasoline   dry mill with
                                                     baseline         dry DGs        baseline         dry DGs
----------------------------------------------------------------------------------------------------------------
                 Lifecycle Stage                             100 yr 2%
                                                             30 yr 0%
----------------------------------------------------------------------------------------------------------------
Net Domestic Agriculture (w/o land use change)..             N/A        -499,029             N/A        -347,365
Net International Agriculture (w/o land use                  N/A         452,118             N/A         314,711
 change)........................................
Domestic Land Use Change........................             N/A          79,547             N/A          92,575
International Land Use Change...................       N/A \310\       1,911,391             N/A       1,910,822
Fuel Production \311\...........................         823,262       1,404,083         573,058         977,358
Fuel and Feedstock Transport....................   (see footnote         174,327  ..............         121,346
                                                            321)
Tailpipe Emissions \312\........................       3,417,311          37,927       2,378,800          26,400
¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤
    Net Total Emissions.........................       4,240,674       3,560,365       2,951,858       3,095,846
----------------------------------------------------------------------------------------------------------------

[[Page 25042]]

    Table VI.C.1-1 demonstrates the importance of the discount rate and
time period analyzed as well as the importance of significance of
including GHG emissions from international land use changes. Assuming
100 years of corn ethanol produced in a basic dry mill ethanol
production facility and using a 2% discount rate results in corn
ethanol having a 16% reduction in GHG emissions compared to the 2005
baseline gasoline assumed to be replaced. In contrast, assuming 30
years of corn ethanol production and use and no discounting of the GHG
emission impacts results in predicting that corn ethanol will have a 5%
increase in GHG emissions compared to petroleum gasoline.
    As discussed in Section VI.B.2.a, EPA's interpretation of the EISA
statute compels us to include significant indirect emission impacts
including those due to land use changes in other countries. The data in
Table VI.C.1-1 indicate that excluding the international land use
change would result in corn ethanol having an approximately 60%
reduction in lifecycle GHG emissions compared to petroleum gasoline
regardless of the timing or discount rate used.\313\
---------------------------------------------------------------------------

    \313\ The treatment of emissions over time is not critical if
international land use change emissions are excluded because the
results without land use change are consistent over time. Therefore
the overall lifecycle GHG results do not vary with time or discount
rate assumptions.
---------------------------------------------------------------------------

    In Table VI.C.1-1, we project a standard dry mill ethanol plant in
2022 using corn as its feedstock, using natural gas for process energy,
and drying the co-product of distillers grains (DGs). Different corn
ethanol production technologies will have different lifecycle GHG
results. For example, due to its high carbon content, using coal as the
process energy source significantly worsens the lifecycle GHG impact of
ethanol produced at such a facility. On the other hand, replacing
natural gas with renewable biomass as the process energy source greatly
improves the GHG assessment.
    Other technology options are available to improve the efficiency of
ethanol facilities. Table VI.C.1-2 shows the impact that different corn
ethanol production process pathways will have on the overall lifecycle
GHG results. Table VI.C.2-2 shows that currently available technologies
could be applied to corn ethanol plants to reduce their net GHG emissions.
    For example, a combined heat and power (CHP) configuration, used in
combination with corn oil fractionation, would result in a GHG
emissions reduction of 27% relative to the 2005 petroleum baseline over
100 years using a 2% discount rate, and a 6% reduction over 30 years
with no discounting. In addition, advanced technologies such as
membrane separation and raw starch hydrolysis could improve the
emissions associated with corn ethanol production even more
substantially. Combining all of these technologies in a state-of-the-
art natural gas powered corn ethanol facility would produce ethanol
that has approximately 35% less lifecycle GHG emissions than an energy
equivalent amount of baseline gasoline displaced over 100 years using a
2% discount rate and, by comparison a 14% reduction when accounting for
30 years of emission changes but applying no discounting. Details on
these different technologies are included in the DRIA Chapter 1.5.
    Table VI.C.1-2 also shows that the choice of drying DGs can have a
significant impact on the GHG emissions associated with an ethanol
plan, since drying the ethanol byproduct is an energy intensive
process. However, wet DGs are only suitable where a local market is
available such as a dairy farm or cattle feedlot, since wet DGs are
highly perishable.

Table VI.C.1-2--Lifecycle GHG Emissions Changes for Various Corn Ethanol
        Pathways in 2022 Relative to the 2005 Petroleum Baseline
------------------------------------------------------------------------
                                          Percent change
                                             from 2005    Percent change
   Corn ethanol production plant type        petroleum       from 2005
                                           baseline (100   baseline (30
                                              yr 2%)          yr 0%)
------------------------------------------------------------------------
Natural Gas Dry Mill with dry DGs.......             -16              +5
Natural Gas Dry Mill with dry DGs and                -19              +2
 CHP....................................
Natural Gas Dry Mill with dry DGs, CHP,              -27              -6
 and Corn Oil Fractionation.............
Natural Gas Dry Mill with dry DGs, CHP,              -30             -10
 Corn Oil Fractionation, and Membrane
 Separation.............................
Natural Gas Dry Mill with dry DGs, CHP,              -35             -14
 Corn Oil Fractionation, Membrane
 Separation, and Raw Starch Hydrolysis..
Natural Gas Dry Mill with wet DGs.......             -27              -6
Natural Gas Dry Mill with wet DGs and                -30              -9
 CHP....................................
Natural Gas Dry Mill with wet DGs, CHP,              -33             -12
 and Corn Oil Fractionation.............
Natural Gas Dry Mill with wet DGs, CHP,              -36             -15
 Corn Oil Fractionation, and Membrane
 Separation.............................
Natural Gas Dry Mill with wet DGs, CHP,              -39             -18
 Corn Oil Fractionation, Membrane
 Separation, and Raw Starch Hydrolysis..
Coal Fired Dry Mill with dry DGs........             +13             +34
Coal Fired Dry Mill with dry DGs and CHP             +10             +31
Coal Fired Dry Mill with dry DGs, CHP,                -5             +15
 and Corn Oil Fractionation.............
Coal Fired Dry Mill with dry DGs, CHP,               -13              +8
 Corn Oil Fractionation, and Membrane
 Separation.............................
Coal Fired Dry Mill with dry DGs, CHP,               -21              -1
 Corn Oil Fractionation, Membrane
 Separation, and Raw Starch Hydrolysis..
Coal Fired Dry Mill with wet DGs........              -9             +12
Coal Fired Dry Mill with wet DGs and CHP             -11             +10
Coal Fired Dry Mill with wet DGs, CHP,               -17              +3
 and Corn Oil Fractionation.............
Coal Fired Dry Mill with wet DGs, CHP,               -25              -4
 Corn Oil Fractionation, and Membrane
 Separation.............................
Coal Fired Dry Mill with wet DGs, CHP,               -30              -9
 Corn Oil Fractionation, Membrane
 Separation, and Raw Starch Hydrolysis..
Biomass Fired Dry Mill with dry DGs.....             -39             -18
Biomass Fired Dry Mill with wet DGs.....             -40             -19
Natural Gas Fired Wet Mill..............              -7             +14

[[Page 25043]]

Coal Fired Wet Mill.....................             +20             +41
Biomass Fired Wet Mill..................             -47             -26
------------------------------------------------------------------------

