Growth Energy respectfully submits these comments on the Environmental Protection Agency’s Proposed Renewable Fuel Standard (RFS) Program: RFS Annual Rules.
Growth Energy is the world’s largest association of biofuel producers, representing 89 biorefineries that produce nearly 9 billion gallons annually of low-carbon renewable fuel and 95 businesses associated with the biofuel production process.
Congress established the RFS program to force the market to increase the use of renewable fuel in the nation’s transportation fuel supply. Congress did so in recognition of the many benefits that increased renewable fuel use would bring: reduction in harmful greenhouse gas (“GHG”) emissions, enhanced energy security and independence, and economic development. Although renewable fuels continue to promote these benefits—in fact, conventional starch ethanol achieves more than double the GHG reduction that Congress initially expected—EPA’s implementation of the RFS program to date has not fully served Congress’s
objectives. EPA has repeatedly failed to issue timely standards, and then has set the standards after the fact to the actual level of renewable fuel use. EPA has long guarded a massive RIN bank ostensibly to provide a safety valve for an emergency, with the consequence of undermining the very incentives that Congress intended the RFS program to provide. And EPA’s annual process of determining standards has often simply aimed to match the standards to what it predicted the market would do anyway, nullifying the RFS program.
Growth Energy Comments on EPA’s
Renewable Fuel Standard (RFS) Program:
RFS Annual Rules
Docket # EPA-HQ-OAR-2021-0324
Emily Skor
Chief Executive Officer
Growth Energy
701 8th Street NW
Suite 450
Washington, DC 20001
(202) 545-4000
February 4, 2022
i
TABLE OF CONTENTS
INTRODUCTION ……………………………………………………………………………………………………………1
DISCUSSION ………………………………………………………………………………………………………………….5
I. OVERVIEW OF STATUTORY FRAMEWORK: WHEN RESETTING RFS VOLUMES,
EPA’S MANDATE IS TO PRIORITIZE CONGRESS’S CORE RFS OBJECTIVES TO THE
EXTENT REASONABLY FEASIBLE, UNLESS DOING SO WOULD CAUSE IMPORTANT
AND SEVERE HARM …………………………………………………………………………………………………5
A. EPA Must Implement the Reset So as to Serve the RFS Program’s Core
Objectives ………………………………………………………………………………………………….5
B. The Statutory Structure Shows That the Reset Serves as a Prospective
Multi-Year Waiver ………………………………………………………………………………………8
C. The Proper Framework for Analyzing the Statutory Reset Factors Is
Spurring Increased Use of Renewable Fuels Except to the Extent That
Such Use Would Be Infeasible or Would Cause Important and Severe
Harm …………………………………………………………………………………………………………9
D. EPA’s Proposed Approach to the Statutory Factors Is Flawed Because It
Disregards the Proper Framework for Conducting a Reset ……………………………..12
II. THE CENTRAL ROLE OF CLIMATE CHANGE TO THE RFS PROGRAM AND TO THE
RESET ANALYSIS ………………………………………………………………………………………………….13
A. Consistent with the Overarching Reset Framework, This Reset Presents
EPA with an Opportunity to Further the Administration’s Climate Goals
Through Full and Effective Implementation of the RFS Program ……………………13
1. The President has prioritized climate action and set ambitious
GHG reduction goals ………………………………………………………………………13
2. The Administration’s climate goals are not attainable without
significant and immediate GHG reductions in the transportation
sector, and these reductions are not feasible without biofuels ……………….14
3. Demand for liquid fuels will persist for the foreseeable future ……………..15
4. EPA and the Administration have repeatedly recognized the
necessity of biofuels to decarbonize the transportation fuel supply
in the United States …………………………………………………………………………17
5. EPA has undervalued the GHG reduction potential of biofuels …………….19
B. Updating the GHG LCA for Ethanol Should Be a Top Priority for EPA …………..19
1. The best available science indicates corn ethanol has more than
double the GHG emissions reductions of EPA’s outdated estimate,
putting it roughly on par with advanced biofuels ………………………………..22
2. Uncertainty in LCA modeling can be managed by comparing
existing credible studies …………………………………………………………………..25
ii
3. EPA’s “illustrative” scenario of GHG impacts of the proposed rule
is flawed and misleading ………………………………………………………………….27
III. EPA OVERESTIMATES OTHER ENVIRONMENTAL IMPACTS ………………………………………….27
A. There Is No Credible Evidence That the Proposed Rule Will Cause
Adverse Impacts to Wetlands, Ecosystems, Species, Habitat, Water
Quality/Availability, or Soils ………………………………………………………………………27
1. Claims that the RFS causes land use change are unsubstantiated, as
EPA has previously acknowledged and must incorporate into its
analysis here …………………………………………………………………………………..28
2. The RFS program has not caused adverse impacts to wetlands,
wildlife, ecosystems; nor will the Proposed Rule ………………………………..37
3. The RFS Program has not caused adverse impacts to soil and water
quality; nor will the Proposed Rule …………………………………………………..39
4. The RFS Program has not caused adverse impacts to water
quantity and availability; nor will the Proposed Rule …………………………..41
5. EPA does not adequately assess the environmental impacts
associated with petroleum-based fuels ……………………………………………….42
B. EPA Should Finalize a “No Effects” Finding Under Section 7 of the
Endangered Species Act …………………………………………………………………………….43
C. Air Quality Impacts of Proposed Rule ………………………………………………………….46
D. Increasing Renewable Fuel Volumes Benefits Communities with
Environmental Justice Concerns ………………………………………………………………….47
1. Environmental justice and climate change …………………………………………47
2. Environmental justice and air quality ………………………………………………..47
3. Environmental justice and water/soil impacts …………………………………….47
IV. EPA’S PROPOSED MODIFICATION OF THE 2020 STANDARDS UNDERMINES THE
RFS PROGRAM, CONTRADICTS THE CLEAN AIR ACT, AND IS IRRATIONAL …………………..48
A. EPA’s Proposal to Retroactively Lower Previously Set Standards to the
Level of Actual Use Negates the RFS and Therefore Is Impermissible …………….49
B. The Reset Power Is Not Available to Reduce Already-Set Standards for a
Past Year or to Reduce Volume Requirements More Than Needed to
Address the Circumstances That Triggered the Reset …………………………………….51
1. EPA cannot use the reset power to retroactively reduce past
standards ……………………………………………………………………………………….52
2. EPA has no statutory authority to use the reset to reduce volumes
further than needed to address the cellulosic production issues
triggering the reset ………………………………………………………………………….54
C. EPA’s Proposed Rationale for Reducing the 2020 Standards Is Irrational
and Contrary to the Statute …………………………………………………………………………55
iii
1. Pandemic-related demand destruction does not justify reducing the
2020 percentage standards ……………………………………………………………….56
2. EPA’s apparent over-projection of SREs and the shortfall in
production of cellulosic biofuel could justify at most only some of
the proposed reduction of the 2020 standards ……………………………………..56
3. The disproportionate decline in gasoline use relative to diesel use
does not justify further retroactive reduction of the 2020 standards ………58
4. EPA’s refusal to allow supposedly “disruptive outcomes” violates
the statute and undermines the RFS program ……………………………………..59
5. EPA’s management of the RIN bank is incoherent and exposes
EPA’s mistaken belief that its role is to manage the fuel market …………..60
V. EPA’S PROPOSAL TO SET THE 2021 STANDARDS TO ACTUAL LEVELS UNLAWFULLY
NEGATES THE RFS PROGRAM …………………………………………………………………………………62
VI. EPA SUBSTANTIALLY UNDERSTATES THE REASONABLY FEASIBLE VOLUME OF
ETHANOL USE IN 2022 …………………………………………………………………………………………..65
A. More Than 15 Billion Gallons of Ethanol Could Easily Be Produced
Domestically …………………………………………………………………………………………….66
B. Substantially More E85 and E15 Could Easily Be Delivered and
Consumed ………………………………………………………………………………………………..67
1. E85 ……………………………………………………………………………………………….67
2. E15 ……………………………………………………………………………………………….68
C. The Principal Impediment to Increased Use of E85 and E15 Is Retail
Price, Which Can Be Addressed Through Higher RFS Standards ……………………69
D. EPA’s Speculative Concerns About Misfuelling Liability Are Unfounded
and Irrelevant ……………………………………………………………………………………………70
E. Storage Infrastructure Compatibility Is Not a Meaningful Barrier to
Increased Use of Ethanol or Expansion of Distribution Infrastructure ………………71
iv
VII. IN SETTING RFS STANDARDS, EPA SHOULD BACKFILL SHORTFALLS WITH ANY
OTHER AVAILABLE QUALIFYING RENEWABLE FUELS ………………………………………………..74
VIII. EPA MUST INCLUDE CARRYOVER CELLULOSIC RINS IN THE AVAILABLE VOLUME
WHEN REDUCING THE CELLULOSIC VOLUME REQUIREMENT ………………………………………76
IX. THE PROPOSED RFS STANDARDS WOULD NOT APPRECIABLY RAISE RETAIL PRICES
FOR FOOD OR GASOLINE ………………………………………………………………………………………..78
X. EPA’S PROPOSED RESPONSE TO ACE REMAND IS NECESSARY AND APPROPRIATE …………79
XI. EPA SHOULD RETAIN THE STANDARD EQUATION AS REVISED IN THE 2020 RULE ………….81
XII. ROBUST RFS REQUIREMENTS PROMOTE RURAL ECONOMIC HEALTH …………………………..84
XIII. EPA IS OBLIGATED TO ADJUST THE 2022 RFS STANDARDS TO MAKE UP FOR PAST
RETROACTIVE SRES ……………………………………………………………………………………………..85
A. EPA’s Refusal Violates Its Statutory Duty to Set Standards That “Ensure”
That the Required Volumes Are Met ……………………………………………………………85
B. EPA’s Refusal Is Arbitrary and Capricious …………………………………………………..87
C. EPA’s Refusal Impermissibly Creates for Itself a Non-textual Waiver
Power ………………………………………………………………………………………………………87
XIV. BIOINTERMEDIATES ………………………………………………………………………………………………88
XV. EPA SHOULD APPROVE KERNEL FIBER REGISTRATIONS AND PENDING BIOFUEL
PATHWAYS ………………………………………………………………………………………………………….90
XVI. EPA SHOULD ADOPT THE PROPOSED APPROACH TO CONFIDENTIAL BUSINESS
INFORMATION ………………………………………………………………………………………………………91
1
INTRODUCTION
Growth Energy respectfully submits these comments on the Environmental Protection
Agency’s Proposed Renewable Fuel Standard (RFS) Program: RFS Annual Rules.
1
Growth
Energy is the world’s largest association of biofuel producers, representing 89 biorefineries that
produce nearly 9 billion gallons annually of low-carbon renewable fuel and 95 businesses
associated with the biofuel production process.
Congress established the RFS program to force the market to increase the use of
renewable fuel in the nation’s transportation fuel supply. Congress did so in recognition of the
many benefits that increased renewable fuel use would bring: reduction in harmful greenhouse
gas (“GHG”) emissions, enhanced energy security and independence, and economic
development. Although renewable fuels continue to promote these benefits—in fact,
conventional starch ethanol achieves more than double the GHG reduction that Congress initially
expected—EPA’s implementation of the RFS program to date has not fully served Congress’s
objectives. EPA has repeatedly failed to issue timely standards, and then has set the standards
after the fact to the actual level of renewable fuel use. EPA has long guarded a massive RIN
bank ostensibly to provide a safety valve for an emergency, with the consequence of
undermining the very incentives that Congress intended the RFS program to provide. And
EPA’s annual process of determining standards has often simply aimed to match the standards to
what it predicted the market would do anyway, nullifying the RFS program.
Although EPA’s current proposal contains several salutary features, it also again reflects
these and other fundamental errors in the agency’s approach to the RFS program. On the one
hand, the proposal includes the conclusion that a non-advanced volume of 15 billion gallons of
renewable fuel is readily achievable, a long overdue remedy of the unlawful 2016 general
waiver, and a long overdue decision to bring small refinery exemption (“SRE”) decisions into
the sunlight. On the other hand, many aspects of the proposal would seriously damage the RFS
program and violate EPA’s legal duties, by, for example, substantially undervaluing the benefits
of conventional ethanol for climate change, relieving obligated parties of their failure to meet
their 2020 obligations (even after accounting for the actual levels of fuel use and SREs in 2020),
and nullifying the program for 2021. Growth Energy, therefore, urges EPA to carefully
reconsider many parts of its proposal to ensure that they accord with the goals Congress set for
the RFS program and the limits Congress placed on EPA’s authority.
Growth Energy also urges EPA to finalize this rulemaking expeditiously. Compliance
year 2022 is well underway and market participants need clear, definite RFS signals. As this
proposal shows, delay translates into lost opportunity to encourage increased renewable fuel use
and fulfill the program’s aims. It is imperative that EPA minimize delay in issuing standards.
More specifically, in this comment, Growth Energy argues as follows:
Part I: EPA must adopt a framework for performing a reset that is faithful to the RFS
Program’s statutory structure and purpose. In proposing standards for 2020, 2021, and 2022,
1 Renewable Fuel Standard (RFS) Program: RFS Annual Rules, Proposed Rule (“NPRM”), 86
Fed. Reg. 72,436 (Dec. 21, 2021).
2
EPA invokes its reset authority for the first time. Contrary to EPA’s proposed approach, the
reset is not a valid mechanism to re-open previously finalized standards, to override
congressional directives and priorities, or to engage in an amorphous balancing of factors as it
sees fit. Rather, Congress intended reset mechanism to be a targeted prospective correction for
the specific conditions that triggered the reset. In conducting a reset, EPA must still establish
volume requirements that, first and foremost, further Congress’s market-forcing policy and
objectives, to the extent that a volume of renewable fuel use is feasible and will not cause
important and severe harm of the type that would trigger another waiver. Further, EPA must
always take into account the best available science when performing a reset.
Part II: EPA should prioritize climate change impacts and must incorporate the best
available science in its analysis. Reducing GHG emissions from the transportation sector is a
core congressional objective of the RFS—indeed the RFS is the only Clean Air Act program
explicitly aimed at reducing GHG emissions—and deserves special emphasis. Congress’s intent
that implementing the RFS’s market-forcing policy will achieve the full measure of available
GHG reductions from transportation fuel aligns with EPA’s and the Administration’s stated
climate goals and efforts to decarbonize the transportation sector. Doing so requires EPA, in this
rulemaking, to update its lifecycle GHG emissions analysis for conventional corn ethanol using
the best currently available science, which is much more favorable than EPA’s existing 2010
lifecycle analysis. Specifically, two expert reports on lifecycle analysis should guide EPA’s
update: (1) Environmental Health and Engineering, a multidisciplinary team of environmental
health scientists and engineers, presents critical analysis of available credible studies deriving a
central best estimate of carbon intensity, taking uncertainty into account; and (2) Stefan Unnasch
of Lifecycle Associates, an independent environmental consulting firm specializing in lifecycle
analysis of fuel production pathways, presents a comprehensive review of the emissions factors
of the ethanol lifecycle and explains why several key elements of EPA’s 2010 lifecycle analysis
are demonstrably no longer valid.
Part III: EPA must correct its overstatement of potential adverse environmental impacts
associated with RFS. There is no credible evidence that the proposed standards will adversely
affect wetlands, ecosystems, species, habitat, water quality/availability, or soils. More
specifically, claims that the RFS causes land use change are unsubstantiated, and EPA’s
environmental assessment should reflect the absence of an established causal relationship
between the RFS and any adverse impacts to wildlife habitat, ecosystems, or species. There is
similarly no sound science that the RFS program has caused or will cause adverse impacts to
water quantity or quality. EPA must correct the record on its treatment of these issues.
Moreover, EPA may, and should, make a finding under Section 7 of the Endangered Species Act
that the proposed 2022 standards would have no effect on listed species or habitat, nor could the
proposed 2020 and 2021 standards conceivably have such an impact because those years are
entirely past. On the other hand, EPA fails to recognize the potentially significant benefits to air
quality associated with ethanol-blended fuels. Finally, EPA’s environmental justice analysis
makes wholly unsupported statements regarding adverse impacts to soil and water that may have
negative impacts on environmental justice communities. EPA should continue to recognize the
important role biofuels may play in mitigating disproportionate impacts of climate change on
low-income and vulnerable communities, as well as the air quality benefits of ethanol-blended
fuels for these communities.
3
Part IV: EPA’s proposed modification of the 2020 standards undermines the RFS
program, contradicts the Clean Air Act, and is irrational. EPA proposes to retroactively reduce
the 2020 volumes to those actually used in 2020. This is plainly unlawful. EPA has no power to
relieve obligated parties of their noncompliance simply because they did not comply. Doing so
nullifies the RFS program. Congress designed the RFS program to force the market to use
increasing volumes of renewable fuel each year, and the threat of penalties for noncompliance is
the mechanism by which the program implements this design. EPA’s proposed retroactive
absolution creates a perverse incentive: obligated parties will have no reason to bother
complying with RFS standards. When they fail, EPA will absolve them, and the more they fail,
the more likely EPA is to save them. Congress did not grant EPA such a counterproductive
power. Certainly, the reset provision does not grant such power. The reset provision was
intended to enable EPA to prospectively issue a multi-year waiver to remedy the circumstances
that triggered the reset. Moreover, EPA’s assessment of the appropriate reduction of the 2020
standards is flawed. The standards automatically account for lower-than-projected demand for
transportation fuel in 2020. Further, EPA should not reduce the standards to account for lowerthan-projected SREs. But even if those adjustments were valid, EPA’s proposed reduction
would still exceed them by more than half a billion RINs. None of EPA’s reasons for the
reduction is well-founded or rational. Having to use carryover RINs, to carry forward RIN
deficits, or to incur noncompliance penalties are precisely how the RFS program provides
compliance flexibility and ultimately ensures compliance; those are not adverse consequences
EPA can or should alter the standards after the fact to avoid. EPA’s approach to the RIN bank is
particularly troubling; EPA has long said the bank would provide a cushion for a major
unforeseen compliance disruption, and yet now EPA proposes not to draw on the bank because
there was an unforeseen compliance disruption, namely, the Covid-19 pandemic. That is
irrational and undermines the RFS program.
Part V: EPA’s proposal to set the 2021 standards to actual levels unlawfully negates the
RFS Program. EPA’s proposal to set 2021 volumes at levels of actual use is also unlawful.
Again, the reset authority cannot be used retroactively or to go beyond remedying the conditions
that triggered the reset. EPA’s proposal also negates the RFS program for 2021. Although the
D.C. Circuit previously approved of an approach similar to what EPA proposes for 2021, that
decision does not condone EPA’s now-routine procedure of delay and negation. The Clean Air
Act cannot be interpreted to grant EPA the power to cancel the program, which is the effect of
EPA’s proposal. In setting 2021 standards, EPA should instead—as it has in some past years—
determine the volumes based on data available as of November 30, 2020, and allow obligated
parties to comply through carryover RINs and deficit carryforward. That approach fulfills
Congress’s clear directive that EPA ensure that the volume requirements be met regardless of
EPA’s delay in issuing standards.
Part VI: EPA substantially understates the reasonably feasible volume of ethanol use in
2022. EPA’s proposed 2022 standards reflect an unjustifiably low expectation of ethanol use.
There is ample production-facility capacity, feedstock, distribution infrastructure, and vehicles to
consume vastly more ethanol than EPA assumes—without adverse environmental or economic
consequences. Indeed, projected ethanol volumes assume that only a miniscule fraction of
existing infrastructure for delivering and consuming E85 and E15 will be used. The so-called
E10 blendwall is not the problem; it is the consequence of market conditions that do not
encourage the market to shift from E10 to higher-ethanol blends. The solution is precisely how
4
Congress intended the RFS program to function: higher RFS requirements would raise RIN
prices, encouraging investments and discounts of higher-ethanol blends relative to E10, thereby
motivating consumers to choose those blends at substantially higher volumes.
Part VII: In setting RFS standards, EPA should backfill shortfalls with any other
available qualifying renewable fuels. The Clean Air Act and principles of reasoned
decisionmaking require that EPA backfill any renewable fuel shortfall with any other types of
reasonably available qualifying renewable fuel, unless doing so could trigger a waiver or
otherwise cause important and severe harm. EPA’s 2022 proposal ignores this by mechanically
reducing the total renewable fuel standard by the same amount as the projected cellulosic
shortfall. The result is to unnecessarily lower total renewable fuel volumes, thereby increasing
GHG emissions from standard fossil fuels that could be replaced with other renewable fuels.
This shortchanges the RFS program’s ability to achieve both Congress’s and the administration’s
environmental goals.
Part VIII: EPA must include carryover cellulosic RINs in the available volume when
reducing the cellulosic volume requirement. The Clean Air Act’s text and purpose require that,
for purposes of exercising the cellulosic waiver, EPA count available carryover cellulosic RINs
toward the projected volume of cellulosic fuel that is available during a calendar year.
Part IX: The proposed RFS standards would not appreciably raise retail prices for food
or gasoline. Prices for 2020 and 2021 are already past and therefore cannot be affected by this
proposal. And EPA’s proposed non-advanced volume of 15 billion gallons for 2022—even if
filled entirely with conventional ethanol—would not divert corn from projected non-ethanol uses
and thus would not increase corn or associated food prices. Further, strong empirical data show
that expected renewable fuel use will actually lower retail gasoline prices.
Part X: EPA’s proposed response to the invalidation of its 2016 general waiver on
remand is necessary and appropriate. Growth Energy appreciates EPA’s proposed remedy for
the unlawful general waiver of the 2016 standards on remand from the D.C. Circuit. That error
has long undermined the RFS program by inflating the RIN bank, which in turn has suppressed
RIN prices and dampened the RFS program’s incentives to increase the use of renewable fuel.
Only by remedying the unlawful error through a make-up obligation does EPA comply with the
court’s judgment and fulfill its statutory duty to ensure that the legally valid applicable volumes
are met. It is imperative, though, that EPA also issue its promised second 250-million-gallon
supplemental requirement for 2023.
Part XI: EPA should retain the standard equation as revised in the 2020 rule. Even if
EPA adopts the standards and findings of its separately proposed denial of pending SRE petitions
(as it should), EPA should still retain the standard equation as revised by the original 2020 rule,
so that the equation adjusts for projected SREs. That would enable EPA to set future standards
that are rationally and reasonably calculated to ensure that the applicable volume requirements
are met, as EPA is statutorily required to do.
Part XII: Robust RFS requirements promote rural economic health. As Congress
expected, a commitment to growth in the use of ethanol promotes meaningful economic benefits,
especially in rural and agricultural areas of the country. The ethanol industry supports hundreds
5
of thousands of jobs and creates billions of dollars of household wealth and GDP. Increasing
ethanol use would help grow these benefits.
Part XIII: EPA is obligated to adjust the 2022 RFS standards to make up for past
retroactive SREs. The massive volume of retroactive SREs that EPA has granted in the past
have ballooned the RIN bank and undermined the market-forcing effect of the RFS program. To
fulfill its statutory duty to ensure that the volumes are met, EPA must adjust the 2022 standards
to offset the past retroactive SREs. Failing to do so also violates principles of reasoned decision
by disregarding a central problem with the standard-setting task and by setting standards that
EPA knows in advance will fail to serve their intended purpose. This failure also converts
exemptions into atextual waiver, which EPA has no authority to do.
Part XIV: It is imperative that renewable fuel producers have flexibility to use
biointermediates in fuel production in order to lower costs and drive innovation. EPA should
ensure that the final biointermediates regulations facilitate use of biointermediates, afford needed
flexibility to producers, and are not unduly burdensome on potential biointermediates or
renewable fuel producers.
Part XV: EPA should act expeditiously to approve the numerous pending registration
applications for simultaneous production of starch and cellulosic ethanol from corn kernel
feedstock. Growth Energy urges EPA to expedite pathway approval for carbon capture,
utilization, and storage, and to approve the pending petition to allow biodiesel and renewable
diesel facilities to use corn oil produced from corn wet mills as feedstock.
Part XVI: EPA should adopt the proposed approach to confidential business
information. Growth energy supports EPA’s proposal not to treat as confidential basic
information relating to SRE petitions and SRE decisions for purposes of the Freedom of
Information Act. EPA thwarts essential oversight and engages in secret national lawmaking
when it conceals its SRE decisions. EPA’s proposal accords with recent case law, Justice
Department guidance, and good government practices. EPA has made similar proposals in the
past; now is the time to finally adopt this important policy change.
DISCUSSION
I. OVERVIEW OF STATUTORY FRAMEWORK: WHEN RESETTING RFS VOLUMES, EPA’S
MANDATE IS TO PRIORITIZE CONGRESS’S CORE RFS OBJECTIVES TO THE EXTENT
REASONABLY FEASIBLE, UNLESS DOING SO WOULD CAUSE IMPORTANT AND SEVERE
HARM
A. EPA Must Implement the Reset So as to Serve the RFS Program’s Core
Objectives
In the Energy Policy Act of 2005 (“EPAct”) and Energy Independence and Security Act
of 2007 (“EISA”), Congress created the RFS and established annual renewable fuel volumes in
order to achieve three principal objectives: (1) reduce GHG emissions to address climate change;
(2) improve U.S. energy security; and (3) support agricultural development. These fundamental
6
purposes are reflected in the statutory text codified at section 211(o) of the Clean Air Act;2
wellsettled by the case law3
and legislative history;4
and EPA’s own statements (e.g., “Congress
created the renewable fuel standard (RFS) program to reduce greenhouse gas emissions and
expand the nation’s renewable fuels sector while reducing reliance on imported oil.”).5
Moreover, as discussed in greater detail below, Congress intended the RFS to be marketforcing—the statutory volumes are not mere projections of what the market could be anticipated
to achieve in the absence of the program and applied in that manner they would serve no
purpose. “Congress intended the Renewable Fuel Program to be a market forcing policy that
would create demand pressure to increase consumption of renewable fuel.”6
The steadily
increasing volumes established by the RFS are designed–by sending appropriate price signals to
the market—to incentivize and accelerate the transition from petroleum-based fuels to biofuels,
2
121 Stat. 1492, Energy Independence and Security Act of 2007.
3 American for Clean Energy v. EPA (“ACE”), 864 F.3d 691, 696 (D.C. Cir. 2017) (“Congress
intended the Renewable Fuel Program to move the United States toward greater energy
independence and to reduce greenhouse gas emissions.”); Growth Energy v. EPA, 5 F. 4th 1, 7
(D.C. Cir. 2021) (“To move the United States towards greater reliance on clean energy, the
Clean Air Act’s Renewable Fuel Standard Program calls for annual increases in the amount of
renewable fuel introduced into the U.S. fuel supply.”); American Fuel & Petrochemical Mfrs. v.
EPA, 937 F.3d 559, 568 (D.C. Cir. 2019) (“Enacted in 2005 and amended in 2007, the
Renewable Fuel Program … was designed ‘[t]o move the United States toward greater energy
independence and security’ and ‘to increase the production of clean renewable fuels.’”).
4
149 Cong. Rec. S5986, 2003, Statement of Sen. Tim Johnson, Co-Sponsor (“Simply put,
adoption of the RFS amendment will help lower our dependence on foreign oil, strengthen
energy security, increase farm income, provide for clean air, and create jobs throughout the
United States, particularly in the rural communities.”); id. at S5985, Statement of Tom Daschle,
Co-Sponsor (“Clean air benefits cannot be understated. In 2002 alone—just last year—ethanol
use in the United States reduced greenhouse gas emissions by 4.3 million tons, which is the
equivalent of removing more than 636,000 vehicles from the road. That is a remarkable
achievement.”); id. at S6048, Statement of George Voinovich, Co-Sponsor (“Importantly,
renewable fuels help to reduce greenhouse gases emitted from vehicles. Including carbon
dioxide, methane, and other gases that contribute to global warming—another answer to the
problem of carbons.”).
5 Renewable Fuel Standard Program, EPA, https://www.epa.gov/renewable-fuel-standardprogram; see also, e.g., NPRM at 72,439 (recognizing congressional “intent [behind the RFS] to
support increasing production and use of renewable fuels, and the potential positive impacts of
renewable fuels on several of the statutory factors such as climate change and energy security”);
EPA, Renewable Fuel Standard Program (RFS2) Summary and Analysis of Comments 1-1 (Feb.
2010), https://nepis.epa.gov/Exe/ZyPDF.cgi?Dockey=P1007GC4.pdf (“As our analysis in
support of the rulemaking demonstrates, we believe that the increase[d] use of renewable fuels in
place of petroleum fuels will provide both greenhouse gas and energy benefits to our nation, as
well as significant economic benefits to our agricultural sector.”).
6 ACE, 864 F.3d at 705 (quotation marks omitted).
7
in order to fully capture their societal, economic, and environmental benefits. Various
statements in EPA’s current proposal recognize the market-forcing intent of the RFS.7
The Proposed Volume Standards for 2020, 2021, and 2022 (“Proposed Rule”) is
significant because it marks the first time EPA has invoked its authority under Section
211(o)(2)(B)(ii) to reset the statutory volumes originally established by Congress.8
The reset is
not, however, an occasion for EPA to reconsider or refashion the purposes behind the RFS
program; nor is it an authorization to retroactively reopen annual RVO rules finalized in previous
years. It is, rather, the opportunity provided by Congress for EPA to conduct a midcourse
correction and to recalibrate the statutory volume requirements in order to pursue full
implementation of the program’s market-forcing function, consistent with the unforeseen
constraints that triggered the reset.
To be sure, Congress identified a set of factors for EPA to consider in establishing new
volumes in the event the reset is triggered.9
And EPA has a certain degree of discretion in the
way it analyzes and applies these factors. But Congress did not intend these factors to be an
excuse for EPA to write on a clean slate, to substitute its own judgment for Congress’s purposes,
or to engage in a free-form weighing and balancing of factors to establish new volumes however
EPA sees fit in pursuit of whatever objectives EPA wishes to prioritize.10
In short, Congress did not intend the reset to be unmoored from the RFS program’s
original intent or from the conditions that triggered the reset. Correspondingly, Congress did not
intend the reset mechanism to authorize a complete reworking of those parts of the RFS program
that are functioning properly. To the contrary, Congress created the reset authority to allow EPA
to make a targeted correction in the implementation of the RFS program under a narrowly
circumscribed set of circumstances, and then to assess a number of factors, but only to the extent
relevant to determining the best way to address the circumstances that triggered the reset in the
first place and while continuing to pursue the program’s purposes. Thus, the reset primarily
empowers EPA to implement a multi-year waiver to spare itself and interested parties the
burdens and uncertainty of annual, ad hoc waiver proceedings, and consideration of the
remaining statutory factors is relevant only to the extent they help EPA improve implementation
7 See, e.g., NPRM at 72,439 (noting EPA’s reliance on “our assessment of the ability for the RFS
program to incentivize increased production and use of renewable fuel in 2022”).
8 See NPRM at 72,443 n.33.
9
42 U.S.C. § 7545(o)(2)(B)(ii).
10 Cf. NPRM at 72,443 (“While the statute requires that EPA base its determination on an
analysis of these factors, it does not establish any numeric criteria, require a specific type of
analysis (such as quantitative analysis), or provide guidance on how EPA should weigh the
various factors.”).
8
of the program to better realize the program’s core objectives. This is evident from the basic
structure of the RFS program and how the reset authority fits within that structure.11
B. The Statutory Structure Shows That the Reset Serves as a Prospective MultiYear Waiver
In the EPAct and EISA, Congress established applicable volumes for various categories
of renewable fuels in a statutory table. The table provides annual applicable volumes through
2022 for total, advanced, and cellulosic biofuels, and through 2012 for biomass-based diesel.12
Congress authorized EPA to “waive” these statutory applicable volumes but only in
circumstances where certain expressly prescribed conditions are met. For example, EPA may
invoke the “general waiver” provision under Section 211(o)(7)(A) to reduce volumes from any
category of renewable fuel if either EPA determines that the statutory volume would severely
harm the economy or environment of a State, a region, or the United States, or if EPA determines
that there is inadequate domestic supply of renewable fuel to meet the statutorily required
volume.13 Further, EPA must invoke the “cellulosic waiver” provision under Section
211(o)(7)(D) to reduce the volume of cellulosic biofuel if EPA determines that cellulosic biofuel
production is projected to be less than the statutory volume.14
EPA’s reset authority under Section 211(o)(7)(F) is triggered only if EPA invokes these
waivers to reduce the statutory volume for a given category of renewable fuel by at least 20
percent for two consecutive years or by at least 50 percent for a single year.15 At that point, EPA
must establish new applicable volumes for the waived renewable fuel category, and it must do so
for all of the remaining years in the statutory table.16
EPA must exercise its reset authority within one year of the Agency action that triggered
the waiver.17 EPA must also finalize the reset volumes at least 14 months prior to the first year to
which they are applicable.18 EPA’s reset must be based on the Agency’s assessment of a set of
statutory factors organized into six groups,19 and EPA must carry out this analysis in
11 See King v. Burwell, 576 U.S. 473, 498 (2015) (holding that a statute’s structure and purpose
are vitally important to its interpretation since “[a] fair reading of legislation demands a fair
understanding of the legislative plan”).
12 42 U.S.C. § 7545(o)(2)(B)(i)(I-IV).
13 Id. at § 7545(o)(7)(A).
14 Id. at § 7545(o)(7)(D).
15 Id. at § 7545(o)(7)(F).
16 Id.
17 Id.
18 Id. at § 7545(o)(2)(B)(ii).
19 Id.
9
coordination with the Secretary of the Department of Energy (“DOE”) and the Secretary of the
U.S. Department of Agriculture (“USDA”).20
Once the last year in the table is reached for a given category of renewable fuel (e.g., the
last year in the table for total, advanced, and cellulosic renewable fuel is 2022), it is then
incumbent on EPA to establish new applicable volumes for all future years, involving the same
set of statutory factors that governs the reset.21 (The process of setting new applicable volumes
for the years beyond the original statutory table is referred to as the “set” rulemaking process.)
Importantly, while both the reset mechanisms rely on the same list of statutory factors that apply
to the “set,” their purposes and functions are distinct because of the different circumstances that
trigger each mechanism, and EPA’s approach to each must reflect this distinction. A set
rulemaking is triggered simply by the passage of time, and EPA’s role is to continue pursuing the
core congressional objectives for the RFS in light of the present circumstances—which is to say,
to set volumes that strive to increase the use of renewable fuel to the extent feasible unless doing
so would cause important and severe harm. A reset rulemaking is similar but, importantly, has a
narrower primary focus: to remedy the specific circumstances that triggered the reset in lieu of
repeated annual waivers.
In the current rulemaking, the reset was triggered by repeated use of the cellulosic
waiver. EPA’s central focus, therefore, should be to adjust the volume requirements to account
for the circumstances that are leading to consistent shortfalls in cellulosic production. EPA
should use the statutory reset factors to anticipate the level of cellulosic production that can
feasibly be achieved without in turn triggering another type of waiver or otherwise causing
important and severe harm, and set the cellulosic volume requirement to that level. Further,
because the cellulosic standard is nested, EPA should reduce the advanced and total volume
requirements correspondingly, except if EPA determines that higher levels of renewable fuel use
can feasibly be achieved without in turn triggering another type of waiver or otherwise causing
important and severe harm. This approach best serves Congress’s purpose of promoting
increased renewable fuel use to reduce GHG emissions, enhance U.S. energy security, and
support economic development, and best accounts for the statutory structure and context of the
reset.
C. The Proper Framework for Analyzing the Statutory Reset Factors Is
Spurring Increased Use of Renewable Fuels Except to the Extent That Such
Use Would Be Infeasible or Would Cause Important and Severe Harm
Once the reset process is triggered, EPA must follow the analysis Congress set forth in
Section 211(o)(2)(B)(ii). EPA is required, first and foremost, to engage in a backward-looking
assessment of how the program has performed to date.22 This gives EPA the opportunity to
consider how program implementation may be improved to better achieve the RFS program’s
20 Id.
21 Id.
22 Id.; acknowledged by EPA at EPA, Draft Regulatory Impact Analysis: RFS Annual Rules
(“DRIA”) 8 (Dec. 2021), https://nepis.epa.gov/Exe/ZyPDF.cgi?Dockey=P1013KOG.pdf.
10
core congressional objectives, particularly by assessing and responding to the conditions that led
to the particular waiver that triggered the reset.
EPA must then engage in a forward-looking assessment/projection of the impacts of
setting volumes at certain levels in future years. This gives EPA the opportunity to consider how
the objectives of the program may best be achieved, given the results of the Agency’s analysis.
In both its backward-looking and forward-looking assessments, EPA must consider a
variety of statutory factors loosely collected into six groups:
I. [T]he impact of the production and use of renewable fuels
on the environment, including on air quality, climate change,
conversion of wetlands, ecosystems, wildlife habitat, water
quality, and water supply;
II. [T]he impact of renewable fuels on the energy security of the
United States;
III. [T]he expected annual rate of future commercial production
of renewable fuels, including advanced biofuels in each
category (cellulosic biofuel and biomass-based diesel);
IV. [T]he impact of renewable fuels on the infrastructure of the
United States, including deliverability of materials, goods,
and products other than renewable fuel, and the sufficiency
of infrastructure to deliver and use renewable fuel;
V. [T]he impact of the use of renewable fuels on the cost to
consumers of transportation fuel and on the cost to transport
goods; and
VI. [T]he impact of the use of renewable fuels on other factors,
including job creation, the price and supply of agricultural
commodities, rural economic development, and food
prices.23
These factors fall roughly into three categories: (1) factors relating to core statutory
objectives; (2) capacity constraints/feasibility; and (3) other potential economic and
environmental harms (or benefits), largely consonant with those considered in the severe
economic or environmental harm general waiver.24 Although the statutory list itself does not
provide EPA with a precise method to weigh the factors, not all factors are of equal importance
or deserve equal weight; the structure and purpose of the statute demonstrate that each group of
factors serves a distinct role in the analytical hierarchy.