    As described in Sections VI.A and VI.B, there are a number of
parameters and modeling assumptions that could impact the overall
renewable fuel GHG results. The estimates in Table VI.C.1-2 are based
on the GHG emissions for a specific change in volumes analyzed in 2022
(12.3 to 15 Bgal). These volumes represent the change in corn ethanol
production that would occur in 2022 without and then with EISA mandates
in place. The GHG impact is then normalized to a per gallon or Btu
basis in relation to gasoline. These values are used to represent every
gallon of corn ethanol produced throughout the program.
    There are several important implications associated with this
methodology. First, this analysis focuses on the average impact of an
increase in fuel produced using a technology pathway and does not
distinguish the emission performance between biofuel production plants
using the same basic production technology and type of feedstock. Thus
it does not account for any incremental differences in facility design
or operation which may affect the lifecycle GHG performance at that
facility. Second, by focusing on 2022, this analysis does not track how
biofuel GHG emission performance may change over time between now and
2022. Third, the results presented here are based on the GHG impacts of
the volumes analyzed.
    For this proposal, we believe that using the emissions assessment
from a typical 2022 facility for each major technology pathway captures
the appropriate level of detail needed to determine whether a
particular biofuel meets the threshold requirements in EISA. To address
whether the GHG emissions vary significantly over time, we also
calculated corn ethanol lifecycle GHG emissions estimates in 2012 and
2017. As shown in Table VI.C.1-3, corn ethanol's lifecycle GHG
emissions reductions are fairly consistent regardless of which base
year is analyzed. This may be due to countervailing forces that
stabilize land use change emissions over the period of our analysis.
Crop yields increase over time (therefore reducing land use pressure),
but there is also increasing production of other renewable fuels that
require land for feedstock production (therefore increasing land use
pressure). Although we are proposing to use 2022 as the base year for
our lifecycle GHG emissions estimates, we invite comments on this
approach.

  Table VI.C.1-3--Corn Ethanol Lifecycle GHG Emissions Changes in 2012,
                             2017, and 2022
------------------------------------------------------------------------
                                          Percent change  Percent change
                                             from 2005       from 2005
          Scenario Description               petroleum       petroleum
                                           baseline (100   baseline (30
                                              yr 2%)          yr 0%)
------------------------------------------------------------------------
Corn Ethanol Natural Gas Dry Mill in                 -16              -3
 2012 with dry DGs......................
Corn Ethanol Natural Gas Dry Mill in                 -13              +9
 2017 with dry DGs......................
Corn Ethanol Natural Gas Dry Mill in                 -16              +5
 2022 with dry DGs......................
------------------------------------------------------------------------

    We also tested the impact of analyzing a larger change in corn
ethanol volumes on the GHG emissions estimates. Table VI.C.1-4 shows
the sensitivity of our analysis to the volume changes analyzed. Based
on this scenario, the GHG emissions estimates associated with a larger
change (6.3 Bgal) in corn ethanol volumes (8.7 Bgal to 15 Bgal) results
in lower GHG emission reductions. Additional details on these
sensitivity analyses are included in the DRIA Chapter 2.

 Table VI.C.1-4--Corn Ethanol Lifecycle GHG Emissions Changes Associated
                      With Different Volume Changes
------------------------------------------------------------------------
                                          Percent Change  Percent Change
                                             from 2005       from 2005
          Scenario Description               Petroleum       Petroleum
                                           Baseline (100   Baseline (30
                                              yr 2%)          yr 0%)
------------------------------------------------------------------------
Corn Ethanol Natural Gas Dry Mill in                 -16              +5
 2022 with dry DGs; 2.7 Bgal change in
 corn ethanol volumes...................
Corn Ethanol Natural Gas Dry Mill in                  -6             +14
 2022 with dry DGs; 6.3 Bgal change in
 corn ethanol volumes...................
------------------------------------------------------------------------

    The results presented in previous tables assume that managed
pasture (i.e., land actively used for livestock grazing) converted from
pasture to cropland would be replaced with new pasture in other areas.
The area of

[[Page 25044]]

managed pasture converted to cropland was estimated using satellite
data from Winrock and land cover data from GTAP. As a sensitivity
analysis, we also analyzed a scenario in which none of the pastureland
converted to cropland would be replaced if, for example, livestock
production could be more intensively developed on the remaining pasture
(see first row in Table VI.C.1-5). Similarly, we also calculated
results assuming that all pasture acres would be replaced (second row
in Table VI.C.1-5). Finally, the third row of Table VI.C.1-5 includes
lifecycle GHG results assuming that all of the land converted to
cropland would come from pasture and that none of that pasture would be
replaced, which is counter to the land use trends identified by the
Winrock satellite data. As can be seen, the assumption of pastureland
replacement can have a significant effect on the results. We ask for
comment on the best assumptions to be made when considering the need to
replace pasture that has been converted to crop production. We note
that the best decision on pasture land replacement may vary by country
or region due to such factors as the current intensity of use of
pasture land as well as trends in demand for pasture. DRIA Chapter 2
includes more details about the treatment of pasture conversion, and
sensitivity analysis of the types land use changes induced by corn
ethanol production.

 Table VI.C.1-5--Corn Ethanol Lifecycle GHG Emissions Changes Associated
             with Different Assumptions on Land Use Changes
------------------------------------------------------------------------
                                          Percent Change  Percent Change
                                             from 2005       from 2005
          Scenario Description               Petroleum       Petroleum
                                           Baseline (100   Baseline (30
                                              yr 2%)          yr 0%)
------------------------------------------------------------------------
Corn Ethanol Natural Gas Dry Mill in                 -34             -19
 2022 with dry DGs; 0% pastureland
 replaced...............................
Corn Ethanol Natural Gas Dry Mill in                  -2             +24
 2022 with dry DGs; 100% pastureland
 replaced...............................
Corn Ethanol Natural Gas Dry Mill in                 -48             -38
 2022 with dry DGs; grassland only
 conversion and 0% pastureland replaced.
------------------------------------------------------------------------

    DRIA Chapter 2 includes results for additional sensitivity analysis
of corn ethanol lifecycle GHG emissions. We also intend to conduct
additional sensitivity analysis for the final rule. We invite comment
on these assumptions.
b. Imported Ethanol
    Table VI.C.1-6 presents the breakout of lifecycle GHG emissions for
sugarcane ethanol compared to a 2005 petroleum baseline under different
discount rate and time horizon scenarios and land use assumptions. This
assessment was based on applying the same methodology as for other
biofuels including the assessment of both direct and indirect impacts
using the combination of FASOM, FAPRI and Winrock modeling results.
Virtually all the ethanol from sugarcane is expected to be imported
from Brazilian production. Applying the proposed FAPRI/Winrock
methodology to sugarcane ethanol production in Brazil predicts a large
increase in new acres planted, which has a relatively large impact on
overall GHG emissions. The impact is from both new sugarcane production
acres in Brazil resulting in land use change but also reduced commodity
exports from Brazil resulting in land use change in other countries.
    The proposed FAPRI/Winrock methodology predicts that new crop
acreage is converted from a range of land types. In contrast, some
studies suggest that sugarcane ethanol production can increase in
Brazil by relying on existing excess pasture lands and will not
significantly impact other land types.\314\ Table VI.C.1-6 provides the
range of lifecycle GHG emission reduction results under these different
assumptions of type conversion patterns. As a sensitivity analysis,
shows results for a scenario where none of the grassland converted to
cropland in Brazil would be replaced if, for example, livestock
production could be more intensively developed on the remaining pasture
(see second row in Table VI.C.1-6). The third row of Table VI.C.1-6
includes lifecycle GHG results assuming that in Brazil all of the land
converted to cropland would come from grassland and that none of that
grassland would be replaced. As can be seen in the table, the
assumption of pastureland replacement can have an important effect on
the results. DRIA Chapter 2 includes more details about the treatment
of pasture conversion, and sensitivity analysis of the types land use
changes induced by sugarcane ethanol production.
---------------------------------------------------------------------------

    \314\ Goldemberg, J.; Coelho, ST.; Guardabassi, PM. The
sustainability of ethanol production from sugarcane. Energy Policy.
2008. doi:10.1016/j.enpol.2008.02.028.

Table VI.C.1-6--Sugarcane Ethanol GHG Emission Changes Under Varied Land
 Use Assumptions and Varied Discount Rates and Time Horizons Relative to
                         2005 Petroleum Baseline
------------------------------------------------------------------------
  Land Use Change Scenario Description      (100 yr 2%)     (30 yr 0%)
------------------------------------------------------------------------
FAPRI/Winrock estimate with managed                  -44             -26
 pasture replacement....................
FAPRI/Winrock estimate with no pasture               -59             -45
 replacement in Brazil..................
Only grassland conversion in Brazil and              -64             -52
 no pasture replacement in Brazil.......
------------------------------------------------------------------------

    We are aware that recent land use enforcement policies in Brazil
may shift cropland expansion patterns (see also Section VI.B.5.b.iii).
We seek comment on both pasture conversion patterns and Brazil land use
enforcement policy impacts. We are conducting more detailed economic
modeling of the Brazilian agricultural sector by state for inclusion in
FAPRI to address pasture, enforcement and other assumptions for the
final rule. State level production data could be used in conjunction
with Winrock's state level satellite data, which may substantially change the

[[Page 25045]]

estimates of the location and type of land being converted in Brazil
for the final rule.
    We have also assumed that sugarcane ethanol production relies on
burning bagasse as an energy source and that the process produces
excess electricity. We factor in credits from this excess electricity
based on offsetting the Brazilian electricity grid. As Brazil
implements limits on field burning of bagasse there may be additional
bagasse used at sugarcane ethanol plants and additional electricity
production. We plan to look at this further for the final rule analysis.
c. Cellulosic Ethanol
    Given that commercially-viable cellulosic ethanol production is not
yet a reality, analysis of this pathway relies upon significant
assumptions regarding the development of production technologies. As
described in the previous section, our analysis assumed corn stover
required no international land use changes, since corn stover does not
compete with other crops for acreage in the U.S. Therefore, using corn
stover as a feedstock for cellulosic biofuel production would not have
an impact on U.S. exports. We assumed some of the nutrients would have
to be replaced through higher fertilizer rates on acres where stover is
removed; however, increased stover removal was also associated with
higher rates of reduced tillage or no tillage practices which results
in soil carbon increase. See Section IX.A for details. In addition,
cellulosic ethanol was assumed to be produced using the biochemical
process which is expected to produce more electricity from the lignin
in the feedstock than is required to power the ethanol plant, so excess
electricity can be sold back to the grid. See DRIA Chapter 2 for
additional details. This electricity provides a GHG benefit, which
results in GHG emissions reductions from fuel production as shown in
Table VI.C.1-7.

   Table VI.C.1-7--Absolute Lifecycle GHG Emissions for Corn Stover Cellulosic Ethanol and the 2005 Petroleum
                                                    Baseline
                                                 [CO2-eq/mmBtu]
----------------------------------------------------------------------------------------------------------------
                                                                    Corn stover                     Corn stover
                                                                      ethanol                         ethanol
                                                       2005          (selling          2005          (selling
                                                     Petroleum        excess         Petroleum        excess
                                                     baseline     electricity to     baseline     electricity to
                                                                       grid)                           grid)
----------------------------------------------------------------------------------------------------------------
                 Lifecycle Stage                            (100 yr 2%)
                                                            (30 yr 0%)
----------------------------------------------------------------------------------------------------------------
Net Domestic Agriculture (w/o land use change)..  ..............         178,862             N/A         124,503
Net International Agriculture (w/o land use       ..............               0             N/A  ..............
 change)........................................
Domestic Land Use Change........................  ..............         -78,448             N/A         -91,925
International Land Use Change...................  ..............               0             N/A               0
Fuel Production.................................         823,262        -875,424         573,058        -609,367
Fuel and Feedstock Transport....................  ..............         107,214  ..............          74,629
Tailpipe Emissions..............................       3,417,311          37,927       2,378,800          26,400
                                                 ---------------------------------------------------------------
    Net Total Emissions.........................       4,240,674        -629,870       2,951,858        -475,130
----------------------------------------------------------------------------------------------------------------

    Although switchgrass must compete with other crops for land in the
U.S., average switchgrass ethanol yields are on average higher than
corn ethanol yields (approximately 580 gallons/acre compared to 480
gallons/acre). Therefore, switchgrass would need approximately 20% less
land to produce the same amount of ethanol compared to corn. In
addition, FASOM predicts that switchgrass would generally be grown on
more marginally productive land. Since switchgrass is not projected to
displace crop acres with high yields, new switchgrass acres generally
would not have a large impact on exports. Therefore, the international
land use change impacts are modest. Like cellulosic ethanol from corn
stover, switchgrass ethanol is also assumed to produce excess
electricity that can be sold to the grid, therefore switchgrass
cellulosic ethanol results in relatively large lifecycle GHG reductions
compared to the replaced petroleum gasoline as shown in Table VI.C.1-8.

    Table VI.C.1-8--Absolute GHG Emissions for Switchgrass Cellulosic Ethanol and the 2005 Petroleum Baseline
                                                 [CO2-eq/mmBtu]
----------------------------------------------------------------------------------------------------------------
                                                                    Switchgrass                     Switchgrass
                                                                      ethanol                         ethanol
                                                       2005          (selling          2005          (selling
                                                     Petroleum        excess         Petroleum        excess
                                                     baseline     electricity to     baseline     electricity to
                                                                       grid)                           grid)
----------------------------------------------------------------------------------------------------------------
                 Lifecycle Stage                            (100 yr 2%)
                                                            (30 yr 0%)
----------------------------------------------------------------------------------------------------------------
Net Domestic Agriculture (w/o land use change)..  ..............        -470,620  ..............        -327,590
Net International Agriculture (w/o land use       ..............        -356,712  ..............        -248,301
 change)........................................
Domestic Land Use Change........................  ..............         -65,318  ..............         -76,015
International Land Use Change...................  ..............         423,097  ..............         424,094
Fuel Production.................................         823,262        -874,599         573,058        -608,793
Fuel and Feedstock Transport....................  ..............         136,663  ..............          95,129
Tailpipe Emissions..............................       3,417,311          37,927       2,378,800          26,400
                                                 ---------------------------------------------------------------

[[Page 25046]]

    Net Total Emissions.........................       4,240,674      -1,169,561       2,951,858        -715,076
----------------------------------------------------------------------------------------------------------------

    Cellulosic ethanol does not have nearly as significant an impact on
land use as other biofuels, therefore we did not calculate sensitivity
impacts of, for example, assuming full replacement of pasture versus no
pasture replacement which could be important in the lifecycle GHG
assessment of other biofuels. As the land use issue is not critical for
the cellulosic feedstock fuels in the scenarios we analyzed, the impact
of timing and discount rates also do not have a significant impact on
the overall results for cellulosic ethanol. Both of the cellulosic
ethanol pathways we examined, switchgrass and corn stover using
enzymatic processing, reduced lifecycle GHG emissions by significantly
more than the 60% threshold for cellulosic biofuel. Table VI.C.1-9
summarizes the lifecycle GHG results for cellulosic ethanol fuel
pathways.