First, certain key factors directly correspond to the core congressional objectives behind
the RFS program: i.e., the impact of renewable fuels on climate change (para. I); energy security
(para. II); and job creation and rural economic development (para. VI).25 These are the
23 42 U.S.C. § 7545(o)(2)(B)(ii).
24 Id. § 7545(o)(7)(A)(i).
25 See, e.g., NPRM at 72,439 (recognizing “the potential positive impacts of renewable fuels on
several of the statutory factors such as climate change and energy security.”).
11
environmental, economic, and societal benefits of increasing renewable fuel use that Congress
expressly sought to promote in enacting the RFS. Under the market-forcing structure of the
RFS, EPA must prioritize achievement of these goals through increased renewable fuel usage to
the extent feasible without causing severe negative harm.26
Second, certain factors relate to the feasibility of achieving a level of renewable fuel use:
the expected rate of commercial production (para. II) and the sufficiency of infrastructure to
deliver and use renewable fuel (para. IV). These factors reflect Congress’s intention that,
although EPA should use the volume requirements to increase the use of renewable fuel, that
effort is limited by how much renewable fuel can feasibly be produced, delivered, and
consumed. These feasibility factors provide the practical cap beyond which the market cannot
be forced.
Third, the remaining factors represent other potential negative (or positive) effects of
increased renewable fuel use: additional environmental impacts other than the key congressional
objective of addressing climate change, including air quality, conversion of wetlands,
ecosystems, wildlife habitat, water quality, water supply (para. I); the deliverability of products
other than renewable fuel (para. IV); fuel prices (para. V); and the prices of other goods and
foods (para. VI).27 Congress recognized the possibility that increasing the use of renewable fuel
beyond a certain point could have other adverse environmental or economic effects, and directed
EPA to account for them. But Congress did not intend for EPA to weigh these effects equally
with the primary goals of the program—that would frustrate the purpose of the RFS and involve
a far greater, and suspect, delegation of policymaking authority to EPA. Rather, the standard
Congress established for the general waiver—“severe” harm28—provides the guidepost for
weighting these additional factors. As the purpose of the reset is to address—and, moving
forward, to avoid—the conditions that result in annual waivers, Congress understandably
intended EPA to consider this third group of reset factors so as to avoid the future need for a
severe environmental or severe economic harm waiver.29 EPA may reduce volume requirements
26 And again, as noted above, EPA is permitted to reduce the volumes in the statutory table only
with respect to the category of renewable fuel that triggered the reset, and only so far as
necessary to address the underlying problem that caused it.
27 As noted elsewhere in these comments, several of these factors illustrate the beneficial impacts
of biofuel use. For example, use of biofuels improves air quality (with respect a number of
conventional and hazardous pollutants) and lowers the cost of gasoline to consumers. See, e.g.,
DRIA at 266.
28 42 U.S.C. § 7545(o)(7)(i).
29 If EPA were to reduce otherwise feasible renewable fuel volumes based on environmental or
economic impacts that are speculative or that fall well below the “severe” threshold, such an
approach would be contrary to law and arbitrary and capricious.
12
below their feasible level only if necessary to avoid severe harms to the considerations Congress
specified.30
D. EPA’s Proposed Approach to the Statutory Factors Is Flawed Because It
Disregards the Proper Framework for Conducting a Reset
EPA’s proposed approach to analyzing and applying the statutory reset factors is flawed,
in that the Agency seems to believe it has discretion to balance the factors as it sees fit, or to
offset important concrete benefits with speculative or minor harms.31 EPA, for example, notes
that the reset “does not establish any numeric criteria” or mathematical formulae that directly
translate into volumes.32 That is true, in part, but is also beside the point. Certain of the reset
factors, like GHG emissions and job creation, are susceptible to quantitative analysis and should
be addressed as such. Other factors may be less quantitative in nature, like energy security, but
are no less capable of being analyzed objectively. And while Congress did not dictate a specific
numeric formula for resetting volumes, it is not true, as EPA asserts, that the statute “does not …
provide guidance on how EPA should weigh the various factors.”33
To the contrary, as discussed above, the text, structure, and purposes of the statute
provide EPA with a simple, coherent framework with which to analyze and apply the factors
together. Again, the ultimate goal of the reset analysis is clear: establish volume requirements
that promote increased use of renewable fuel to the extent feasible without causing important and
severe harm. The factors that relate to these benefits should be analyzed within this framework.
30 While the purpose of these factors in the reset analysis is to prevent severe harm, several of
these factors show beneficial impacts of biofuel use. For example, use of biofuels improves air
quality (with respect a number of conventional and hazardous pollutants) and lowers the cost of
gasoline to consumers. See, e.g., DRIA at 266.
31 NPRM at 72,447 (“[S]ome of the statutory factors assessed for conventional renewable fuel
favor the implied statutory volume (15 billion gallons) or higher volumes, while other factors
favor lower volumes.”); NPRM at 72,443 (stating that neither the statute nor legislative history
“provide[s] guidance on how EPA should weigh the various factors”); RFS Volume Rule
Overview for Interagency Review (Aug 27, 2021) (Docket #EPA-HQ-OAR-2021-0324-0315)
(relegating statutory factors to an appendix and making no reference to the factors when
describing the basis for annual volumes).
32 NPRM at 72,443.
33 Id.
13
II. THE CENTRAL ROLE OF CLIMATE CHANGE TO THE RFS PROGRAM AND TO THE RESET
ANALYSIS
A. Consistent with the Overarching Reset Framework, This Reset Presents EPA
with an Opportunity to Further the Administration’s Climate Goals
Through Full and Effective Implementation of the RFS Program
1. The President has prioritized climate action and set ambitious GHG
reduction goals
The President has repeatedly recognized that climate change poses an “existential
threat.”34 As EPA has explained, “[t]he impacts of climate change are affecting people in every
region of the country, threatening lives and livelihoods and damaging infrastructure, ecosystems,
and social systems in communities across the nation.”35 For that reason, failure to utilize all
available pathways of reducing carbon emissions increases the likelihood of “climate disaster”
with “devastating” impacts, particularly on the most vulnerable communities.36
The Administration has therefore made combatting climate change one of its highest
priorities. Among other commitments, it has adopted the Intergovernmental Panel on Climate
Change’s (IPCC’s) goal of limiting global temperature increases to 1.5 degrees Celsius.37 And
the Administration has adopted highly ambitious national goals to address climate change by
slashing GHG emissions, including a 50-52% reduction in GHG emissions by 2030 and a net34 Remarks by President Biden Before Signing Executive Actions on Tackling Climate Change,
Creating Jobs, and Restoring Scientific Integrity, White House Briefing Room (Jan 27, 2021),
https://www.whitehouse.gov/briefing-room/speeches-remarks/2021/01/27/remarks-by-presidentbiden-before-signing-executive-actions-on-tackling-climate-change-creating-jobs-and-restoringscientific-integrity/ (“I’m signing today an executive order to supercharge our administration
ambitious plan to confront the existential threat of climate change. And it is an existential
threat.”).
35 U.S. Environmental Protection Agency Policy Statement on Climate Change Adaptation (May
26, 2021), in U.S. ENVIRONMENTAL PROTECTION AGENCY CLIMATE ADAPTATION ACTION PLAN
(Oct. 2021), https://www.epa.gov/system/files/documents/2021-09/epa-climate-adaptation-planpdf-version.pdf.
36 U.S. Dep’t of State, The Long-Term Strategy of the United States (Nov. 2021),
https://unfccc.int/sites/default/files/resource/US_accessibleLTS2021.pdf.
37 See generally IPCC, Climate Change 2021: The Physical Science Basis (Aug. 2021); Fact
Sheet: President Biden Sets 2030 Greenhouse Gas Pollution Reduction Target Aimed at
Creating Good-Paying Union Jobs and Securing U.S. Leadership on Clean Energy
Technologies, White House Briefing Room (Apr. 22, 2021) (referring to “the President’s goal of
achieving net-zero greenhouse gas emissions by no later than 2050 and of limiting global
warming to 1.5 degrees Celsius, as the science demands.”),
https://www.whitehouse.gov/briefing-room/statements-releases/2021/04/22/fact-sheet-presidentbiden-sets-2030-greenhouse-gas-pollution-reduction-target-aimed-at-creating-good-payingunion-jobs-and-securing-u-s-leadership-on-clean-energy-technologies/.
14
zero economy by 2050.38 The Administration has adopted a whole-of-government approach to
addressing this problem, with all three of the federal agencies involved in the reset analysis and
in renewable fuels policy generally—EPA, DOE, and USDA—playing prominent roles.39
2. The Administration’s climate goals are not attainable without significant
and immediate GHG reductions in the transportation sector, and these
reductions are not feasible without biofuels
“The transportation sector is the biggest contributor to greenhouse gases in our economy–
which means it can and must be a big part of the climate solution.”40 Indeed, as the White House
recently acknowledged, “we must reduce” greenhouse emissions from the transportation sector
“to ensure we meet President Biden’s goals to create a net-zero economy by 2050.”41
Within the transportation sector, an essential tool for reducing GHGs and combating
climate change is transitioning from petroleum-based fuels to renewable fuels to the greatest
extent feasible and continuing to lower the carbon intensity of the renewable fuels that displace
petroleum.42 Indeed, the RFS program is the only Clean Air Act regulatory program aimed
38 Id.; Exec. Order 14008, Tackling the Climate Crisis at Home and Abroad (Jan. 27, 2021).
39 Fact Sheet: President Biden Takes Executive Actions to Tackle the Climate Crisis at Home
and Abroad, Create Jobs, and Restore Scientific Integrity Across Federal Government, White
House Briefing Room (Jan 27, 2021), https://www.whitehouse.gov/briefing-room/statementsreleases/2021/01/27/fact-sheet-president-biden-takes-executive-actions-to-tackle-the-climatecrisis-at-home-and-abroad-create-jobs-and-restore-scientific-integrity-across-federalgovernment; About the Bioenergy Technologies Office, DOE,
https://www.energy.gov/eere/bioenergy/about-bioenergy-technologies-office (“[B]ioenergy
technologies will help decarbonize the transportation sector, while mitigating greenhouse gas
emissions to combat climate change.”); Bioenergy in a Changing Climate, USDA,
https://www.climatehubs.usda.gov/bioenergy-changing-climate (“Bioenergy can reduce
dependence on fossil fuel, reduce reliance on foreign oil, lower emissions of greenhouse gases
and bring business to rural economies.”).
40 NHTSA Advances Biden-Harris Administration’s Climate & Jobs Goals, Nat’l Highway
Traffic Safety Admin. (Apr. 22, 2021), https://www.nhtsa.gov/press-releases/nhtsa-advancesbiden-harris-administrations-climate-jobs-goals; see also Carbon Pollution from Transportation,
EPA, https://www.epa.gov/transportation-air-pollution-and-climate-change/carbon-pollutiontransportation (“Greenhouse gas (GHG) emissions from transportation account for about 29
percent of total U.S. greenhouse gas emissions, making it the largest contributor of U.S. GHG
emissions.”).
41 Fact Sheet: The Bipartisan Infrastructure Investment and Jobs Act Advances President
Biden’s Climate Agenda, White House Briefing Room (Aug. 5, 2021)
https://www.whitehouse.gov/briefing-room/statements-releases/2021/08/05/fact-sheet-thebipartisan-infrastructure-investment-and-jobs-act-advances-president-bidens-climate-agenda/.
42 See Int’l Energy Agency, Net Zero by 2050: A Roadmap for the Global Energy Sector 78
(2021) (projecting that to reach net zero by 2050, global liquid biofuel consumption rises from
1.6 mboe/d to 6 mboe/d in 2030, before levelling off around 7mboe/d in 2050).
15
explicitly at reducing GHG emissions. While the Administration is pursuing initiatives to
improve the fuel efficiency of gasoline and diesel-powered vehicles and to increase the number
of electric vehicles on the road, those policies can only go so far toward countering the pace of
climate change; they face political, economic, and structural challenges, and may be on a long
and slow path to implementation. The fact is that liquid fuels will be needed to keep the
American transportation system running for the foreseeable future. And, given that fact, the
Administration will not be able to realize its climate goals unless it harnesses the full potential of
biofuels to substitute for the petroleum that will otherwise dominate the liquid fuels market for
years to come.
3. Demand for liquid fuels will persist for the foreseeable future
Even if demand for gasoline and diesel decreases due to increases in fuel efficiency and
electrification, the downward trend is projected to be slow and occur over several decades. The
International Energy Agency’s most recent medium-term projections for gasoline demand in the
United States, for example, show a steep recovery from pandemic levels in 2020,43 followed by
only a mild decrease in gasoline demand over the next four years.44
Longer-term projections similarly show that demand for liquid fuels as a source of energy
will continue to be substantial. Bloomberg’s economic transition scenario, which projects global
gasoline demand out to 2050 while incorporating increases in electric vehicle usage from
technological advances and market trends, predicts that demand for gasoline and diesel will not
43 The U.S. Energy Information Agency’s short-term projections also show U.S. liquid fuel
consumption increasing through 2023. Short-Term Energy Outlook: U.S. Liquid Fuels, U.S.
Energy Info. Admin. (Jan. 11, 2022), https://www.eia.gov/outlooks/steo/report/us_oil.php.
44 Int’l Energy Agency, Oil 2021: Analysis and Forecast to 2026 (Mar. 2021),
https://iea.blob.core.windows.net/assets/1fa45234-bac5-4d89-a532-768960f99d07/Oil_2021-
PDF.pdf (data converted from million barrels per day).
16
significantly decrease until the 2030’s, and will retain a major share of transportation sector
energy consumption through 2050.45
Moreover, the U.S. Energy Information Administration projects liquid fuel consumption
in the U.S. transportation sector to stay nearly constant through 2050, with gradual declines in
motor gasoline offset by gradual increases in renewable jet fuel.46
45 Electric Vehicle Outlook 2021, BloombergNEF (Aug. 2021), https://bnef.turtl.co/story/evo2021/page/1.
46 Annual Energy Outlook 2021, Total Energy Use: Liquid Fuels, U.S. Energy Info. Admin.,
https://www.eia.gov/outlooks/aeo/data/browser/#/?id=1-AEO2021®ion=0-
0&cases=ref2021&start=2019&end=2050&f=A&linechart=~~~~ref2021-d113020a.30-1-
AEO2021&ctype=linechart&sourcekey=0.
17
These projections demonstrate that even as the light duty vehicles market transitions to
electrification, liquid fuels will remain a substantial component of transportation sector energy
usage for a long time. The concentration of renewable fuels as a portion of the liquid fuels
supply must steadily increase if climate goals are to be realized, especially if overall demand for
liquid fuels declines over the long term.
4. EPA and the Administration have repeatedly recognized the necessity of
biofuels to decarbonize the transportation fuel supply in the United States
As EPA has repeatedly stated, “[t]he fundamental objective of the RFS provisions under
the Clean Air Act is clear: To increase the use of renewable fuels in the U.S. transportation
system every year through at least 2022 in order to reduce greenhouse gases (GHGs) and
increase energy security.”47 Moreover, EPA recognizes that “[r]enewable fuels represent an
opportunity for the U.S. to move away from fossil fuels towards a set of lower GHG
transportation fuels, and a chance for a still-developing low GHG technology sector to grow,”
and that “[t]hese lower GHG renewable fuels include corn starch ethanol.”48
In the Administration’s “Long-Term Strategy of the United States, Pathways to Net-Zero
Greenhouse Gas Emissions by 2050,” released in advance of the Glasgow Climate Conference,
47 Renewable Fuel Standard Program: Standards for 2014, 2015, and 2016 and Biomass-Based
Diesel Volume for 2017, 80 Fed. Reg. 77,420 (Dec. 14, 2015) (to be codified at 40 C.F.R. pt.
80).
48 NPRM at 77,421; see also ACE, 864 F.3d at 696 (“By mandating the replacement—at least to
a certain degree—of fossil fuel with renewable fuel, Congress intended the [RFS] Program to
move the United States toward greater energy independence and to reduce greenhouse gas
emissions.”); Am. Petroleum Inst. v. E.P.A., 706 F.3d 474, 476 (D.C. Cir. 2013) (“In
establishing the RFS program, Congress made commercial production of cellulosic biofuel, an
advanced biofuel derived from sources of lignocellulose such as switchgrass and agricultural
wastes, central to the program’s objective of reducing greenhouse gas emissions.”).
18
for example, alternative fuels including biofuels play a necessary role in the transition away from
fossil fuels in the transportation sector. Recognizing that “[o]ver time, electricity, carbon
beneficial biofuels, and hydrogen will become increasingly clean,” the Long-Term Strategy
emphasizes that use of these energy sources must be accelerated in order to decrease carbon
emissions in the face of a projected increase to transportation sector demand.49 Indeed, the
Long-Term Strategy projects that net-zero in the transportation sector will be achieved by
accelerating the use of both alternative fuels and electricity to power a transition away from
fossil fuel consumption:
Across multiple net-zero scenarios, increased use of alternative fuel, including
biofuel, and electricity will be needed to offset decreases in fossil fuel usage.50
To accomplish these goals the President has, among other things, called upon the
agricultural sector to maximize its carbon reduction potential, noting that “America’s farmers,
ranchers, and forest landowners have an important role to play in combating the climate crisis
and reducing greenhouse gas emissions, by sequestering carbon in soils, grasses, trees, and other
vegetation and sourcing sustainable bioproducts and fuels.”51
49 U.S. Dep’t of State, The Long-Term Strategy of the United States 42 (Nov. 2021),
https://unfccc.int/sites/default/files/resource/US_accessibleLTS2021.pdf.
50 Id. Figure 8.
51 Exec. Order 14008, Tackling the Climate Crisis at Home and Abroad (Jan. 27, 2021)
(emphasis added).
19
5. EPA has undervalued the GHG reduction potential of biofuels
As discussed in greater detail below, EPA’s failure to update its methodology for
assessing the lifecycle GHG emission benefits of ethanol causes EPA to substantially undervalue
those GHG benefits, contrary to the best science available.52 This undervaluation can lead the
Agency to falter in its administration of the RFS program, implementing the program in a way
that shortchanges (or eliminates) its market-forcing potential, and undermining the program’s
ability to incentivize increasingly higher renewable fuel volumes as compared to what would
have occurred without the program.
In the reset and in its implementation of the RFS program generally, EPA should seek to
advance the RFS program’s unique climate objectives and better align the RFS with the
Administration’s climate policies. Among other things, as part of the Agency’s backwardlooking review of the implementation of the RFS program, EPA should fully and accurately
account for the role of ethanol in reducing GHG emissions on a lifecycle basis. While shortfalls
in cellulosic biofuels have resulted in an inability to meet the ambitious statutory goal of 36
billion gallons by 2022, U.S. ethanol production has nonetheless quadrupled between 2005 and
2019, and ethanol has substantially exceeded the threshold expectations for GHG reductions.53
As a result, studies have shown that roughly 544 million metric tons of CO2e emissions have
been avoided54—the equivalent of taking 8.5 million passenger vehicles off the road each year.55
EPA’s proposal and DRIA give short shrift to the GHG accomplishments of biofuels under the
RFS, noting generally that renewable fuels “have the potential to reduce GHGs and influence
climate change if their use displaces petroleum derived fuels,” but failing to acknowledge studies
showing how they have proven their value as a tool in the fight against climate change.56
B. Updating the GHG LCA for Ethanol Should Be a Top Priority for EPA
As EPA and the courts have long recognized, where EPA is engaged in rulemaking to
protect the environment and public health, EPA must rest its decisions on the best available
52 See infra Part II.B.
53 U.S. Bioenergy Statistics, USDA (Jan. 21, 2022), https://www.ers.usda.gov/data-products/u-sbioenergy-statistics.
54 Usung Lee et. al Retrospective Analysis of the U.S. Corn EthanolIndustry for 2005–2019:
Implications for Greenhouse Gas Emission Reductions (May 4, 2021),
https://doi.org/10.1002/bbb.2225.
55 Greenhouse Gas Equivalencies Calculator, EPA (Mar. 2021),
https://www.epa.gov/energy/greenhouse-gas-equivalencies-calculator.
56 DRIA at 63.
20
science.57 For that reason, Growth Energy applauds EPA’s announcement to hold a workshop
later this year “to incorporate the best available science into an update of our lifecycle analysis
(LCA) of biofuels.”58 Nevertheless, in light of this fundamental principle, EPA’s failure over the
last decade to incorporate best available science in its LCA methodology for corn ethanol–and its
failure to do so in the current proposal–is perplexing and if continued in the final rule would be
arbitrary and capricious. Instead, EPA disregards evidence in the record and continues to
employ an admittedly outdated and inaccurate LCA for ethanol. While EPA attempts to justify
this failure on “uncertainty” regarding ethanol’s LCA, EPA neglects to use available scientific
methods commonly used in other contexts to address any uncertainty here.
To accurately quantify the GHG impacts of EPA’s reset rulemaking–both in terms of the
GHG increase resulting from EPA’s proposed reduction of the 2020 volumes and resetting 2021
volumes to reflect actual production, as well as the benefits to be realized by setting ambitious
but feasible 2022 volumes–EPA must update its corn ethanol LCA.59 EPA in 2010 projected that
lifecycle GHG emissions from corn ethanol would be 21% less than the representative 2005
petroleum baseline, and still relies on this value in the current rulemaking.60 In contrast, the best
available and most recent science–including studies published by the DOE’s Argonne National
Lab and USDA, the two agencies that EPA is required by law to consult with in conducting the
reset analysis–place the lifecycle GHG reductions from corn ethanol in the range of 39-46%
below the petroleum baseline. These results are bolstered by other studies, including the expert
analyses of Environmental Health & Engineering, Inc. (EH&E) and Life Cycle Associates, LLC
attached to this comment letter and described briefly below.61 EPA cannot disregard this
57 Physicians for Social Resp’y v. Wheeler, 956 F.3d 634 (D.C. Cir. 2020); Our Mission and
What We Do, https://www.epa.gov/aboutepa/our-mission-and-what-we-do (EPA must ensure
that national “efforts to reduce environmental risks are based on the best available scientific
information.”); Exec. Order No. 13990 (Jan. 20, 2021) (“the Federal Government must be guided
by the best science”).
58 Notice of Workshop on Biofuel Greenhouse Gas Modelling, 86 Fed. Reg. 73757 (“Through
this workshop, we will initiate a public process for getting input on (i) how to incorporate the
best available science into an update of our lifecycle analysis (LCA) of biofuels.”) (emphasis
added).
59 Growth Energy is encouraged by EPA’s announcement of a February Workshop on Biofuel
Greenhouse Gas Modelling, but it is crucial that EPA follow through with the efforts necessary
to promulgate an accurate and updated lifecycle analysis based on the best available science as
soon as possible.
60 2010 Renewable Fuel Standard Program (RFS2) Regulatory Impact Analysis, US Envt’l Prot.
Agency, Report No.: EPA-420-R–10–006 (“2010 RIA”).
61 See Scully, et. al., Carbon intensity of corn ethanol in the United States: state of the science
(2021) (showing reduction of 46%); Lee, et. al., Retrospective analysis of the U.S. corn ethanol
industry for 2005–2019: implications for greenhouse gas emission reductions (2021) (showing
reduction of 44%%); Rosenfeld, et. al. A Life-Cycle Analysis of the Greenhouse Gas Emissions
from Corn-Based Ethanol (Sept. 5, 2018) (showing reduction of 39%).
21
information in its current rulemaking, regardless of its laudable commitment to initiate a process
outside of this rulemaking to consider this information for other, future purposes.
To be sure, the importance of updating EPA’s LCA for corn ethanol extends beyond the
current rulemaking; it will also be relevant to a range of agency decisions, including, for
example, the forthcoming set rulemaking and any future rulemakings designed to facilitate the
use of E15, flex fuel vehicles, and use of higher-level ethanol blends like E85, and will be critical
to the development of sound policy related to sustainable aviation fuel (SAF). With respect to
SAF, a memorandum of understanding between DOE, USDA and DOT adopted ambitious
targets, calling for production of 3 billion gallons by 2030 and 35 billion gallons by 2050.62
Accomplishing these volume goals will not be feasible without harnessing the potential of the
U.S. ethanol industry, which produces 86% of the nation’s biofuels.63 However, EPA’s current
LCA for ethanol poses a regulatory barrier to SAF produced from ethanol because it would be
unlikely to generate RINs under the RFS program64 and would be unlikely to qualify for an SAF
tax credit as currently proposed in Congress.65
Consistent with Administration policy, the current proposed RFS rule, as well as other
new policies EPA may pursue to achieve the Administration’s climate goals, will rely on the
social cost of carbon (SCC) in combination with ethanol’s lifecycle analysis to monetize the
benefits of anticipated GHG emission reductions in the Agency’s regulatory cost-benefit
analyses.66 (And we note that EPA has emphasized the efforts taken to ensure that the SCC is
developed in accordance with “best available science.”67) But use of a flawed ethanol LCA as an
62 Sustainable Aviation Fuel Grand Challenge, DOE,
https://www.energy.gov/eere/bioenergy/sustainable-aviation-fuel-grand-challenge
63 EIA Expands Data Coverage of Biofuels in our Monthly Energy Review, U.S. Energy Info.
Admin. (Nov. 19, 2021) (“Fuel ethanol accounted for 86% of total U.S. biofuels production in
July 2021, biodiesel for 9%, renewable diesel fuel for 5%, and other biofuels for less than 1%.”).
64 42 U.S.C. § 7545 (o)(2)(A)(i) (requiring renewable fuel to achieve a 20% reduction in
lifecycle GHG emissions)
65 Build Back Better Bill Act, H.R. 5376, 117th Cong. § 136203 (2021) (requiring SAF to
achieve a 50% reduction in lifecycle GHG emissions).
66 Exec. Order No.13990, Protecting Public Health and the Environment and Restoring Science
to Tackle the Climate Crisis (Jan. 20, 2021).
67 DRIA at 77 (“Specifically, in 2009, an interagency working group (IWG) that included the
EPA and other executive branch agencies and offices was established to ensure that agencies
were using the best available science and to promote consistency in the SC-CO2 values used
across agencies.”); id. at 78 (“IWG announced a National Academies of Sciences, Engineering,
and Medicine review of the SC-CO2 estimates to offer advice on how to approach future updates
to ensure that the estimates continue to reflect the best available science and methodologies.”)
(emphasis added); id. (“On January 20, 2021, President Biden issued Executive Order 13990,
which reestablished the IWG and directed it to ensure that the U.S. Government’s estimates of
the social cost of carbon and other greenhouse gases reflect the best available science and the
recommendations of the National Academies.”) (emphasis added).
22
input into SCC cost-benefit calculations skews the resulting monetized benefits, and undermines
EPA’s efforts to ensure scientific integrity in its analyses of the costs and benefits of policies to
mitigate GHGs and address climate change. EPA cannot justify such self-contradiction in this
rulemaking on the grounds that it has not gotten around to updating its LCA yet; the information
has been available, is presented in these comments and part of the record, and ignoring it for
another day in another context would be arbitrary.
Moreover, utilizing the best available science on corn ethanol’s carbon intensity has a
substantial impact on cost-benefit analyses. For example, using the interim SCC, a calculation of
the monetized societal benefits of GHG reductions from actual and projected corn starch ethanol
consumption in 2020-2022, relying on EPA’s 2010 carbon intensity (“CI”) score of 73.2
gCO2e/MJ, would show benefits of about $3.6 billion. However, if using EH&E’s central
estimate of 51.4 gCO2e/MJ68 and keeping all other factors equal, the calculation would show
more than twice the benefits, i.e., about $7.6 billion. This $4 billion gap should then have a
substantial effect on regulatory decision-making.69 EPA must adopt the best available science in
its LCA to ensure that the most currently well-founded GHG benefits of biofuels are disclosed to
the public and relied on by policymakers.
1. The best available science indicates corn ethanol has more than double the
GHG emissions reductions of EPA’s outdated estimate, putting it roughly
on par with advanced biofuels
It has been over a decade since EPA modeled the lifecycle GHG impacts of corn ethanol
in the original 2010 RFS rule. In the intervening years, the models used in EPA’s original
analysis have been substantially updated with respect to the models’ methodologies, designs,
data, and parameters. Most significantly, modeled emissions associated with indirect land use
change (“iLUC”) are dramatically lower than previously understood. Because iLUC is a
substantial component of EPA’s estimated GHG emissions for corn starch ethanol, it is critical
that EPA address the latest science that coalesces around a substantially-reduced iLUC
estimate.70 Indeed, in 2010, EPA acknowledged that land use change modeling, which was
relatively new science at the time, would need to be reassessed as modeling techniques
68 Environmental Health & Engineering, Inc., Response to 2020, 2021, and 2022 Renewable Fuel
Standard (RFS) Proposed Volume Standards 5 (Figure 1) (Feb. 3, 2022) (“EH&E Report”)
(attached as Ex. 1).
69 If the final social cost of carbon increases from than the interim value, the gap between results
using EPA’s 2010 CI score and EH&E’s central best estimate would widen even further. See
DRIA at 83 (noting that “the SC-GHG estimates used in this proposed rule likely underestimate
the damages from GHG emissions”).
70 DRIA at 66. EPA might also consider that reputable models used in other jurisdictions, such as
the GHGenius model used in Canada, that do not include indirect LUC given the speculative
nature of iLUC emissions.
23
improved.71 The early stage of this modeling is demonstrated by EPA’s own reduction in its
LUC estimate by more than 50% from the proposed rule in 2009 to the final in 2010.72 LUC
models did not suddenly stop improving in 2010, and the Agency’s 2010 estimate is now an
outlier from the most recent and best available science:
71 Regulation of Fuels and Fuel Additives: Changes to Renewable Fuel Standard Program, 75
Fed. Reg. 14,670, 14,678 (Mar. 26, 2010) (to be codified at 40 C.F.R. pt. 80) (“the Agency is
also committing to further reassess these determinations and lifecycle estimates “); Cong. Rsch.
Serv., Calculation of Lifecycle Greenhouse Gas Emissions for the Renewable Fuel Standard
(RFS) 13 (“EPA acknowledged that a transparent and scientific analysis of the GHG emission
impact of renewable fuels going forward will be further refined as additional data sources and
models become available.”); EPA Off. of Inspector Gen., EPA Has Not Met Certain Statutory
Requirements to Identify Environmental Impacts of Renewable Fuel Standard pdf p. 3 (Aug. 18,
2016) (“[T]he EPA committed to update this analysis as lifecycle science evolves, but does not
have a process for initiating an update.”).
72 EPA, Lifecycle Greenhouse Gas (GHG) Emissions Results Spreadsheets (Oct. 30 2008),
Docket ID EPA-HQ-OAR-2005-0161, https://www.regulations.gov/document/EPA-HQ-OAR2005-0161-0938; EPA, Renewable Fuel Standard Program (RFS2) Regulatory Impact Analysis
(Feb. 2010), https://19january2017snapshot.epa.gov/sites/production/files/2015-
08/documents/420r10006.pdf
24
Comparison of USEPA’s iLUC estimates with relevant most recent studies from the U.S. and Europe73
iLUC estimates have decreased over time because both modeling techniques and data
inputs have improved. Specifically, over the past 12 years, iLUC models have improved by (1)
addition of new modules that allow for more accurate simulation of real-world agricultural
practices; (2) addition of more spatially resolved information on land cover; and (3) tuning of
parameters that describe rates of land conversion and land.74
The attached report by EH&E, “Response to 2020, 2021, and 2022 Renewable Fuel
Standard (RFS) Proposed Volume Standards,” (Exhibit 1) addresses these improvements in
greater detail. EH&E is a multi-disciplinary team of environmental health scientists and
engineers with expertise in measurements, models, data science, LCA, and public health, who
recently published a study titled “Carbon intensity of corn ethanol in the United States: state of
the science.”75 This study concluded that the carbon intensity corn starch ethanol has decreased
73 EH&E Report at 5 (Figure 1).
74 EH&E Report at 5-8.
75 Scully, et. al.
25
by approximately 50% over the past 30 years to a current central estimate of 51gCO2e/MJ, or
46% below the 2005 petroleum baseline.76
As explained in more detail in EH&E’s expert report (attached), updates to two models in
particular, GTAP-BIO and FAPRI, warrant EPA’s close attention as EPA previously relied on
these models for its 2010 analysis and enhanced data inputs yield substantially different results a
decade on.77 Taking these updates into account, most credible studies using these models reflect
an iLUC estimate for corn ethanol of 1.3 to 11 gCO2e/MJ.78 Analyses by European investigators
of iLUC using different models similarly arrive at a substantially-reduced iLUC estimate of 8 to
9 gCO2e MJ−1
.
79 EPA cannot disregard this information in the current rulemaking for which it is
highly relevant.
In addition, the attached expert report by Life Cycle Associates, LLC, “Review of GHG
Emissions of Corn Ethanol under the EPA RFS2” (Exhibit 2), similarly concludes that, using a
methodology similar to EPA’s in the 2010 RIA with updates for best available data and
modeling choice, parameters, and inputs, an appropriate corn ethanol CI is in the range of 40 to
55 g CO2e/MJ, or approximately 45 to 55% lower than a representative 2005 gasoline baseline.80
As addressed extensively in the report, the key factors that result in a substantially lower CI than
EPA projected in 2010 include:
Adjustments for iLUC, taking into account modeling that better reflects realworld observations regarding reduced rates of deforestation;
Ethanol plants’ reduced energy consumption and the lower GHG intensity of the
electric grid;
Increased corn oil extraction and substantial use of corn oil in BBD production;
Displacement of iLUC and nitrogen emissions associated with soybeans;
A higher baseline CI for 2005 petroleum fuels.
EPA must take this analysis, along with the substantial scientific literature on ethanol’s GHG
LCA, into account in this rulemaking.
2. Uncertainty in LCA modeling can be managed by comparing existing
credible studies
EPA’s DRIA repeatedly references “considerable” uncertainty in LCA modeling.81 It was
true in 2009 and 2010 that there was uncertainty with respect to GHG lifecycle assessments of
76 Id.
77 EH&E Report at 5-8.
78 Scully, et. al. at 7.
79 EH&E Report at 5.
80 Life Cycle Associates, LLC, Review of GHG Emissions of Corn Ethanol under the EPA RFS2
48 (Feb. 4, 2022) (attached as Ex. 2).
81 DRIA at 65.
26
fuels, especially as the models were less mature, but that was no impediment to EPA then. Nor
should it be now. Reliance on uncertainty is not a substitute for grappling with the best currently
available information, especially where EPA recognizes that such information is available but
would defer engaging with it. And even then, after undertaking thorough consideration of the
best available information, remaining uncertainty is a justifiable rationale for an agency to
withhold action only if the uncertainty is so “unusually profound” at that juncture that the agency
“could not form” a reasoned judgment about the evidence.82 Here, the uncertainty is neither
unusually profound nor prevents EPA from forming reasoned judgment on the LCA of ethanol.
More specifically, analyses of iLUC confront four main categories of uncertainty: (1)
methodology, (2) model design, (3) data, and (4) parameters.83 EPA can manage these
uncertainties by relying upon the existing literature to derive an updated central estimate of iLUC
emissions from available estimates in the numerous credible studies using generally accepted or
commonly used models.84 This approach addresses and narrows each category of uncertainty by
comparing estimates that are the product of models with different methods, designs, data, and
parameter values. Examples of this approach exist, for example, in Scully et al.
85 and
Lewandrowski et al.,86 and EPA should build upon these methodologies and the existing
literature to incorporate the best available science in an updated GHG LCA for ethanol.
An analogous example of this approach exists in the social cost of carbon (SCC) metric.
In developing the SCC, the interagency working group had to address an issue with significant
scientific uncertainty by utilizing central estimates, embracing the principle of best available
science, and adopting an interim approach to provide the best available values without causing
undue delay.87 EPA is legally obligated to adopt similar techniques in the context of updating the
LCA for corn ethanol for purposes of this rulemaking, taking into account the best current
information, and only then addressing remaining uncertainties using established methods such as
central estimates.
82 Murray Energy Corp. v. EPA, 936 F.3d 597, 620 (D.C. Cir. 2019); see also Motor Vehicle
Mfrs. Ass’n of U.S., Inc. v. State Farm Mut. Auto. Ins. Co., 463 U.S. 29, 52 (1983).
83 ICAO. CORSIA Supporting Document: CORSIA Eligible Fuels – Life Cycle Assessment
Methodology, (June 2019).
84 EH&E Report at 8-11.
85 Scully et al 2021.
86 Jan Lewandrowski, et. al. The Greenhouse Gas Benefits of Corn Ethanol –Assessing Recent
Evidence, 11 Biofuels 361, (2019).
87 DRIA at 77 (“As a member of the IWG involved in the development of the February 2021 SCGHG TSD, the EPA agrees that the interim SC-GHG estimates represent the most appropriate
estimate of the SC-GHG until revised estimates have been developed reflecting the latest, peerreviewed science.”).