 Table VI.C.1-9--Cellulosic Ethanol GHG Emission Changes From Different
 Feedstocks and Varied Discount Rates and Time Horizons Relative to 2005
                           Petroleum Baseline
                              [In percent]
------------------------------------------------------------------------
    Assumption--feedstock type         (100 yr 2%)         (30 yr 0%)
------------------------------------------------------------------------
Corn Stover.......................               -115               -117
Switchgrass.......................               -128               -121
------------------------------------------------------------------------

d. Biodiesel
    EPA's modeling predicts that soybean-based biodiesel production has
a large land use impact for two major reasons. Soybean biodiesel has a
relatively low gallon per acre yield (approximately 65 gal/acre for
soybean biodiesel versus 480 gal/acre for corn ethanol). Thus, the
impact of any land-use change tends to be magnified with soybean
biodiesel. Even when the higher Btu value of biodiesel is taken into
consideration, Btu/acre yields are still significantly lower for
biodiesel than for ethanol (approximately 97 gal/acre ethanol
equivalent). Furthermore, our analysis suggests that due to high world
wide demand for soybeans for food, cooking and other non-biofuel uses,
soybean and other edible oils used for biofuel are generally replaced
by production in other countries including production in tropical
climates where the GHG emissions released per acre of converted land
are highest. This indicates that soy-based biodiesel lifecycle GHG
emissions could be greatly reduced with the adoption of policies and
agricultural practices that limit the amount of tropical deforestation
induced by soy-based biodiesel production. DRIA Chapter 2 includes
sensitivity analyses about the types of land converted to crops as a
result of soy-based biodiesel production. Table VI.C.1-10 presents the
breakout of the absolute lifecycle GHG emissions for soybean biodiesel
and the petroleum diesel fuel baseline by lifecycle stage.

     Table VI.C.1-10--Absolute Lifecycle GHG Emissions for Soybean Biodiesel and the 2005 Petroleum Baseline
                                                 [CO2-eq/mmBtu]
----------------------------------------------------------------------------------------------------------------
                                                  2005 Petroleum      Soybean     2005 Petroleum      Soybean
                                                     baseline        biodiesel       baseline        biodiesel
----------------------------------------------------------------------------------------------------------------
                 Lifecycle Stage                            (100 yr 2%)
                                                            (30 yr 0%)
----------------------------------------------------------------------------------------------------------------
Net Domestic Agriculture (w/o land use change)..  ..............        -423,206  ..............        -294,586
Net International Agriculture (w/o land use       ..............         195,304  ..............         135,948
 change)........................................
Domestic Land Use Change........................  ..............          -8,980  ..............         -10,451
International Land Use Change...................  ..............       2,474,074  ..............       2,469,574
Fuel Production.................................         749,132         838,490         521,458         583,658
Fuel and Feedstock Transport....................  ..............         149,258  ..............         103,896
Tailpipe Emissions..............................       3,424,635          30,169       2,383,828          21,000
                                                 ---------------------------------------------------------------
    Net Total Emissions.........................       4,173,768       3,255,109       2,905,286       3,009,039
----------------------------------------------------------------------------------------------------------------

    Our analysis is based on a change in biodiesel volumes from 0.4
Bgal to 0.7 Bgal. Similar to the analysis we conducted for corn-
ethanol, we plan to run a sensitivity analysis on the impact of using
different volumes for the final rule.

[[Page 25047]]

    As discussed in Section VI.B.2.a, EPA's interpretation of the EISA
statute compels us to include significant indirect emission impacts
including those due to land use changes in other countries. The data in
Table VI.C.1-10 indicate that excluding the international land use
change would result in soy-based biodiesel having an approximately 80%
reduction in lifecycle GHG emissions compared to petroleum gasoline
regardless of the timing or discount rate used. The treatment of
emissions over time is not critical if international land use change
emissions are excluded because the results without land use change are
consistent over time. Therefore the overall lifecycle GHG results do
not vary with time or discount rate assumptions.
    In contrast, GHG emissions from waste oil and greases are assumed
to have no land use impacts. We assumed any land use change was
attributed to the original use of the feedstock, for example, soy oil
was produced for the purpose of using for cooking and the land required
to produce this cooking oil is properly attributed to that use.
Gathering and re-using the left over waste cooking oil would have no
additional land use impact. This lack of land use impact greatly
influences the lifecycle GHG analysis. Table VI.C.1-11 presents the
breakout of the absolute lifecycle GHG emissions for waste grease
biodiesel and the petroleum diesel fuel baseline by lifecycle stage.

  Table VI.C.1-11--Absolute Lifecycle GHG Emissions for Waste Grease Biodiesel and the 2005 Petroleum Baseline
                                                 [CO2-eq/mmBtu]
----------------------------------------------------------------------------------------------------------------
                                                       2005                            2005
                                                     Petroleum     Waste grease      Petroleum     Waste grease
                                                     baseline        biodiesel       baseline        biodiesel
----------------------------------------------------------------------------------------------------------------
                 Lifecycle Stage                            (100 yr 2%)
                                                            (30 yr 0%)
----------------------------------------------------------------------------------------------------------------
Net Domestic Agriculture (w/o land use change)..  ..............               0  ..............               0
Net International Agriculture (w/o land use       ..............               0  ..............               0
 change)........................................
Domestic Land Use Change........................  ..............               0  ..............               0
International Land Use Change...................  ..............               0  ..............               0
Fuel Production.................................         749,132         658,198         521,458         458,160
Fuel and Feedstock Transport....................  ..............         149,258  ..............         103,896
Tailpipe Emissions..............................       3,424,635          30,169       2,383,828          21,000
                                                 ---------------------------------------------------------------
    Net Total Emissions.........................       4,173,768         837,626       2,905,286         583,056
----------------------------------------------------------------------------------------------------------------

    Table VI.C.1-12 summarizes the lifecycle GHG results for biodiesel
fuel pathways. As the waste grease biodiesel is not assumed to have any
land use impact the choice of timing or discount rate does not impact
the waste grease biodiesel results. However, as the soybean biodiesel
is found to have a large land use impact the choice of timing and
discount rate has a big impact on the soybean biodiesel results.

Table VI.C.1-12--Biodiesel Lifecycle GHG Emission Changes From different
 Feedstocks and Varied Discount Rates and Time Horizons Relative to 2005
                           Petroleum Baseline
------------------------------------------------------------------------
    Assumption--feedstock type         (100 yr 2%)         (30 yr 0%)
------------------------------------------------------------------------
Soybean...........................               -22%                +4%
Waste Grease......................               -80%               -80%
------------------------------------------------------------------------

    Table VI.C.1-13 shows the sensitivity of our assessment for soy oil
biodiesel assuming 100% of the grassland converted to cropland is
replaced compared to an assumption that none of this grassland is
replaced for livestock grazing. DRIA Section 2.8.2.4 provides more
information about sensitivity analysis for the pasture replacement assumptions.

 Table VI.C.1-13--Soy-Based Biodiesel GHG Emission Changes Under Varied
    Land Use Assumptions and Varied Discount Rates and Time Horizons
                   Relative to 2005 Petroleum Baseline
------------------------------------------------------------------------
 Assumption--land types available
          for conversion               (100 yr 2%)         (30 yr 0%)
------------------------------------------------------------------------
100% Pasture Replacement..........                -4%               +27%
No Pasture Replacement............               -45%               -27%
------------------------------------------------------------------------

2. Treatment of GHG Emissions Over Time
    As described in Section VI.B.5, changes in indirect land use
associated with increased biofuel production result in GHG emissions
increases that accumulate over a long time period. Since there is a
large release of carbon in the first year of land conversion, it can
take many years for the benefits of the biofuel to make up for these
early carbon emissions, depending on the specific biofuel in question.
Table VI.C.2-1 contains the payback period associated with several
types of biofuels and fuel production pathways. A payback period of 0
indicates that these pathways do not have land use change impacts and
therefore reduce emissions in the first year that they are produced.
Assessments are made in comparison to

[[Page 25048]]

the baseline transportation fuel used in 2005 in the U.S. as mandated
by EISA. The percent reduction goal is the lifecycle GHG emissions of
the biofuel compared to the baseline petroleum fuel it is replacing.

                                         Table VI.C.2-1--Payback Period
                                                   [in years]
----------------------------------------------------------------------------------------------------------------
                                                                      Payback period (years)
                                                 ---------------------------------------------------------------
                    Fuel type                        Reduction       Reduction       Reduction       Reduction
                                                     goal: 0%        goal: 20%       goal: 50%       goal: 60%
----------------------------------------------------------------------------------------------------------------
Corn Ethanol 2022 Base Dry Mill NG \315\........              33              54       \316\ N/A             N/A
Corn Ethanol 2022 Best Case Dry Mill NG \317\...              23              31             N/A             N/A
Corn Ethanol 2022 Base Dry Mill Coal \318\......              75            >100             N/A             N/A
Corn Ethanol 2022 Base Dry Mill Biomass \319\...              22              31             N/A             N/A
Soybean Biodiesel...............................              32              46             105             N/A
Waste Grease Biodiesel..........................               0               0               0             N/A
Sugarcane Ethanol...............................              18              26              61             N/A
Switchgrass Ethanol.............................               3               3               4               5
Corn Stover Ethanol.............................               0               0               0               0
----------------------------------------------------------------------------------------------------------------

    
---------------------------------------------------------------------------

    \315\ Dry Mill corn ethanol plant using natural gas with 2022
energy use and dry DDGS.
    \316\ Payback periods were not calculated for ethanol made from
corn starch for the advanced biofuel reduction goals of 50% and 60%
since this corn ethanol does not qualify under EISA as a potential
advanced biofuel.
    \317\ Dry Mill corn ethanol plant using natural gas with 2022
energy use and w/CHP, Fractionation, Membrane Separation, and Raw
Starch Hydrolysis (wet DGS).
    \318\ Dry Mill corn ethanol plant using coal with 2022 energy
use and dry DDGS.
    \319\ Dry Mill corn ethanol plant using biomass with 2022 energy
use and dry DDGS.
---------------------------------------------------------------------------

    As described in Section VI.B.5, we have focused our lifecycle GHG
analysis on two ways of accounting for GHG emissions over time. In one
set of results we consider lifecycle GHG emissions over 100 years and
discount future emissions with a 2% discount rate. In the other set of
results we consider 30 years of GHG emissions with no discounting of
future emissions (i.e., 0% discount rate). Whereas the discussion
immediately above focused on lifecycle GHG impacts assuming 100 years
with a 2% discount rate and 30 years with no discount rate, Table
VI.C.2-2 shows the lifecycle GHG emissions reductions estimates with a
variety of time periods and discount rates.

                       Table VI.C.2-2--Lifecycle GHG Emissions Changes of Select Biofuels Relative to the 2005 Petroleum Baseline
--------------------------------------------------------------------------------------------------------------------------------------------------------
                                                Lifecycle GHG emissions changes of select biofuels relative to the 2005 petroleum baseline
--------------------------------------------------------------------------------------------------------------------------------------------------------
          Time horizon                           30 Years                                50 Years                                100 Years
--------------------------------------------------------------------------------------------------------------------------------------------------------
          Discount rate              0%        2%        3%        7%        0%        2%        3%        7%        0%        2%        3%        7%
--------------------------------------------------------------------------------------------------------------------------------------------------------
Corn Ethanol                            5%       18%       25%       54%      -17%       -2%        7%       44%      -36%      -16%       -4%       41%
Dry Mill NG
Corn Ethanol                          -14%       -1%        6%       35%      -36%      -21%      -12%       25%      -55%      -35%      -23%       22%
Best Case Dry Mill NG
Corn Ethanol                           34%       46%       53%       83%       11%       27%       35%       72%       -8%       13%       24%       69%
Dry Mill Coal
Corn Ethanol                          -18%       -6%        1%       31%      -41%      -25%      -17%       20%      -60%      -39%      -28%       16%
Dry Mill Biomass
Soybean Biodiesel...............        4%       20%       29%       68%      -24%       -4%        7%       55%      -48%      -22%       -7%       51%
Waste Grease Biodiesel..........      -80%      -80%      -80%      -80%      -80%      -80%      -80%      -80%      -80%      -80%      -80%      -80%
Sugarcane Ethanol...............      -27%      -17%      -11%       12%      -45%      -32%      -26%        3%      -61%      -44%      -35%        1%
Switchgrass Ethanol.............     -124%     -122%     -121%     -115%     -128%     -125%     -124%     -117%     -131%     -128%     -126%     -117%
Corn Stover Ethanol.............     -116%     -117%     -117%     -118%     -115%     -116%     -116%     -117%     -114%     -115%     -115%     -117%
--------------------------------------------------------------------------------------------------------------------------------------------------------

D. Thresholds

    EISA established GHG thresholds for each category of renewable fuel
that it mandates. EISA also provided EPA with the authority to adjust
the threshold levels for each category of renewable fuels if certain
requirements are met. Renewable fuels must achieve a 20% reduction in
lifecycle greenhouse gas emissions compared to the average lifecycle
greenhouse gas emissions for gasoline or diesel sold or distributed as
transportation fuel in 2005. Due to the grandfathering provisions of
EISA as adopted in this rule, this threshold only pertains to renewable
fuel produced at plants to be constructed in the future. EPA is
permitted to adjust this threshold to as low as 10%, based on the
``maximum achievable level, taking cost into consideration, for natural
gas fired corn-based ethanol plants allowing for the use of a variety
of technologies.'' Based on our analysis, there are a number of corn
ethanol natural gas plant configurations that could meet the 20%
reduction in GHG emissions thresholds in 2022 if modeling emission over
a 100 year time frame and then discounting these emissions 2%.
Therefore, based on this assessment, we believe that an adjustment to
the 20% threshold would be unnecessary and we are proposing to maintain
it at the 20% level if we adopt the 100 year, 2% discounting methodology.
    On the other hand, based on our current analyses, if we adopt an
assessment methodology which assesses emissions over just 30 years,
then no currently analyzed natural gas-fired corn ethanol pathway will
meet the 20% threshold. However, some of the natural gas corn ethanol
pathways do