27
3. EPA’s “illustrative” scenario of GHG impacts of the proposed rule is
flawed and misleading
In its proposal, EPA both relies upon an outdated approach to lifecycle assessment, and
applies that outdated LCA in a faulty manner. Specifically, in the “illustrative GHG scenario”
EPA uses the volume of renewable fuel consumed in 2020 (which was depressed due to the
pandemic and associated economic downturn) as the baseline against which to measure the
proposed 2021 and 2022 volumes.88 Using this lens, increases in renewable fuels volumes in
2021 and 2022 are portrayed as a demand shock that induces land use change in the
“illustrative” scenario. However, use of the most currently accurate baseline (e.g., 2019 actual
or proposed volumes or a no RFS scenario) would show that increases in conventional volumes
in 2021 and 2022 are merely returning to 2019 levels after an anomalous gasoline demand
collapse in 2020 caused by the COVID-19 pandemic.89 Thus, the purported GHG emissions
increases in 2021 and 2022 associated with the land use change are illusory, and a product of the
depressed baseline volumes in 2020, rather than reflective of real-world conditions. EPA itself
notes that “once the cost of clearing and converting land is incurred, it seems likely that land
will continue to be used for agricultural purposes in the future.”90 Yet EPA contradicts this
assertion by concluding that returning to 2019 renewable fuel volumes will result in the
conversion of new land. In reality, as addressed below, even when EPA has increased
renewable fuel volumes in the past, there have not been concomitant conversion of land for
agricultural uses. EPA must correct the illustrative scenario to remove GHG emissions
increases associated with the misleadingly assumed “pulse “of land use change in 2021 and
2022.
III. EPA OVERESTIMATES OTHER ENVIRONMENTAL IMPACTS
A. There Is No Credible Evidence That the Proposed Rule Will Cause Adverse
Impacts to Wetlands, Ecosystems, Species, Habitat, Water
Quality/Availability, or Soils
Under the proper framework for evaluating the statutory factors in Section
211(o)(2)(B)(ii), EPA should “reset” volumes so as to maximize achievement of the core
congressional objectives behind the RFS (including reducing GHG emissions), tempered only by
capacity constraints and the need to avoid unacceptable negative effects, i.e., those that would
produce severe environmental or economic consequences. One of the main areas in which EPA
must consider the possibility of indirect costs (and/or benefits) from the RFS program is the
88 DRIA at 68-69.
89 Robert Rapier, Gasoline Demand Collapses to a 50-Year Low, Forbes (Apr. 9, 2020) (noting
that the COVID-19 pandemic resulted in the lowest gasoline demand levels in over half a
century); EH&E Report at 11-14.
90 DRIA at 67.
28
potential for indirect impacts to land and water, including to wetlands, ecosystems, species,
habitat, water quality/availability, and soils.91
On this score, the Proposed Rule and DRIA contain a number of vague, conclusory, and
unsupported statements suggesting that the proposed 2022 renewable fuel volumes (and the
implied conventional biofuels volumes in particular) could result in adverse indirect
environmental impacts. For example, EPA suggests that the Proposed Rule would result in
“increased corn production in the U.S.,” which in turn “could result in greater conversion of
wetlands, adverse impacts on ecosystems and wildlife habitat, adverse impacts negative impacts
on water quality and supply, and increased prices for agricultural commodities and food
prices.”92 However, as explained below, a wide body of scientific literature demonstrates that
there is no established causal connection between the RFS and adverse impacts to the
environment, a fact which EPA has acknowledged in other contexts but overlooks here. No
studies discussed in the DRIA, or any other studies to date, have definitively established a causal
link between the RFS and impacts to land or water. In fact, the existing data indicate that the
RFS does not result in land use change or adverse water availability/quality impacts. This is not
news to EPA as it has reached the same conclusions in other contexts. For the reasons described
in greater detail below, EPA should find that the record evidence fails to establish indirect
environmental impacts to land or water as a reason to depart downward from the renewable fuel
volumes proposed for 2020, 2021, or 2022, including the implied conventional volume of 15
billion gallons proposed for 2022. Indeed, EPA should acknowledge that there is room for
volumes to be set at even higher levels without triggering any unacceptable adverse
environmental effects.
1. Claims that the RFS causes land use change are unsubstantiated, as EPA
has previously acknowledged and must incorporate into its analysis here
EPA’s analysis of the potential for indirect impacts to wetlands, ecosystems, habitat,
water quality and supply, and soils—indirect impacts that would result from additional lands
being converted to agricultural purposes—fails to adequately address the critical threshold
question of whether the RFS program causes land use conversion in the first place. EPA relies
on its 2018 Second Triennial Report in asserting that “[e]vidence from observations of land use
change suggests that some of [the] increase in acreage and crop use is a consequence of
increased biofuel production.”93 But EPA also asserts that these effects cannot reasonably be
estimated or quantified.94 Further, it is unclear how these particular “observations” of land use
change shed light on whether there has been an increase in total acreage and crop use, or whether
biofuel production is the cause of any such increase. Unable to take the analysis beyond
speculation that “some” increase in acreage and crop use is attributable to biofuel production,
91 See 42 U.S.C. § 7545(o)(2)(B)(ii)(I). Impacts on climate change and air quality, which are
also listed in section 211(o)(2)(B)(ii)(I), are addressed separately in Parts II and III of these
comments.
92 NPRM at 72,447.
93 DRIA at 88.
94 Id. at 88, 127‒28.
29
EPA does not identify potential impacts to land or water as a reason to adjust downward the
volumes it proposes for 2020, 2021, or 2022. Indeed, at the end of the day, EPA insists that in
devising its proposed total renewable fuel volumes, it has “take[n] into consideration the
potential negative impacts of renewable fuels produced from crops such as corn or soybeans on
environmental factors such as the conversion of wetlands, ecosystems, and wildlife habitat.”95 A
closer examination of the record evidence, however, would allow EPA to bolster the support for
its proposal and to be less equivocal in analyzing the land and water-related factors in Section
211(o)(2)(B)(ii)(I).
Rather than creating the impression that some unquantifiable amount of negative
environmental impacts exists, EPA should acknowledge that there is simply no credible evidence
for the proposition that the RFS program has caused, or that the Proposed Rule will cause, land
conversion notwithstanding EPA’s vague hypothesizing to the contrary. Because EPA fails to
address the complex chain and multiple variables relevant to addressing the question of causality,
much of its analysis of environmental impacts are rooted in suppositions and inapposite. Further,
the DRIA’s reliance on studies cited in the Second Triennial Report for evidence of causation is
perplexing as EPA has acknowledged in other contexts, including certain sections of the DRIA,
that “a causal connection [between the RFS and land use change] is difficult to make with
confidence.”96 Below we first explain the significance of understanding causality with respect to
the RFS and land use change before turning to EPA’s assessment of impacts to wetlands,
wildlife, habitat, soil, and water.
a. Understanding the causal chain
The relationship between the RFS and impacts to land and water involves a complex
causal chain comprised of the following significant links, each of which have multiple
interrelated variables:97
95 NPRM at 72,451.
96 DRIA at 88.
97 Ramboll, Supplemental Analysis Regarding Allegations of Potential Impacts of the RFS on
Species Listed Under the Endangered Species Act at 2 (Nov. 29, 2019) (“Nov. 2019 Ramboll
Supplemental Report”).
30
Take, for example, the third and fourth links in the chain. Farmers’ decisions about
whether to plant more corn, and whether to convert more land for that purpose, are driven by
many factors having nothing to do with the RFS Program. These include including fluctuating
food prices, other government policies, existing equipment and infrastructure, and improvements
in crop yields, pests and disease, and the costs of energy and fuel.98 Future prices of different
crops relative to each other also help farmers determine the crop planting mix; however, there are
limitations on farmers’ ability to switch between crops (e.g., crop rotation schedules, limitations
on machinery needed for particular crops).99 Thus, the RFS is only one of many variables that
drive planting decisions.
98 Ramboll, The RFS and Ethanol Production: Lack of Proven Impacts to Land and Water at 5
(Aug. 18, 2019) (“Aug. 2019 Ramboll Report”).
99 Nov. 2019 Ramboll Supplemental Report at 4.
31
Further, even to the extent farmers’ decisions are driven in part by corn prices, there is no
direct causal connection between the RFS program and corn prices. The complex economic and
policy factors that determine corn prices include oil prices, currency exchange rates, economic
growth (and demand for food) in developing countries, market speculation, U.S. agricultural
policies, trade restrictions, and macroeconomic shocks.100 In addition, most of the increase in the
price of corn (as well as other crops like soy and wheat) since 2005 has been attributed to higher
oil prices, rather than any other factor.101 A recent study that analyzed food price inflation and
land use classification concluded that food inflation since 1973 was actually the lowest during
biofuel increases in the years 1991–2016, and was most highly correlated with the price of oil.102
100 Id. at 6.
101 Id. In recent years the strongest correlation to corn prices has been oil prices because costs of
producing corn include costs for fuel, lubricants, electricity, and fertilizer. See D.S. Shrestha et
al., Biofuel Impact on Food Prices Index and Land Use Change, 124 Biomass & Bioenergy 43
(2019), https://ethanolrfa.org/file/1829/Shreshtha-et-al-Biofuel-impact-on-food-prices-indexand-land-use-change-03-2019.pdf; Energy for Growing and Harvesting Crops Is a Large
Component of Farm Operating Costs, U.S. Energy Info. Admin. (Oct. 17, 2014),
https://www.eia.gov/todayinenergy/detail.php?id=18431.
102 Net Gain, Analysis of EPA’s Proposed Rulemaking for 2020, 2021, and 2022 RVOs,
Regarding Land Use Change, Wetlands, Ecosystems, Wildlife Habitat, Water Resource
Availability, and Water Quality at 7 (Feb. 3, 2022) (“Net Gain Report”) (attached as Ex. 3).
32
Indeed, the most recent data available demonstrate that there is no predictable effect between
recent RFS obligations and corn prices.103
Moreover, even if a farmer does decide to plant corn, there is no reason to assume that he
or she would alter land from non-agricultural use to agricultural use to do it. If a farmer decides
to increase production of a certain crop, this can be accomplished by either producing more of
the crop on existing land (intensification) or putting new land into production (extensification,
which could result in land use change).104 All else being equal, extensification is the least
preferred option as it is the option most likely to involve additional expenditures such as land
clearing and other preparation.105 This decision is also driven by a variety of factors unrelated to
ethanol demand which include weather conditions, soil quality, crop output, and input prices,
innovations in equipment, crop insurance, disaster insurance, and marketing loans.106
Further, as a factual matter, nearly a century of USDA data illustrates that corn yields per
acre have steadily increased while total corn acreage has plateaued.107
103 See Prices Received by Month, USDA (2019)
https://www.nass.usda.gov/Charts_and_Maps/graphics/data/pricecn.txt.
104 Aug. 2019 Ramboll Report at 4.
105 Id.
106 Nov. 2019 Ramboll Supplemental Report at 4, 23.
107 Net Gain Report at 9; Stillwater Associates LLC, The RFS Reset: A Look at Corn Land Use
and Conventional Ethanol Production at 4, 7–10 (Aug. 30, 2019) (“2019 Stillwater Report”).
33
Relative Change in Acres of Corn Planted and Yield (1926-2021)108
The increase in demand has largely been met by an approximately 7-fold increase in yield
in bushels per acre.109 In other words, increased production of corn is driven by intensification—
the production of more corn on the same acreage of land—which does not result in land
conversion.110 The USDA further anticipates changes in corn production to result in an increase
of approximately 16.1 more bushels per acre by 2028 without a substantial increase in farmed
acres (and with a corresponding reduction in the use of water resources and fertilizer).111
Finally, critical to understanding the issue of whether the RFS causes land use change is
the basic question of whether the RFS drives demand for ethanol (the first link in the chain
above). As EPA itself has noted, ethanol demand in the United States has been driven by factors
wholly unrelated to the RFS, such as the ubiquity of E10 in the domestic gasoline market and the
demand for ethanol for export markets.112 Because ethanol is less expensive than gasoline to
produce and enhances the octane of fuel at lower cost to the consumer, “[w]ith or without the
108 Net Gain Report at 9.
109 Id.
110 Id.
111 Aug. 2019 Ramboll Report at 29.
112 Id. at 15; 2019 Stillwater Report at 3-4, 9; Initial Br. for Respondent U.S. EPA at 95–96,
Growth Energy v. U.S. EPA (No. 19-1023) (Jan. 9, 2020) (“EPA Br.”); EPA, Endangered
Species Act No Effect Finding for the 2020 Final Rule at 7 (Dec. 2019) (“2020 No Effect
Memo”).
34
RFS, the industry will continue to produce E10 in substantial quantities.”113 EPA has thus
previously reasonably concluded in the context of the 2019 Renewable Volume Obligation
(RVO) that the RFS program “is not driving corn ethanol production or corn cultivation (much
less farmers’ decisions on when, where, and how to plant crops in 2019),” and therefore could
not be driving LUC or impacts to habitat, wildlife, and ecosystems, as discussed below.114
In sum, careful analysis and consideration of the causal chain makes clear that the RFS
program is not driving conversion of land for production of corn crop for ethanol for the RFS
program. Nor will the 2022 proposed volumes have this effect as the implied conventional
renewable fuel volumes are the same as three years ago in the 2019 RVO, e.g., 15 billion gallons.
There is no evidence that new, previously uncultivated land would be put into corn production as
a result of EPA reverting back to the 2019 implied conventional volumes in the 2022 RVO.
Indeed, the analysis above suggests that this is a wholly unrealistic outcome because (1) even in
2020, when gasoline demand dropped precipitously due to the pandemic-related economic
downturn and ethanol production was idled, there is no evidence that this shock caused farmers
to change planting behaviors for 2021 to allow corn fields in rotation to fallow;115 (2) even if
farmers had reduced corn plantings in 2021 as a result of the economic downturn, increased corn
demand has been met for decades through intensification, not extensification, meaning it is
highly unlikely there would be LUC as a result of 2022 volumes returning to 2019 levels; (3) the
RFS is only one of myriad factors impacting farmers’ planting decisions; and (4) in any event,
the Proposed Rule likely will not be finalized in time to impact crop planting decisions where
planning decisions are made well in advance of the spring planting season.116 Further, EPA
should apply these real-world understandings to its “illustrative” GHG scenario that alleges there
will be an “initial pulse of land use change emissions” in 2021 and 2022 due to use of a
misleading 2020 baseline, as addressed in Part II above.117
b. Studies that have asserted that the RFS causes land use change are
flawed
Recent literature has uncovered serious flaws in studies cited by the DRIA that attempt to
implicate the RFS in land use change, ecosystem impacts, grassland losses, or adverse impacts to
water quality tied to land use change.118 This is mainly because previous analyses (such as Lark
et al, 2015, and Wright et al., 2017) were based on the Crop Data Layer (CDL) data set, which
has been found to be inaccurate. Recent studies note that reliance on CDL data results in
113 EPA Br. at 96.
114 Id.
115 Stillwater Associates LLC, Comments to EPA on 2020-2022 RFS Rule at 3 (Feb. 4, 2022)
(“2022 Stillwater Report”).
116 See Nov. 2019 Ramboll Supplemental Report at 5–6.
117 DRIA at 73.
118 Net Gain Report at 4–6.
35
misleading levels of land use change by a wide margin and even contradicts USDA data.119 EPA
has previously acknowledged the methodological and analytical flaws in these studies and should
do so here.120 Moreover, many studies addressing LUC and the RFS fail to recognize (1) the
extent to which cropping practices contribute substantially to meeting increased demand for
corn, and (2) that production of dried distillers grains (DDGS) has offset substantial demand for
corn as livestock feed.121 EPA should also acknowledge a recent study with countervailing
findings that assessed the local impact of ethanol plants on cropland transitions and concluded
that ethanol plant expansions actually reduce the probability of cropland conversion by 0.5% on
average. It found further that fields near ethanol plants are 10% less likely to be converted from
non-agricultural land into cropland than fields farther away.122
c. EPA should acknowledge, as it has in other contexts, that there is
no established causal link between the RFS and land use change
EPA’s discussion of environmental impacts in the DRIA is fundamentally flawed in that
it fails to adequately address the causal issues described above. As discussed further below with
respect to specific environmental media, throughout the DRIA, EPA refers to potential
environmental impacts associated with biofuels production rather than direct causal impacts of
the RFS program or the Proposed Rule. The Second Triennial Report suffered from these same
analytical shortcomings, which EPA has taken pains in the years since the report’s release to
clarify its analysis, but it is unfortunately repeating the same errors here. Specifically, EPA
explained that:
The [Second Triennial] Report explored general associations
between biofuel production (much of which has no relation to RFS
rules), crop cultivation, and environmental impacts. Although the
Report also inferred some land-use impacts related to biofuel
production were linked to the RFS program, it did not causally link
specific RFS annual rules with land-use impacts. The Report
instead emphasized the difficulties in making such a causal
attribution. And, critically, the Report did not attempt an analysis
of impacts caused by the later released 2019 Rule.123
119 See Jennifer B. Dunn. et al, Measured Extent of Agricultural Expansion Depends on Analysis
Technique, 11 Biofuels, Bioproducts & Biorefining 247 (2017); Joshua Pristola & Randall
Pearson, Critical Review of Supporting Literature on Land Use Change in the EPA’s Second
Triennial Report to Congress (2019); 2019 Stillwater Report; Nov. 2019 Ramboll Supplemental
Memo; Net Gain Report at 4–5.
120 See 2020 No Effect Memo at 15–16; EPA Br. at 92.
121 See Aug. 2019 Ramboll at 23–24.
122 Net Gain Report at 4–5 (citing Pates and Hendricks, Estimating the Local Impact of Ethanol
Plants on Cropland Transitions, Selected Poster prepared for Presentation at the 2019
Agricultural & Applied Economics Association Annual Meeting, Atlanta, GA (2019)).
123 EPA Br. at 101 (internal citations omitted) (emphasis added).
36
Further, EPA’s 2020 No Effect Memo clarified the findings in the 2018 Triennial Report
suggesting that biofuels grown for the RFS program have an impact on land use change or water
quality and quantity:
This report did not specifically address the impacts of the 2020 RFS
standards…. [and] while the report did discuss literature that
generally relates biofuels to crop cultivation, such statements are of
limited relevance to the 2020 RFS standards as they did not purport
to establish any causal link between the 2020 RFS standards and
increased crop cultivation…. [T]he report did not purport to
establish a causal connection between the 2020 RFS standards or
any other RFS annual rule and land use changes…. Thus, the report
is of limited utility in assessing the environmental impacts of the
2020 RFS standards.124
The 2020 No Effect Memo further explains that the Second Triennial Report did not consider
necessary factors including complex regulatory and market factors that are relevant to evaluating
a causal relationship, which results in an analysis that “is not accurate and leads to incorrect
attribution of land use change and biofuels.”125 EPA has similarly concluded that “[e]vidence
shows that recent RFS rules, like the 2019 Rule [which establishes the same conventional
volumes as the Proposed 2022 Rule] are not associated with increased corn and soybean demand
or cultivation in the United States.”126 However, EPA does not discuss in the DRIA the
substantial evidence that supported this conclusion with respect to the prior RVO rulemakings
and applies equally to its analysis of environmental impacts of the Proposed Rule.
The DRIA also overlooks EPA’s own critiques of Lark and others that assert a causal
connection between the RFS and land use change based on flawed methodologies, assumptions,
and data sets. For example, EPA previously criticized Lark’s failure to establish the necessary
causal links to support an assertion that the RFS causes LUC, or to make a defensible link
124 2020 No Effect Memo at 14 (emphasis added).
125 Id. at 15.
126 Id. at 87.
37
between any particular planting or land conversion decisions that may impact habitat or species
and the RFS program.127
2. The RFS program has not caused adverse impacts to wetlands, wildlife,
ecosystems; nor will the Proposed Rule
With the necessary background and context set forth above, we now turn to EPA’s
discussion of environmental impacts to wetlands, wildlife, and ecosystems in the DRIA. As an
initial matter, the introductory discussion to this section of the DRIA makes numerous specious
claims that must be corrected. First, it reiterates the Triennial Report’s claim that there is “an
observed increase in acreage planted with soybeans and corn between the decade leading up to
the enactment of EISA and the decade following enactment. Evidence from observations of land
use change suggests that some of this increase in acreage and crop use is a consequence of
increased biofuel production.”128 These statements ignore the data that total acreage planted in
corn has actually remained stable over time despite substantial increases in corn yields and
ethanol production.129 It implies that EISA and the RFS cause increased corn plantings while
failing to take into account other factors that influence each link in the causal chain described
above, such as ethanol demand external to the RFS (i.e., use in E10 to boost octane and the rise
in foreign ethanol exports) and the complex variables that impact individual farmers’ decisions
regarding what to plant, where, and how much.
EPA then asserts without support that “[i]t is likely that the environmental and natural
resource impacts associated with land use change are, at least in part, due to increased biofuel
production and use.”130 To what land use change is EPA referring that is relevant here? Acres
planted in corn have been stable for over a decade. Further, biofuels production is driven by
many factors, of which the RFS program is only one, and EPA has previously established that
the program, as implemented, is not driving demand.131 EPA should acknowledge that the best
available science supports that there is no causal connection between the RFS and LUC, and
127 See EPA Br. at 91–92 (“Dr. Lark generalizes that, across the United States, only 27% of
uncultivated lands that unregulated farmers converted to croplands were planted to corn. This
means most of the lands are converted for other reasons. Even when farmers plant corn on
converted land, most of this corn is grown for non-biofuel uses. Even for corn that is ultimately
used for biofuel production, Environmental Petitioners present no evidence that such biofuels are
ultimately used for RFS compliance (as opposed to being exported). And assuming some corn
grown on some undisclosed plot of converted land was used to produce biofuels satisfying the
RFS volumes, Environmental Petitioners present no evidence that the unregulated farmer would
not have converted the specific parcel or planted corn but for the 2019 Rule.”) (internal citation
omitted).
128 DRIA at 88.
129 See Net Gain Report at 9; 2019 Stillwater Report at 7–10; 2022 Stillwater Report at 2.
130 DRIA at 88.
131 2020 No Effect Memo at 6; EPA Br. at 96.
38
therefore there are no established impacts to wetlands, wildlife, and ecosystems caused by the
RFS program in general or the Proposed Rule in particular.132
In particular with respect to wetlands, EPA’s discussion of losses of wetlands cite to
studies that do not even attempt to link wetlands losses to the RFS or even to biofuels production
in general.133 Rather, the reports explore wetlands losses on cropland and rangeland without any
particular nexus to biofuels, ethanol, or the RFS. EPA’s conclusion is that it cannot “determine
the portion of wetlands acres lost in order to grow feedstocks for biofuels,” but there is no reason
to link the loss of wetlands to biofuels in the first instance.134 EPA’s discussion thus misleads
the reader into thinking there is some established connection between wetlands loss and biofuels
production when there is not.
135 EPA later suggests that simply because there has been wetlands
loss since 2007 it is possible the 2022 volumes will exacerbate those losses without the necessary
causal nuances described in detail above.136 These conclusions are unfounded and misleading,
especially when offered in the context of the Draft Regulatory Impact Analysis that is supposed
to explore the regulatory impacts of the RFS historically and the Proposed Rule’s potential future
effects.
Similarly, with respect to ecosystems, EPA acknowledges at the outset that “attributing
the fraction of these changes [to ecosystems] to biofuels is not currently possible with any degree
of confidence” and therefore attributing changes to the RFS program and the Proposed Rule are
not possible.137 EPA nonetheless discusses at-length studies and data exploring reductions in
rangeland and grassland that have no identifiable link to the RFS program or the Proposed Rule,
without devoting any attention to explaining the many complex reasons why attribution is not
possible and is not supported by the scientific literature.
With respect to wildlife, on the one hand, EPA acknowledges that “[w]hile the impacts of
land use and management on wildlife have been studied, the impacts of biofuels generally and
the annual volume requirements under the RFS program specifically have not.”138 But it also
asserts that biofuels production is driving conversion of grasslands to cropland with potential
risks for grassland species including birds and insects.139 The DRIA then summarizes a variety
of scientific literature on losses of bird species and insect pollinators such as bees.140 Consistent
with its conclusion on wildlife, EPA asserts that it cannot “confidently estimate the fraction of
wildlife habitat loss or of corn or soy production that is attributable to biofuel production or use,”
132 Net Gain Report at 12–13.
133 DRIA at 89.
134 Id. (emphasis added).
135 See Net Gain Report at 12–13.
136 DRIA at 91.
137 Id. at 92.
138 Id. at 97.
139 Id. at 96.
140 Id.
39
much less to the RFS program or a particular RVO rulemaking.141 EPA should acknowledge that
the scientific literature does not support that there is any relationship between the RFS program
and adverse wildlife impacts, rather than leaving readers with the distinct impression that there
are impacts of bird species and bees tied to the program, but they are not quantifiable or
“confidently” estimable.
Further, claims by parties (e.g., Lark) that have sought to tie the RFS to impacts to
wildlife have been thoroughly disproven. For example:
There is no evidence that the whooping crane is affected by annual RFS rules. The
population has been increasing over time and has grown at an accelerated rate after the
RFS was implemented.
There is no evidence that the Black-footed ferret is affected by annual RFS rules.
Populations have been rapidly increasing since 2000, with no dip apparent in the years
after the RFS was implemented.
There is no evidence that annual RFS rules are impacting Gulf Sturgeon by exacerbating
the Gulf of Mexico dead zone. The Gulf Sturgeon’s critical habitat is located east of the
Mississippi River delta, while the Gulf of Mexico hypoxic zone is exclusively to the
west. Moreover, as addressed below, there is no evidence that land use tied to the RFS
has impacted nutrient loading in the Gulf of Mexico because nutrient loading has
remained relatively constant from 1980 through present day.142
3. The RFS Program has not caused adverse impacts to soil and water
quality; nor will the Proposed Rule
In the section of the DRIA titled “Drivers” in the soil and water quality section, EPA
asserts that because corn-grain ethanol and soy biodiesel account for most biofuels “produced to
date[,] … the majority of soil and water quality impacts from biofuels thus far have come from
the production of corn and soybeans.”143 EPA then extensively discusses adverse soil and water
quality effects that are tied to agricultural activities in general and alleged, but unquantified,
extensification of corn and soybean plantings, in particular.144 EPA concludes that “[a]ssumed
increases in biofuel production associated with higher RFS volumes would likely lead to an
increase in land for agriculture …. [and] an increase in cropland acreage would generally be
expected to lead to more negative soil and water quality impacts.”145 This discussion is incorrect
and misleading for a number of reasons. First, nowhere does EPA explain that there is no
evidence linking adverse soil and water quality impacts to the RFS program, as the latest
scientific literature confirms.146 Second, EPA simply assumes any increase in biofuel production
141 Id. at 98.
142 Nov. 2019 Ramboll Supplemental Report at 7–12.
143 DRIA at 100.
144 Id. at 101–12.
145 Id. at 113‒14.
146 Net Gain Report at 14, 16–17.
40
would drive extensification when there is no support for this claim, as addressed extensively
above. Increases in ethanol production have occurred for the entire duration of the RFS program
without need for new cropland for corn plantings given efficiency and yield improvements.147
Further, EPA reliance on the Second Triennial Report to suggest that corn grown for
ethanol is a contributor to eutrophication in downstream water bodies is misplaced.148 The
Second Triennial Report fails to consider that these watersheds are composed of a complex mix
of urban and rural uses, where agriculture runoff may be only one component—and where there
is still no attempt to link corn grown for ethanol to the RFS.149 Significantly, nutrient loading
into the Gulf of Mexico has remained relatively stable in the last 40 years, despite corn yield
increasing dramatically during this time frame.150 Furthermore, the data demonstrates that
regional hypoxic conditions in Western Lake Erie and the Gulf of Mexico were increasing in
frequency and severity long before ethanol production in the United States increased.151 In
addition, there has been a reduction in total nitrogen concentrations in surface water bodies in
Iowa, which is the highest corn producing state, a fact which further refutes that there is a link
between expanded corn production for any reason and increased nutrient loading.152 Though
more progress can always be made on nutrient efficiency and minimizing runoff, the data shows
that eutrophication is not worsening, even with higher levels of corn production, and that
increased production of corn does not result in increased usage of nutrients—in fact, the most
recent data show that there has been a downward trend in nutrients required on a per-bushel
basis. In other words, even an increased production of corn does not result in increased usage of
nutrients and has actually resulted in a reduction in nutrient requirements on a per gallon of
ethanol basis.153
EPA also does not address recent farming technologies and techniques that reduce the use
of fertilizers and nutrients. For example, fertilizers have been reduced through the use of
precision agriculture, variable-rate application, and GPS- and sensor-based mapping which
restrict the addition of fertilizer to the area immediately around the plant.154 Seed improvements
have also produced plants with increased efficiency at utilizing available nitrogen, thus further
lowering fertilizer application requirements.155 Advances in sustainable farm management,
including substantial improvements in nutrient formulation and use, and technological
improvements in pesticide and fertilizer application, will continue to reduce the potential for
147 Id. at 9.
148 DRIA at 100, 108.
149 Net Gain Report at 18–19.
150 Id.
151 Id. at 18, Figure 6 (using data from the U.S. Geological Survey, Nutrient Loading for the
Mississippi River and Subbasins, https://nrtwq.usgs.gov/mississippi_loads/#/).
152 Aug. 2019 Ramboll Report at 14.
153 2019 Stillwater Report at 29.
154 Id.
155 Id.
41
impacts to water quality in regional watersheds near corn growing areas regardless of the cause
of historical water quality impacts.156
4. The RFS Program has not caused adverse impacts to water quantity and
availability; nor will the Proposed Rule
EPA’s discussion of water quantity and availability suffers from the same flaws as its
analysis of the environmental media discussed above. Specifically, EPA fails to meaningfully
distinguish between potential impacts to water availability that are tied to the RFS and the
Proposed Rule and potential impacts associated with biofuel production at any volumes and due
to any number of drivers (e.g., export demand, demand for ethanol for E10 blending). The wide
body of scientific literature on water quantity does not support a causal nexus between the RFS
and strained water resources.157 Further, in the Proposed Rule and DRIA, EPA ignores that
advances in farming practices and technology have reduced the potential impacts of biofuel
production on water resource availability. EPA states that “the primary driver of impacts to
water quantity is the water used for irrigation of the biofuel feedstocks.”158 Yet, today, most of
the corn grown in the United States is non-irrigated. There has been an increased use of
precision agriculture methods and best practices to retain soil moisture and reduce tilling, all of
which mitigate any impacts to water resource availability.159 In addition, technological
advancements such as genetic engineering for improved drought tolerant corn cultivars result in
corn that is able to tolerate reductions in water without affecting yield.160 The chart below
clearly demonstrates a decreasing trend in water used for growing corn over the past few
decades, in part as a result of these practices.161
156 Aug. 2019 Ramboll Report at 7.
157 Net Gain Report at 14–17.
158 DRIA at 115.
159 Nov. 2019 Ramboll Supplemental Report at 30.
160 Net Gain Report at 16–17.
161 Id. at 17, Figure 5.
42
Moreover, EPA places too little emphasis on the reduction in water use accomplished by
ethanol refineries in recent years. Ethanol refineries have made great strides to reduce water
consumption by employing practices such as process optimization, capture of water vapor from
dryers, and boiler condensate recycling.162 EPA should correct its discussion of water quantity
and availability to, at a minimum, make clear that the scientific literature does not suggest that
the RFS program causes any adverse impacts.
5. EPA does not adequately assess the environmental impacts associated
with petroleum-based fuels
Any possible effects of biofuel production and refining on the environment should not be
viewed in a vacuum, but should be viewed with respect to the alternative—production of
gasoline and diesel fuel. Although EPA acknowledges that a comparison between the effects of
biofuel and petroleum production should be undertaken, it states that “such an assessment would
be expansive and could not be performed on the timeline of this rulemaking.”163 However, there
is a wealth of readily-available scientific literature that addresses petroleum impacts, which
clearly document direct adverse effects on land use change, water quality, ecosystems (including
wetlands), and wildlife.164 Instead EPA presents an incomplete assessment of water-related
impacts of petroleum, without consideration of petroleum production’s impact on land use
change, habitat, wildlife, and ecosystems. In short, in assessing the factors in Section
211(o)(2)(B)(ii)(I), EPA needs to consider the potential for indirect environmental impacts
biofuels as compared to the potential for such impacts from the petroleum-based fuels that the
biofuels are displacing. This comparative analysis is at the heart of lifecycle GHG analysis,
which assesses the lifecycle GHG emissions from a gallon of renewable fuel compared with the
162 Id. at 15; Aug. 2019 Ramboll Report at 33.
163 DRIA at 112.
164 See Aug. 2019 Ramboll Report at 37–43.
R² = 0.4346
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1994 1998 2003 2008 2013 2018
Average Volume of Water Applied to Corn for
Grain or Seed (Acre‐feet/Ac.)
Pressure Systems (ac‐ft/ac) Gravity Systems (ac‐ft/ac)
43
equivalent gallon of petroleum fuel. Logically, the same comparative analysis should be
undertaken with respect to all of the environmental factors in Section 211(o)(2)(B)(ii)(I).
B. EPA Should Finalize a “No Effects” Finding Under Section 7 of the
Endangered Species Act
The Proposed Rule suggests that EPA is considering U.S. Fish and Wildlife Service
(“USFWS”) consultation pursuant to Section 7(a)(2) of the Endangered Species Act (ESA) for
the Proposed Rule based on two D.C. Circuit cases.165 Specifically, the Proposed Rule states that
based on the D.C. Circuit’s decisions finding that EPA failed to make an adequate effects
determination under Section 7 of the ESA for the 2018 and 2019 RFS rules, EPA would consider
initiating consultation “as appropriate” for the Proposed Rule.166 Here, consultation is not
appropriate for a number of reasons. At a minimum, to the extent that EPA does engage in
informal consultation with the wildlife services (USFWS and/or the National Marine Fisheries
Service), these agencies should easily conclude that the Proposed Rule is not “likely” to
adversely affect listed species or habitat and thus determine that a formal consultation process is
not required.167
As a preliminary matter, the 2020 and 2021 proposed volumes will be entirely
retroactive; therefore, EPA is amply supported in finding that the proposed volumes for those
years have no effect on listed species or habitat. In addition, EPA may, and should, make a
finding that the 2022 RVO volumes would have no effect on listed species or habitat. EPA need
not consult with USFWS because it can make a reasoned showing that there is no evidence of
any impact. First, there is no evidence that the RFS incentivizes ethanol production at the
implied conventional volumes proposed, which are the same as the 2019 and 2020 RVOs. As
EPA corrected concluded in the 2020 No Effect Memo, even if EPA were “to waive all of the
RFS requirements for 2020, [it] would expect domestic use of ethanol in the U.S. to only
decrease slightly” due to the ubiquity of E10 blending.168 Nor is there evidence that
demonstrates land use change or water impacts as a result of crops grown for renewable fuel for
the RFS program, either in past years or at the volumes proposed for 2022. Even further
removed, without impacts on land and/or water, there can be no impact on listed species or
habitat. And as EPA concluded in the DRIA, “it is not possible to …. confidently estimate the
impacts to … wildlife … generally nor from the annual volume requirements, specifically.”169
165 See Am. Fuel & Petrochemical Mfrs. v. EPA, 937 F.3d 559 (D.C. Cir. 2019); Growth Energy
v. EPA, 5 F. 4th 1 (D.C. Cir. 2021).
166 NPRM at 72,442.
167 See 50 C.F.R. 402.13(c).
168 2020 No Effect Memo at 6.
169 DRIA at 98–99.
44
This is because there is no evidence of such a causal relationship. To be sure, none of the studies
cited in EPA’s DRIA demonstrates any link.170
As further explained below, EPA may explain and support a no effect finding with ample
evidence in the record, including the Ramboll Report and Supplemental Memo and the Net Gain
Report (and literature cited therein), all of which undertake an exhaustive effort to address what
evidence exists regarding corn grown and ethanol produced for the RFS and impacts to the
environment and wildlife, and find none.
In any event, if EPA still determines that the Proposed Rule may affect listed species or
habitat, all that is required is informal consultation with USFWS to determine whether the action
“is not likely to adversely affect” a listed species or critical habitat.171 The record clearly
supports a finding that the RFS is not likely to adversely affect listed species or habitat. As set
forth in Ramboll’s November 2019 Supplemental Memo and Net Gain’s Expert Report, there is
no evidence of any causal relationship between corn grown for the RFS and impacts to
endangered or threatened species or habitat.
Specifically, there is no evidence that (1) the RFS increases the demand for ethanol or the
volumes in the Proposed Rule would increase demand for biofuels in a market-forcing manner;
(2) the increased demand for ethanol determines the prices of corn; (3) the price of corn
determines a farmer’s decision to plant corn; (4) the farmer would clear previously uncultivated
land to plant corn, resulting in land use change; (5) this increased corn production would cause
nutrient loading and worsen hypoxia in downstream water bodies (whether due to intensification
or extensification); and (6) the geographic locations at which land use change or intensification
may occur and/or impacts to water quantity or quality affect any particular listed species or
habitat. Consistent with EPA’s determination in the original 2020 RVO, “[g]iven the highly
attenuated causal chain between the [Proposed Rule] and potential impacts on listed species and
critical habitat, any such impacts would be ‘only reached through a lengthy causal chain that
involves so many steps as to make the consequence not reasonably certain to occur.’”172
As EPA further correctly explained in the original 2020 No Effect Memo and is equally
applicable here:
[The Proposed Rule] do[es] not require, authorize, fund, or carry out
the production of any specific biofuel or crop, the use of any land
that is critical habitat, or the taking of any listed species or other
activity that may affect any listed species. Decisions on what type
of feedstock to use for biofuel production, where such feedstocks
are grown, the types and volumes of agricultural inputs such as
fertilizer or pesticide to use in growing the feedstocks, and what
types of renewable fuel will ultimately be produced, are made by
170 See id. §§ 3.3.3, 3.4.2.3 (citing various studies on terrestrial and aquatic species, none of
which connect impacts to corn grown for the RFS to adverse impacts).
171 50 C.F.R. 402.13(c).
172 2020 No Effect Memo at 4 (quoting 50 C.F.R. 402.17(b)(3)).
45
third parties, and any on-the-ground activities to implement and
carry out those decisions are undertaken by such third parties.
Moreover, some third parties, notably farmers who decide how
much and where they plant crops, are not regulated by the RFS
program at all.173
In addition, EPA has previously comprehensively addressed that the 2018 Second
Triennial Report “did not purport to establish a causal connection between the 2020 RFS
standards or any other RFS annual rule and land use changes,” and is therefore of “limited utility
in assessing the environmental impacts of the … RFS standards.”174 That finding is equally
applicable to EPA’s fulfillment of its obligations under the ESA here.
Moreover, the declaration of Tyler Lark, which was relied upon in both of the D.C.
Circuit decisions referenced above, does not serve as a foundation for a finding that the Proposed
Rule is “likely to adversely affect” a listed species or critical habitat. This declaration has
subsequently been widely critiqued, with much of the evidence cited in the declaration actually
refuting the declaration’s own assertions. As just a few examples, an analysis of several of the
Lark Declaration’s claims of land conversion in specific areas (which allegedly affected species
such as the whooping crane, Powesheik skipperling, and yellow-billed cuckoo), found that those
areas had been converted long before the inception of the RFS, therefore negating any causal
connection.175 Similarly, the Declaration summarily concluded that corn and soy production
worsened the Gulf of Mexico dead zone thus impacting the Gulf Sturgeon; however, the dead
zone does not overlap temporally or geographically with habitat for the Gulf Sturgeon.176
EPA has previously acknowledged the limitations of the Lark Declaration. Specifically,
EPA considered the evidence discussed in the Lark Declaration and found shortcomings,
including that the underlying studies were based on inconclusive temporal and spatial
associations using satellite imagery.177 As EPA explained, “there is no way to determine if the
crops grown on a particular parcel were used for biofuel production versus some other use ….
[the studies cited in the Lark Declaration] remain probabilistic and limited in scope, and
[insufficient to] identify impacts on particular parcels of land.”178 In sum, in the context of an
informal consultation with USFWS for this rulemaking, EPA should carefully review and
173 Id. at 3.
174 Id. at 14.
175 Nov. 2019 Ramboll Supplemental Report at 7–12.
176 Id. at 15 (concluding that “the presumption in the Lark Declaration that the RFS has resulted
in impacts to Gulf sturgeon is unsubstantiated”).
177 See 2020 No Effect Memo at 15.
178 Id. at 8.
46
consider the body of literature that found the Lark Declaration to rely on unsupported
assumptions and speculation, used faulty data sets, and is otherwise unreliable.179
Thus, EPA may explain and support a no effect finding with ample evidence in the
record, but at a minimum, EPA, in informal consultation with USFWS, should conclude that the
RFS “is not likely to adversely affect” listed species or habitat.
C. Air Quality Impacts of Proposed Rule
EPA’s brief analysis of air quality impacts in the Proposed Rule, relying in part on the
Anti-Backsliding Study, concludes that the overall impacts to air quality of the biofuels volumes
in the Proposed Rule will be minor and therefore do not favor higher or lower volumes.180 EPA
finds that the overall concentration of ethanol in gasoline is likely to remain constant at
approximately 10% as a result of the rulemaking, rather than the rule spurring increased
production and use of higher-level ethanol blends.181 EPA’s analysis in the DRIA overlooks the
air quality benefits of ethanol-blended fuels.
In particular, EPA should acknowledge the benefits of ethanol-blended fuel in reducing
emissions of potent air toxics such as benzene and 1,3 butadiene, as well as particulate matter
(PM) and carbon monoxide.182 Specifically with respect to primary PM2.5, a new study finds
substantial cold start emissions reductions associated with increased ethanol blending and the
dilution of aromatics in the final blend. The attached report from Drs. Kazemiparkouhi,
MacIntosh, and Suh summarizes the study’s findings and its implications, namely that ethanolblended fuels can drive air quality improvements in communities with environmental justice
concerns that, due to proximity to congested roadways, are more adversely impacted by tailpipe
emissions from mobile sources, including primary PM2.5.
183 In sum, EPA should consider the
potential emissions and air quality benefits of ethanol-blended fuels in setting volumes of
renewable fuels under its reset authority and in future RFS rulemakings.
179 See Joshua Pristola & Randall Pearson, Critical Review of Supporting Literature on Land Use
Change in the EPA’s Second Triennial Report to Congress (2019); 2019 Stillwater Report; Nov.
2019 Ramboll Supplemental Report; Net Gain Report.
180 See DRIA at 58.
181 Id. at 58, 61.
182 See Growth Energy Comments on Proposed Anti-Backsliding Determination for Renewable
Fuels and Air Quality, Docket ID No. EPA-HQ-OAR-2020-0240 (July 1, 2020),
https://www.regulations.gov/comment/EPA-HQ-OAR-2020-0240-0012.
183 Comments of Drs. Fatemeh Kazemiparkouhi, David MacIntosh, Helen Suh, EPA-HQ-OAR2021-0324 (Feb. 3, 2022) (attached as Ex. 4).
47
D. Increasing Renewable Fuel Volumes Benefits Communities with
Environmental Justice Concerns
1. Environmental justice and climate change
Growth Energy strongly agrees with EPA that the proposed increases in the renewable
fuel volumes in 2022 will reduce GHG emissions and therefore may mitigate disproportionate
impacts of climate change on low income and vulnerable communities.184 As addressed above,
biofuels such as corn ethanol contribute substantially to reducing GHG emissions in the
transportation sector. For example, recent analysis finds that nationwide use of E15 in lieu of
E10 could reduce U.S. GHG emissions by over 17 million tons per year, the equivalent of
removing 3.85 million vehicles from the roads.185 Although it may be difficult to quantify with
precision the benefits to vulnerable communities associated with reductions in GHG emissions,
EPA is correct to take these known benefits into account in a qualitative manner.
2. Environmental justice and air quality
EPA’s discussion of potential air quality impacts of the Proposed Rule on communities
with environmental justice concerns overlooks the extent to which these communities may
experience improvements in local air quality associated with combustion of gasoline-ethanol
blends, especially at higher concentrations. Combustion of the fossil fuel component of
gasoline and diesel results in harmful primary particulates and toxic aromatics like benzene and
toluene. As EPA notes, low income, minority, and vulnerable communities are often proximate
to major roadways where these pollutants are more concentrated.186 Increased biofuel-blending
can mitigate these emissions. In particular, a recent study conducted by the University of
California, Riverside found that greater use of ethanol-blended fuels can reduce carbon
monoxide, ozone, and primary particulate matter levels relative to the use of gasoline-only
fuels.187 In addition, as discussed above, primary PM2.5 emissions from gasoline-ethanol blends
are lower than non-blended fuels. Primary PM2.5 emissions have substantial human health
impacts as discussed above and in Exhibit 4. Because communities of concern face
disproportionate impacts of such emissions, we encourage EPA’s environmental justice analysis
to take into account the ability of increased biofuel-blending to ease the pollution burdens these
communities bear, including through reductions in primary particulates and the toxic
constituents in gasoline.
3. Environmental justice and water/soil impacts
EPA suggests in the DRIA that the proposed 2022 volumes may “have
disproportionately severe negative impacts on environmental justice communities within
184 See DRIA at 221.
185 Air Improvement Resources, Inc., GHG Benefits of 15% Ethanol (E15) Use in the United
States 2 (Nov. 30, 2020), http://www.airimprovement.com/reports/national-e15-analysisfinal.pdf.
186 DRIA at 222.
187 University of California, Riverside, Center for Environmental Research and Technology (“CE-CERT”) Study.
48
American Indian tribes and other low income populations that rely on local fisheries as a source
of food or income or that may not be able to afford costly water filtration systems to address
nitrate contamination in their drinking water.”188 This sweeping statement is wholly
unsupported by EPA’s analysis of past and potential environmental impacts of the RFS program
in general and the 2022 proposed volumes in particular. First, as addressed extensively above,
the scientific and economic literature on these issues does not indicate that the RFS drives
farmers’ choices to plant particular crops and raises corn prices.189 There are many complex
and interrelated variables that impact planting decisions and commodity decisions. Further,
even if the RFS may drive more demand for corn or soybeans, the literature does not support
that demand would be met in a manner that entails negative impacts to water quality, e.g., by
greater application of fertilizers that may cause additional agricultural runoff, nutrient leaching,
and algae blooms.
EPA’s conclusory and alarmist statements regarding endangerment to human
populations that rely on fisheries entirely overlook that over the last four decades as demand for
biofuels has increased substantially, fertilizer use has gone down as farming technologies have
improved over time.190 Further, nutrient loading to the Gulf of Mexico has remained stable since
the 1980s,191 and as EPA acknowledged in the Second Triennial Report, total nitrogen
concentrations in surface water bodies in Iowa (where corn growth has intensified) has gone
down.192 Moreover, EPA does not geographically associate hypothesized adverse impacts to
water quality to environmental justice communities that may rely on local fisheries as a food
source or communities that cannot afford water filtration. There is no evidence to support that
proposed 2022 volumes may adversely impact such communities.
IV. EPA’S PROPOSED MODIFICATION OF THE 2020 STANDARDS UNDERMINES THE RFS
PROGRAM, CONTRADICTS THE CLEAN AIR ACT, AND IS IRRATIONAL
“Congress intended the Renewable Fuel Program to be a ‘market forcing policy’ that
would create ‘demand pressure’ to increase consumption of renewable fuel.”193 EPA’s proposal
to retroactively lower previously set standards to the point of actual use—and thereby relieve
obligated parties of their failure to comply with their legal obligations under those standards—
nullifies Congress’s RFS policy. EPA’s proposed retroactive absolution creates perverse
incentives for future compliance: no longer will there be any reason for obligated parties to
188 DRIA at 224.
189 As addressed in Part VI, our analysis supports that the Proposed Rule will not have adverse
impacts on food prices. However, to the extent the proposed volumes do have an impact on
prices, we agree with EPA that such impacts will be minimal and will not adversely impact low
income and vulnerable communities. See DRIA at 225.
190 See Net Gain Report.
191 Aug. 2019 Ramboll Report at 27.
192 EPA, Biofuels and the Environment: Second Triennial Report to Congress 69 (June 2018),
https://cfpub.epa.gov/si/si_public_file_download.cfm?p_download_id=542063&Lab=IO.
193 ACE, 864 F.3d at 705.
49
bother trying to comply with RFS standards; if they fail, they can count on EPA to bail them out.
And the worse they fail, the more likely EPA will be to save them. Obviously, Congress did not
authorize EPA to do this. The reset provision invoked by EPA certainly does not; that provision
permits only prospective adjustments, and requires that any adjustments be focused on
addressing the circumstances that triggered the reset. Moreover, contrary to EPA’s assertions,
the 2020 volumes do affect future use of renewable fuel through the carryover RIN mechanism
(as EPA acknowledges elsewhere in its proposal). At most, EPA could reduce the 2020
standards to account for the actual level of SREs and cellulosic production but no further.
A. EPA’s Proposal to Retroactively Lower Previously Set Standards to the
Level of Actual Use Negates the RFS and Therefore Is Impermissible
EPA proposes “to retroactively reduce the 2020 volumes to those actually used.”194
Although EPA admits that “retroactively adjusting the 2020 standards will disrupt market
expectations created by the prior final rule, for instance on the part of biofuel producers,” it states
that it is “relieving burdens on obligated parties, and in some cases, the potentially onerous
burden of noncompliance with the RFS program and the possibility of penalty payments.”195
EPA’s reasoning is baffling. For one thing, EPA admits that its proposal will harm
renewable fuels producers—the entities that Congress expected to benefit economically from the
RFS program—for the benefit of obligated parties—the entities on whom Congress placed the
burden of compliance. EPA provides no explanation—nor could it—for why it is permissible or
appropriate to choose to transfer the cost of obligated parties’ noncompliance onto renewable
fuels producers.
More fundamentally, Congress designed the RFS program “to force the market to create
ways to produce and use greater and greater volumes of renewable fuel each year.”196 The threat
of penalties for noncompliance is precisely how the program implements this market-forcing
policy.197 If compliance was to be required only when compliance is not burdensome, there
would have been no point in requiring compliance.
EPA’s proposed approach will teach obligated parties that they need not bother trying to
comply with their congressionally mandated obligations in the future because if they fail to, EPA
will relieve them of those obligations—and the worse they fail, the more likely EPA will be to do
so, creating an insidious incentive.198 This time, EPA invokes the Covid-19 pandemic; next
time, it might invoke some other unusual circumstance. This approach conflicts with Congress’s
194 NPRM at 72,449.
195 Id.
196 ACE, 864 F.3d at 710.
197 Id. at 704-705 (noting that Congress “place[d] any compliance burdens” on obligated parties
to “ensur[e] that the renewable fuel volume requirements are met”); see Duquesne Light Co. v.
EPA, 791 F.2d 959, 960 (D.C. Cir. 1986) (Clean Air Act’s penalties are “designed to alter …
economic behavior by changing the costs of” noncompliance (quotation marks omitted).
198 NPRM at 72,449.
50
intent and is an irrational way to manage the RFS program. Indeed, the Supreme Court recently
rejected it, explaining that interpreting the RFS statute such that “the least compliant refineries
[w]ould be the most favored” would have “the strange”—and impermissible—“effect of
disincentivizing … refineries from ever trying to comply.”199
EPA asserts that “this rulemaking has no ability to affect actual production, imports, and
use of renewable fuel in 2020,” since “2020 has already passed.”200 EPA’s reliance on this
truism disregards how the RFS program functions, as EPA itself admits elsewhere in its
proposal. The 2020 standards will have such effects in future years through the mechanisms of
the carryover RIN bank and deficit carryforwards. As EPA correctly says, the original “higher
volumes for 2020 … would cause some combination of a drawdown of the carryover RIN bank
[and] carryforward deficits.”201 And EPA correctly explains that the carryover RIN bank and
carryforward deficits “operate such that … any RIN deficits [in one year] can impact the market
for RINs and renewable fuels in the next year. As such, compliance with the RFS standards for
one year is inherently intertwined with compliance for the prior year.”202 In other words, as EPA
says, the “actual market effects” of one year’s RFS standards are “mediated through the
carryover RIN bank” from the prior year’s compliance performance.203 EPA’s disregard of this
obvious reality when crafting its modification of the 2020 standards therefore renders its
proposal internally incoherent and irrational.
EPA insists on taking an expansive, atextual view of its statutory authority—a view that
allows it to nullify the very statutory program it was charged with administering. This is not how
agencies are supposed to interpret their authority. EPA cannot disregard the fundamental nature
of the RFS program as a binding, market-forcing policy that operates through enforced
compliance obligations, or the reality that enforcement of the obligations for past years affect the
course of the program. As the Supreme Court has held, it is fundamental to the constitutional
separation of powers and the constitutional process of bicameralism and presentment that the
Executive Branch cannot nullify a duly enacted statute—even if the statute expressly says it
can.204 Even an agency’s power to cancel a specific statutory program can exist only where
Congress has clearly granted that power and the cancellation “execut[es] the policy that
Congress had embodied in the statute.”205 That standard is not met here. Again, the statute does
not authorize cancellation, and it makes no sense to think that Congress implicitly intended EPA
to be able to use the reset provision to cancel the program when that would defeat the central
congressional purpose of the RFS program: forcing the market to increase its use of renewable
199 HollyFrontier Cheyenne Ref., LLC v. Renewable Fuels Ass’n, 141 S. Ct. 2172, 2182 (2021).
200 NPRM at 72,448.
201 Id. at 72,457.
202 Id. at 72,454.
203 Id. at 72,456.
204 Clinton v. City of New York, 524 U.S. 417, 443 (1998) (invalidating Line Item Veto Act);
Ethyl Corp. v. EPA, 51 F.3d 1053, 1060 (D.C. Cir. 1995) (“We refuse, once again, to presume a
delegation of power merely because Congress has not expressly withheld such power.”).
205 Clinton, 524 U.S. at 444.
51
fuel through binding RFS obligations. As the Supreme Court has said, “Congress … does not
alter the fundamental details of a regulatory scheme in vague terms or ancillary provisions—it
does not, one might say, hide elephants in mouseholes.”206
That is doubly true for an interpretation that would give the agency the power to nullify a
congressional program “of vast economic and political significance,” like the RFS.207 When it
comes to “authorizing an agency to exercise [such] powers,” courts “expect Congress to speak
clearly.”208 Congress has not clearly given EPA the power to cancel the RFS standards, and
therefore EPA does not have that power. On the contrary, Congress imposed on EPA “a
statutory mandate to ‘ensure[]’ that [the volume] requirements are met.”209
B. The Reset Power Is Not Available to Reduce Already-Set Standards for a
Past Year or to Reduce Volume Requirements More Than Needed to
Address the Circumstances That Triggered the Reset
The Clean Air Act provides that if EPA “waives” an “applicable volume requirement set
forth in [the statutory] table” by “at least 20 percent … for 2 consecutive years” or by “at least 50
percent … for a single year,” EPA “shall promulgate a rule … that modifies the applicable
volumes set forth in the [statutory] table … for all years following the final year to which the
waiver applies.”210 Because EPA used a cellulosic waiver to reduce the total and advanced
volume requirements by more than 20 percent in consecutive years, EPA triggered its authority
to reset the total and advanced volume requirements.211 EPA now proposes to use this power to
reduce the standards it set for 2020 far beyond the level needed to offset the cellulosic shortfall
that materialized in 2020.
EPA lacks the authority to use the reset to alter previously set standards for a past year, to
reduce volume requirements by more than needed to address the circumstances that triggered the
reset, and certainly to do both together. EPA asserts that the reset provision “provides EPA
broad discretion to modify the renewable fuel volumes and to establish biofuel volume
requirements at the volumes actually consumed.”212 In particular, EPA says that the reset
provision “give[s] EPA considerable discretion to weigh and balance the various factors required
by statute” and “implied authority to consider factors that inform our analysis of the statutory
206 Whitman v. American Trucking Ass’n, 531 U.S. 457, 468 (2001).
207 Alabama Ass’n of Realtors v. HHS, 141 S. Ct. 2485, 2489 (2021) (quotation marks omitted);
see, e.g., Massachusetts v. EPA, 549 U.S. 497, 532-534 (2007); Whitman, 531 U.S. at 485; North
Carolina v. EPA, 531 F.3d 896, 910-911 (D.C. Cir. 2008).
208 Alabama Ass’n of Realtors, 141 S. Ct. at 2489.
209 ACE, 864 F.3d at 698-699 (quotation marks omitted); see 42 U.S.C. § 7545(o)(2)(A)(i),
(3)(B)(i).
210 42 U.S.C. § 7545(o)(7)(F).
211 NPRM at 72,443.
212 Id. at 72,449.
52
factors.”213 EPA’s expansive view of its discretion is substantially overstated. EPA has
mistakenly read the list of reset factors in isolation, divorced from the rest of the statute. The
reset power is in fact a narrow one whose scope reflects its triggering circumstances: it enables
EPA to adjust the volume requirements to account for a systemic supply shortage or severe harm
relative to Congress’s original expectations, and to do so prospectively across multiple years, for
the purpose of relieving EPA and interested parties of the burdens and uncertainties of relying on
ad hoc annual waivers. In other words, the reset provides EPA the power to anticipatorily issue a
multi-year waiver.
1. EPA cannot use the reset power to retroactively reduce past standards
EPA proposes to use the “reset” power to retroactively reduce the previously set 2020
standards. Those standards instructed market participants how to structure their conduct during
2020 and that afforded obligated parties ample time to comply. EPA may not unwind them now.
Courts are “reluctant to find … authority [for retroactive rulemaking] absent an express statutory
grant.”214 Here, there is nothing close to an express grant to use the reset power retroactively.
On the contrary, interpreting the statute that way would, as explained, allow EPA to nullify the
RFS program—something Congress did not permit. The reset provision in the statute can only
be read to permit prospective modifications.
The statute prescribes prospective sequencing: EPA is statutorily required to issue a reset
“rule []within 1 year after issuing [the triggering] waiver”215 and “no later than 14 months before
the first year for which [the reset] applicable volume will apply.”216 When using the reset power,
EPA is to “modif[y] the applicable volumes … for all years following the final year to which the
[triggering] waiver applies.”217 And the statute directs EPA to consider factors that are expressly
forward-looking and make sense only that way. For example, EPA must consider “the expected
annual rate of future commercial production of renewable fuels.”218 All these provisions convey
Congress’s intent that the reset be performed before the compliance years for any standards that
are reset.
Moreover, the statute specifies that EPA may use the reset to modify only “the applicable
volumes set forth in the [statutory] table.”219 That precludes using the reset to modify standards
that have already been modified through a waiver, as EPA proposes here, since the 2020
standards already reflect EPA’s prior cellulosic waivers.
EPA invokes judicial decisions (reluctantly) interpreting the Clean Air Act to permit EPA
213 Id. at 72,443.
214 Bowen v. Georgetown Univ. Hosp., 488 U.S. 204, 208-209 (1988).
215 42 U.S.C. § 7545(o)(7)(F).
216 Id. § 7545(o)(7)(F); id. § 7545(o)(2)(B)(ii).
217 Id. § 7545(o)(7)(F) (emphasis added).
218 Id. § 7545(o)(2)(B)(ii)(III) (emphasis added).
219 Id. § 7545(o)(7)(F); id. § 7545(o)(2)(B)(ii).
53
to promulgate RFS standards after a statutory deadline.220 That authority does not support EPA’s
proposed retroactive reset of the 2020 standards. For one thing, the issue here is not whether
EPA can set volume requirements after the statutory deadline but whether it can alter ones it has
already set after the compliance year. The issue in EPA’s cited cases was whether, by missing
the statutory deadline to issue percentage standards, EPA “forfeited its authority” to do so.221
Only because that result would have undermined the entire RFS program did the D.C. Circuit
find that the strong presumption against retroactive rulemaking was overcome. Emphasizing
“Congress’ focus on ensuring the annual volume requirement was met regardless of EPA delay,”
the D.C. Circuit reasoned that denying EPA the ability to issue RFS standards after it has missed
the statutory deadline would “lead to the drastic and somewhat incongruous result of precluding
EPA from fulfilling its statutory mandate” to ensure that the requirements are met,222 which
would thus be “flatly contrary to Congress’ intent and would turn agency delay into a windfall
for” obligated parties.223
Nothing remotely like that is at stake from missing the statutory deadline for a reset. In
fact, the circumstances here are the reverse. The reset power is not essential to EPA’s
implementation of the program or, specifically, its fulfillment of its duty to ensure that the
volume requirements are met. Even without a reset, there would be volume requirements and
EPA could use them to set percentage standards—exactly as EPA did in the original 2020
rulemaking. Rather, the reset is merely a convenience to relieve interested parties and EPA of
the burdens and uncertainties of having to use the ad hoc waiver process serially. This is evident
from the facts that the reset is triggered only by large or repeated waivers, and that the reset
provision directs EPA to reset the volume requirements for all subsequent years once
triggered.224 Thus, the reset provision reflects Congress’s recognition that if its original volume
expectations were too far off the market, it would be more convenient administratively and
provide more long-term certainty to the market to allow EPA to adjust the statutory volumes for
all remaining years in one fell swoop than to go through an inevitable waiver process each year.
In short, the reset power provides EPA the ability to issue an advance multi-year waiver.
To be clear, even if EPA can exercise its reset authority after the statutory deadline, it can
do so only for future years, not for a past year. Like the statutory interpretation rejected by the
D.C. Circuit in EPA’s cited cases, allowing EPA to reset already-set standards for past years
would interfere with EPA’s overarching statutory duty to ensure that the volume obligations are
met and would afford obligated parties a windfall, relieving them of apparently legally binding
obligations they already had ample opportunity to comply with. EPA’s view here would also
pervert the reset: instead of using the reset to increase market certainty by averting the need for
ad hoc waivers, EPA’s proposal undermines market certainty by unsettling already-finalized
220 NPRM at 72,438 & n.5.
221 Monroe Energy, LLC v. EPA, 750 F.3d 909, 919 (D.C. Cir. 2014).
222 Id. at 920 (quotation cleaned).
223 National Petrochemical & Refiners Ass’n v. EPA (“NPRA”), 630 F.3d 145, 157 (D.C. Cir.
2010) (quotation marks omitted).
224 42 U.S.C. § 7545(o)(7)(F).
54
standards and altering the effect of actions taken in reliance on those standards.
EPA’s general authority to reconsider finalized RFS standards is no answer.225 An
agency’s power of reconsideration is not unfettered; it must still be exercised pursuant to and
within the bounds of the agency’s statutory authority. And as explained, EPA lacks statutory
authority to reset already-set standards for a past year.
2. EPA has no statutory authority to use the reset to reduce volumes further
than needed to address the cellulosic production issues triggering the reset
As just explained, the reset is intended to function as an anticipatory, multi-year waiver.
Accordingly, the circumstances that lead to the triggering waiver also constrain the scope of the
reset authority: EPA may use its reset authority only to adjust the applicable volumes to account
for the conditions that led to the triggering waivers and, but for a reset, would continue to do so.
As it turned out, the reset was triggered by cellulosic waivers due to shortfalls in cellulosic
biofuel production, and therefore EPA may use the reset to reduce the cellulosic, advanced, and
total volume requirements only to the extent needed to account for future shortfalls in cellulosic
production. Insofar as EPA’s proposal would use the reset to reduce the 2020 volumes below
that level, it exceeds EPA’s authority and is unlawful.
The statute ties the scope of the reset to the triggering conditions. The reset power is
triggered only by a “waive[r],”226 and a waiver is available “only in limited circumstances”227: if
“implementation” of an RFS volume requirements “would severely harm the economy or
environment of a State, a region, or the United States”; if “there is an inadequate domestic
supply” of renewable fuel to meet a requirement; if “the projected volume of cellulosic biofuel
production is less than the minimum applicable volume”; or if “there is a significant renewable
feedstock disruption or other market circumstances that would make the price of biomass-based
diesel fuel increase significantly.”228 Further, the reset provision—which is presented in the
statutory subsection defining the “waivers”—is triggered only by significant waivers in
consecutive years or a large waiver in one year.229 And the reset provision directs EPA to
modify the volumes “for all years following the final year to which the [triggering] waiver
applies.”230 All this together shows that the reset serves as an alternative to the ordinary annual
waiver process; when it becomes evident that repeated waivers will be inevitable because
Congress’s initial expectations were substantially incorrect, instead of going through the waiver
process each subsequent year, EPA can effect a multi-year waiver, modifying the volume
requirements for all subsequent years in fell swoop, simplifying the process for EPA and all
interested parties and providing the market with greater certainty about future RFS obligations.
225 NPRM at 72,444.
226 42 U.S.C. § 7545(o)(7)(F).
227 NPRA, 630 F.3d at 149.
228 42 U.S.C. § 7545(o)(7)(A), (D), (E).
229 Id. § 7545(o)(7)(F).
230 Id. (emphasis added).
55
Accordingly, the purpose of the reset is to remedy the circumstances that led to the
triggering waivers and that would continue to lead to future waivers but for a reset, and that is
the extent of EPA’s power. EPA’s view, in contrast, is that once there is a triggering waiver for
the reset, EPA can rewrite the relevant applicable volume to any degree it wants, as long as it
considers the statutorily specified factors.231 But there is no reason to suppose—in fact, it is
absurd to suppose—that Congress intended the reset provision to grant EPA more power than
necessary to address the condition that triggers the reset authority in the first place. “‘Modify,’”
the D.C. Circuit has held, “connotes moderate change.”232 And neither the power to modify nor
the duty to consider various statutorily specified factors in conducting the reset requires greater
authority. Rather, EPA is to assess those factors for the purpose of determining the expected
extent of the triggering condition in the future and the appropriate future volume, accounting for
what is feasible and potential important and severe adverse consequences. That is, as explained
above, EPA is to consider the statutory reset factors for the purpose of determining how to set
new volume requirements that address the reset-triggering conditions while trying to avoid
causing conditions for other waivers in the future. That is it; again, Congress does not sneak
major powers into obscure technical provisions.
Finally, the reset provision’s use of the word “any” has no bearing on this issue. The
statute states, “[f]or any of the tables in paragraph (2)(B),” if EPA “waives” a requirement “in
any such table” by the requisite amount, it may reset.233 Thus, “any” merely refers to the
applicable volumes whose waiver can trigger a reset: any of them—cellulosic, BBD, advanced,
or total—not just one or a subset of them. The word “any” does nothing to define the scope of
the modification that EPA may make to the volume requirement.
C. EPA’s Proposed Rationale for Reducing the 2020 Standards Is Irrational
and Contrary to the Statute
EPA proposes to reduce the 2020 standards in response to two “significant and
unanticipated events”: (1) “The COVID–19 pandemic and the ensuing fall in transportation fuel
demand, especially the disproportionate fall in gasoline demand relative to diesel demand, which
significantly reduced the production and use of biofuels in 2020 below the volumes we
anticipated could be achieved”; and (2) “The potential that the volume of gasoline and diesel
exempted from 2020 RFS obligations through small refinery exemption (SREs) will be far lower
than projected in the 2020 final rule.”234 EPA says that if it were “to simply leave the original
volumes from the 2020 final rule in place, we would expect some combination of potentially
disruptive outcomes: (1) A reduction in the quantity of carryover RINs; (2) obligated parties
carrying deficits into 2021; and/or (3) obligated parties being out of compliance with their RFS
obligations.”235
231 See NPRM at 72,442-72,443.
232 MCI Telecomms. Corp. v. American Tel. & Tel. Co., 512 U.S. 218, 228 (1994).
233 42 U.S.C. § 7545(o)(7)(F).
234 NPRM at 72,438.
235 Id. at 72,448.
56
At every turn, this analysis is unreasonable and contrary to the Clean Air Act. At most,
EPA could lower the 2020 standards to account for the actual level of SREs and the actual level
of cellulosic production, but no further.
1. Pandemic-related demand destruction does not justify reducing the 2020
percentage standards
As EPA acknowledges, RFS obligations are “self-adjusting” to the actual level of
demand for transportation fuel because the obligations “are applied as a percentage to an
obligated party’s gasoline and diesel fuel production; the obligation to acquire RINs for
compliance rises and falls along with gasoline and diesel fuel production volume.”236 In other
words, the percentage standards automatically reduce the volume of renewable fuel required to
be used proportionally to the reduction in demand for transportation fuel. Accordingly, lowerthan-projected demand for transportation fuel cannot generate compliance difficulties, and
retroactively reducing the original volume requirements proportionally to the subsequent
reduction in demand serves no purpose because the percentage standards already do that
automatically. And reducing the original volume requirements beyond the proportional
reduction already accounted for by the percentage standards to account for lower-than-projected
demand for transportation fuel is obviously unjustified. In sum, the only permissible adjustment
for lower actual fuel use is pointless one.
Yet, EPA proposes to reduce the 2020 standards by a disproportionately large amount.
This is evident in its proposed revised percentage standards. For example, EPA would reduce
the total standard from 11.56% to between 10.78% and 11.36%.237 If EPA were proposing to
reduce the volume requirements proportionally to the lost demand, the percentage standards
would remain unchanged.
Although EPA acknowledges that the percentage standards “self-adjust[]” for “a shortfall
in gasoline and diesel fuel consumption relative to the projected volumes results,” EPA insists
that 2020 is “different” from prior years when that happened because “the shortfalls in 2020
were … significantly larger than in any previous year.”238 This is irrelevant and irrational. The
function of a percentage standard is that it self-adjusts in perfect proportionality regardless of
how much the fuel use changes. Indeed, EPA admits that “the decrease in transportation fuel
demand in 2020 proportionally decreased the required renewable fuel volume.”239
2. EPA’s apparent over-projection of SREs and the shortfall in production of
cellulosic biofuel could justify at most only some of the proposed
reduction of the 2020 standards
EPA points to the possibility that SREs will be lower than the projection on which the
2020 standards were based. Adjusting RFS standards retroactively to account for inaccurate
236 Id.
237 Id. at 72,465.
238 Id. at 72,448.
239 Id.
57
SRE projections is a bad policy because it promotes market uncertainty. Producers and obligated
parties knew well in advance what the RFS requirements were for 2020 and were able to
conform their conduct to those requirements. Producers held up their end of the bargain: there
was ample renewable fuel to meet the 2020 standards. Obligated parties did not. Relieving them
now for the SRE shortfall disrupts the planning and investments that producers have made. It is
distressing, as well as arbitrary and capricious, for EPA to consider only the benefits to obligated
parties and disregard the cost to producers when considering a retroactive action. EPA must
“consider[] the benefits and the burdens attendant to its approach.”240 In any event, if EPA will
reduce standards retroactively when actual SREs are lower than projected, its policy will be
irrational and therefore unlawful unless it also commits to raising standards retroactively when
actual SREs are higher than projected.
Assuming EPA can adjust the standards to account for actual SREs, then if no SREs are
granted for 2020—as EPA rightly proposes—the SRE shortfall would account for some but not
all of the projected 2020 RIN shortfall. The appropriate way to adjust the standards to
retroactively account for actual SREs and actual transportation fuel use is to recompute the
percentage standards using these actual figures instead of their projections, along with the
original volume requirements. Setting G and D to their actual figures (123.25 and 50.49) and
setting GE and DE to zero, while keeping RFV-rf set to 20.09, yields a Total percentage standard
of 11.09%—well below the original requirements of 11.56%—and a total volume requirement of
17.64 billion RINs.241 The most recent EMTS data (rather than the August 2021 data EPA used
for its proposal) indicate that 17.06 billion RINs were separated in 2020.242 Therefore,
accounting for lower transportation fuel demand and zero SREs still leaves a RIN shortfall of
0.58 billion (17.64 minus 17.06).
Additionally, EPA points to the shortfall in cellulosic production, and invokes the
cellulosic waiver authority to reduce the 2020 cellulosic standard.243 Even putting aside the fact
that EPA could issue cellulosic waiver credits to cover this shortfall, this shortfall could justify
only an extremely small reduction to the 2020 cellulosic standard. The percentage standards’
self-adjustment already accounts for some of this shortfall. The additional shortfall in cellulosic
production beyond the amount accounted for automatically by the percentage standards is only
0.02 billion RINs. Using the method for recomputing the percentage standards just described—
which maintains the original volume requirement but uses actual fuel use and zero SREs—yields
a revised cellulosic standard of 0.33%, or 0.52 billion RINs.244 The latest EMTS data show that
240 ACE, 864 F.3d at 718 (quotation cleaned; emphasis added).
241 See Stillwater Associates LLC, Comments to EPA on 2020-2022 RFS Rule, Prepared for
Growth Energy 18-19 (Feb. 4, 2022) (“2022 Stillwater Report”) (attached as Ex. 5).
242 See id. at 19.
243 NPRM at 72,438 n.4.
244 See 2022 Stillwater Report at 18-19.
58
0.50 billion cellulosic RINs were separated in 2020.245 Therefore, the additional cellulosic
shortfall that could be the subject of a cellulosic waiver is only 0.02 billion (0.52 minus 0.50).
In sum, adjusting the standards to account for actual transportation fuel use, for zero
SREs, and actual cellulosic production leaves a 2020 RIN shortfall of 0.56 billion (0.58 minus
0.02). EPA’s proposal to set the 2020 standards to actual levels would erase that shortfall. But
as explained in the next sections, EPA’s reasons for doing that contradict the statute and the duty
of reasoned decisionmaking.
3. The disproportionate decline in gasoline use relative to diesel use does not
justify further retroactive reduction of the 2020 standards
EPA says that the decline in fuel demand “disproportionately affected gasoline more than
diesel fuel.”246 “This is important,” EPA states, “because on average finished gasoline contains
more renewable content than finished diesel.”247 That does not matter. A disproportionate
decline in demand for gasoline and diesel can happen in any year; it is an inherent risk of the
program. Demand for gasoline and demand for diesel are affected by different factors, so there
is no reason to expect that they will necessarily move together. If such a divergence justified
adjustment for 2020, it could equally justify adjustment for any compliance year. But EPA has
never even considered making such adjustments, and they are unwarranted: Obligated parties
can manage their compliance performance daily, and many do, and the RIN market is liquid.
Therefore, obligated parties have ample opportunity to ensure that they meet their compliance
obligations even as demand for gasoline and demand for diesel diverge.
Even if this disproportionate demand loss provided a theoretically valid justification for
retroactively reducing the 2020 percentage standards, it would not suffice to lower the 2020
standards because EPA has not adequately explained and justified its specific proposal. EPA has
not quantified the extent of this disproportionate loss or tied that to the extent of its proposed
reduction in the standards. In other words, EPA needs to determine how much of the 0.56 billion
RIN shortfall is attributable to this disproportionate decline in gasoline use. Such an analysis
would need, at a minimum, to account for the fact that the actual volume of BBD exceeded the
amounts required for compliance and thus can offset at least some of the disproportionately large
loss of gasoline demand. Specifically, if the standards are recomputed as described here—using
the original volume requirements, the actual transportation fuel used, and zero SREs—the BBD
standard would be 2.06 billion (2.10%).248 But according to the most recent EMTS data, 2.48
billion BBD RINs were separated, leaving an excess of 0.42 billion BBD RINs.249 The most
direct and appropriate way to account for this excess is simply to ignore the disproportionate
245 See id. at 19.
246 NPRM at 72,448.
247 Id.
248 2022 Stillwater Report at 18-19.
249 Id.
59
decline in demand for gasoline relative to diesel, allowing all the excess BBD RINs to offset the
shortfall in RINs associated with gasoline (particularly D6 RINs).
(EPA states that “the projections in the 2020 final rule overestimated the use of biodiesel
and renewable diesel, even if we adjust those projections by the shortfall in diesel demand.”250
That assessment does not reflect an adjustment for the over-projection of SREs. But even
without that adjustment, EPA’s statement would still be incorrect in light of the most recent
EMTS data. Without an adjustment for the over-projection of SREs, the actual BBD obligation
according to the original percentage standard is 2.15 billion RINs, but again 2.48 billion BBD
RINs were separated.251)
Finally, EPA could rationally adopt its proposed position to reduce the standards given
the disproportionately larger decline in demand for gasoline only if it committed to also raising
the standards to account for a disproportionately larger decline in demand for diesel in a future
year.