[[Page 25049]]

have lifecycle GHG emission benefits in the 10% to 20% range. Corn
ethanol is expected to be the major biofuel contributing to meeting the
renewable fuel standards through at least the middle of the next
decade. Therefore, if we adopt a 30 year timeframe for emissions
assessment and do not discount the results, we may adjust the renewable
fuel thresholds to the minimum level as necessary to incorporate at
least a few of the best GHG pathways for corn ethanol. While this
adjusted threshold level could be revised based on pathway analyses
done for the final rule, at this time we would intend to allow a full
10% adjustment of the renewable fuel threshold, down to a threshold
value of 10% reduction compared to the 2005 gasoline baseline.
    Cellulosic biofuels must meet a 60% reduction in GHG emissions
relative to the petroleum baseline. EPA is permitted to adjust this
threshold to as low as 50% if it is ``not commercially feasible for
fuels made using a variety of feedstocks, technologies, and processes''
to achieve the 60% threshold. Our initial analysis indicates that
cellulosic biofuels from corn stover, switchgrass, and bagasse will all
meet the 60% threshold regardless of whether we use to 100 year, 2%
discount methodology or the 30 year analysis time frame without
discounting. Furthermore, we believe most fuels made from other
cellulosic feedstocks would as well. Therefore we do not believe it is
necessary to adjust the threshold for cellulosic biofuel at this time.
    Biomass-based diesel must achieve a 50% reduction in GHG emissions
relative to petroleum-based diesel. EPA is permitted to adjust this
threshold to as low as 40% if it is ``not commercially feasible for
fuels made using a variety of feedstocks, technologies, and processes''
to meet the 50% level. For biomass-based diesel, our analysis indicates
that biodiesel from waste oils such as yellow grease and tallow would
meet the 50% threshold, and we anticipate that biodiesel from chicken
waste and non-food grade corn oil fractionation would as well
regardless of whether we use a 100 year, 2% discount methodology or the
30 year analysis time frame without discounting. However, our current
analysis indicates that there is insufficient feedstock from waste
grease and fats to meet the one billion gallon volumetric requirement
under EISA. Biodiesel from soy oil (and we believe biodiesel from other
food grade vegetable oils) would reduce GHG emissions by no more than
22% using a 100 year, 2% discount methodology and would be estimated to
increase GHG emissions if we analyze emission impacts over 30 years
whether the emissions are discounted or not. Even if EPA adjusted the
biomass-based diesel standard to the minimum allowable level of 40%,
soybean-based biodiesel would still not meet the GHG emissions
reductions threshold for biomass based diesel fuel. One option for
meeting the volumetric requirement and the emissions reduction
threshold, assuming a 100 year timeframe and a 2% discount rate for GHG
emission impacts would be to allow biodiesel producers to average the
emissions reductions from a blend of soy oil or food grade vegetable
oil-based biodiesel with waste oil based biodiesel, as discussed in
more detail in Section VI.E. However, this approach may still be
insufficient to ensure that the required volumes of biomass-based
diesel can be produced unless other sources of biomass-based diesel
become available. Therefore, we invite comments on whether it be
appropriate to both reduce the threshold to 40% and allow biodiesel
producers to average their emissions to meet the one billion gallon
volumetric requirement as discussed below in Section VI.E.3.c.
    Advanced biofuels must achieve a 50% reduction in GHG emissions.
EPA is permitted to adjust this threshold to as low as 40% if it is
``not commercially feasible for fuels made using a variety of
feedstocks, technologies, and processes'' to achieve the 50% threshold.
Our current lifecycle analysis suggests that sugarcane based ethanol
only offers an estimated 44% reduction in GHG emissions relative to the
gasoline it replaces when assessing 100 years of emission impacts and
discounting these emissions 2%, and an estimated 27% reduction when
assessing 30 years of emission impacts with no discounting. Therefore,
it would not qualify as an advanced biofuel if we did not adjust the
50% GHG threshold. We are also unaware of other renewable fuels that
may be available in sufficient volumes over the next several years to
allow the statutory volume requirements for advanced biofuel to be met.
As a result, we are proposing that the GHG threshold for advanced
biofuels be adjusted to 44% or potentially as low as 40% depending on
the results from the analyses that will be conducted for the final
rule. Based on our current analysis of the lifecycle GHG impacts of
sugarcane ethanol, such an adjustment would help ensure that the volume
mandates for advanced biofuel can be met.
    We invite comments on these proposed thresholds and our basis for them.

E. Assignment of Pathways to Renewable Fuel Categories

    The lifecycle analyses that we conducted for a variety of fuel
pathways formed the basis for our determination of which pathways would
be permitted to generate RINs, and to which of the four renewable fuel
categories (cellulosic biofuel, biomass-based diesel, advanced biofuel,
and renewable fuel) those RINs should be assigned. This determination
involved comparing the lifecycle GHG performance estimates to the GHG
thresholds associated with each renewable fuel category, discussed in
Section VI.D above. In addition, each of the four renewable fuel
categories is defined in EISA to include or exclude certain types of
feedstocks and production processes, and these definitions also played
a role in determining the appropriate category for each pathway. This
section describes our proposed assignments of pathways to one of the
four renewable fuel categories. The GHG lifecycle values used in this
assignment of fuel pathways to the four renewable fuel categories were
based on the lifecycle analysis results over a 100-year timeframe and
using a 2% discount rate, as described in Section VI.C. Different
assignments of pathways to the four renewable fuel categories would
occur with different lifecycle results, but we propose that the same
assignment methodology would be followed regardless.
1. Statutory Requirements
    EISA establishes requirements that are common to all four
categories of renewable fuel in addition to requirements that are
unique to each of the four categories. The common requirements
determine which fuels are valid for generating RINs under the RFS2
program. For instance, all renewable fuel must be made from renewable
biomass, which defines the types of feedstocks that can be used to
produce renewable fuel that is valid under the RFS2 program, and also
defines the types of land on which crops can be grown if those crops
are used to produce valid renewable fuel under the RFS2 program. See
Section III.B.4 for a more detailed discussion of renewable biomass.
Moreover, all renewable fuel must displace fossil fuel present in
transportation fuel, or be used as home heating oil or jet fuel.
    The requirements that are unique to each of the four categories
provide a basis for assigning each pathway to a category. For each of
the four categories of renewable fuel, EISA provides a definition,
specifies the associated GHG

[[Page 25050]]

thresholds, lists the allowable feedstocks and/or fuel types, and in
some cases provides exclusions. Table VI.E.1-1 summarizes these
requirements as we are applying them under the proposed RFS2 program.

                           Table VI.E.1-1--Requirements for Renewable Fuel Categories
----------------------------------------------------------------------------------------------------------------
                                                         Biomass-based
                                  Cellulosic biofuel        diesel         Advanced biofuel     Renewable fuel
----------------------------------------------------------------------------------------------------------------
GHG threshold...................  60%...............  50% \a\...........  40-44% \a\........  20% a, b.
Eligible Inclusions.............  Renewable fuel      Any renewable fuel  All cellulosic      All advanced
                                   made from           that is a diesel    biofuel and         biofuel, and any
                                   cellulose,          fuel substitute.    biomass-based       other fuel made
                                   hemicellulose, or                       diesel, as well     from renewable
                                   lignin.                                 as other            biomass that is
                                                                           renewable fuels     used to replace
                                                                           including ethanol   or reduce the
                                                                           from sugar,         quantity of
                                                                           starch, or waste    fossil fuel
                                                                           materials,          present in a
                                                                           biogas, and         transportation
                                                                           butanol and other   fuel.
                                                                           alcohols.
Exclusions......................  ..................  Any renewable fuel  Ethanol derived
                                                       made from           from corn starch.
                                                       coprocessing with
                                                       petroleum.
----------------------------------------------------------------------------------------------------------------
\a\ As discussed in Section VI.D, we are seeking comment on the need to adjust the thresholds, and are proposing
  that the GHG threshold for advanced biofuels be adjusted to as low as 40%.
\b\ 20% threshold does not apply to grandfathered volumes. See discussion in Section III.B.3.