4. EPA’s refusal to allow supposedly “disruptive outcomes” violates the
statute and undermines the RFS program
Obligated parties could close the RIN gap and achieve compliance by using carryover
RINs or carrying their RIN deficits into 2021 (or both). Or they could elect to incur the penalties
for noncompliance. EPA deems these “disruptive outcomes.”252 But the idea that such outcomes
could justify a reduction in the standards is nonsense. They are simply the natural consequences
of obligated parties’ failure to stay on top of their legally binding RFS obligations. If they
“disrupt” anything, it is only to disrupt the preference of some obligated parties—many of which
are integrated with or closely tied to petroleum producers—not to increase their use of renewable
fuel as required. Such disruption is precisely what Congress intended the RFS program to
cause—it is a “market forcing policy.”253 Carryover RINs allow obligated parties to make up
their deficiency in the same compliance year, deficit carrying gives obligated parties extra time
to make up their deficiency, and the imposition of penalties is the backstop to ensure that
obligated parties do not simply disregard their program obligations.254
If EPA could reduce the RFS standards retroactively to avoid these outcomes, there
would be no point in having binding RFS obligations at all and the program would be worthless.
Congress could not possibly have intended EPA to manage the program this way. In fact,
Congress was clear that EPA could not reduce the requirements merely to avoid these outcomes.
Congress expected that compliance with the RFS would sometimes be challenging or unpleasant
for obligated parties, but expressly decided to allow EPA to reduce the nationwide volume
250 NPRM at 72,448.
251 2022 Stillwater Report at 18.
252 NPRM at 72,448.
253 ACE, 864 F.3d at 705.
254 42 U.S.C. §§ 7545(d)(1), (o)(5)(D).
60
requirements “only in limited circumstances”255: in particular, if “there is an inadequate domestic
supply” of renewable fuel or if implementation “would severely harm the economy or
environment of a State, a region, or the United States.”256 EPA’s proposal in effect would grant a
nationwide waiver based on less serious difficulties—difficulties that EPA does not suggest
would meet the statutory waiver requirements. The D.C. Circuit has already rejected this
approach, explaining that there is no reason to think “Congress would have established the
severe-harm waiver standard only to allow waiver” under another provision “based on lesser
degrees of economic harm.”257 As explained above, this principle constrains even the reset
power, which is properly viewed as a power to issue an anticipatory, multi-year waiver.
5. EPA’s management of the RIN bank is incoherent and exposes EPA’s
mistaken belief that its role is to manage the fuel market
EPA’s refusal now to intentionally maintain standards that could lead to a drawdown of
carryover RINs exposes EPA’s management of the bank as irrational, counterproductive, and
beyond its authority under the Clean Air Act. Specifically, EPA’s proposal reveals that EPA
actually views its role as managing the fuels market by maintaining RIN prices within a narrow
band that EPA, in its inscrutable judgment, deems appropriate.
According to EPA, having a “bank of carryover RINs is extremely important in providing
a liquid and well-functioning RIN market upon which success of the entire program depends,
and in providing obligated parties compliance flexibility in the face of substantial uncertainties in
the transportation fuel marketplace.”258 Neither of these functions, however, supports EPA’s
proposal for 2020.
With respect to maintaining liquidity and a well-functioning RIN market, EPA explains
that “[c]arryover RINs enable parties ‘long’ on RINs to trade them to those ‘short’ on RINs
instead of forcing all obligated parties to comply through physical blending.”259 That is true, but
it explains only the role of tradeable credits; it does not justify maintaining a large bank of RINs
carried over from a past year for compliance in a future year. The tradability of RINs enables
obligated parties to shuffle their RINs for a given year as that year’s compliance demonstration
deadline approaches, enabling efficient compliance with obligations that apply to the year in
which the credits were generated. For example, if one obligated party separated 10 million fewer
RINs than needed to meet its 2020 obligations, and another obligated party separated 14 million
more RINs than needed to meet its 2020 obligation, the short party could buy 10 million RINs
from the long party, and both could meet their 2020 obligations. That is sufficient for the RIN
market to provide the flexibility needed to facilitate compliance. It allows all obligated parties to
achieve compliance in the most efficient way possible: more-efficient parties can separate excess
255 NPRA, 630 F.3d at 149 (emphasis added).
256 42 U.S.C. § 7545(o)(7)(A), (D), (E).
257 ACE, 864 F.3d at 712.
258 NPRM at 72,454.
259 Id.
61
RINs, which less-efficient parties can purchase for less than the marginal cost of increasing their
own separation of RINs directly.
But even if there can be a perpetual RIN bank, it is not justified by a desire for RIN
liquidity. And certainly, liquidity does not justify the massive RIN bank that EPA insists on
maintaining. Maintaining a RIN bank into a future year does not help create a well-functioning
RIN market—again, that already exists without a perpetual bank. What it actually does is
suppress RIN prices, lowering the marginal cost of compliance and thereby deterring the market
from making the very investments that Congress intended to the RFS program to incentivize. As
the D.C. Circuit has recognized, “higher RIN prices” are how the RFS program achieves its
market-forcing policy of promoting increased renewable-fuel use: they “incentivize precisely the
sorts of technology and infrastructure investments and fuel supply diversification that the RFS
program was intended to promote.”260
As for using the bank to provide compliance flexibility in the face of substantial
uncertainties in the transportation fuel market, EPA’s proposed modification belies that purpose.
EPA proposes that, because of the Covid-19 pandemic and associated economic disruptions—an
“unforeseeable circumstance[] that [supposedly] could limit the availability of RINs”—EPA
should reduce the 2020 obligations so that obligated parties do not need to use the RIN bank for
compliance.261 In a nutshell, EPA’s position is: the bank provides RINs that obligated parties
can use in an emergency, but in an emergency, the obligations should be reduced so that
obligated parties do not need to use the bank of RINs. That is facially irrational. The message
EPA’s conduct sends is that it wants to maintain the bank to suppress RIN prices. Indeed, D6
RINs have never exceeded $2 and have rarely exceeded $1.262 But again, as the D.C. Circuit has
recognized, higher RIN prices are the essential mechanism by which the RFS program was
designed to achieve its goal of forcing the market to use rapidly escalating volumes of renewable
fuel.263 By taking actions to suppress RIN prices, EPA undermines Congress’s intent.
Moreover, EPA offers no explanation for why the RIN bank needs to be so big to
accomplish its purported goals. EPA says that there will be about 1.85 billion carryover RINs in
the bank for 2020.264 If obligated parties used some of those carryover RINs to satisfy the 0.56
billion RIN shortfall in 2020 (computed above), there would still be 1.29 billion carryover RINs
for 2021 and thereafter. Under EPA’s proposal, no drawdown will be needed for 2021—
because EPA would set those standards to the actual levels of renewable use—or for 2022—
because EPA would set those standards to levels EPA believes could be achieved entirely
through physical blending in 2022.265 Even if carryover RINs are needed to meet the 250-
260 Monroe Energy, 750 F.3d at 919.
261 NPRM at 72,454.
262 EPA, RIN Trades and Price Information, https://www.epa.gov/fuels-registration-reportingand-compliance-help/rin-trades-and-price-information.
263 Monroe Energy, 750 F.3d at 919.
264 NPRM at 72,455.
265 Id. at 72,451, 72,455.
62
million supplemental obligation that EPA will set for 2022 to begin remedying the unlawful
2016 general waiver, there would still be 1.04 billion carryover RINs for compliance in 2023 and
beyond, with ample opportunity for obligated parties to build the bank back up with more excess
renewable-fuel use in 2022 and subsequent years.266 And even if obligated parties relied
exclusively on carryover RINs to meet the second 250-million supplemental obligation imposed
in 2023 to remedy the unlawful 2016 general waiver, that would still leave at least 0.79 billion
carryover RINs. That would provide a sizeable cushion for 2022, in the event that physical
blending falls short of the 2022 requirements, or for future standards.
EPA must—but fails to—“articulate a satisfactory explanation” as to why obligated
parties need more carryover RINs than that.267 EPA mentions “the uneven holding of carryover
RINs among obligated parties,” but that makes no sense. Because RINs are highly tradeable, any
“uneven” holdings can be evened out so that all obligated parties can comply with their
obligations.268 Again, as EPA itself notes, parties short in RINs can readily buy from parties that
are long. Beyond that, EPA’s only remark on the subject is the vague speculation that, “were
market disruptions to occur with an insufficient carryover RIN bank, it could force the need for a
new waiver of the standards, undermining the market certainty so.”269 But EPA does not explain
why any particular amount of carryover RINs less than 1.85 billion would be “insufficient.”
Moreover, by proposing to retroactively reduce the 2020 standards, especially when there are
sufficient RINs to comply with them, EPA introduces a more fundamental uncertainty: when and
to what extent finalized RFS standards will be binding in the future or will be negated by an
agency that mistakenly views its role as managing the RIN market rather than simply setting
standards sufficient to “ensure” that the applicable volumes are met year after year.270
V. EPA’S PROPOSAL TO SET THE 2021 STANDARDS TO ACTUAL LEVELS UNLAWFULLY
NEGATES THE RFS PROGRAM
EPA proposes to set the 2021 standards equal to the volumes of renewable fuel actually
used in 2021.271 EPA claims this is justified because, due to its own delay, the 2021 standards
cannot affect renewable fuel production in 2021, which will have already passed by the time
standards are finalized.272 EPA adds that setting 2021 standards to actual use will “mitigat[e]
burdens on obligated parties” by ensuring they “have sufficient RINs to comply.”273 These
concerns are misplaced and dangerous to the RFS program, and EPA lacks authority to set the
266 Id.
267 Motor Vehicle Mfrs. Ass’n of U.S., Inc. v. State Farm Mut. Auto. Ins. Co., 463 U.S. 29, 43
(1983).
268 NPRM at 72,455.
269 Id. at 72,454.
270 42 U.S.C. § 7545(o)(2)(A)(i), (3)(B)(i); ACE, 864 F.3d at 698-699, 710.
271 NPRM at 72,450.
272 Id.
273 Id.
63
2021 standards on that basis. EPA’s “concerns” reflect a failure to grasp the implications of how
the RFS program functions, which EPA recognizes in other contexts, revealing the irrationality
of its analysis for 2021. More fundamentally, other than lowering the cellulosic standard to the
actual level, EPA’s proposed approach for 2021 reflects a power to cancel the RFS program. But
as discussed above, EPA has no authority to do that. Certainly, that is not what the reset
provision authorizes; as explained above, the reset can be used only prospectively, consistent
with its purpose as a multi-year waiver. There is no benefit from using that power for a past
year. Notably, again, the reset power is not essential to EPA’s implementation of the program
or, specifically, its fulfillment of its duty to ensure that the volume requirements are met, so there
is no special need to be able to use the reset retroactively.
Even if EPA could use the reset to determine a past year’s standards, EPA’s proposal
would exceed its reset authority. As explained above, EPA’s focus in using the reset must be
confined to modifying the standards to address the conditions that triggered the reset. Thus, at
most EPA could use the reset to adjust the 2021 cellulosic, advanced, and total standards to
account for the shortfall in cellulosic production. Similarly, even if EPA can use its cellulosic
waiver power on a past year, that power is limited to reducing the volume requirements to
account for the shortfall in cellulosic production. And as explained later, EPA cannot and should
not reflexively reduce the advanced and total requirements by amounts equal to the cellulosic
shortfall. Rather, EPA must backfill any shortfalls with other types of available qualifying
renewable fuel.
EPA has the technical ability to set appropriate standards for 2021 other than by simply
matching the actual levels. For example, it could do so using relevant data that was available as
of November 30, 2020, as it would have done had it issued the 2021 standards on time, limited
only by the volume of carryover RINs and a manageable level of carryforward RIN deficits.
Although setting the standards that way now could not cause greater use of renewable fuel in
2021, compliance could be achieved in several ways and thereby could promote increased use of
renewable fuel in future years. EPA could directly set higher 2021 requirements; EPA could
combine the higher 2021 volumes with the 2022 volumes to create a combined standard for
2022; or EPA could issue a 2021 standard equal to actual use and add a supplemental obligation
to the 2022 standards equal to the difference between the actual 2021 volumes and what it would
have set the 2021 standards to had it acted timely. The first approach would likely lead obligated
parties to draw down the RIN bank, and the latter two approaches would allow obligated parties
to choose between drawing down the bank and increasing actual use of renewable fuel in 2022.
Any of these approaches would increase renewable fuel use over the course of the RFS
program—as EPA acknowledges elsewhere and as discussed above, the “market effects” of RFS
standards are “mediated” across years “through the carryover RIN bank” (drawing down the
bank in one-year increases the need for actual use in future years because obligated parties
cannot turn to those retired carryover RINs for compliance)—consistent with Congress’s
intent.274 And as explained above, setting standards to intentionally draw down the bank is
entirely appropriate.
274 Id. at 72,456.
64
This is, in fact, exactly what EPA did in prior years when it missed the deadline for
issuing RFS standards, and the D.C. Circuit approved of those actions as furthering Congress’s
intent. For instance, EPA did not issue the 2009 BBD standard until well into 2010. Yet, instead
of writing off that year by retroactively setting the 2009 BBD standard to the actual level, EPA
“combined” the 2009 and 2010 volumes “into a single requirement” to “ensure that these two
year[s’] worth of [fuel] will be used.”275 This combined approach, EPA said, best fulfilled what
“Congress expected and intended.”276 The D.C. Circuit sustained EPA’s “combined” approach,
concluding that it fulfilled Congress’s intent, and EPA’s statutory duty, of “ensuring the annual
volume requirement[s are] met regardless of EPA delay.”277 “EPA could not ignore the 2009
mandate” due to its own delay.278 Not maintaining the statutory requirements for both past and
current years, the Court said, would thus have been “‘flatly contrary to Congress’ intent and
would turn agency delay into a windfall for the regulated entities.’”279 Similarly attending to
“Congress’ focus on ensuring the annual volume requirement was met regardless of EPA delay,”
EPA issued the 2013 standards at the statutory levels, and the D.C. Circuit affirmed.280
Although EPA then set the late 2014 and 2015 RFS standards to reflect actual levels, and
the D.C. Circuit affirmed that action in ACE, the court’s decision is flawed and does not support
EPA’s proposal for 2021. ACE declared that in issuing late RFS requirements, EPA must
“consider[] the benefits and the burdens attendant to its approach.”281 Yet, the court did not
adhere to that in reviewing EPA’s 2014 and 2015 standards. ACE failed to consider that EPA
could have mitigated any burdens in other ways, such as through carryover RINs or by adding
the late standards to a future year’s requirements. ACE also mistakenly ignored the
overwhelming cost of setting standards equal to actual use: that doing so nullifies the RFS
program for the late year. A proper analysis that accounted for these additional dimensions of
the situation yields the conclusion that setting late standards to the level of actual use is irrational
and contrary to the statute.
Moreover, even if ACE’s approval of EPA’s approach for 2014-2015 were reasonable at
the time, EPA’s subsequent delay in proposing the 2021 standards—and the 2022 standards,
which EPA says will not be finalized before June 2022, almost halfway through the year—shows
that delay and late standards set to actual levels have become EPA’s modus operandi, and thus
that ACE’s approval cannot be extended beyond the 2015 standards. EPA’s now-routine
invocation of this approach displays the very concern ACE assured was not yet present in 2015:
275 Regulations of Fuels and Fuel Additives: Changes to Renewable Fuel Standard Program
(“2010 Rule”), 75 Fed. Reg. 14,670, 14,718 (Mar. 26, 2010).
276 Renewable Fuel Standard Program (RFS2) Summary and Analysis of Comments 3-186-188,
EPA (Feb. 2010),
https://nepis.epa.gov/Exe/ZyPDF.cgi/P1007GC4.PDF?Dockey=P1007GC4.PDF.
277 NPRA, 630 F.3d at 163.
278 Id. at 157.
279 Id. at 156-157 (quoting EPA brief).
280 Monroe Energy, 750 F.3d at 919-921.
281 864 F.3d at 718 (quotation cleaned).
65
that EPA could “us[e] its delay as an excuse to shirk its statutory duties.”282 If EPA can simply
wait until a year is over to set standards and then set them at actual use, an Administrator who
wishes to nullify the program need only do that year after year. And EPA’s pattern of behavior
shows that this is happening and will continue to happen. That is clearly not what ACE
envisioned or approved. Nor is it what the reset provision envisions or permits. Again,
Congress could not have intended to implicitly grant EPA the discretion to cancel a major
statutory program. There is no reason why EPA could not have issued the 2021 standards on
time. Its delay suggests a disregard for the RFS program’s effects. And correspondingly, EPA’s
late proposal suggests a desire to prop up the RIN bank at the expense of Congress’s marketforcing aims, as well as a commitment by EPA to protect obligated parties from binding RFS
obligations to use more renewable fuel than they otherwise would without the RFS and the
compliance consequences of failing to do so. Nothing in the Clean Air Act or ACE purports to
afford EPA that authority.
VI. EPA SUBSTANTIALLY UNDERSTATES THE REASONABLY FEASIBLE VOLUME OF
ETHANOL USE IN 2022
EPA projects that 13.788 billion gallons of ethanol will be used in 2022.283 This is far
below the country’s capacity to produce, distribute, and consume ethanol. EPA primarily blames
the so-called E10 blendwall, declaring that it is “a deciding factor in limiting growth in domestic
consumption of ethanol.”284 That is incorrect. The blendwall is the result, not the cause, or the
gap between actual and potential ethanol use. The cause is economic incentives—at bottom,
consumers are not widely incentivized to select higher-ethanol blends over E10. The design of
the RFS, however, is to provide incentives to increase the use of renewable fuel. Higher RFS
requirements increase RIN values, which in turn lower the price of transportation fuel in inverse
proportion to the concentration of renewable fuel. Thus, EPA can and should, consistent with
Congress’s intent, strive to encourage increased use of ethanol above the blendwall by increasing
the RFS requirements, and especially the implied non-advanced requirement.
Instead, however, EPA proposes to set the 2022 standard based on “the projection of
ethanol concentration derived from EIA reports for 2022 as a reasonable estimate of what level
can be achieved in 2022.”285 In other words, EPA appears to view its task as predicting how the
market will behave independently and then to match the volume requirements to that prediction.
That approach contradicts Congress’s clear intent that the RFS standards would be market
forcing.
282 ACE, 864 F.3d at 719.
283 DRIA at 51.
284 Id. at 35.
285 DRIA at 184.
66
A. More Than 15 Billion Gallons of Ethanol Could Easily Be Produced
Domestically
Domestic producers of ethanol could easily meet demand of more than 15 billion gallons
of ethanol in 2022. Existing facilities have the nameplate capacity to produce about 17.38 billion
gallons per year, according to EIA.286 To date, actual domestic ethanol production peaked in
2018, when 16.061 billion gallons were produced domestically.287 Although it appears that only
about 14.87 billion gallons of ethanol were produced domestically in 2021, that output reflected
the lower demand for transportation fuel caused by the Covid-19 pandemic and is not indicative
of feasible capacity in 2022.288 In any event, it is still substantially more than the 13.788 billion
gallons that EPA assumes will be used in 2022.
Moreover, a fairly conservative assessment of the domestic production capacity in light
of feedstock supply and non-ethanol demand for feedstock shows that about 15.565 billion
gallons could be produced. For this assessment, Stillwater Associates assumed that the number
of planted corn acres would remain at the 2007 level: 93.5 million.289 Stillwater Associates next
assumed that the percentage of planted corn acres that would be harvested would equal the
average percentage over the past decade: 91.3%.290 Then Stillwater assumed that the corn yield
would remain at its 2021 level—177.0 bushels / acre (“bu/ac”)—even though it has increased at
a virtually constant rate of 1.9 bu/ac annually since 1936, and 1.8 bu/ac annually since 2008.291
Stillwater also determined that ethanol conversion rates have increased at an annual rate of 0.01
gallons / bushel of corn between 1982 and 2020.292 Finally, Stillwater assumed that demand for
corn for non-ethanol uses, including feed, food, seed, and other industrial uses, would increase
consistent with population growth, and that corn imports and exports would equal USDA’s latest
estimates.293 Putting this all together, and accounting for the feed co-products produced at
ethanol plants, Stillwater calculates that 15.565 billion gallons of ethanol could be produced in
2022.294
Finally, past export demand for ethanol is irrelevant when projecting the feasibly
available supply of ethanol in the future. As long as there is domestic demand for ethanol to
comply with RFS obligations, domestically produced ethanol will be used to meet that demand
rather than exported. That is because exported ethanol lacks the value of the associated D6 RIN;
without that value, producers receive less for their ethanol in the export market than they do in
286 2022 Stillwater Report at 1; see DRIA at 183.
287 2022 Stillwater Report at 1
288 Id.
289 Id. at 1, 4.
290 Id. at 4.
291 Id. at 3.
292 Id.
293 Id. at 4-5.
294 Id. at 5-8.
67
the domestic market. No rational producer, therefore, will choose to export ethanol when there is
existing demand for the ethanol under the RFS program. The only reason that producers have
exported ethanol in recent years is that there was no demand for that fuel to comply with RFS
obligations.295
B. Substantially More E85 and E15 Could Easily Be Delivered and Consumed
Analysis by Stillwater shows that EPA’s expected volume of ethanol use in 2022—
13.788 billion gallons—will use only a small fraction of the existing infrastructure to deliver and
consume E85 and E15. And of course, the market can expand that infrastructure in response to
appropriate incentives.
1. E85
With respect to E85, Stillwater determines that in 2020, the utilization rate for the
country’s existing E85 distribution infrastructure was between only 8% and 14%, depending on
assumptions regarding the number of dispensers. EPA states that there were about 3,947 stations
offering E85 in 2020.296 More recent and reliable data, however, indicate that there are 4,125
stations selling E85.297 Notably, these are existing dispensers that are compatible with and
approved for use with E85.298 On average, stations that sell E85 have 1.8 E85 dispensers.299
These figures yield a range of existing E85 dispensers: from 3,947 dispensers (based on 3,947
stations with 1.0 E85 dispensers each) to 7,425 dispensers (based on 4,125 stations with 1.8 E85
dispensers each).300 A typical dispenser can deliver 45,000 gallons of E85 per month through
normal use, containing 33,000 gallons of ethanol.301 Therefore, existing E85 infrastructure can
deliver between 2.31 billion and 4.0 billion gallons of E85 per year.302 But EIA projects that
only 320 million gallons of E85 will be consumed in 2022.303 That implies a utilization rate of
8-14%.304
Were the existing capacity used to typical full capacity for dispensers, between and 1.27
billion and 2.36 billion gallons of additional ethanol could be consumed above the ethanol in the
295 Id. at 16.
296 DRIA at 192.
297 2022 Stillwater Report at 8.
298 Id.
299 Id.
300 Id.
301 Id.
302 Id. at 8-9.
303 DRIA at 38; 2022 Stillwater Report at 9.
304 8% = 320 million / 4 billion; 14% = 320 million / 2.31 billion.
68
E10 that would be replaced.305 Put another way, each five-percentage point increase in
utilization delivers an additional 116 million to 200 million gallons of E85, and an additional 85
million to 148 million gallons of ethanol above the ethanol in the replaced E10.306 This analysis
shows just how much room there is to increase the delivery of E85 right now. And it does not
even account for the market’s ability to add new E85 infrastructure, which can be done at low
cost through ordinary upgrade cycles.
As explained above, ethanol producers could largely meet this additional potential
demand. For 2022, EPA expects 13.788 billion gallons of ethanol to be used, but producers
could produce 15.565 billion gallons or more. That difference of at least 1.777 billion gallons of
ethanol is in the middle of the range of additional incremental ethanol that could be dispensed
through existing infrastructure (1.27 bil gal to 2.36 bil gal).
There are also vastly more vehicles than needed to use this additional ethanol. EIA
estimates that there will be 20.4 million FFVs on the road in 2022.307 Based on projections of
vehicle miles driven, each FFV could use 588 gallons of E85 per year, and the fleet of FFVs
could use 12.65 billion gallons of E85 in a year.308 At EIA’s projected 2022 E85 volume of 320
million gallons, each FFV would use an average of 15.7 gallons of E85 for the year, which
represents a vehicle utilization rate of 2.67%.309 Each five-percentage point increase in vehicle
utilization consumes an additional 376.32 million incremental gallons above the E10 it would
replace. Thus, it is obvious that the vehicle fleet could easily consume the potential additional
1.777 billion gallons of ethanol production.
2. E15
The story is similar with respect to E15. According to EPA, there are 2,300 stations
selling E15.310 Stillwater estimates that these stations average 3.3 E15 dispensers each, meaning
that there are about 7,540 E15 dispensers across the country.311 Notably, these are existing
dispensers that are compatible with and approved for use with E15.312 Given that a typical
dispenser can deliver 45,000 gallons of E15 per month through normal use, this existing E15
infrastructure can deliver about 2.9 billion gallons of E15 per year containing 0.145 billion
gallons of ethanol above the ethanol in the replaced E10, accounting for the regulatory barrier to
305 2022 Stillwater Report at 8-9.
306 Id. at 9.
307 Id.
308 Id. at 9-10.
309 Id. at 10.
310 DRIA at 196.
311 2022 Stillwater Report at 11.
312 Id. at 10-11.
69
selling E15 during the summer ozone season.313 There is, as noted, certainly sufficient ethanol
supply to meet that demand level.
And there are more than enough vehicles to use the maximize level of E15. More than
96% of gasoline-using vehicles, accounting for more than 98% of miles traveled, will be E15-
compatible in 2022.314
C. The Principal Impediment to Increased Use of E85 and E15 Is Retail Price,
Which Can Be Addressed Through Higher RFS Standards
For the reasons just discussed, there is no infrastructure impediment to increased
distribution or consumption of E85 or E15. The primary reason that the use of those fuels has
not increased faster is that they have not been priced low enough relative to E10 to entice
consumers to switch to them from E10. Most gas consumers are highly price sensitive and
creatures of habit. And higher-ethanol blends have lower energy content than E10.
Consequently, as EPA observes, higher-ethanol blends must be priced below the point of E10
parity on an energy-equivalent basis to be widely competitive.315
In practice, however, this relative pricing has not been achieved. Analysis of historical
E85 prices shows that on an energy-equivalent (or gasoline gallon equivalent) basis, E85 is
priced at or below parity nowhere and generally is priced substantially above parity.316 At such
prices, E85 appears to be used largely only mandated fleets and consumers who are especially
committed to ethanol—that is, price-insensitive consumers.317 In other words, retailers are
marketing E85 as a niche product sold at a premium price. In order to markedly increase its use,
it would need to be marketed as a mass-market, lower-margin product.318
The RFS program provides a mechanism to redress this relative-pricing problem and spur
greater conversion from E10 to higher-ethanol blends. More demanding RFS standards would
reduce the supply of RINs and thereby raise their price. Because RINs function as a discount, or
coupon, on transportation fuel, the higher the ethanol blend, the greater the RIN-based discount
to the consumer. Consequently, higher RFS standards lower the prices of higher-ethanol blends
313 Id. at 11.
314 Air Improvement Resource, Inc., Analysis of Ethanol-Compatible Fleet for Calendar Year
2022, at 2 (Nov. 16, 2021), https://growthenergy.org/wp-content/uploads/2021/12/Analysis-ofEthanol-Compatible-Fleet-for-Calendar-Year-2022-16Nov21.pdf (attached as Ex. 11); see also
2022 Stillwater Report at 12.
315 DRIA at 37.
316 2022 Stillwater Report at 14-15.
317 Id. at 15.
318 Id.
70
relative to E10, eventually enabling them to be priced below parity with E10 on an energyequivalent basis.319 This is, in fact, precisely how the RFS is supposed to work.
D. EPA’s Speculative Concerns About Misfuelling Liability Are Unfounded and
Irrelevant
EPA’s Draft Regulatory Impact Analysis states that a barrier to increased E15 and E85
use is a fear among some retailers about liability for misfuelling incompatible vehicles.320 EPA,
however, cites no evidence that misfuelling commonly occurs or is otherwise a legitimate
concern. Indeed, EPA’s only citation is a 2015 Comment from the Petroleum Marketers
Association of America that itself offers zero citations in its one conclusory paragraph that floats
this concern.321 As EPA itself has repeatedly explained, regulations already “require pump
labeling, a misfuelling mitigation plan, surveys, product transfer documents, and approval of
equipment configurations, providing consumers with the information needed to avoid
misfuelling.”322 These regulatory safeguards largely negate any risk of misfuelling.
And even if all these safeguards were to fail, the risk remains negligible because, as EPA
again notes, “the portion of vehicles not designed and/or approved for E15 use continues to
decline.”323 As noted, more than 96% of gasoline-using vehicles, representing more than 98% of
vehicle miles travelled will be compatible with E15 in 2022.324 Because all post-2001 models
are E15 compatible, the risk will only continue to decrease each year as older vehicles exit the
marketplace. There will also be an estimated 20.4 million E85 compatible FFVs in use in
2022.325 Billions of miles have been driven on E15, reflecting millions of retail transactions.
Misfuelling issues have not materialized despite such widespread use. And in any event, E15 is
covered under retailers’ liability insurance, like any other federally approved fuel.
Accordingly, even EPA has acknowledged that misfuelling fears are speculative and
likely unfounded, but has said they remain relevant because, “warranted or not,” “some retailers
will continue to have concerns.”326 But it is unreasonable to cater to these unfounded concerns,
using them as an excuse to avoid setting volumes high enough to incentivize readily available
infrastructure growth—a goal Congress clearly intended the RFS to advance. The fundamental
purpose of the RFS is to force more renewable fuel use than the industry would otherwise be
willing to provide. EPA therefore abdicates its congressionally mandated role if it capitulates to
319 Id. at 15-16.
320 DRIA at 196.
321 Id. (citing Petroleum Marketers Association of America Comment 3, EPA-HQ-OAR-2015-
0111-1197 (July 27, 2015)).
322 Id. at 197.
323 Id.
324 Air Improvement Resource Report at 2.
325 Air Improvement Resource Report at 1.
326 Renewable Fuel Standard Program -Standards for 2019 and Biomass-Based Diesel Volume
for 2020: Response to Comments 106-07, EPA-420-R-18-019 (Nov. 2018).
71
such a speculative industry concern, thereby nullifying the RFS’s ability to nudge the industry
into its next logical phase where higher ethanol blends are the norm. If EPA believes its
regulations and the composition of the vehicle fleet mitigate misfuelling concerns, as they clearly
do, then EPA must act accordingly. Surrendering to an unfounded misfuelling concern floated
by the oil industry would be arbitrary and capricious because it is not only a concern “Congress
has not intended [EPA] to consider,” but also clearly “runs counter to the evidence before the
agency.”327 Only once the RFS appropriately incentivizes retailers to adopt higher ethanol
blends will retailers test and move past this unfounded concern.
E. Storage Infrastructure Compatibility Is Not a Meaningful Barrier to
Increased Use of Ethanol or Expansion of Distribution Infrastructure
EPA’s Draft Regulatory Impact Analysis presents the concern that E15 and E85
infrastructure growth is hindered by retailers’ inability to demonstrate that their underground
storage tank systems (USTs) are compatible with higher ethanol blends.328 To be clear, even if
this concern were well-founded, it would not be a meaningful barrier to increased use of ethanol
because, as explained above, existing infrastructure, which is already compatible with and
approved for use with these blends, can deliver vastly more ethanol than has historically been the
case. There is no need to expand infrastructure to increase use. In any event, this concern about
compatibility is not a serious impediment to infrastructure expansion.
Pursuant to EPA regulations promulgated in 2015,329 owners and operators of fuel
stations must be approved to use higher ethanol blends by demonstrating to state implementing
agencies that their “UST system (which includes but is not limited to tanks, pumps, ancillary
equipment, lines, gaskets, and sealants)”330 is “compatible with the fuel stored to prevent releases
to the environment.”331 Specifically, implementing agencies must at least require retailers to
provide proof of compatibility through: 1) “Certification or listing of UST system equipment or
components by a nationally recognized, independent testing laboratory,” 2) written “Equipment
327 State Farm, 463 U.S. at 43.
328 DRIA 196-197.
329 Revising Underground Storage Tank Regulations—Revisions to Existing Requirements and
New Requirements for Secondary Containment and Operator Training, 80 Fed. Reg. 41,566
(July 15, 2015).
330 This generally covers everything but the above-ground fuel dispenser, which is not regulated
by the EPA. UST System Compatibility with Biofuels at 5 (July 2020), EPA Report number EPA
510-K-20-001, available in docket EPA-HQ-OAR-2021-0324. Dispensers are regulated at the
state and local levels. Id.
331 Id.
72
or component manufacturer approval,” or 3) “another option … no less protective of human
health and the environment.”332
The concern about storage compatibility is largely irrelevant because E15-compatible
tanks are often unnecessary today. As EPA itself acknowledges, most E15 is now produced by
blender pumps, which do not need to be connected to E15-compatible storage systems.333 In any
event, virtually all new tanks made today are not only compatible, but also laboratory-listed or
manufacturer-approved and therefore ready for regulatory approval.334 Focusing on the storage
systems needed for non-blender pumps (rare though they are), EPA warns that the required
compliance “documentation for … the types of materials used, and … installation dates, is often
unavailable” even if equipment is compatible.335 These concerns are misplaced and overstated.
To begin, EPA’s concerns are self-imposed by regulations that it has recognized are
unnecessarily burdensome and should change. EPA has in fact already proposed a rulemaking to
ease these compliance regulations in ways that would greatly alleviate its concerns here.336 As
EPA states, the new UST proposal “will make it easier for owners and operators to meet
compatibility requirements with their current infrastructure, if unable to demonstrate
compatibility.”337 First and foremost, this proposal would immediately secure compatibility for
an expected “24 percent” of the nation’s fuel stations by approving any USTs that “have
secondarily contained tanks and piping (including safe suction piping) and use interstitial
monitoring.”338 And the percentage is likely much higher in many states, “including those in
New England, New York, California, and Florida,” which have “required full or partial
secondary containment prior to … the Energy Policy Act of 2005 (EPAct).”339 This safe and
easy regulatory fix would alone usher in a huge expansion in the nation’s approved E85 and E15
infrastructure. But EPA has completely failed to account for this in its Regulatory Impact
Analysis for the RFS rulemaking at issue here.
Second, EPA’s own UST proposal also finds that all steel tanks, all post-2005 fiberglass
tanks, and all flexible reinforced plastic piping are E15-compatible and warrant no further
332 See EPA, Emerging Fuels and Underground Storage Tanks (USTs),
https://www.epa.gov/ust/emerging-fuels-and-underground-storage-tanks-usts#existing
(summarizing applicable regulations).
333 DRIA at 195; see 2022 Stillwater Report at 11.
334 Stillwater Associates LLC, Infrastructure Changes and Cost to Increase Consumption of E85
and E15 in 2017, at 30 (“2016 Stillwater Report”) (July 11, 2016) (attached as Ex. 6).
335 DRIA at 197.
336 E15 Fuel Dispenser Labeling and Compatibility With Underground Storage Tanks, 86 Fed.
Reg. 5,094 (Jan. 19, 2021).
337 Id. at 5,099.
338 Id. at 5,099-5,100.
339 Id.
73
demonstration requirements.340 The proposal would also ensure all newly installed UST systems
are compatible with up to 100% ethanol.341 Growth Energy supports finalization of this
proposal, which would go a long way toward removing the alleged infrastructure barriers EPA
notes in this rulemaking. Data confirm that all steel tanks and all double-walled fiberglass tanks
since 1990 are designed to store up to 100% ethanol.342 Moreover, EPA could easily go further
in exempting other safe equipment from compliance demonstration requirements. For example,
based on robust compatibility analyses conducted by Oak Ridge National Laboratory, all metal
and fiberglass UST system piping and the vast majority of flexible plastic piping is E15-
compatible.343 Growth Energy also recommends that EPA modify the existing regulations to
allow a retailer to forgo demonstration if it conducts semi-annual third-party UST inspections
and reports inspection results to its regulating agency.344
But even under the existing regulations EPA admits should change, EPA’s concerns are
overstated. The concern about whether stations with older tanks have the necessary
documentation is largely unfounded, as “EPA’s rule has created a cottage industry of consultants
willing to help the station owner meet the documentation requirements.”345 Multiple online
databases have also compiled ready-to-use compliance letters for common equipment, by
manufacturer.346 Moreover, a station can be approved as long as one tank is compatible and
documented, and most stations have at least three tanks, and many have four.347
When retrofitting is necessary, average costs are not as high as EPA projects. As even
EPA admits, “[m]any owners may already be able to demonstrate compatibility for the tanks and
piping in their UST systems. These components are often the largest expenses …. In this
situation, owners may be able to upgrade other components of their UST system with less
operational downtime and less cost because they will not need to break the concrete pad over the
340 Id.
341 Id. at 5,100-5,101.
342 National Renewable Energy Laboratory, E15 and Infrastructure (May 2015),
https://afdc.energy.gov/files/u/publication/e15_infrastructure.pdf.
343 Oak Ridge National Laboratory, Analysis of Underground Storage Tank System Materials to
Increased Leak Potential Associated with E15 Fuel, ORNL/TM-2012/182 (Jul. 2012) (attached
as Ex. 7).
344 See Growth Energy, Comments on EPA’s Proposed E15 Fuel Dispenser Labeling and
Compatibility with Underground Storage Tanks Regulations, Docket # EPA–HQ–OAR–2020–
0448 (Apr. 19, 2021) (attached as Ex. 8).
345 2016 Stillwater Report at 5.
346 See, e.g., Petroleum Equipment Institute, UST Component Compatibility Library,
https://www.pei.org/ust-component-compatibility-library (attached as Ex. 9).; Association of
State and Territorial Solid Waste Management Officials, Compatibility Tool,
https://astswmo.org/ust-compatibility-tool/ (attached as Ex. 10).
347 2016 Stillwater Report at 28.
74
UST system to replace tanks or piping.”348 Recognizing this, Stillwater has found that an
incompatible station can generally offer E85 with just $30,000 in costs: $15,000 for an E85-
capable dispenser, and $15,000 for underground infrastructure work.349 Those costs can be
reduced even further by taking advantage of the industry’s regular cycle for replacing dispensers
every seven years. By upgrading during the ordinary replacement cycle, the station’s marginal
cost of upgrading to E85 is just $20,000: $15,000 for the underground work and an incremental
$5,000 for the E85-compatible dispenser over the $10,000 for an E10 dispenser.350 Furthermore,
as EPA’s Regulatory Impact Analysis notes, government funds are also available to mitigate
these costs. Specifically, in 2020, the USDA initiated its Higher Blends Infrastructure Incentive
Program (HBIIP), which provides funds to help retail service station owners to upgrade or
replace their equipment to offer higher ethanol blends.351
VII. IN SETTING RFS STANDARDS, EPA SHOULD BACKFILL SHORTFALLS WITH ANY
OTHER AVAILABLE QUALIFYING RENEWABLE FUELS
Whenever EPA sets volume requirements, whether through a waiver, the reset, or
otherwise, the statute and principles of reasoned decisionmaking require that backfill any
renewable fuel shortfall with any other types of reasonably available qualifying renewable fuel
unless doing so could trigger a waiver or otherwise cause important and severe harm.
As the D.C. Circuit and EPA have previously recognized, “Congress intended the
Renewable Fuel Program to move the United States toward greater energy independence and to
reduce greenhouse gas emissions.”352 All renewable fuel that qualifies for compliance under the
RFS program reduces lifecycle greenhouse gas emissions by at least 20%—and usually far
more—relative to the “baseline” lifecycle GHG emissions of gasoline or diesel.353 Whenever
there is a shortfall of a given category of renewable fuel relative to the volume that Congress
expected for a given year, EPA faces a choice: it can allow fossil fuel to fill in the gap or it can
call upon obligated parties to fill the gap with other qualifying renewable fuels. As long as there
is additional renewable fuel that is reasonably available for compliance, the choice is clear: EPA
must use that renewable fuel to backfill the shortfall and thereby fulfill the objectives of the RFS
program.
348 UST System Compatibility with Biofuels at 16.
349 2016 Stillwater Report at 5-6.
350 Id.
351 DRIA at 195-196.
352 ACE, 864 F.3d at 696; see Renewable Fuel Standard Program: Standards for 2014, 2015 and
2016 and Biomass-Based Diesel Volume for 2017 (“2014-2016 Rule”), 80 Fed. Reg. 77,420,
77,421 (Dec. 14, 2015).
353 42 U.S.C. § 7545(o)(2)(A)(i).
75
EPA undoubtedly has the power to do that under both the cellulosic waiver354 and the
reset power.355 In fact, EPA’s prior position was that it would use that power to backfill
cellulosic shortfalls with available renewable fuel. In setting the 2016 standards, EPA said:
“[W]e do not believe that it would be consistent with the energy security and greenhouse gas
reduction goals of the statute to reduce the applicable volumes of renewable fuel set forth in the
statute absent a substantial justification for doing so. When using the cellulosic waiver authority,
we believe that there would be a substantial justification to exercise our discretion to lower
volumes of total and advanced renewable fuels in circumstances where there is inadequate
projected production or import of potentially qualifying renewable fuels, or where constraints
exist that limit the ability of those biofuels to be used for purposes specified in the Act (i.e., in
transportation fuel, heating oil or jet fuel). In particular, we believe that the cellulosic waiver
authority is appropriately used to provide adequate lead time and a sufficient ramp-up period for
non-cellulosic biofuels to be produced and constraints on their use for qualifying purposes
eliminated, so they can fill the gap presented by a shortfall in cellulosic biofuels.”356
That approach (whether under the cellulosic waiver power or the reset power) is
compelled not only by Congress’s purpose but also by EPA’s duty to act rationally.357 To satisfy
that duty, EPA must account for any “important aspect of the problem” and “articulate a
satisfactory explanation for” its action.358 Clearly, the availability of renewable fuel that would
serve the congressional goals of the RFS program is an important consideration in any standard
setting under the RFS.
EPA’s proposal for 2022 does not consider backfilling the implied non-advanced volume
(which would mean increasing it above 15 billion gallons). Instead, EPA proposes mechanically
to reduce the total standard by the same amount as the advanced.359 As explained elsewhere in
this comment, conventional ethanol reduces GHG emissions by more than 40% relative to
baseline emissions and promotes energy security and independence, and even grandfathered
biodiesel may reduce GHG emissions relative to the baseline. The is no concrete evidence of
serious harms that could outweigh these benefits; as also explained elsewhere in this comment,
for example, increasing the use of ethanol would not have significant adverse consequences for
the environment or the economy. And as explained elsewhere in this comment, additional
ethanol would be available to meet a higher total RFS requirement. In sum, the analysis is clear:
EPA is obligated, consistent with Congress’s purpose in creating the RFS program, to backfill
354 See 42 U.S.C. § 7545(o)(7)(D)(i) (allowing cellulosic waiver by “lesser” amount); Alon
Refining Krotz Springs, Inc. v. EPA, 936 F.3d 628, 663 (D.C. Cir. 2019) (“the discretionary
waiver provision necessarily empowers EPA to depart upward from the statutory level of noncellulosic advanced biofuel for a given year”).
355 42 U.S.C. § 7545(o)(2)(B)(ii)-(iv).
356 2014-2016 Rule at 77,434.
357 Monroe Energy, 750 F.3d at 916.
358State Farm, 463 U.S. at 43.
359 NPRM at 72,451.
76
the projected 2022 cellulosic shortfall with available conventional ethanol (and other renewable
fuel that could meet the implied non-advanced requirement).
Certainly, nothing in the statute requires EPA to apply the same cellulosic waiver to the
total volume requirement that it applies to the advanced biofuels requirement. The statute says
that if EPA reduces the cellulosic standard, it “may also reduce the applicable volume of
renewable fuel and advanced biofuels requirement established under paragraph (2)(B) by the
same or a lesser volume.”360 In the past, EPA has stressed the word “and,” and asserted that the
statutorily implied non-advanced volume of 15 bil gal is a hard cap. That misreads the statute.
The total volume requirement could be reduced by a lesser amount than the cellulosic standard
“and” the advanced volume requirement could be reduced by a lesser amount than the cellulosic
standard, even if those two reductions are different. And nothing in the text of the statute says
that the implied volume cannot exceed 15 bil gal after the application of waivers. On the
contrary, for the reasons just discussed, congressional intent and statutory structure require the
opposite. Indeed, the statute directs EPA to “ensure” that “at least” the specified amount of each
category of renewable fuel is used.361 Using the cellulosic waiver to reduce the total requirement
by less than the advanced requirement, to backfill a cellulosic shortage with available
conventional ethanol, accords with that directive.
VIII. EPA MUST INCLUDE CARRYOVER CELLULOSIC RINS IN THE AVAILABLE VOLUME
WHEN REDUCING THE CELLULOSIC VOLUME REQUIREMENT
Regardless of how EPA otherwise manages the RIN bank, it must at least count carryover
cellulosic RINs toward the “projected volume” of cellulosic fuel “available during [each]
calendar year” for purposes of determining the extent to which it exercises its cellulosic waiver
power.362 And although the reset provision does not contain the same language, because the
reset is, as explained above, an advance multi-year waiver, it is subject to the same constraint.
The cellulosic waiver provision’s term “the projected volume available during [a]
calendar year” means all cellulosic volume obligated parties may use to comply with their RFS
obligations.363 That includes carryover RINs. Indeed, the D.C. Circuit has said that carryover
RINs are “available for compliance.”364 This accords with the basic concept of the waiver: it
“authorizes EPA to ease the … Program’s requirements when complying with those requirements
would be infeasible,” and it is feasible to meet the requirements to the extent there is production
plus carryover RINs.365
360 42 U.S.C. § 7545(o)(7)(D)(i).
361 Id. § 7545(o)(2)(A)(i).
362 Id. § 7545(o)(7)(D)(i).
363 Id.
364 Monroe Energy, 750 F.3d at 918 (emphasis added); see American Fuel & Petrochemical
Manufacturers v. EPA, 937 F.3d 559, 582 (D.C. Cir. 2019) (similar).
365 ACE, 864 F.3d at 708 (emphasis added).
77
Comparison with earlier language in the same provision reinforces this conclusion. The
waiver is triggered if “the projected volume of cellulosic biofuel production is less than” the
statutory volume.366 That phrase is not the reference of the later “projected volume available.”
“Where Congress uses certain language in one part of a statute and different language in another,
it is generally presumed that Congress acts intentionally”367—“especially” where different words
“are used in the same sentence.”368 This presumption is appropriate here because Congress
easily could have re-used the phrase “of cellulosic biofuel production” instead of “available”:
“For any calendar year for which the projected volume of cellulosic biofuel production is less
than the minimum applicable volume …, the Administrator shall reduce the applicable volume of
cellulosic biofuel … to the projected volume of cellulosic biofuel production during that calendar
year.” Or Congress could have written: “… shall reduce … to that projected volume during that
calendar year.” Either of those approaches would be a more natural and clear way to direct EPA
to set the cellulosic standard to the level of projected production, without regard to available
carryover RINs.
Including carryover RINs in “the projected volume available” also furthers Congress’s
intent that the RFS program serve as a market forcing policy to increase the use of renewable
fuels. Excluding carryover RINs from the “projected volume available” inflates the supply of
RINs, depressing RIN prices and discouraging the very investment Congress intended to
incentivize. EPA’s proposal even recognizes this dynamic: “despite the continued rapid growth
in cellulosic biofuel volumes, excess carryover cellulosic RINs in 2018 and 2019 resulted in low
cellulosic RIN prices, which in turn may have negatively affected investment in cellulosic
biofuel production.”369
ACE is not to the contrary. ACE held that EPA need not consider carryover RINs when
determining whether there is “inadequate domestic supply” for purposes of the general waiver
under § 7545(o)(3)(B)(ii). That holding is irrelevant here because the cellulosic waiver provision
differs from the general waiver provision in two important ways. First, the relevant text here is
“the projected volume available.” This differs from “inadequate domestic supply” especially in
its use of “available” as a modifier of “volume” rather than “production” or “supply.” In fact,
the phrase “the projected volume available” has no analogue in the general waiver. The phrase
“inadequate domestic supply” in the general waiver defines the trigger for the waiver, and its
analogue in the cellulosic waiver is “the projected volume of cellulosic biofuel production,” a
term that admittedly does not encompass carryover RINs. In contrast to the cellulosic waiver,
the general waiver contains no language instructing EPA about the level to which EPA must
reduce the applicable volume.
Second, ACE credited EPA’s explanation that, “were [EPA] to consider carryover RINs
as a supply source …, the number of carryover RINs would be reduced to almost zero,”
366 42 U.S.C. § 7545(o)(7)(D)(i).
367 National Fed’n of Indep. Bus. v. Sebelius, 567 U.S. 519, 544 (2012).
368 Northeast Hosp. Corp. v. Sebelius, 657 F.3d 1, 12 (D.C. Cir. 2011) (quotation marks omitted).
369 NPRM at 72,455-72,456.
78
eliminating critical “flexibility and liquidity provided by carryover RINs.”370 That rationale—
though dubious for reasons discussed above—does not apply to cellulosic RINs because the
Clean Air Act also directs EPA to enable obligated parties to satisfy their cellulosic volume
obligations by purchasing cellulosic waiver credits, whenever the cellulosic waiver is
triggered.371 These credits provide the “market liquidity[,] transparency,” and “certainty” that
EPA otherwise relies on RINs to provide.372 EPA nonetheless insists that “the benefits of
carryover RINs … also apply to cellulosic carryover RINs,” without any explanation of how that
could be so given the availability and role of the waiver credits.373 That kind of “ipse dixit
conclusion … epitomizes arbitrary and capricious decisionmaking.”374
Finally, although including carryover cellulosic RINs in “the projected volume available”
when exercising the cellulosic waiver power would represent a change in EPA policy, that is
permissible. “Agencies are free to change their existing policies as long as they provide a
reasoned explanation for the change.”375 EPA could easily satisfy this requirement by adopting
the explanation offered here for the change in policy.
IX. THE PROPOSED RFS STANDARDS WOULD NOT APPRECIABLY RAISE RETAIL PRICES
FOR FOOD OR GASOLINE
As EPA observes, its proposed volume requirements would have negligible effects on
food and retail gas prices.
With respect to food prices, EPA “estimate[s] that the proposed volumes would have
minimal impacts … (increases of total food expenditures of 0.15% and 0.40% in 2021 and 2022
respectively).”376 EPA is incorrect with respect to 2021; its proposal cannot affect food prices in
2021 at all because 2021 is already over. As for 2022, EPA’s assessment is substantially
overstated, but in any event, there is no reason to expect that the effect could be greater than
EPA’s assessment. As Stillwater explains, the amount of corn needed to produce the volume of
ethanol EPA assumes—or even to meet the full 15-billion gallon implied non-advanced
volume—would not increase the demand for corn relative to recent prior years and therefore
would not be expected to raise corn prices at all. Moreover, as EPA notes, “corn and soy are a
relatively small proportion of most foods purchased and consumed in the United States,”377 and
370 864 F.3d at 715.
371 42 U.S.C. § 7545(o)(7)(D)(ii)-(iii).
372 Id. § 7545(o)(7)(D)(iii).
373 NPRM at 72,456.
374 American Pub. Commc’ns Council v. FCC, 215 F.3d 51, 53 (D.C. Cir. 2000) (citation
omitted).
375 Encino Motorcars, LLC v. Navarro, 579 U.S. 211, 221 (2016).
376 DRIA at 225.
377 DRIA at 225.
79
only a fraction of corn and soybean are used for renewable fuel. So, any effect of the RFS
standards on food prices will be limited to the margin.
With respect to retail gas prices, EPA “project[s] relatively small price impacts of the
proposed volumes …, with slight decreases in the price of gasoline (–0.11 and –0.02 cents per
gallon in 2021 and 2022 respectively) and increases in the price of diesel (0.70 and 3.22 cents
per gallon in 2021 and 2022 respectively).”378 Again, EPA’s proposal cannot affect gas prices in
2021 at all. As for 2022, EPA’s expectation that its proposal would slightly decrease gasoline
prices is sound and well-supported by its evidence.
X. EPA’S PROPOSED RESPONSE TO ACE REMAND IS NECESSARY AND APPROPRIATE
Growth Energy applauds EPA’s proposed remedy for the unlawful general waiver of the
2016 standards on remand from the D.C. Circuit in ACE.
379 In setting the percentage standards
for 2016, EPA invoked its “inadequate domestic supply” general-waiver power to reduce the
total volume by 500 million gallons.380 But the D.C. Circuit invalidated that action because EPA
impermissibly considered “demand-side constraints that affect the consumption of renewable
fuel by consumers.”381 Accordingly, the court “vacate[d] EPA’s decision to reduce the total
renewable fuel volume requirements for 2016 through use of its ‘inadequate domestic supply’
waiver authority, and remand[ed] the rule to EPA for further consideration in light of [its]
decision.”382 EPA thus had an unlawful 500-million-gallon deficit, which it has now proposed to
cure through 250-million-gallon supplemental requirements in 2022 and 2023. Growth Energy
stresses, however, that the proposed 250-million-gallon supplemental requirement would fulfill
EPA’s remedial duty only if EPA also finalizes its promised 250-million-gallon supplemental
requirement for 2023.
EPA has a duty to adjust future standards—as it has proposed to do—to remedy the
unlawful 2016 waiver. ACE held that EPA lacked statutory authority to waive the 500 million
gallons, and EPA is “without power to do anything which is contrary to either the letter or spirit
of the mandate construed in the light of the opinion” rendered in ACE.
383 Pursuant to the
mandate and EPA’s general duty to “ensure” that the statutory volume requirements are met,
378 Id.
379 864 F.3d 691 (D.C. Cir. 2017).
380 Renewable Fuel Standard Program: Standards for 2014, 2015, and 2016 and Biomass-Based
Diesel Volume for 2017 (“2016 Rule”), 80 Fed. Reg. 77,420, 77,439 (Dec. 14, 2015); ACE, 864
F.3d at 701-702.
381 ACE, 864 F.3d at 696.
382 Id. at 696-697.
383 City of Cleveland v. Federal Power Comm’n, 561 F.2d 344, 346 (D.C. Cir. 1977) (quotation
marks omitted); accord U.S. Postal Serv. v. Postal Regulatory Comm’n, 747 F.3d 906, 910 (D.C.
Cir. 2014).
80
EPA must cure its adjudicated legal error.384
EPA also has the power to remedy its prior error by setting future supplemental
requirements standards, as it has proposed. Congress directed EPA to “ensure” that the
transportation fuel sold in the United States “contains at least” the statutorily specified amount of
renewable fuel.385 Invoking this power, EPA on two prior occasions made up a prior year’s
requirements by adding it to, or supplementing, a future year’s requirements.386 EPA can use
this power again to impose the proposed remedial supplemental obligations.
EPA’s proposal raises no issues of retroactive rulemaking. The proposed supplemental
obligations are not retroactive. Instead of “attach[ing] new legal consequences to events
completed before its enactment,”387 the supplemental obligations would apply only to future
conduct by obligated parties, namely, to their activity in 2022 (and 2023). Moreover, whether
viewed through the retroactive lens or not, the supplemental obligation would be reasonable and
therefore permissible because it would not unsettle legitimate “expectation[s].”388 Obligated
parties were always legally bound to meet the 2016 statutory volume requirement except to the
extent EPA validly waived it; obligated parties could not have had settled “expectation[s]” in an
ultra vires waiver.389 Further, by proposing a future obligation, the proposal affords obligated
parties ample notice and opportunity to plan their future activity to achieve compliance. And
even if that were not sufficient to render the proposal reasonable, obligated parties could rely on
carryover RINs to satisfy the supplemental obligation, as EPA notes.390
384 See WildEarth Guardians v. EPA, 830 F.3d 529, 535 (D.C. Cir. 2016) (“The necessary
consequence of vacating the Implementation Rule on the ground that it failed adequately to
adhere to Subpart 4 would be some kind of corrective EPA action strictly implementing that
Subpart ….”); Multicultural Media, Telecom & Internet Council v. FCC, 873 F.3d 932, 936
(D.C. Cir. 2017) (A “decision that the agency’s action was substantively unreasonable generally
means that, on remand, the agency must exercise its discretion differently and reach a different
bottom-line decision”); Chicago & S. Air Lines, Inc. v. Waterman S.S. Corp., 333 U.S. 103, 113
(1948) (“Judgments, within the powers vested in courts by the Judiciary Article of the
Constitution, may not lawfully be revised, overturned or refused faith and credit by another
Department of Government.”).
385 42 U.S.C. § 7545(o)(2)(A)(i); see also ACE, 834 F.3d at 698-699 (quoting 42 U.S.C. §
7545(o)(3)(B)(i)) (emphasis added) (quotation marks omitted).
386 NPRA, 630 F.3d at 156-157, 163; Monroe Energy, 750 F.3d at 919-921.
387 Landgraf v. USI Film Prods., 511 U.S. 244, 269-270 (1994); NPRA, 630 F.3d at 159.
388 Monroe Energy, 750 F.3d at 920.
389 Id.
390 NPRM at 72,455-72,456.
81
XI. EPA SHOULD RETAIN THE STANDARD EQUATION AS REVISED IN THE 2020 RULE
As Growth Energy explains in a separate comment on EPA’s proposed denial of all
pending SRE applications,391 EPA should deny all such applications and, for the same reasons,
should reverse and deny all SREs that were remanded to EPA by the D.C. Circuit in Sinclair
Wyoming Refining Co. v. EPA.
392
Regardless of whether EPA denies the pending and remanded SRE applications, it should
retain the standard equation as revised in the 2020 rule, so that the equation adjusts for projected
SREs. The revised equation does not prescribe a particular prediction method; it may be
appropriate for EPA to update its method in light of new standards or empirical data affecting the
likelihood of granting SREs for a given year (such as the standards and data underlying the
proposed denial of all pending SREs). Maintaining the standard equation as revised in the 2020
rule would thus enable EPA to set standards that realistically reflect the obligations needed to
achieve full compliance with the applicable volume requirements, especially should EPA change
course and decide to grant SREs in some future year.
In fact, as Growth Energy has explained before, EPA must adjust the percentage
standards to account for SREs. First, Congress designed the RFS program “to force the market
to create ways to produce and use greater and greater volumes of renewable fuel each year.”393
Thus, in setting annual volume requirements EPA has a “statutory mandate to ‘ensure[]’ that …
[volume] requirements are met,”394 as well as a statutory mandate to promulgate general rules for
the RFS program that “ensure that transportation fuel sold or introduced into commerce in the
United States … contains at least the applicable volume of renewable fuel.”395 If EPA does not
“adjust renewable fuel obligations to account for exemptions,” it creates a “renewable-fuel
shortfall,” “imped[ing] attainment of overall applicable volumes.”396 Or, as EPA put it in the
2020 rule, “any SREs granted after we issued the annual rule containing the percentage standards
for that year effectively reduced the required volume of renewable fuel for that year.”397 Thus,
as EPA recognized in modifying the standard equation in the 2020 rule, raising the percentage
standards to account for SREs has “the effect of ensuring that the required volumes of renewable
fuel are met when small refineries are granted exemptions”398; refusing to make this adjustment
391 See EPA, EPA Proposes to Deny All Pending RFS Small Refinery Exemption Petitions (Dec.
2021), https://www.epa.gov/sites/default/files/2021-12/documents/420f21065.pdf.
392 Order, No. 19-1196, ECF No. 1925942 (D.C. Cir. Dec. 8, 2021).
393 ACE, 864 F.3d at 710.
394 Id. at 698-699 (quoting 42 U.S.C. § 7545(o)(3)(B)(i)).
395 42 U.S.C. § 7545(o)(2)(A)(i); see also id. § 7545(o)(2)(A)(iii)(I).
396 American Fuel, 937 F.3d at 571, 588.
397 Renewable Fuel Standard Program: Standards to 2020 and Biomass-Based Diesel Volume for
2021 and Other Changes: Final Rule, 85 Fed. Reg. 7016, 7050 (Feb. 6, 2020).
398 Id.
82
violates EPA’s duty to set percentage standards that “ensure” that the volume requirements are
met.
Second, in setting the standards, EPA must “consider [all] important aspect[s] of the
problem” and “examine the relevant data and articulate a satisfactory explanation for its action
including a rational connection between the facts found and the choice made.”399 Because SREs
have the effect of lowering the required volumes and preventing the required volumes from
being met, how to account for SREs is an integral part of the problem EPA faces whenever it sets
percentage standards; EPA cannot set standards by blinding itself to them.
Third, refusing to account for retroactive extensions impermissibly converts its
exemption power into a waiver power, in contradiction of the statute’s plain text and structure.
In several provisions of the statute, Congress explicitly granted EPA the power to reduce the
required nationwide volumes, and labeled those powers “waivers.”400 These “waiver” powers
may be exercised “only in limited circumstances,” namely, the circumstances specified in the
statute.401 In contrast, the provisions allowing EPA to exempt small refineries contain neither of
those features: they do not say that EPA may reduce the nationwide volume requirements or use
the label “waiver”; rather, they are labeled “exemption,” and they authorize EPA to determine
merely that the compliance obligation “shall not apply to” the specific applicant refinery because
of special circumstances relating to that refinery.402 There is no reason to depart from “the usual
rule that when the legislature uses certain language in one part of the statute and different
language in another, [courts and agencies must] assume[] different meanings were intended.”403
Refusing to account for retroactive SREs in effect converts SREs into waivers of the nationwide
volume requirements. That is pernicious because it would expand EPA’s waiver power to
situations that would not meet the statutorily specified triggers for a waiver. As EPA has
acknowledged, “small refinery exemptions are held to a different standard than a waiver,”
including a waiver for “severe economic harm.”404 There is no reason to think “Congress would
have established the severe-harm waiver standard ‘only to allow waiver’” under the small
refinery exemption provision “based on lesser degrees of economic harm.”405 If Congress had
intended to grant EPA a power to waive nationwide volume requirements based on findings that
individual refineries will suffer “disproportionate economic hardship” if they must comply, it
399 State Farm, 463 U.S. at 43 (quotation marks omitted).
400 See 42 U.S.C. § 7545(o)(7)(A) & (D)-(E), (8)(D).
401 NPRA, 630 F.3d at 149 (emphasis added).
402 42 U.S.C. § 7545(o)(9).
403 United States v. Monzel, 641 F.3d 528, 533 (D.C. Cir. 2011).
404 Renewable Fuel Standard Program-Standards for 2019 and Biomass-Based Diesel Volume
for 2020: Response to Comments 19, EPA (Nov. 2018),
https://nepis.epa.gov/Exe/ZyPDF.cgi?Dockey=P100VU6V.pdf.
405 ACE, 864 F.3d at 712.
83
would have said so—it knew how to. EPA has no authority to rewrite the statute or create a new,
non-textual waiver power.406
Contrary to the views of some commentors, the fact that Congress, in 42 U.S.C.
§ 7545(o)(3)(C)(ii), directed EPA to account for the behavior of small refineries in one particular
respect does not mean that Congress intended implicitly to forbid EPA from taking account of
small refineries in any other way. The expressio unius canon does not “mean that anything not
required is forbidden.”407 Rather, the question is whether Congress’s silence on a specific issue
is so closely related to issues it addressed explicitly that it is appropriate to infer from the
different treatment that Congress intended to withhold authority from EPA on the silent issue.408
No such inference is warranted here. As EPA recognized during the 2020 rulemaking,
section 7545(o)(3)(C)(ii) and the Supplemental 2020 NPRM represent two different solutions to
two distinct and opposite problems: § 7545(o)(3)(C)(ii) tells EPA what to do when a refinery that
was exempt during the prior compliance year nonetheless used renewable fuel that year, i.e., it
directs EPA to account for potential overcompliance.
409 In contrast, the revised standard
equation addresses the problem of near certain undercompliance. It is not surprising that
Congress saw fit to address the specific situation where renewable fuel was used by exempt
refineries (a problem that EPA’s RIN system quickly rendered a nonissue),410 while at the same
time failing to mention the (now) much more pressing problem of unaccounted for exempt fossil
fuels. Indeed, Congress likely did not consider the problem of retroactive exemptions at all,
given its expectation that SREs would exist only on a “[t]emporary” basis, would be
“extend[ed]” only upon a showing of “disproportionate economic hardship,”411 and would
accordingly fade away within a few years of the program’s start. In fact, it would be nonsensical
406 See, e.g., In re Sealed Case, 237 F.3d 657, 670 (D.C. Cir. 2001) (“Agencies are not
empowered to carve out exceptions to statutory limits on their authority.”).
407 2A Sutherland, Statutes and Statutory Construction § 47:25 (7th ed. updated Nov. 2021).
408 See, e.g., Barnhart v. Peabody Coal Co., 537 U.S. 149, 168 (2003) (“[T]he canon expressio
unius est exclusio alterius does not apply to every statutory listing or grouping; it has force only
when the items expressed are members of an ‘associated group or series,’ justifying the inference
that items not mentioned were excluded by deliberate choice, not inadvertence.” (quoting United
States v. Vonn, 535 U.S. 55, 65 (2002)))
409 See Renewable Fuel Standard Program-Standards for 2020 and Biomass-Based Diesel
Volume for 2021 and Other changes: Response to Comments 167-168 (“2020 Response”), EPA
(Dec. 2019), https://nepis.epa.gov/Exe/ZyPDF.cgi?Dockey=P100YAPQ.pdf; 42 U.S.C. §
7545(o)(3)(C)(ii) (“In determining the applicable percentage for a calendar year, the
Administrator shall make adjustments … to account for the use of renewable fuel during the
previous calendar year by small refineries that exempt under paragraph (9).”).
410 See Regulation of Fuels and Fuel Additives: Changes to Renewable Fuel Standard Program:
Notice, 74 Fed. Reg. 24,904, 24,954 (May 26, 2009) (explaining that the volume of renewable
fuel generated by exempt refineries is “expected to be very small” and that EPA’s RIN-credit
trading system accounts for such volumes in any event).
411 42 U.S.C. § 7545(o)(9).
84
to infer that, by addressing overcompliance, Congress intended to bar EPA from addressing
undercompliance given the statutory command that EPA “ensure” that the standards are met and
the overarching objective to compel increased annual use of renewable fuel.412
Moreover, the expressio unius argument proves far too much. It would mean that any
adjustment of percentage standards to account for exempt volumes would violate the statute,
even if the exemptions had already been granted at the time of EPA’s rulemaking. If that were
true, EPA’s percentage standard formula would have been invalid from the start,413 and EPA
would be powerless to extend SREs in appropriate cases while still “ensuring” the target volumes
are met. This result flouts the statutory structure, as Growth Energy has already explained at
length. Whatever expressio unius inference may exist is far too weak to justify such a
disharmonious result.
Finally, some commentors have argued that the revised standard equation violates the
statutory requirement to avoid “redundant obligations.”414 It does not. That statutory section
merely provides that EPA should not require an obligated party to “satisfy the RFS standards
more than once for the same volume of gasoline and diesel.”415 For instance, redundancy could
arise if an obligated party that functioned as both a refinery and an importer were required to
meet its obligation twice for the same volume of fuel it introduced into commerce. The revised
standard equation requires no such result. It merely calls for the percentage obligation that
applies to all obligated parties to be adjusted in light of the exempt volume of transportation, and
still requires each (non-exempt) obligated party to meet that obligation only once.
XII. ROBUST RFS REQUIREMENTS PROMOTE RURAL ECONOMIC HEALTH
EPA is correct that “[h]igher volumes of … renewable fuel could result in more domestic
jobs in the biofuels industry” and “rural economic development.”416 Moreover, maintaining prepandemic levels of demand for renewable fuel is imperative for protecting the economic health
of the renewable fuel industry and the jobs of those who work in the industry or adjacent
industries. The ethanol industry supported nearly 305,000 jobs, created $19 billion in household
income, and contributed over $34 billion in GDP in 2020.417 And that was 19% below 2019’s
GDP contribution due to the pandemic.418 These contributions to rural economies could grow
significantly if appropriate management of the RFS transitions the nationwide market to E15.
412 ACE, 864 F.3d at 710.
413 See 40 C.F.R. § 80.1405 (requiring EPA at minimum to account for nonretroactive SREs
when setting percentage standards).
414 42 U.S.C. § 7545(o)(3)(C)(i).
415 See 2020 Response at 182.
416 NPRM 72,447, 72,451.
417 Renewable Fuels Association, Contribution of the Ethanol Industry to the Economy of the
United States in 2020 10 (Feb. 2, 2021) (attached as Ex. 12).
418 Id. at 8-9.
85
Specifically, transitioning to E15 would create more than 180,000 new jobs, spurring $10.5
billion in household income and $17.8 billion in additional GDP.419
XIII. EPA IS OBLIGATED TO ADJUST THE 2022 RFS STANDARDS TO MAKE UP FOR PAST
RETROACTIVE SRES
EPA has previously refused to adjust standards to account for the retroactive exemptions
granted for prior years. In its current proposal, EPA again omits any adjustment for the billions
of RINs lost to prior retroactive exemptions. Without such an adjustment, EPA violates its
statutory duties to set RFS standards that will “ensure” that the volume requirements are met and
to engage in reasoned decisionmaking. EPA’s refusal reflects a failure to address an important
issue—how the standards would be affected by past retroactive exemptions—or to rationally
connect the standards to the evidence before it. EPA’s refusal also converts the exemptions into
a waiver, contrary to the statute. So voluminous are the past retroactive exemptions that EPA’s
refusal to account for them undermines Congress’s intent that the RFS standards force the market
to use increasing amounts of renewable fuel annually.
A. EPA’s Refusal Violates Its Statutory Duty to Set Standards That “Ensure”
That the Required Volumes Are Met
“After EPA determines the volume requirements for the various categories of renewable
fuel” by considering whether any statutory waivers are appropriate, “it has a ‘statutory mandate’
to ‘ensure[]’ that those requirements are met” by setting percentage standards that will achieve
those volumes.420 EPA violated this duty by refusing to adjust the 2020 standards to account for
past retroactive exemptions.
If EPA does not “adjust renewable fuel obligations to account for exemptions,” it creates
a “renewable-fuel shortfall,” “imped[ing] attainment of overall applicable volumes.”421
Recognizing that fact, EPA “raises the percentage standard” for a given year to account for the
exemptions “that were granted … before [it] established the percentage standard for that year.”422
That solution, however, is “only partial” because it does “not … account for small refinery
exemptions granted after [EPA] promulgates percentage standards for that year—so-called
retroactive exemptions.”423
In the 2020 Rule, EPA finally recognized that to fulfill its “ensure” duty, it must also
adjust the standards to account for retroactive exemptions—but it did so only with respect to the
retroactive exemptions it projected it would grant for 2020.424 EPA correctly explained that
419 ABF Economics, Economic Impact of Nationwide E15 Use 1-2 (June 10, 2021) (attached as
Ex. 13).
420 ACE, 864 F.3d at 698-699 (quoting §7545(o)(3)(B)(i)).
421 American Fuel, 937 F.3d at 571, 588.
422 Id. at 588.
423 Id.
424 85 Fed. Reg. at 7,049.
86
“should [it] grant [exemptions] without accounting for them in the percentage formula, those
exemptions would effectively reduce the volumes of renewable fuel required by the RFS
program, potentially impacting renewable fuel use in the U.S.”425 Raising the standards to
account for projected retroactive exemptions, EPA declared, has “the effect of ensuring that the
required volumes of renewable fuel are met.”426
That is only partially true because granting retroactive exemptions in past years without
accounting for them in any annual volume standards also—as EPA acknowledged—“effectively
reduced the required volume of renewable fuel for th[ose] year[s].”427 Thus, the sound premises
of EPA’s analysis implied more remediation: they required EPA to also adjust the 2020
standards to account for past retroactive exemptions. EPA’s refusal to do that violated its
“ensure” duty.
EPA is not relieved of that duty just because the compliance years for which those prior
retroactive exemptions were granted are past; their depressive effect on the RFS program’s
volume requirements continue today.
Because EPA did not account for past retroactive exemptions granted when it set the
standards for those years, those exemptions freed up RINs corresponding to the exemption
volumes for compliance in a future year. By “effectively reduc[ing]” the volume of renewable
fuel that obligated parties were required to use in the covered years, the past retroactive
exemptions provided obligated parties with a RIN windfall—RINs that should have been needed
to meet the required volumes but, because of the retroactive exemptions, were not. Using the
mechanism of the RIN bank, obligated parties carried their RIN windfall forward for compliance
in a future year, in effect transferring the renewable-fuel shortfall caused by the retroactive
exemptions to a later year. By disregarding the shortfall from past retroactive exemptions that
was embedded in the RIN bank, EPA sets standards that cannot “ensure” that the market would
use the required amount of renewable fuel in 2022.
This conclusion follows from how carryover (or banked) RINs and RFS compliance
work. As EPA recognized when it set the 2020 standards (and many times before), and as it
recognized again in this rulemaking, obligated parties will necessarily use all available carryover
RINs to comply with their RFS obligations because any unused carryover RINs would expire
and become worthless.428 To meet their RFS obligations, obligated parties first apply their
carryover RINs to their fullest extent and then use renewable fuel produced in the compliance
year until they meet their volume obligation. Thus, the effective volume requirement—the
amount of renewable fuel that the standards actually require obligated parties to use—is the
nominal volume requirement minus the available carryover RINs, i.e., minus the RIN bank. If
425 Id. at 7,050.
426 Id.
427 Id.
428 DRIA at 43; 85 Fed. Reg. at 7,021 n.15; see also 77 Fed. Reg. 70,759, 70,775 (“[T]he
availability of rollover RINs can significantly affect the potential impact of implementation of
the RFS volume requirements.”).
87
and to the extent that obligated parties use more than the effective volume requirement, that is
only because of purely voluntary decisions, not because the standards actually required them to
do so—and the RINs from such additional use are banked for the next year.429 That the bank
appears to maintain a “balance” from one year to the next thus is a fiction, concealing the reality
that the bank is regenerated—often enabled by the carryover RINs from the prior year. See id.
So, to the extent the RIN bank contains RINs because of unaccounted-for past retroactive
exemptions, the standards will not force the market to use the required renewable fuel. And
accordingly, to the extent EPA sets the 2022 standards without accounting for such RINs, EPA
sets standards that cannot “ensure” that the required volumes of renewable fuel would be met.430
B. EPA’s Refusal Is Arbitrary and Capricious
EPA’s refusal to account for past retroactive exemptions in setting RFS standards is
arbitrary and capricious, rather than the product of reasoned decisionmaking. In setting the
standards, EPA must “consider [all] important aspect[s] of the problem” and “examine the
relevant data and articulate a satisfactory explanation for its action including a rational
connection between the facts found and the choice made.”431 EPA’s refusal does not do so.