2. Assignments for Pathways Subjected to Lifecycle Analyses
    There are a wide variety of pathways (unique combinations of
feedstock, fuel type, and fuel production process) that could result in
renewable fuel that would be valid under the RFS2 program. As described
earlier in this section, we conducted lifecycle analyses for some of
these pathways, and these analyses allowed us to determine if the GHG
thresholds shown in Table VI.E.1-1 would be met under the assumption of
a 100-year timeframe and discount rate of 2%. For other pathways that
we have not yet subjected to lifecycle analyses, there were some cases
in which we could nevertheless still make moderately confident
determinations as to the likely GHG impacts by making comparisons to
the pathways that we did analyze. A discussion of these other
determinations is provided in Section VI.E.3 below.
    For pathways that we subjected to lifecycle analysis, we were able
to assign each pathway to one of the four renewable fuel categories
defined in EISA by comparing the descriptions of each pathway and its
associated GHG performance to the requirements shown in Table VI.E.1-1.
The results are shown in Table VI.E.2-1.

   Table VI.E.2-1--Proposed Assignment of Pathways to One of the Four
 Renewable Fuel Categories for Pathways Subjected to Lifecycle Analyses
------------------------------------------------------------------------

------------------------------------------------------------------------
Cellulosic biofuel pathways.......  Ethanol produced from corn stover or
                                     switchgrass in a process that uses
                                     enzymes to hydrolyze the cellulose
                                     and hemicellulose.
Biomass-based diesel pathways.....  Biodiesel (mono alkyl esters)
                                     produced from waste grease and
                                     waste oils.
Advanced biofuel pathways.........  Ethanol produced from sugarcane
                                     sugar in a process that uses
                                     sugarcane bagasse for process heat.
                                     \a\
Renewable fuel pathways...........  Ethanol produced from corn starch in
                                     a process that uses biomass for
                                     process heat.
                                    Ethanol produced from corn starch in
                                     a process that includes:
                                       --Dry mill plant.
                                       --Process heat derived from
                                        natural gas.
                                       --Combined heat and power (CHP).
                                       --Fractionation of feedstocks.
                                       --All distillers grains are
                                        dried.
                                    Ethanol produced from corn starch in
                                     a process that includes:
                                       --Dry mill plant.
                                       --Process heat derived from
                                        natural gas.
                                       --All distillers grains are wet.
                                    Ethanol produced from corn starch in
                                     a process that includes:
                                       --Dry mill plant.
                                       --Process heat derived from coal.
                                       --Combined heat and power (CHP).
                                       --Fractionation of feedstocks.
                                       --Membrane separation of ethanol.
                                       --Raw starch hydrolysis.
                                       --All distillers grains are
                                        dried.
                                    Ethanol produced from corn starch in
                                     a process that includes:
                                       --Dry mill plant.
                                       --Process heat derived from coal.
                                       --Combined heat and power (CHP).
                                       --Fractionation of feedstocks.
                                       --Membrane separation of ethanol.
                                       --All distillers grains are wet.

[[Page 25051]]

                                    Biodiesel (mono alkyl esters)
                                     produced from soybean oil.
------------------------------------------------------------------------
\a\ Our current analysis concludes that ethanol from sugarcane sugar
  would have a GHG performance of 44% in comparison to gasoline under
  our assumed 100-year timeframe and 2% discount rate. Since this falls
  short of the 50% GHG threshold for advanced biofuel, we have
  categorized it as general renewable fuel. However, we request comment
  on lowering the applicable GHG threshold for advanced biofuel so that
  ethanol from sugarcane sugar could be categorized as advanced biofuel.
  See further discussion in Section VI.D.

    In addition, our lifecycle analyses also identified pathways that
did not meet the minimum 20% GHG threshold under an assumed 100-year
timeframe and 2% discount rate, and thus would be prohibited from
generating RINs unless a facility met the prerequisites for
grandfathering as described in Section III.B.3. These prohibited
pathways all involved the production of ethanol from corn starch in a
process that uses natural gas or coal for process heat, but which does
not meet any of the process technology requirements listed in Table
VI.E.2-1. Our proposal for temporary D codes in Sec.  80.1416 would
explicitly prohibit the generation of RINs for these pathways.
    The proposed assignments of individual pathways to one of the four
renewable fuel categories shown in the table above assumed a 100-year
timeframe and discount rate of 2% for lifecycle GHG emission impacts.
The assignments would be different if we had assumed a different
timeframe and discount rate. By comparing the relative GHG emission
reductions shown in Table VI.C.1-2 to the thresholds in Table VI.E.1-1,
a variety of different assignments is possible covering timeframes of
30, 50, and 100 years, and discount rates of 0%, 2%, 3%, and 7%. For
instance, under the assumption of 30 years and no discounting,
switchgrass ethanol and corn stover ethanol would continue to be
categorized as cellulosic biofuel and biodiesel made from waste grease
would continue to be categorized as biomass-based diesel. However,
sugarcane ethanol could no longer be potentially categorized as
advanced biofuel but instead would be categorized as renewable fuel.
Moreover, some pathways would not meet the minimum threshold of 20% for
renewable fuel, and so could not generate RINs if the volume was not
grandfathered. This would include soybean biodiesel and all of the corn
starch ethanol pathways shown in Table VI.E.2-1 produced from newly
constructed plants not meeting the grandfathering criteria discussed in
Section III.B.3.
3. Assignments for Additional Pathways
    We were not able to conduct lifecycle modeling for all potential
pathways in time for this proposed rulemaking. Instead, we focused the
lifecycle GHG emissions analysis on the feedstocks that, based on FASOM
predictions and other information, we anticipate could contribute the
largest volumes to the renewable fuel pool and the production processes
representing the largest shares of the market. As more information
becomes available, we anticipate that we will be updating the lifecycle
methodology and expanding the list of emission factors.
    Beyond the pathways that we explicitly subjected to lifecycle
analysis, there are additional pathways that may not currently be
significant contributors to the volume of renewable fuel produced, but
their volumes could increase in the future. Moreover, we believe it is
important that as many pathways as possible be included in the lookup
table in the regulations to help ensure that the volume requirements in
EISA can be met and to encourage the development of new fuels. To this
end, we evaluated these additional pathways to determine if they could
be deemed valid for generation of RINs, and if so which of the four
renewable fuel categories they would fall into. This section describes
our evaluation of these additional pathways and the resulting proposed
assignment to one or more of the four renewable fuel categories.
a. Ethanol From Starch
    Our lifecycle analysis focused on ethanol from corn starch.
However, there are a variety of other sources of starch that use or
could use a very similar process for conversion to ethanol. These
include wheat, barley, oats, rice, and sorghum. Some existing corn-
ethanol facilities already use small amounts of starch from these other
plants along with corn in their production of ethanol.
    Although we have not explicitly analyzed the land use or processing
impacts of these other starch plants on their lifecycle GHG
performance, we believe it would be reasonable to assume similar
impacts to corn in terms of the types of land that would be displaced
and other aspects of producing and transporting the feedstock.
Therefore, we propose that the pathways shown in Table VI.E.2-1 for
ethanol produced from corn starch also be applied to ethanol produced
from other sources of starch.
    The lifecycle analyses conducted for this proposal only examined
cases in which a corn-ethanol facility dried 100% of its distiller's
grains or left 100% of its distiller's grains wet. The treatment of the
distiller's grains for corn-ethanol facilities impacts the
determination of whether the 20% GHG threshold for renewable fuel has
been met. However, in practice some facilities may dry only a portion
of their distiller's grains and leave the remainder wet. As described
in Section III.D.3, we are proposing that a facility that dried only a
portion of its distiller's grain would be treated as if it dried 100%
of its grains, and would thus need to implement additional GHG-reducing
technologies as described in the lookup table in order to qualify to
generate RINs. However, we are also taking comment on whether a
selection of pathways should be included in the lookup table that
represent corn-ethanol facilities that dry only a portion of their
distiller's grains. We also request comment on whether RINs could be
assigned to only a portion of the facility's ethanol in cases wherein
only a portion of the distiller's grains are dried.
b. Renewable Fuels from Cellulosic Biomass
    In analyzing the lifecycle GHG impacts of cellulosic ethanol, we
determined that ethanol produced from corn stover or switchgrass
through a process using enzymatic hydrolysis followed by fermentation
of the resulting sugars met the GHG threshold of 60% for cellulosic
biofuel by a wide margin (regardless of the discount rate and the time
period over which the lifecycle GHG emissions are discounted). However,
there are many other potential sources of cellulosic biomass, and other
processing mechanisms to convert cellulosic biomass into fuel. For some
of these cases, we believe that we can make determinations regarding
whether the GHG thresholds shown in Table VI.E.1-1 are likely to be
met. In addition, as the forestry component of the FASOM model is
incorporated into the analysis,