As explained above, when EPA sets standards, EPA knows it has a duty to set standards
that will ensure that the required volumes of renewable fuel will be used, and it knows—or at
least should know given its experience with the RFS program and the data it has regarding the
volume of past retroactive exemptions and the size of the RIN bank—that if it does not adjust the
standards to account for those past retroactive exemptions, the standards will not ensure that the
required volume of renewable fuel are used and the inflated supply of carryover RINs will create
disincentives for new production. EPA thus blinds itself to a critical problem with the standards
it is setting, and sets standards that cannot rationally be justified by the evidence before it. EPA
in fact is proposing to set standards it knows will not achieve their goal, even if all of its
projections for 2022 prove 100% accurate. That is arbitrary and capricious.
Moreover, EPA has the tools to make the necessary adjustment. As discussed above,
EPA has at least two available options for accounting for past retroactive exemptions, both of
which it has used before. First, EPA could apply a lesser discretionary cellulosic waiver to the
advanced and total volume requirements. Second, EPA could increase the nominal 2022 volume
requirements or impose a supplemental requirement, much as it did with the 2009 and 2010
standards.
C. EPA’s Refusal Impermissibly Creates for Itself a Non-textual Waiver Power
By refusing to adjust the standards to account for past retroactive exemptions, EPA also
impermissibly converts the exemptions into waivers, contrary to the statute’s text. As discussed,
the effect of the unaccounted-for retroactive exemptions is to reduce the nationwide volume
429 85 Fed. Reg. at 7021 n.15.
430 See National Petrochemical, 630 F.3d at 153 (“ensure” means “make sure, [or] certain”).
431 State Farm, 463 U.S. at 43 (quotation marks omitted).
88
requirement, but Congress authorized EPA to reduce that requirement only through a duly issued
“waiver.”
Congress expressly granted EPA the power to “reduce” nationwide volume requirements,
but labeled those powers “waiver,”432 and permitted EPA to use them “only” in the “limited
circumstances” specified in the statute,433 such as where “implementation of the [statutory
volume] requirement would severely harm the economy … of a State, a region, or the United
States.”434 In contrast, the “exemption” provision—as Congress labeled it—does not say that
EPA may “reduce” the volume requirements, but rather authorizes EPA to determine merely that
the compliance obligation “shall not apply to” a specific refinery because of a different special
circumstance, namely, that compliance would cause the refinery “disproportionate economic
hardship.”435
“[T]he usual rule [is] that when the legislature uses certain language in one part of the
statute and different language in another”—here, exemption rather than waiver—courts and
agencies must “assume[] different meanings were intended,”436 and there is no reason to depart
from that rule here. Indeed, as EPA has acknowledged, exemption petitions “are held to a
different standard”—“economic hardship”—“than a waiver under severe economic harm.”437
Congress would not have “established the severe-harm waiver standard only to allow waiver”
under the small-refinery exemption provision “based on lesser degrees of economic harm.”438
EPA has no authority to rewrite the statute to convert its “exemption” power into a new “waiver”
power.439
XIV. BIOINTERMEDIATES
Growth Energy appreciates EPA’s efforts to advance regulatory clarity regarding the use
of biointermediates in the production of renewable fuels. Since the 2016 proposed “Renewables
Enhancement and Growth Support Rule” (REGS proposal), it has become even more important
for renewable fuel producers to have the flexibility to use biointermediates in fuel production in
order to lower costs and drive innovation. EPA should ensure that the final biointermediates
regulations facilitate use of biointermediates for production of specific advanced renewable
fuels such as SAF, while also ensuring sufficient flexibility as renewable fuel producers
continue to innovate to produce low carbon fuels using various feedstocks and biointermediate
products. Though we appreciate the importance of preserving the integrity of the RFS program,
432 42 U.S.C. § 7545(o)(7), (8)(D).
433 National Petrochemical, 630 F.3d at 149.
434 42 U.S.C. § 7545(o)(7)(A).
435 42 U.S.C. § 7545(o)(9)(A)(i), (B)(i).
436 United States v. Monzel, 641 F.3d 528, 533 (D.C. Cir. 2011).
437 EPA, 2020 RFS Response to Comments at 14.
438 ACE, 864 F.3d at 712 (quotation marks omitted); see also 42 U.S.C. § 7545(o)(7)(A).
439 See, e.g., Mingo Logan Coal Co. v. EPA, 829 F.3d 710, 721 (D.C. Cir. 2016); Ethyl Corp. v.
EPA, 51 F.3d 1053, 1061 (D.C. Cir. 1995).
89
we believe modifications of the proposed regulations are warranted to ensure the framework is
not unduly burdensome on potential biointermediates and renewable fuel producers without
compromising program integrity. Below we briefly address our primary concerns with the
proposed framework.
Definition of “Biointermediate”: In order to preserve flexibility, we support
EPA reverting back to its approach in the REGS proposal of defining types of
products that constitute biointermediates while excluding other substances that
only undergo form changes. We are concerned that if EPA must undertake an
entirely new rulemaking any time renewable fuel producers wish to utilize a
different type of biointermediate in fuel production, it will be impose
unnecessary regulatory burdens on the Agency and create substantial time delays
for parties wishing to make investments in development and introduction of
innovative renewable fuels to the market. Rather than specifying particular
biointermediates and requiring rulemaking to add more, we suggest that EPA
has adequate flexibility through the registration process to impose conditions on
types of biointermediates that it views as posing greater compliance oversight
concerns. Specifically, consistent with EPA’s practice of including conditions
on approval of a petition submitted pursuant to the Efficient Producer process, it
could require as part of the registration approval process particular conditions
applicable both to the biointermediate producer and the renewable fuel producer
to address any such concerns.
In addition, to the extent the Agency is concerned that the previously-proposed
definition may be too broad and inadvertently encompass substances that it does
not intend to regulate as biointermediates, it could clarify these issues with
producers as questions arise (e.g., encourage the industry to inquire and then to
inform a producer it need not register as a biointermediate producer). Finally, if
EPA decides to proceed with the current specifically delineated list of
biointermediates, Growth Energy supports inclusion of the identified substances,
recommends adding cellulosic sugars (such as those derived from paper
production processes) , and recommends that EPA develop a streamlined
approval process to address new biointermediates. A model for that streamlined
process is the Efficient Producer Petition Process, which provides producers
clarity as to the information EPA needs to approve registration under a pathway,
facilitates expedited EPA review, and allows EPA to tailor conditions on
approval as necessary.
Relatedly, Growth Energy encourages EPA to confirm that corn oil produced as a
byproduct from the ethanol production process that may undergo form changes
(such as the addition of water, physical separation, and drying) prior to sale to a
biodiesel producer would not be considered a biointermediate. EPA could do so
by expressly including pre-processing of corn oil as an acceptable form change
that does not result in treatment of the corn oil as a biointermediate.
Voluntary QAP: EPA should allow participation in the Quality Assurance Plan
(QAP) process on a voluntary basis, rather than imposing an additional
90
mandatory cost and burden on biointermediates producers and renewable fuel
producers. To be sure, many parties will likely participate, but other parties may
view contractual arrangements and/or alternative oversight mechanisms as
sufficient assurances that RINs will be valid. Given the breadth of the proposed
liability provisions applicable to all regulated parties in a fuel production chain
that involves biointermediates, the proposed regulations have ample safeguards
even without requiring mandatory QAP.
Necessary Update to Ethanol Carbon Intensity: As addressed in detail above, it
is imperative that EPA update its GHG LCA for corn starch-based ethanol. The
current value, which greatly underestimates the GHG benefits of ethanol, could
adversely affect the use of undenatured ethanol as a biointermediate in advanced
fuel production. Production of other advanced fuels such as SAF using ethanol
may have its own energy requirements that impact overall lifecycle GHG
emissions of the resulting fuel. It is therefore very important that SAF and other
advanced fuels that may utilize ethanol as a biointermediate should accurately
reflect ethanol’s full GHG benefits. Failure to account for all such benefits
could improperly disqualify such fuels from appropriate treatment under the
RFS program.
We welcome the opportunity for further engagement on these issues in order to advance
EPA’s finalization of a regulatory path for use of biointermediates in renewable fuel production.
XV. EPA SHOULD APPROVE KERNEL FIBER REGISTRATIONS AND PENDING BIOFUEL
PATHWAYS
Growth Energy urges EPA to act speedily to approve the numerous pending registration
applications for simultaneous production of starch and cellulosic ethanol from corn kernel
feedstock. Through the RFS program, Congress especially sought to encourage the production
of cellulosic biofuel, which achieves the greatest reduction in GHG emissions relative to
gasoline. This Administration has underscored the importance of advanced and cellulosic
biofuels in helping the United States to achieve its ambitious and necessary GHG reduction
goals. Removal of regulatory barriers and prompt approval of the pending kernel fiber
registrations is important to encourage and reward investment in technology to convert cellulose
to ethanol.
Additionally, to further producer innovation and the production of advanced biofuels, we
urge EPA to prioritize and expedite pathways for carbon capture, utilization, and storage. We
also urge EPA to expedite approval of the pending petition from the Corn Refiners Association
(CRA) to allow biodiesel and renewable diesel facilities to utilize corn oil produced from corn
wet mills as a feedstock.
91
XVI. EPA SHOULD ADOPT THE PROPOSED APPROACH TO CONFIDENTIAL BUSINESS
INFORMATION
Growth Energy supports EPA’s proposal to not withhold basic information from smallrefinery exemption (“SRE”) petitions and decisions as confidential business information (“CBI”)
pursuant to exemption 4 of the Freedom of Information Act (“FOIA”). 440
Transparency over SRE petitions and decisions is necessary so that the public can
monitor the production of renewable fuel in compliance with the RFS program. Starting with the
2016 compliance year, the number of SREs dramatically increased, which resulted in a
“renewable-fuel shortfall” where millions of gallons of “renewable fuel simply [went]
unproduced.”441 Many of these SRE decisions were highly suspect.442 Yet, the decisions were
made in secret, and basic information about the exemptions was withheld under FOIA. As the
D.C. Circuit observed, the story of EPA’s administration of SREs “paint[s] a troubling picture of
intentionally shrouded and hidden agency law” that leaves “those aggrieved by the agency’s
actions without a viable avenue for judicial review.”443 Transparency is necessary because EPA
cannot lawfully make secret exemptions that undermine the RFS program goals.
None of the information covered by EPA’s proposal plausibly qualifies as exempt from
disclosure under FOIA. As EPA explains in its proposal, the Supreme Court addressed the
meaning of “confidential” in its decision in Food Marketing Institute v. Argus Leader Media,
139 S. Ct. 2356 (2019) (“Argus Leader”).444 The Court held that information is “confidential”
under Exemption 4 “[a]t least” where the information is “both customarily and actually treated as
private by its owner and provided to the government under an assurance of privacy.”445 The
Court then identified two potential conditions for satisfying this standard.
First, the information must be “customarily kept private, or at least closely held, by the
person imparting it.”446 Here, small refineries do not do that. They have publicly disclosed the
same or similar facts. For example, HollyFrontier has disclosed all these facts (and more) in its
securities filings, including: the fact of exemption extensions for two of its refineries, their names
and locations, the years for which the refineries received extensions, when the extensions were
granted, the effects of those extensions (e.g., “RINs cost reduction”), and how EPA effectuated
the extensions (e.g., providing “vintage RINs to replace the RINs previously retired” or
440 NPRM at 72,476-72,478; Growth Energy only comments on the portion of the Proposed
Rule’s “Public Access to Information” section proposing to disclose certain basic information
relating to small refinery exemptions. For convenience and readability, however, we use the
Proposed Rule as a shorthand to refer to that portion.
441 American Fuel, 937 F.3d at 571
442 See Renewable Fuels Ass’n v. EPA, 948 F.3d 1206, 1214 (10th Cir. 2020).
443 Advanced Biofuels Ass’n v. EPA, 792 F. App’x 1, 5 (D.C. Cir. 2019).
444 NPRM at 72,476.
445 Argus Leader, 139 S. Ct. at 2366.
446 Id. at 2363.
92
“reinstat[ing] the RINs previously submitted”).447 A news article also reported that a particular
company (Husk Energy) told the reporter that it inherited a small refinery exemption for 2017
when it acquired a plant in Superior, Wisconsin, and that it will seek an exemption for the
Superior plant for 2018.448
Second, Argus Leader suggested that a second condition may also have to be met to
qualify for Exemption 4: that “the party receiving [the information] provide[d] some assurance
that it will remain secret.”449 The Department of Justice has issued guidance regarding the
second condition.450 DOJ’s guidance clarifies that Exemption 4 cannot be claimed by an agency
when “the Government provides an express or implied indication to the submitter prior to or at
the time the information is submitted to the Government that the Government would publicly
disclose the information.”451 EPA’s proposal is therefore an appropriate way to preclude
Exemption 4 protection by conclusively establishing that EPA will deny such assurances of
confidentiality: it will “not consider certain basic information incorporated into EPA actions on
petitions and submissions … to be entitled to treatment as CBI under Exemption 4 of the
FOIA.”452 Nothing more is required to disqualify the information for Exemption 4. This
approach accords with the Department of Justice’s guidance,453 EPA’s proposal would be an
express statement that EPA will not treat the information as confidential.
Taking steps to ensure that this basic information relating to SREs is not withheld under
Exemption 4 is also long overdue. EPA previously proposed a similar rule twice before without
adopting it. In 2016, EPA first proposed a similar rule.454 EPA explained then “basic
information related to [EPA’s] decisions on small refinery/refiner exemption petitions is not
entitled to treatment as CBI.”455 In April 2019, EPA appeared ready to adopt the a similar rule
447 HollyFrontier Corp., SEC Form 10-K, at 41, 43, 77-78 (Feb. 20, 2019) (“HollyFrontier 2018
10-K”); HollyFrontier Corp., SEC Form 10-K, at 40-41, 77 (Feb. 21, 2018) (“HollyFrontier 2017
10-K”).
448 Jarrett Renshaw & Chris Prentice, Exclusive: Chevron, Exxon Seek ‘Small Refinery’ Waivers
From U.S. Biofuels Law, Reuters (Apr. 12, 2018), https://www.reuters.com/article/us-usabiofuels-epa-refineries-exclusive/exclusive-chevron-exxon-seek-small-refinery-waivers-from-us-biofuels-law-idUSKBN1HJ32R (attached as Ex. 14).
449 139 S. Ct. at 2363.
450 See Office of Information Policy, U.S. DOJ, Exemption 4 After the Supreme Court’s Ruling
in Food Marketing Institute v. Argus Leader Media and Accompanying Step-by-Step Guide,
(Oct. 4, 2019), https://www.justice.gov/oip/exemption-4-after-supreme-courts-ruling-foodmarketing-institute-v-argus-leader-media.
451 NPRM at 72476.
452 NPRM at 72,477.
453Id.
454 Renewables Enhancement and Growth Support Rule, 81 Fed. Reg. 80,828, 80,909 (Nov. 16,
2016).
455 Id.
93
again, but inexplicably abandoned it once more.456 In the interim, EPA has publicly disclosed
only the aggregate number of extensions and renewable fuel volumes exempted despite
numerous requests for further transparency,457 and even refused to provide any specific
information on the exemption extensions to members of Congress.458 This has led to litigation
over the EPA’s opposition to providing basic information about SRE petitions—information that
EPA had previously determined in 2016 was not CBI.459 Now is the time to finalize this rule and
expose its SRE decisions to some sunlight. As EPA notes, the proposed approach will “provide
certainty” regarding how EPA will treat the information,460 and will likely lead to fewer lawsuits
and reduce unnecessary compliance costs.
For example, in past litigation, EPA was forced to evaluate whether information in SRE
petitions were confidential on a case-by-case basis.461 Refineries submitted substantiations of
their confidentiality claims, and EPA then had to perform its own exhaustive search of public
sources to confirm whether the information was previously published.462 As to seven decision
documents, the D.C. District Court found EPA’s CBI determination “unpersuasive” after in
camera review.463 Notwithstanding the inadequacy of EPA’s past CBI determinations, the
process of reviewing and purportedly substantiating each claim by each refinery undoubtedly
introduces unnecessary compliance costs, especially given the dubious confidentiality over the
information in the first place.
Growth Energy further believes that EPA should not withhold additional categories of
information in connection with its decisions on exemption extensions, including: (i) the specific
standards EPA actually applied to decide whether to grant or deny the extension; (ii) EPA’s final
analysis of whether to grant or deny the extension; and (iii) if an extension is granted, the means
by which EPA effectuated the extension, such as allowing the refinery to unretire RINs. Records
456 See Renewables Enhancement and Growth Support (REGS) Rule (Apr. 11, 2019).
457 See, e.g., Growth Energy FOIA Request (Apr. 9, 2018), EPA-HQ-2018-006398; Growth
Energy FOIA Request (Apr. 12, 2018), EPA-HQ-2018-006524; Growth Energy FOIA Request
(July 23, 2018), EPA-HQ-2018-009898; Growth Energy FOIA Request (Mar. 19, 2019), EPAHQ-2019-004370; see also Growth Energy 2019 Comment at 17-22.
458 Letter from Senator Charles E. Grassley, et al. to Administrator of EPA, Scott Pruitt, at 2
(Apr. 12, 2018),
https://www.grassley.senate.gov/imo/media/doc/Pruitt%20Small%20Refinery%20Letter%204.1
2.18.pdf; see Letter from Assistant Administrator of EPA, William L. Wehrum, to Senator
Charles E. Grassley, at 1 (July 12, 2018), http://www.ascension-publishing.com/EPA-RINWaivers-071818.pdf.
459 See, e.g., Def’s Opp. Mot. Summ. J., Renewable Fuels Ass’n v.EPA, No. 18-2031, ECF #58
(D.C. Cir. Dec. 15, 2020).
460 NPRM at 72,478.
461 Renewable Fuels, No. 18-2031, ECF #58.
462 Id.
463 Opinion, Renewable Fuels, No. 18-2031, ECF #62 at 25, (D.C. Cir. Feb. 4, 2021).
94
embodying the standards EPA uses to grant an exemption extension, its final analysis on a
refinery’s entitlement to an extension, and the means EPA uses to effectuate an extension are all
“interpretations” or “considered statements” of EPA’s policy on SRE extensions, including on
the scope of EPA’s statutory authority to grant an extension and to allow retroactive remedies
using RINs.464 This information is not covered by the deliberative process privilege or
predecisional. Such SRE decisions are not “advisory opinions, recommendations,” or “personal
opinions of the writer” that “reflect internal deliberations on the advisability of any particular
course of action.”465 Instead, they are what EPA actually applied or decided—actions to which
the deliberative process privilege “can never apply.”466
464 Public Citizen v. Office of Mgmt. and Budget, 598 F.3d 865, 874 (D.C. Cir. 2009); Tax
Analysts v. IRS, 117 F.3d 607, 609, 617 (D.C. Cir. 1997); Coastal States Gas Corp. v. Dep’t of
Energy, 617 F.2d 854, 869 (D.C. Cir. 1980); Sterling Drug, Inc. v. FTC, 450 F.2d 698, 708 (D.C.
Cir. 1971).
465 Public Citizen, 598 F.3d at 875 (“an agency’s application of a policy to guide further
decision-making does not render the policy itself predecisional”).
466 NLRB v. Sears, Roebuck & Co., 421 U.S. 132, 153-154 (1975).
Exhibit List
Growth Energy Comments on EPA’s
Proposed Renewable Fuel Standard Program:
Renewable Fuel Standard Annual Rules
Docket # EPA-HQ-OAR-2021-0324
Volume 1
Exhibit
Number
Title of Exhibit
1 Environmental Health & Engineering, Inc., Response to 2020, 2021, and 2022
Renewable Fuel Standard (RFS) Proposed Volume Standards (Feb. 3, 2022)
2 Life Cycle Associates, LLC, Review of GHG Emissions of Corn Ethanol under
the EPA RFS2 (Feb. 4, 2022)
3 Net Gain, Analysis of EPA’s Proposed Rulemaking for 2020, 2021, and 2022
RVOs, Regarding Land Use Change, Wetlands, Ecosystems, Wildlife Habitat,
Water Resource Availability, and Water Quality (Feb. 3, 2022)
4 Comments of Drs. Fatemeh Kazemiparkouhi, David MacIntosh, Helen Suh,
EPA-HQ-OAR-2021-0324 (Feb. 3, 2022)
5 Stillwater Associates, LLC, Comments to EPA on 2020-2022 RFS Rule,
Prepared for Growth Energy (Feb. 4, 2022)
6 Stillwater Associates LLC, Infrastructure Changes and Cost to Increase
Consumption of E85 and E15 in 2017 (July 11, 2016)
7 Oak Ridge National Laboratory, Analysis of Underground Storage Tank System
Materials to Increased Leak Potential Associated with E15 Fuel, ORNL/TM2012/182 (Jul. 2012)
8 Growth Energy, Comments on EPA’s Proposed E15 Fuel Dispenser Labeling
and Compatibility with Underground Storage Tanks Regulations, Docket #
EPA–HQ–OAR–2020–0448 (Apr. 19, 2021)
9 Petroleum Equipment Institute, UST Component Compatibility Library
10 Association of State and Territorial Solid Waste Management Officials,
Compatibility Tool
11 Air Improvement Resource, Inc., Analysis of Ethanol-Compatible Fleet for
Calendar Year 2022 (Nov. 16, 2021)
12 Renewable Fuels Association, Contribution of the Ethanol Industry to the
Economy of the United States in 2020 (Feb. 2, 2021)
13 ABF Economics, Economic Impact of Nationwide E15 Use (June 2021)
14 Jarrett Renshaw & Chris Prentice, Exclusive: Chevron, Exxon seek ‘small
refinery’ waivers from U.S. biofuels law, Reuters (Apr. 12, 2018)
15 Stillwater Associates LLC, Potential Increased Ethanol Sales through E85 for
the 2019 RFS (Aug. 17, 2018)
ActiveUS 192803838v.1
Growth Energy Comments on EPA’s
Proposed Renewable Fuel Standard Program:
Renewable Fuel Standard Annual Rules
Docket # EPA-HQ-OAR-2021-0324
Exhibit 1
Environmental Health & Engineering, Inc.
180 Wells Avenue, Suite 200
Newton, MA 02459-3328
TEL 800-825-5343
781-247-4300
FAX 781-247-4305
www.eheinc.com
Environmental Health & Engineering, Inc. | 22493.1 | www.eheinc.com 1
February 3, 2022
U.S. Environmental Protection Agency
1200 Pennsylvania Avenue NW
Washington, DC 20460
Docket Number: EPA-HQ-OAR-2021-0324
Comments of David MacIntosh1,2, Tania Alarcon1,3, Brittany Schwartz1
1 Environmental Health & Engineering, Inc., Newton MA
2 Harvard T.H. Chan School of Public Health, Boston, MA
3 Tufts University, Boston, MA
RE: Response to 2020, 2021, and 2022 Renewable Fuel Standard (RFS) Proposed Volume
Standards
We are writing to comment on topics raised by the proposed RFS rule dated December 7, 2021,
and the associated Draft Regulatory Impact Analysis (DRIA) dated December 2021 (EPA-420-
D-21-002). We provide information to further the discussion of the impacts of renewable fuels
on land use change (LUC) and greenhouse gas emissions (GHG). Our comments are based on
the literature for corn starch ethanol published since the U.S. Environmental Protection Agency’s
(USEPA) most recent life cycle analysis (LCA) of GHG impacts for renewable fuels that was
released in 2010.1
We at Environmental Health & Engineering, Inc. (EH&E) are a multi-disciplinary team of
environmental health scientists and engineers with expertise in measurements, models, data
science, LCA, and public health. Members of our team authored a manuscript titled “Carbon
intensity of corn ethanol in the United States: state of the science”2
which was published in
Environmental Research Letters in January 2021. We conducted a state of the science review of
the carbon intensity (CI) for corn ethanol in the United States (U.S.), applied objective criteria
limited to the U.S. regulatory context, and derived an evidence-based central CI estimate and
credible range as of 2020. Our manuscript concludes that assessments of GHG intensity for corn
ethanol have decreased by approximately 50% over the prior 30 years and converged on a
1
US Environmental Protection Agency (USEPA) 2010 Renewable Fuel Standard program (RFS2) regulatory
impact analysis (RIA) Report No.: EPA-420-R–10–006 (Washington, DC: United States Environmental
Protection Agency, Assessment and Standards Division Office of Transportation and Air Quality US
Environmental Protection Agency). 2
Scully MJ, Norris GA, Alarcon Falconi TM, MacIntosh DL. 2021a. Carbon intensity of corn ethanol in the United
States: state of the science. Environmental Research Letters, 16(4), pp.043001.
Environmental Health & Engineering, Inc. | 22493.1 | www.eheinc.com 2
current central estimate value of approximately 51 grams of carbon dioxide equivalent emission
per megajoule (gCO2e/MJ). That continues to be our view. Later in these comments, we discuss
our approach for determining a credible range for the CI of corn ethanol and draw upon our
published reply to commentary on our paper. Our experience informs the comments that follow
regarding the proposed RFS rule.
Our comments focus on three major topics within the DRIA: 1) LUC estimates, 2) LUC
uncertainties, and 3) the Agency’s illustrative scenario. Our detailed comments on those topics
are presented following the summary.
SUMMARY
A critical evaluation of the best available science on indirect LUC (iLUC) shows that current
estimates of GHG impacts from LUC associated with corn starch ethanol are substantially lower
than findings published in the USEPA 2010 RIA. The change in estimated iLUC impact results
from improvements in agroeconomic models and data used to forecast impacts of supply and
demand for agricultural products. The Agency’s own work during the 2010 rulemaking process
demonstrates that advancement of LUC models can impact their output and lead to substantial
changes in estimated GHG impacts for iLUC. By relying only on its 2010 GHG analysis for the
current rulemaking, the Agency is not giving due consideration to the best available science.
Uncertainties in estimates of LUC, while extant, can be managed in a time-sensitive manner by
incorporating central best estimates from the existing generally accepted and commonly used
LUC models. This approach would allow a range of credible iLUC values to be determined
based on existing models and literature without conducting a full new analysis at this time.
We also comment on the validity of the Agency’s illustrative GHG scenario for corn ethanol.
That scenario presumes the proposed volumes will cause new demand for cropland in 2021 and
2022, but that assumption appears to conflict with actual levels of U.S. corn and ethanol
production available from the U.S. Department of Agriculture (USDA) and Department of
Energy (DOE). Observations of these data suggest existing U.S. corn production capacity is
sufficient to meet the modest increase in demand for ethanol projected by the Agency. Hence, we
encourage the Agency to evaluate the premise of its illustrative scenario in the context of
available U.S. corn and ethanol production data.
We look forward to continued engagement with the Agency on the important issue of LCA
modeling for biofuels, including at the upcoming Workshop on Biofuel Greenhouse Gas
Modeling.3
3
Announcing Upcoming Virtual Meeting on Biofuel Greenhouse Gas Modeling, 86 Fed. Reg. 73756 (Dec. 28,
2021).
Environmental Health & Engineering, Inc. | 22493.1 | www.eheinc.com 3
LAND USE CHANGE ESTIMATES
GHG impacts of corn ethanol projected for 2022 by the USEPA in its 2010 RIA agree
reasonably well with current estimates of impacts, with one notable exception: LUC. Table 1
shows the results of our published study alongside the Agency’s 2010 projection for 2022 and
the latest estimates from the Greenhouse Gases, Regulated Emissions, and Energy Use in
Technologies Model (GREET) based on the Global Trade Analysis Project-biofuel model
(GTAP-BIO). As our study was focused on the U.S. regulatory landscape for renewable fuels,
we reviewed U.S.-based studies and selected a 2004 baseline year to reflect the impacts of the
RFS. Table 1 shows that total LUC represents the largest difference between the Agency’s GHG
emission estimates made in 2010 and the current comparison analyses.
Table 1 Comparison of Reported Life Cycle Analysis Emission Categories of Corn Ethanol in the U.S.
Emission Category
USEPA 2010
2022 Projectiona
EH&E 2020
Topical Reviewc GREET 2021d
Farming Net Co-product 14.1 13.2 16.3
Fuel Production 26.5 29.6 28.1
Other Inputs (fuel & feedstock transport, rice methane*,
livestock*, denaturant, tailpipe) 6.4 4.7 3.7
Total Land Use Change 26.1 3.9 7.4
Total Carbon Intensity Value (gCO2e/MJ) 73.1 51.4 55.6
Lower Bound of Total Carbon Intensity Value (gCO2e/MJ) 51.2b 37.6 —
Upper Bound of Total Carbon Intensity Value (gCO2e/MJ) 91.9b 65.1 —
* Broken out explicitly in EPA analysis.
USEPA United States Environmental Protection Agency
GREET Greenhouse Gases, Regulated Emissions, and Energy Use in Technologies Model
gCO2e/MJ Gram carbon dioxide equivalent emission per megajoule
a US Environmental Protection Agency (USEPA) 2010 Renewable Fuel Standard program (RFS2) regulatory impact analysis (RIA)
Report No.: EPA-420-R–10–006 (Washington, DC: United States Environmental Protection Agency, Assessment and Standards
Division Office of Transportation and Air Quality US Environmental Protection Agency). Figure 2.6-2.
b USEPA. Federal Register 40 CFR Part 80 Regulation of Fuels and Fuel Additives: Changes to Renewable Fuel Standard Program;
Final Rule. March 26, 2010. Table V.C-1.
c Scully, MJ, Norris, GA, Alarcon Falconi, TM. and MacIntosh, DL 2021a. Carbon intensity of corn ethanol in the United States: state of
the science. Environmental Research Letters.
d Argonne National Laboratory. The Greenhouse Gases, Regulated Emissions, and Energy Use in Technologies Model. 2021.
iLUC is a substantial component of the USEPA 2010 analysis to estimate GHG impacts
associated with corn ethanol and warrants an update according to the best available
science. The 2021 DRIA that informs the proposed RFS rule states that “LUC emissions have
the potential to constitute a large portion of the total emissions attributable to crop-based
biofuels, but such estimates also have significant uncertainty.”4
In the Agency’s 2010 RIA, LUC
accounted for 29% (28.2 of 98.2 kilograms carbon dioxide equivalent emissions per million
4
USEPA 2010 RFS2 RIA, page 66.
Environmental Health & Engineering, Inc. | 22493.1 | www.eheinc.com 4
British thermal units [kgCO2e/mmBtu])5
of the total estimated GHG emissions. With estimated
impacts of 32 kgCO2e/mmBtu for iLUC and -3.8 kgCO2e/mmBtu for domestic LUC (dLUC),
iLUC was the dominant component of total LUC impacts in the Agency’s analysis. As described
below, the Agency’s 2010 forecast of LUC impacts is substantially different from most recent
estimates.
Most current iLUC estimates for corn starch ethanol, including models developed in the
U.S. and Europe, are substantially lower than findings published by USEPA in 2010.
Several publications recognize that estimates of iLUC impacts for corn starch ethanol over the
last decade have trended downward (See Attachment A).6,7,8,9
In fact, most estimates of iLUC
GHG impact from U.S. demand for corn ethanol are 2-fold to 4-fold lower than USEPA
estimates published in 2010 as illustrated by Figure 2 in Scully et al.10 and an updated version of
that figure presented below as Figure 1. This updated figure includes iLUC estimates from the
most current relevant and applicable modeling efforts in the U.S. (shown in blue) and in Europe
(shown in red). The four commonly relied upon models—GTAP-BIO, Food and Agricultural
Policy Research Institute- Center for Agricultural and Rural Development (FAPRI-CARD),
MIRAGE, and Global Biosphere Management Model (GLOBIOM)—provide estimates that are
lower than the USEPA central estimate and lower bound value published in 2010.
5
USEPA 2010 RFS2 RIA.. 6
Lee U, Hoyoung K, Wu M, Wang M. 2021. Retrospective analysis of the U.S. corn ethanol industry for 2005-
2019: implications for greenhouse gas emission reductions. Biofuels, Bioproducts & Biorefining, 15(5), pp.1318-
1331. 7
Dunn JB, Mueller S, Kwon H-Y and Wang MQ. 2013. Land-use change and greenhouse gas emissions from
corn and cellulosic ethanol. Biotechnology for Biofuels, 6(1), pp.1-3. 8
Taheripour F, Mueller S and Kwon H. 2021. Appendix A: supplementary information to response to ‘How
robust are reductions in modeled estimates from GTAP-BIO of the indirect land use change induced by
conventional biofuels?’ Journal of Cleaner Production., 310, pp.127431. 9
Carriquiry M, Elobeid A, Dumortier J and Goodrich R. 2020. Incorporating sub-national Brazilian agricultural
production and land-use into US biofuel policy evaluation. Applied Economic Perspectives and Policy, 42,
pp.497-523. 10 Scully et al. 2021a.
Environmental Health & Engineering, Inc. | 22493.1 | www.eheinc.com 5
Figure 1 Comparison of USEPA’s iLUC estimates with relevant most recent studies from the U.S. and Europe
LUC estimates have changed because the models and data that go into them have improved
over time. Our publication on LCA of corn ethanol11 and our reply to comments on the paper,12
as well as a recent retrospective analysis of corn ethanol LCAs by Lee et al.,13 summarize the
enhancements made to the two iLUC models, FAPRI and GTAP-BIO, used in U.S. regulatory
contexts for evaluation of corn ethanol. Briefly, the changes in estimates of iLUC impacts are
attributable to: (1) addition of new modules that allow for more accurate simulation of real-world
agricultural practices, (2) addition of more spatially resolved information on land cover, and (3)
tuning of parameters that describe rates of land conversion and land transformation (See
11 Scully et al. 2021a. 12 Scully MJ, Norris GA, Alarcon Falconi TM, MacIntosh DL, 2021b. Reply to comment on ‘Carbon intensity of
corn ethanol in the United States: state of the science. Environmental Research Letters, 16(11), pp.118002. 13 Lee et al. 2021.
Environmental Health & Engineering, Inc. | 22493.1 | www.eheinc.com 6
Attachment A). Details on important changes made over time to FAPRI and GTAP-BIO are
available in the literature.14,15,16,17,18,19
Enhancements to FAPRI and GTAP-BIO are particularly relevant to the current RFS rulemaking
because USEPA relied on both models for its 2010 forecast of GHG impacts from iLUC and
relies on that former analysis in the current rulemaking. Here we summarize key literature on
enhancements to the models and examples of their effect on estimated iLUC and GHG
emissions.
An early example of refinements to models and data that lead to substantial changes in estimated
GHG impacts for iLUC is found within the Agency’s 2010 rulemaking process itself. After
review of comments on the proposed rule for the 2010 RFS2, the Agency made updates to iLUC
estimates.20 These changes were made possible by the availability of updated studies, including
numerous improvements to the FAPRI-CARD model that are detailed in the 2010 RFS2 final
rule.21 Table 2 shows that the Agency’s central estimate of iLUC emissions decreased by
approximately 50% when using the updated version of FAPRI-CARD. Thus, the Agency’s own
prior work demonstrates that advancement of models can impact their output and lead to
substantial changes in estimated GHG impacts for iLUC.
14 USEPA 2010 RFS2 RIA. 15 Babcock BA, Iqbal Z. 2014. Using recent land use changes to validate land use change models. Staff report 14-
SR 109. Center for Agricultural and Rural Development, Iowa State University. 16 Carriquiry et al. 2020. 17 Taheripour F, Zhao X and Tyner W E. 2017. The impact of considering land intensification and updated data on
biofuels land use change and emissions estimates. Biotechnology for biofuels, 10(1), pp.1-16. 18 Taheripour F and Tyner W. 2013. Biofuels and land use change: applying recent evidence to model estimates.
Applied Science, 3(1), pp.14–38. 19 Kwon H, Liu X, Dunn J B, Mueller S, Wander MM and Wang M. 2020. Carbon calculator for land use and land
management change from biofuels production (CCLUB) Argonne National Library, Division ES September
2020. 20 USEPA 2010 RFS2 RIA. 21 USEPA Federal Register 40 CFR Part 80 Regulation of Fuels and Fuel Additives: Changes to Renewable Fuel
Standard Program; Final Rule. March 26, 2010.
Environmental Health & Engineering, Inc. | 22493.1 | www.eheinc.com 7
Table 2 USEPA’s Central Estimates of International Land Use Change Associated with Corn Ethanol for Biofuel
Over 30 Years, 2022a
Author Study Year
Land Use Change
Model Model Adjustments
Central Estimate of
International LUC Emissions
(g CO2e per MJ)
USEPA
2009 (original RFS2
analysis)
FAPRI NA 60.37a
2010 (revised RFS2
analysis)
Updated FAPRICARD, including Brazil
modulec
• Price-induced crop yields
• Animal feed replacements
• Improved satellite data
30.13b
USEPA U.S. Environmental Protection Agency
g CO2e per MJ gram carbon dioxide equivalent emissions per megajoule
RFS Renewable Fuel Standard
FAPRI Food and Agricultural Policy Research Institute
NA not applicable
FAPRI-CARD Food and Agricultural Policy Research Institute-Center for Agricultural and Rural Development
a US Environmental Protection Agency (USEPA). Lifecycle Greenhouse Gas (GHG) Emissions Results Spreadsheets (30 October
2008). Docket: EPA-HQ-OAR-2005-0161.
b USEPA 2010 Renewable Fuel Standard program (RFS2) regulatory impact analysis (RIA) Report No.: EPA-420-R–10–006
(Washington, DC: United States Environmental Protection Agency, Assessment and Standards Division Office of Transportation and
Air Quality US Environmental Protection Agency).
c Per RFS2 RIA (February 2010), Section 5.1.2.6.
Carriquiry et al.22 presents a more recent example using FAPRI, the same modeling platform
used by USEPA in the 2010 LCA for corn ethanol. The authors use a 2016 version of FAPRICARD that includes effects of demand for ethanol on the price and supply of corn and other
agricultural products, multiple cropping, and conversion of pasture area in Brazil to cropland.