[[Page 25052]]

we will analyze pathways using planted trees, tree residue, and slash
and pre-commercial thinnings from forestland, as qualify under the
renewable biomass definition, for feedstock.
    Cellulosic biomass sources include waste biomass such as corn
stover, and crops grown specifically for fuel production such as
switchgrass. While cellulosic crops grown for the purpose of fuel
production could have land use implications in a lifecycle GHG
analysis, waste materials produced during the harvesting of some other
type of crop would not. Given that the GHG impacts of a fermentation-
based fuel production process are likely to be very similar for
cellulose from a variety of feedstocks, we believe it would be
reasonable to conclude that any cellulosic feedstock from a waste
source that is subjected to enzymatic hydrolysis followed by
fermentation of the resulting sugars would be very likely to meet the
60% GHG threshold for cellulosic biofuel. Therefore, we propose that
cellulosic ethanol produced through an enzymatic hydrolysis process
followed by fermentation using any eligible waste cellulosic feedstock
would be determined to meet the 60% GHG threshold for cellulosic
biofuel. This would include such wastes as wheat straw, rice straw,
sugarcane bagasse, forest slash and thinnings, and yard waste.
    As stated earlier, cellulosic crops grown for the purpose of fuel
production could have land use implications in a lifecycle GHG
analysis. However, the only cellulosic crop that we subjected to
lifecycle analysis was switchgrass which had a relatively small impact
of land-use. Other cellulosic crops that have been considered for fuel
production include miscanthus and trees such as poplar and willow. It
is possible that the land use impacts of miscanthus and planted trees
could be different from that for switchgrass. For instance, while
switchgrass can be grown on marginal lands, planted trees may require
more arable land to thrive. However, according to our lifecycle
analysis for switchgrass, the land use impacts could significantly
increase and the 60% threshold for cellulosic biofuel would still be
met. Therefore, we propose that the pathways shown in Table VI.E.2-1
for ethanol produced from switchgrass through an enzymatic hydrolysis
process followed by fermentation also be applied to ethanol produced
from miscanthus and planted trees. We intend to examine this pathway
more closely for the final rule to determine if this categorization is
appropriate, and request comment on the land use impacts of miscanthus
and planted trees.
    Renewable fuels can also be produced from cellulosic biomass
through various thermochemical processes rather than enzymatic
hydrolysis followed by fermentation. One example of such thermochemical
processes would be biomass gasification to produce ``syngas'' (a
mixture of hydrogen and carbon monoxide) which is then catalytically
synthesized through a Fischer-Tropsch process to produce ethanol,
diesel, gasoline, or other transportation fuels. Another example would
be a catalytic depolymerization process in which the biomass is first
catalytically cracked to smaller molecules and then polymerized under
specific combinations of temperature, pressure, and residence time to
produce a transportation fuel. We have not conducted a lifecycle
analysis of these pathways, but we believe that we can nonetheless make
a reasonable determination regarding the appropriate renewable fuel
category. For instance, we would expect that the GHG emissions produced
during fuel production would be higher for a thermochemical process
than for enzymatic hydrolysis due to the need for greater process heat
produced through the combustion of fossil fuels. However, the yield of
fuel produced per ton of biomass is likely to be greater for
thermochemical processing due to the conversion of the lignin to fuel
in addition to the cellulose and hemicellulose. Thus, while the
lifecycle GHG analyses we conducted for corn stover and switchgrass
demonstrated that the 60% GHG threshold for cellulosic biofuel would be
met by a wide margin, this margin may be smaller if a thermochemical
process was used. While we intend to conduct further analyses of this
family of pathways for the final rule, we believe that a change from
enzymatic hydrolysis to a thermochemical process would be expected to
meet the 60% GHG threshold associated with cellulosic biofuel.
Therefore, we propose that the use of corn stover or other waste
cellulosic biomass, switchgrass, or planted trees in a thermochemical
process would qualify as cellulosic biofuel under the RFS2 program.
This would include pathways that produce ethanol, cellulosic diesel, or
cellulosic gasoline. Since cellulosic diesel fuel produced in this way
would also meet the requirements for biomass-based diesel, we propose
to allow it to be categorized as either cellulosic biofuel or biomass-
based diesel at the producer's discretion. See further discussion of
this issue in Section III.D.2.a. We request comment on our proposed
assignment of categories for renewable fuels produced through a
thermochemical process, as well as data and other information relating
to the various types of thermochemical fuel production processes.
c. Biodiesel
    Our lifecycle analysis of biodiesel (mono alkyl esters) produced
from waste greases/oils demonstrated that the 50% GHG threshold for
biomass-based diesel would be met. Much of the GHG benefit of these
waste greases/oils derives from the fact that they have no land use
impacts. While we did not subject corn oil that is non-food grade to
lifecycle analysis, it is likely that it would also have no land use
impacts. Moreover, such non-food grade corn oil would require nearly
the same process energy to convert it into biodiesel. Therefore, we
propose that the pathway shown in Table VI.E.2-1 for biodiesel produced
from waste greases/oils also be applied to biodiesel produced from non-
food grade corn oil. We intend to analyze this pathway in more depth
for the final rule.
    Our lifecycle analysis of biodiesel produced from soybean oil may
also be applicable to biodiesel produced from other types of virgin
(not waste) oils. This would include canola oil, rapeseed oil,
sunflower oil, and peanut oil. While we have not conducted a detailed
assessment of the land use impacts of these other virgin oils, it is
possible that they would meet the 20% threshold for generic renewable
fuel. Therefore, we propose that the pathway shown in Table VI.E.2-1
for biodiesel produced from soybean oil also be applied to biodiesel
produced from other these virgin oils. We request comment on whether
this is appropriate.
    Although our proposed list of RIN-generating pathways would allow
biodiesel made from waste greases/oils to qualify as biomass-based
diesel, it is likely that there would be insufficient quantities of
these feedstocks to reach the 1.0 billion gallon requirement by 2012.
Biodiesel produced from soybean oil would not qualify as biomass-based
diesel, but instead would be categorized as generic renewable fuel
based on our current analysis of its lifecycle GHG performance.
However, biodiesel production facilities can process either soybean oil
or waste grease with relatively minor changes in operations, and many
facilities that formerly used soybean oil have recently switched to
waste grease due to its more favorable economics. Since the GHG performance

[[Continued on page 25053]]
Notices For
2009
2008
2007
2006
2005
2004
2003
2002
2001
2000
1999
1998
1997
1996
1995
1994
Regulation of Fuels and Fuel Additives: Changes to Renewable Fuel Standard Program

Local Navigation