Improvements made between the 2008 GHG Model and the 2016 GHG Model used to determine
emission factors include enhanced quality of spatial data, a refined relationship between crop
yield and crop price, and other adjustments. As shown in Table 3, the enhanced data and
additional detail contributed to the model result in up to a 44% reduction of total LUC emissions.
Table 3 CARD/FAPRI Central Estimates of Total Land Use Change Associated with Corn Ethanol for Biofuel Over
30 Years, ending in 2021/2022a
Author
Land Use
Change Model Emissions Factors
Central Estimate of
LUC Emissions
(g CO2e per MJ)
Carriquiry et al. FAPRI-CARD
2008 model 23.2
2016 model without sub-national land use data and inputs for Brazil 18.2
2016 model with sub-national land use data and inputs for Brazil 13.1
FAPRI-CARD Food and Agricultural Policy Research Institute-Center for Agricultural and Rural Development
LUC land use change
g CO2e per MJ gram carbon dioxide equivalent emissions per megajoule
a Carriquiry, M., Elobeid, A., Dumortier, J. and Goodrich, R., 2020. Incorporating sub-national Brazilian agricultural production and land-use
into US biofuel policy evaluation. Applied Economic Perspectives and Policy, 42(3), pp.497-523.
22 Carriquiry et al. 2020.
Environmental Health & Engineering, Inc. | 22493.1 | www.eheinc.com 8
Updates to GTAP-BIO since the Agency’s use of this model in 2010 provide another example of
decreases in estimated iLUC that result from model development and refinement. Taheripour et
al.23 describe updates to the land use module of GTAP-BIO that among other items included land
transformation elasticities tuned to trends in regional land cover data across the globe observed
from 2003 – 2013. As shown in Table 4, the updated version of GTAP-BIO, which was tuned to
observed land cover change for 2003 – 2013, produced estimates of LUC GHG impacts
approximately 40% lower than those from the prior (untuned) version of GTAP-BIO.
Table 4 GTAP-BIO Central Estimates of Total Land Use Change Associated with Corn Ethanol Biofuel Over
30 Years. a Modeled greenhouse gas emissions were estimated with an older version of GTAP-BIO
(“Untuned land use module”) and a newer version (“Updated land use module”) that has parameters
tuned to observed changes in cropland and harvested area in the U.S., Brazil, and other regions of the
world.
GTAP-BIO Model Version
GTAP-BIO
Economic Database
(Baseline Year)
Ethanol Expansion
(billion gallons)
Land Use Change
Emissions
(g CO2e per MJ) Reference
Untuned land use module Version 7 (2004) 11.59 BG
(3.41 to 15 BG)
13.4 a, b
Updated land use module 8.7 a, b
Untuned land use module Version 9 (2011) 1.07 BG
(13.93 to 15 BG)
23.3 a, b
Updated land use module 12.0 b
GTAP-BIO Global Trade Analysis Project-biofuel model
GREET Greenhouse Gases, Regulated Emissions, and Energy Use in Technologies model
g CO2e per MJ gram carbon dioxide equivalent emissions per megajoule
BG billion gallons
a Taheripour F, Cui H, Tyner WE. 2016. An exploration of agricultural land use change at the intensive and extensive margins: implications for
biofuels induced land use change. In: Qin Z, Mishra U, Hastings A, editors. Bioenergy and land use change, pp.19-37. American
Geophysical Union (Wiley). b Taheripour F, Zhao X, Tyner WE. 2017. The impact of considering land intensification and updated data on biofuels land use change and
emissions estimates. Biotechnology for Biofuels 10(1), pp.1-16.
In consideration of the preceding discussion and examples, we find strong evidence that current
versions of iLUC models produce substantially lower estimates of GHG emissions compared to
earlier versions of these same models used by USEPA in 2010 to prepare its estimate of iLUC
impacts. By relying only on its 2010 LUC analysis for the current rulemaking, the Agency is
not giving due consideration to the best available science. We encourage USEPA to consider
recent LUC modeling tools, data, and/or results for purposes of the current rulemaking.
UNCERTAINTY OF LAND USE CHANGE ESTIMATES
The 2021 DRIA contains several statements regarding the uncertainty of LUC estimates for
biofuels, characterizes the extent of that uncertainty as “considerable,” and concludes that the
Agency is unable to perform the extensive modeling needed to assess GHG effects of the
23 Taheripour et al. 2017.
Environmental Health & Engineering, Inc. | 22493.1 | www.eheinc.com 9
proposed volumes.24 The Agency should consider alternative approaches for quantifying
LUC impacts that would allow it to incorporate the best available science in lieu of
conducting extensive new modeling for this rulemaking.
As described by Plevin et al.,25 both lack of knowledge about the correct model (model
uncertainty) and the correct values for input parameters to a model (parameter uncertainty)
contribute to uncertainty over the true but unknown value of LUC, especially iLUC. Similarly,
the authors of a report issued by the International Civil Aviation Organization (ICAO) identify
four main categories of uncertainty for iLUC analyses: (1) methodology, (2) model design,
(3) data, and (4) parameters.26 Descriptions of the types of uncertainty provide a useful
framework for prioritizing research into its causes and resolution. Here, we comment on
quantitative indicators of uncertainty about GHG impacts of iLUC demonstrated in the literature.
The literature demonstrates that current, widely used central estimates of iLUC impacts
are of similar magnitude despite being the product of models with different methods,
designs, data, and parameter values. In addition, our recent published review of the literature
found that variability of central estimates of iLUC impacts has decreased over time.27,28 Several
examples from the literature indicate that uncertainty about central estimates of iLUC impacts
produced from different models and research teams are in reasonably good agreement with each
other.
• Carriquiry et al. stated that contemporary iLUC estimates from FAPRI-CARD of 13 g
CO2e/MJ were in line with a contemporary estimate of 13 g CO2e/MJ from GTAP-BIO.29
• iLUC central estimates from analyses conducted by European investigators with the
MIRAGE model30 and GLOBIOM model31 agree to within 1 gCO2e/MJ of each other,
reporting 8 gCO2e/MJ and 9 gCO2e/MJ, respectively32
• Based on the preceding information, iLUC central estimates from MIRAGE and GLOBIOM
are within 5 gCO2e/MJ of the FAPRI-CARD and GTAP-BIO estimates referred to by
24 For example, see pp. 66 – 68. 25 Plevin RJ, Beckman J, Golub AA, Witcover J, O’Hare M. 2015. Carbon accounting and economic model
uncertainty of emissions from biofuel-induced land use change. Environmental Science and Technology, 49(5),
pp.2656-26564. 26 ICAO. 2019. CORSIA Supporting Document: CORSIA Eligible Fuels – Life Cycle Assessment Methodology,
International Civil Aviation Organization, Montreal. 27 Scully et al. 2021a. 28 Scully et al. 2021b. 29 Carriquiry et al. 2020. 30 Laborde D, Padella M, Edward R, Marelli L. 2014. Progress in Estimates of iLUC With MIRAGE Model.
Report EUR 27119 EN. European Commission Joint Research Center, Institute for Energy and Transport. 31 Valin H, Peters D, van den Berg M, Frank S, Havlik P, Forsell N, Hamelinck C. 2015. The Land Use Change
Impact of Biofuels Consumed in the EU. Ref. Ares(2015)4173087 – 8/10/2015, ECOFYS, Utrecht. 32 Original values adjusted to a 30-year averaging period to be comparable to the results from GTAP-BIO and
FAPRI-CARD.
Environmental Health & Engineering, Inc. | 22493.1 | www.eheinc.com 10
Carriquiry et al. (2020). Notably, the MIRAGE results provide the scientific basis for
European Commission policy on GHG emissions from iLUC associated with biofuels from
corn and cereal grains.33,34
• Separate analyses of iLUC impacts of sustainable aviation fuel produced from corn starch
ethanol conducted with GLOBIOM and GTAP-BIO produced identical results.35,36
While more research would be helpful to refine model parameters and data further, the
distribution of uncertainty around central estimates of iLUC is reasonably well
characterized in the literature. The Agency and others have used Monte Carlo and similar
simulation methods to characterize the effects of parameter uncertainty on estimated iLUC
impacts.37,38 These analyses report distributions of uncertainty about iLUC estimates for corn
ethanol that are approximately symmetric or moderately right skewed with a coefficient of
variation of approximately 20%. Other studies have conducted uncertainty analyses with the aim
of determining the relative influence of individual parameter uncertainty on overall uncertainty.39
Those studies show that iLUC uncertainty is dominated by lack of knowledge about crop yield
elasticity with respect to price (YDEL). Experts disagree on the ‘correct’ value and there likely is
no single true value for YDEL around the globe or over time.40 Model developers have
addressed this uncertainty in part by developing values for YDEL and other parameters for
distinct geographic regions of the world.41 Nonetheless, uncertainty about key model parameters
remains and some analysts have suggested that these uncertainties will not be resolved in the
foreseeable future.42 Thus, we encourage the Agency to support continued refinement of iLUC
model parameters, but not let uncertainty deter it from using the best available science to derive a
current central estimate of iLUC impacts. In fact, in the 2010 RIA, the Agency relied upon a
central estimate of iLUC for rulemaking and related policy despite the uncertainty characterized
in its own analysis.43
33 Directive (EU) 2015/1513 of the European Parliament and of the Council of 9 September 2015 amending
Directive 98/70/EC relating to the quality of petrol and diesel fuels and amending Directive 2009/28/EC on the
promotion of the use of energy from renewable sources. Official Journal of the European Union. L 239/1. 34 Directive (EU) 2018/2001 of the European Parliament and of the Council of 11 December 2018 on the
promotion of the use of energy from renewable sources. Official Journal of the European Union. L 328/82. 35 ICAO. 2019. 36 Zhao X, Taheripour F, Malina R, Staples M, Tyner W. 2021. Estimating induced land use change emission for
sustainable aviation biofuel pathways. Science of the Total Environment, 779, pp.146238. 37 USEPA 2010 Final Rule, Table V.C-5. 38 Laborde et al. 2014. 39 Plevin et al. 2015. 40 Plevin et al. 2015. 41 For example, Taheripour et al. 2017. 42 Hertel TW, Tyner WE. 2013. Market-mediated environmental impacts of biofuels. Global Food Security, 2(2),
pp. 131-137. 43 USEPA 2010 RFS2 RIA.
Environmental Health & Engineering, Inc. | 22493.1 | www.eheinc.com 11
Given the availability of numerous estimates of iLUC from generally accepted or commonly
used models, we believe the Agency could conduct a systematic review of the literature to derive
an updated central estimate of iLUC-related GHG emissions for the proposed rulemaking. Scully
et al.44 and Lewandrowski et al.45 provide examples of authoritative reviews of the LUC
literature that the Agency could build upon to conduct its own systematic review of central
estimates for iLUC of corn ethanol and the associated uncertainty.
THE AGENCY’S ILLUSTRATIVE SCENARIO
In this section, we comment on the illustrative scenario for GHG impacts that is presented in the
2021 proposed rule and DRIA published by the Agency. On page 68 of the DRIA, the Agency
states “This scenario is not EPA’s assessment of the likely greenhouse gas impacts of this
proposed rule” but instead “an illustrative scenario of the GHG impacts of biofuel consumption
following the implementation of the proposed standards.” We encourage the Agency to
evaluate the premise of its illustrative scenario in light of available U.S. corn and ethanol
production data.
As described in the DRIA, the illustrative scenario uses corn starch ethanol consumption of 12.5
billion gallons in 2020 as the baseline condition. This baseline reflects a 13% reduction in
ethanol consumption due to the pandemic, compared to the average of the prior five years.46 The
scenario assumes that ethanol consumption increases by 953 million gallons to 13.453 billion
gallons in 2021 and increases by an additional 335 million gallons to 13.788 billion gallons in
2022. Data published by the DOE Energy Information Agency (EIA) report that actual U.S. fuel
ethanol consumption in 2020 was 12.7 billion gallons, which is 0.2 billion gallons higher than
the baseline in EPA’s illustrative scenario.47 EIA data indicates that U.S. fuel ethanol
consumption is likely to be approximately 13.8 billion gallons in 2021, in good agreement with
the Agency’s scenario for total consumption in 2021.48 As shown in Table 5, the Agency’s
estimates of ethanol fuel consumption in 2021 and 2022 are lower than consumption reported by
EIA for 2015 – 2019, the five year period preceding the COVID-19 pandemic. These data
indicate that existing U.S. ethanol production capacity is sufficient to meet the demand projected
by the Agency.
44 Scully et al. 2021a. 45 Lewandrowski J, Rosenfeld J, Pape D, Hendrickson T, Jaglo K, Moffroid K. 2019. The greenhouse gas benefits
of corn ethanol – assessing recent evidence. Biofuels 11(3), pp. 361-375. 46 We calculated this 13% by comparing the average consumption between 2015 and 2019 with the 12.5 billion
gallons of the USEPA baseline. 47 U.S. Department of Energy, Energy Information Administration. Total Energy, Table 10.3 Fuel Ethanol
Overview. https://www.eia.gov/totalenergy/data/browser/?tbl=T10.03#/?f=M. 48 Ibid.
Environmental Health & Engineering, Inc. | 22493.1 | www.eheinc.com 12
Table 5 Annual U.S. Ethanol Production, 2015 – 2021a
Calendar Year Fuel Ethanol Production (Billion gallons) Fuel Ethanol Consumption (Billion gallons)
2015 14.8 13.9
2016 15.4 14.4
2017 15.9 14.5
2018 16.1 14.4
2019 15.8 14.6
2020 13.9 12.7
2021* 14.7 13.8
* Ethanol data for the missing 2021 Q4 was estimated as a function of consumption in Q1, Q2,and Q3.
a U.S. Department of Energy, Energy Information Administration. Total Energy, Table 10.3 Fuel Ethanol Overview.
https://www.eia.gov/totalenergy/data/browser/?tbl=T10.03#/?f=M.
Similarly, records from USDA show that domestic corn feedstock production is also sufficient to
meet the increase in demand for 2021 and 2022 contained in the Agency’s illustrative scenario.
As shown in Table 6, U.S. corn production for ethanol during the four market years49 (2015 –
2018) that preceded the COVID-19 pandemic ranged from 5.2 – 5.6 billion bushels.50,51
Production dropped to 4.9 billion bushels in market year 2019 and increased to 5.0 billion
bushels in market year 2020, still below production levels for market years 2015 – 2018. Based
on yield of 2.85 gallons of ethanol per bushel,52 we estimate that 0.12 billion bushels of corn
production are required to produce the additional 953 million gallons projected for 2021 in the
Agency’s illustrative scenario. An increment of 0.12 billion bushels on top of the 4.9 billion
bushels produced in market year 2019 (total of 5.02 billion bushels) or on top of the 5.0 billion
bushels produce in market year 2020 (total of 5.12 billion bushels) are both approximately 7%
lower than average production during the four years that preceded the COVID-19 pandemic.
49 A market year is defined as September 1 – August 31. For example, market year 2018 is September 1, 2018 –
August 31, 2019, which preceded the COVID-19 pandemic. Market year 2019 is September 1, 2019 – August
31, 2020, which includes approximately the first 6 months of the COVID-19 pandemic. All of market year 2020
occurred during the COVID-19 pandemic. 50 US Department of Agriculture (USDA), Economic Research Service. Feed Grains Custom Query.
https://data.ers.usda.gov/FEED-GRAINS-custom-query.aspx. 51 USDA, Farm Service Agency. Crop Acreage Data. https://www.fsa.usda.gov/news-room/efoia/electronicreading-room/frequently-requested-information/crop-acreage-data/index. 52 Argonne National Laboratory. The Greenhouse Gases, Regulated Emissions, and Energy Use in Technologies
Model. 2021.
Environmental Health & Engineering, Inc. | 22493.1 | www.eheinc.com 13
Table 6 Annual U.S. Corn Production, Market Years 2015 – 2020a,b
Market
Year*
Corn Production
(Billion bushels)
Corn Used for Fuel
Ethanol (Billion bushels)
Corn Export
(Billion bushels)
Corn Planted
Acreage/10
(Million acres)
Corn Harvested
Acreage/10
(Million acres)
2015 13.6 5.2 1.9 8.8 8.1
2016 15.1 5.4 2.3 9.4 8.7
2017 14.6 5.6 2.4 9.0 8.3
2018 14.3 5.4 2.1 8.9 8.1
2019 13.6 4.9 1.8 9.0 8.1
2020 14.1 5.0 2.8 9.1 8.2
* The U.S. corn market year runs from September 1 of the listed year to August 31 of the following year.
a US Department of Agriculture (USDA), Economic Research Service. Feed Grains Custom Query. https://data.ers.usda.gov/FEED-GRAINS-customquery.aspx.
b USDA, Farm Service Agency. Crop Acreage Data. https://www.fsa.usda.gov/news-room/efoia/electronic-reading-room/frequently-requestedinformation/crop-acreage-data/index.
As shown in Figure 2, cropland acreage planted and harvested for corn has remained stable since
2015, including during the pandemic. These data suggest that the ethanol volumes proposed by
the Agency for 2021 were met by existing U.S. corn production capacity and that the same corn
production capacity will meet the volumes proposed for 2022.
Figure 2 Trends in Annual U.S. Corn and Fuel Ethanol Production
Environmental Health & Engineering, Inc. | 22493.1 | www.eheinc.com 14
In consideration of these data, we encourage the Agency to evaluate whether the proposed
volumes for corn ethanol contained in the illustrative GHG scenario of the DRIA are likely to
induce LUC and generate the estimated GHG impact presented in the 2021 DRIA. We
emphasize LUC impacts since historically this effect has been estimated to be a major
contributor of GHG emissions of corn ethanol. In the Agency’s illustrative scenario, iLUC
contributes over 92% of the estimated GHG emissions in the first year.53 In brief, we encourage
the Agency to reconsider the validity and usefulness of the illustrative scenario.
Given (1) the limitations of the Agency’s illustrative scenario, (2) its stated inability to carry out
an in-depth uncertainty analysis of LUC impacts for corn ethanol at this time, and (3) the
deviation between USEPA 2010 LUC results and the current best available science, we
encourage the Agency to apply current central estimates of LUC produced by modeling efforts in
the U.S. and Europe to the current proposed rule. Our prior work provides an example of an
authoritative review of the literature that the Agency could build upon to determine its own
current best central estimate of iLUC and total GHG emissions of corn ethanol.54,55
CONCLUSION
We thank the Agency for this opportunity to comment on the proposed RFS rule. Throughout
this letter, we have discussed improvements of data and models that estimate emissions from
land use change associated with corn ethanol. We encourage the Agency to incorporate these
refined data and advanced models into the calculations that determine GHG impacts and thus
inform the proposed volume standards for biofuels. By relying only on its 2010 analysis for the
current rulemaking, the Agency is not giving due consideration to the best available science. As
noted previously, our published research concludes that assessments of GHG intensity for corn
ethanol have decreased by approximately 50% over the prior 30 years and converged on a
current central estimate value of approximately 51 gCO2e/MJ. Even if EPA intends to solicit
additional data and analysis in its forthcoming workshop on LCA, this is not a reason to continue
to rely on outdated information. There is sufficient updated information available for EPA to
adopt, at least on an interim basis, a carbon intensity estimate that is closer to current central
estimates and relies on the current state of the science.
Enclosures
Attachment A—Downward Trend in Land Use Change GHG Estimates
53 USEPA Lifecycle Greenhouse Gas (GHG) Emissions Results Spreadsheets (30 October 2008). Docket: EPAHQ-OAR-2005-0161. 54 Scully et al. 2021a. 55 Scully et al. 2021b.
eheinc.com 1
Attachment A
Downward Trend in Land Use Change GHG Estimates
– Lewandrowski et al (2019): “across studies, estimates of corn ethanoldriven iLUC emissions trend down over time”
– Malins et al (2020): “various academic and working papers have,
however, tended to decrease iLUC emissions compared to previous
estimates”
– Lee et al (2019): “the downtrend in simulated LUC emissions is a result
of better developed and calibrated economic models and better modeling
of GHG emissions from LUC”
– Taheripour et al (2021): “reduction in land use emissions due to model
and data improvements,” which “is not limited to the GTAP-BIO model
but is a common finding of the literature.”
– Carriquiry et al (2019): “The addition of detailed modeling in Brazil, e.g.,
double-cropping, reduced estimates considerably and highlights the
importance of continuous improvements in global agricultural models.”
“The concept of an observed downward trend in corn ethanol LUC modeling is not novel and we are not
the first to acknowledge it” (Scully et al 2021. Reply to Comment on ‘Carbon intensity of corn ethanol in the United States: state of the science’)
Market
mediated
no yield
response
Market
mediated
with yield
response
Corrections
in Brazil
land use
data
Market
mediated
& other
factors Regional
extensive
margins by
AEZ
2004 data
base and
tuning
2004 data
base and
multiple
cropping
2011 data
base and
multiple
cropping
Charts reproduced from Taheripour et al 2021. Supplementary information to Response to ‘How robust are reductions
in modeled estimates from GTAP‐BIO of the indirect land use change induced by conventional biofuels?’
ActiveUS 192803838v.1
Growth Energy Comments on EPA’s
Proposed Renewable Fuel Standard Program:
Renewable Fuel Standard Annual Rules
Docket # EPA-HQ-OAR-2021-0324
Exhibit 2
Review of GHG Emissions of Corn Ethanol
under the EPA RFS2
LCA.8120.200.2022
4 February 2022
Prepared by:
Stefan Unnasch
DISCLAIMER
This report was prepared by Life Cycle Associates, LLC for Growth Energy. Life Cycle Associates is not
liable to any third parties who might make use of this work. No warranty or representation, express or
implied, is made with respect to the accuracy, completeness, and/or usefulness of information
contained in this report. Finally, no liability is assumed with respect to the use of, or for damages
resulting from the use of, any information, method or process disclosed in this report. In accepting this
report, the reader agrees to these terms.
ACKNOWLEDGEMENT
Life Cycle Associates, LLC performed this study under contract to Growth Energy.
Contact Information:
Stefan Unnasch
Life Cycle Associates, LLC
1.650.461.9048
unnasch@LifeCycleAssociates.com
www.LifeCycleAssociates.com
Recommended Citation: Unnasch. S. (2022). Review of GHG Emissions of Corn Ethanol under the EPA
RFS2, Life Cycle Associates Report LCA.8120.200.2022, Prepared for Growth Energy.
i |
CONTENTS
TABLES…………………………………………………………………………………………………………………………………………………………… II
EXECUTIVE SUMMARY …………………………………………………………………………………………………………………………………….. V
1. INTRODUCTION ………………………………………………………………………………………………………………………………………. 1
1.1 LIFE CYCLE GHG ANALYSIS ………………………………………………………………………………………………………………………………… 2
1.2 LAND USE CHANGE …………………………………………………………………………………………………………………………………………. 3
1.3 MODELING APPROACHES ………………………………………………………………………………………………………………………………….. 3
1.3.1 Approach for Revised GHG Analysis ……………………………………………………………………………………………………… 4
1.4 GLOBAL WARMING POTENTIAL …………………………………………………………………………………………………………………………… 4
2. DOMESTIC AND INTERNATIONAL LAND USE CHANGE …………………………………………………………………………………… 6
2.1 EPA RIA APPROACH FOR LAND USE CHANGE ………………………………………………………………………………………………………….. 6
2.1.1 EPA Modeling Approach …………………………………………………………………………………………………………………….. 6
2.1.2 Challenges with 2010 RIA Land Use Change Analysis ……………………………………………………………………………… 9
2.2 NEW FINDINGS ON LAND USE CHANGE ………………………………………………………………………………………………………………… 10
2.2.1 CCLUB and GTAP ……………………………………………………………………………………………………………………………… 11
2.2.2 Other Corn Ethanol Studies ……………………………………………………………………………………………………………….. 12
2.2.3 Empirical Data …………………………………………………………………………………………………………………………………. 13
2.2.4 Modeling Results ……………………………………………………………………………………………………………………………… 15
2.2.5 Summary of LUC Effects ……………………………………………………………………………………………………………………. 17
3. CORN FARMING…………………………………………………………………………………………………………………………………….. 19
3.1 CORN FARMING …………………………………………………………………………………………………………………………………………… 19
3.2 SENSITIVITY ANALYSIS OF FARM INPUTS ……………………………………………………………………………………………………………….. 22
4. IMPACT OF CO-PRODUCTS ON CORN ETHANOL CI ………………………………………………………………………………………. 24
4.1 DGS CO-PRODUCT ……………………………………………………………………………………………………………………………………….. 24
4.2 CORN DISTILLERS OIL …………………………………………………………………………………………………………………………………….. 26
4.2.1 Corn Oil as Coproduct of Ethanol Production in EPA RIA ……………………………………………………………………….. 27
4.2.2 CDO Under Various Allocation Methods ……………………………………………………………………………………………… 27
4.3 REPLACEMENT FEED ………………………………………………………………………………………………………………………………………. 28
5. BIOREFINERY TECHNOLOGIES ………………………………………………………………………………………………………………….. 29
5.1 CORN ETHANOL YIELD ……………………………………………………………………………………………………………………………………. 30
5.1.1 Ethanol Yield in EPA 2010 RIA ……………………………………………………………………………………………………………. 30
5.1.2 Plant Debottlenecking ………………………………………………………………………………………………………………………. 31
5.1.3 Enzymes and Chemicals ……………………………………………………………………………………………………………………. 31
5.2 ENERGY CONSUMPTION ………………………………………………………………………………………………………………………………….. 31
5.3 CO2 FROM CORN ETHANOL ……………………………………………………………………………………………………………………………… 35
6. PROCESS FUELS……………………………………………………………………………………………………………………………………… 36
6.1 EPA RIA FUEL PRODUCTION ……………………………………………………………………………………………………………………………. 36
6.2 PHASE OUT OF COAL ……………………………………………………………………………………………………………………………………… 36
6.3 NATURAL GAS PRODUCTION AND METHANE EMISSIONS ……………………………………………………………………………………………. 36
6.4 BIOGAS AND BIOMASS PROCESS FUEL …………………………………………………………………………………………………………………. 37
6.5 ELECTRIC POWER ………………………………………………………………………………………………………………………………………….. 38
6.5.1 Grid Carbon Intensity ……………………………………………………………………………………………………………………….. 38
6.5.2 Renewable Power…………………………………………………………………………………………………………………………….. 39
6.6 SUMMARY OF ETHANOL GHG ANALYSIS ISSUES ……………………………………………………………………………………………………… 40
ii |
7. PETROLEUM BASELINE EMISSIONS FOR 2005 ARE LARGER THAN PROJECTED. ………………………………………………… 41
7.1 EPA 2010 RIA APPROACH IN ESTIMATION OF PETROLEUM BASELINE ……………………………………………………………………………. 41
7.2 NEW FINDINGS ON PETROLEUM BASELINE…………………………………………………………………………………………………………….. 41
8. ESTIMATED GHG EMISSIONS FROM CORN ETHANOL …………………………………………………………………………………… 44
9. CONCLUSIONS ………………………………………………………………………………………………………………………………………. 48
10. APPENDIX A – NITROGEN APPLICATION RATES ………………………………………………………………………………………….. 50
REFERENCES …………………………………………………………………………………………………………………………………………………. 53
TABLES
Table 1.1. Global Warming Potential (100-year time horizon). ………………………………………………………………………………………………………… 5
Table 2.1. Addressing Uncertainties in LUC Assessments ………………………………………………………………………………………………………………. 10
Table 2.2. Life Cycle Studies Examining Corn Ethanol. …………………………………………………………………………………………………………………… 13
Table 2.3. Change in GHG Emissions Due to Land Use Change (g CO2e/MMBtu). ……………………………………………………………………………… 18
Table 3.1. Farming Inputs of Corn in the U.S. ………………………………………………………………………………………………………………………………. 22
Table 4.1. The CI of DGS Using Displacement Method. …………………………………………………………………………………………………………………. 26
Table 4.2. The Effect of Displacement Method of CDO on CI of Corn Ethanol. …………………………………………………………………………………. 28
Table 5.1. Ethanol Plant Performance Parameters. ………………………………………………………………………………………………………………………. 29
Table 6.1. Effect of Biogas on Carbon Intensity of Corn Ethanol. ……………………………………………………………………………………………………. 37
Table 6.2. Evaluation Issues related to GHG Analysis. …………………………………………………………………………………………………………………… 40
Table 7.1. Carbon Intensity of 2005 Gasoline from Well to Wheel (WTW). ……………………………………………………………………………………… 41
Table 7.2. Petroleum Gasoline Carbon Intensity. …………………………………………………………………………………………………………………………. 42
Table 8.1. CI of Corn Ethanol for Dry Mill, Natural Gas Operation with Corn Oil Extraction. ………………………………………………………………. 46
iii |
FIGURES
Figure 1.1. EPA’s Analysis of Corn Ethanol GHG emissions. (EPA, 2010) …………………………………………………………………………. 1
Figure 1.2. Modeling Flow for Determination of Total Biofuel Lifecycle Carbon Intensity, Including Both Direct and Indirect
Effects. ………………………………………………………………………………………………………………………………………………………………….. 3
Figure 1.3. System Boundary Diagram for Corn Ethanol Production………………………………………………………………………………. 4
Figure 2.1. System Boundary Used in EPA RIA Study. (EPA, 2010) …………………………………………………………………………………. 7
Figure 2.2. Approaches to LUC Modeling. ………………………………………………………………………………………………………………… 11
Figure 2.3. Comparison of Brazilian Deforestation and U.S. Corn Ethanol Production. …………………………………………………… 14
Figure 2.4. Carbon loss following cropland pasture conversion using Winrock, CENTURY and AEZ-EF emission factor models.
(Malins, et al., 2020). …………………………………………………………………………………………………………………………………………….. 15
Figure 2.5. South Dakota Top Soil Organic Matter. (ACE, 2018) ………………………………………………………………………………….. 15
Figure 2.6. International Land Use Change Estimated by Several Studies. …………………………………………………………………….. 16
Figure 3.1. Corn Yield Over Time. (USDA NASS, 2020)………………………………………………………………………………………………… 19
Figure 3.2. Changes in Corn Production Practices from 2005 to 2010. (Rosenfeld et al., 2018) ……………………………………….. 20
Figure 3.3. Corn Ethanol Production by State. (USDA NASS, 2018) ………………………………………………………………………………. 20
Figure 3.4. Average Corn Grain Yield vs. Production for 14 States. ………………………………………………………………………………. 21
Figure 3.5. Nitrogen Fertilizer Use Rate in the Three Largest Corn Producer States. ………………………………………………………. 21
Figure 3.7. Sensitivity Analysis of Farm Inputs. …………………………………………………………………………………………………………. 23
Figure 4.1. Historical Prices of DDGS and Soybean Meal. (USDA ERS, 2018; World Bank, 2018) ………………………………………. 25
Figure 4.4. Corn Distillers’ Oil Production in the U.S. (USDA NASS, 2018; RFA, 2019) …………………………………………………….. 27
Figure 5.1. Dry Mill Corn Ethanol Yield Data and Projections. ……………………………………………………………………………………… 30
Figure 5.2. CI of Corn Ethanol with Various Technologies Registered under CARB …………………………………………………………. 33
Figure 5.3. Distribution of Natural Gas Usage Among Ethanol Production Facilities. ……………………………………………………… 33
Figure 5.4. Decrease in Natural Gas Usage Since 2004 (EPA RIA combination denotes dry mill plant with only natural gas
which produces 63% dry DGD and 37% wet DGS). …………………………………………………………………………………………………….. 34
Figure 5.5. Electricity Conusmption in Corn Ethanol. (ACE, 2018) ………………………………………………………………………………… 34
Figure 6.1. Well to Wheel (WTW) Carbon Intensity of Natural Gas plus Boiler Emission Factor in GREET. (ANL, 2018, GREET
versions from 1.8b to 2018)……………………………………………………………………………………………………………………………………. 37
Figure 6.2. Carbon Intensity of Electric Power (U.S. Average). …………………………………………………………………………………….. 38
Figure 6.3. Change in Carbon Intensity of Electricity. (Schivley et al., 2018, EIA) ……………………………………………………………. 39
Figure 7.1. CI of Gasoline Estimated by Several Studies. (Unnasch et al., 2018) …………………………………………………………….. 43
Figure 8.1. CI of Corn Ethanol for Dry Mill, Natural Gas Operation with Corn Oil Extraction. …………………………………………… 47
Figure A.1. FASOM Average Nitrogen Fertilizer Use by Crop. (EPA, 2010, not updated in EPA 2021) ……………………………….. 50
Figure A.2. N2O emissions per acre from crop production. …………………………………………………………………………………………. 51
Figure A.3. Effect of Nitrogen Application Rate on Soybean Yield. (Mourtzinis et al., 2018) ……………………………………………. 52
iv |
TERMS AND ABBREVIATIONS
aLCA Attributional Life Cycle Analysis
ANL Argonne National Laboratory
ARB California Air Resources Board
CA California
CA-GREET The standard GREET model modified for use in CA LCFS
CH4 Methane
CI Carbon intensity
cLCA Consequential Life Cycle Analysis
CO Carbon monoxide
CO2 Carbon dioxide
DOE U.S. Department of Energy
EPA U.S. Environmental Protection Agency
g CO2e Grams of carbon dioxide equivalent
GHG Greenhouse Gas
GREET The Greenhouse gas, Regulated Emissions, and Energy use in
Transportation model
GWP Global Warming Potential
HC Hydrocarbon
IPCC Intergovernmental Panel on Climate Change
LCA Life Cycle Analysis or Life Cycle Assessment
LCFS Low Carbon Fuel Standard
LCI Life Cycle Inventory
LCFS Low Carbon Fuel Standard
LHV Lower Heating Value
N2O Nitrous oxide
NG Natural Gas
RIA Regulatory Impact Analysis
RFS2 Revised Federal Renewable Fuels Standard
UN United Nations
UNFCC United Nations Framework Convention on Climate Change
VOC Volatile Organic Compound
WTT Well-To-Tank
WTW Well-To-Wheel
v |
EXECUTIVE SUMMARY
Twelve years of experience and improved analysis methods have provided new insight into the life
cycle greenhouse gas (GHG) emissions from corn ethanol. This study reviews the key factors that affect
the life cycle emissions from corn ethanol production as well as the most recent agricultural data.
Some of the key factors affecting corn ethanol have evolved as predicted in EPA’s 2010 Regulatory
Impact Analysis (2010 RIA), while other factors point towards substantially lower life cycle GHG
emissions.
EPA developed a consequential LCA approach that estimated the emissions associated with the
incremental ethanol capacity induced by the RFS policy as well as the incremental crop production
required to make up for the net effect of corn crops diverted to ethanol production and distiller’s
grains sold as animal feed. The modeling approach involved a combination of the FASOM model that
has been used to develop the U.S. inventory for agricultural emissions, the FAPRI model, which
estimates the effect of the use of agricultural products on global agricultural production, and the
GREET model, which estimates life cycle GHG emissions from the fuel used in ethanol plants. EPA’s
analysis aligned the economic modeling of the FASOM and FAPRI modeling and calculated emission
impacts that are tied to the model predictions including changes in rice and beef consumption as well
as deforestation associated with new crop production.
The 2010 RIA overestimated the GHG impact of corn ethanol due largely to overestimating indirect
land use conversion (ILUC) emissions as well as numerous small details associated with the life cycle of
corn ethanol. EPA’s agro-economic models rely on economic projections to attribute land use change
to crop production without considering factors such as changes in farming and cattle production
practices. Recent data on deforestation has shown that land ownership is much more important in
affecting deforesting than the macro-economic pressure or crop prices. Burning in the Amazon has
declined and increased due to policies associated with land ownership. A more accurate
representation of the effect of crops on pasture conversion is represented in more recent publications
based on the GTAP model and EPA would generate similar results if its ILUC modeling tools included an
accurate representation of factors such as flexibility in changing cattle stocking rates. The analysis
inputs to GTAP modeling would yield similar results in the FASOM/FAPRI modeling system. If EPA
continues to use the FAPRI results for its international LUC analysis, the results could be scaled to
reflect the values from GTAP that more accurately represent the interaction between pasture and
cropland.
Several other factors affecting corn ethanol have also changed since the publication or the 2010 RIA.
Corn ethanol uses about 0.7 kWh to produce one gallon of ethanol and the GHG intensity of electric
power has declined substantially with increased natural gas production, a reduction in coal-based
power, and growth in renewable power. The RIA also underestimated the adoption of low emission
technologies that have resulted in lower emissions from ethanol plants and many small details
associated with each step of the ethanol life cycle.
vi |
More significantly, EPA underestimated the effect of distiller’s grains and corn oil. Much of the corn
used for ethanol production has resulted in the displacement of soybean production. The same acre of
land that was producing soybeans and converted to corn for ethanol produces the same amount of
feed via the distiller’s grains from the ethanol plant. Therefore, any change in net feed requirements is
subtle at best. The GREET model also underestimates the displacement effect of both soybeans and
urea that would otherwise be fed to cattle. Even though soybeans fix nitrogen1
, USDA data shows that
they have required more nitrogen fertilizer than projected in the RIA. Also, the emissions associated
with urea feed in the GREET model omit the displacement of fossil carbon2
in urea. Corn ethanol plants
have produced significant quantities of corn oil as predicted in the 2010 RIA. However, about half of
the corn oil is used as biodiesel which corresponds to about 2.5% of the energy output of an ethanol
plant. The GHG emissions associated with corn production and any ILUC should be partially assigned to
biodiesel.
These factors should be incorporated in EPA’s GHG analysis of corn ethanol in this 2020, 2021, 2022
Renewable Volume Obligation (RVO) rulemaking, including the following considerations:
ILUC and soil carbon storage should reflect the latest research.
o ANL soil carbon storage modeling (CCLUB) shows increased soil carbon storage with
corn farming that was not taken into account in the 2010 RIA.
o New analysis based on the GTAP shows the effect of pasture intensification which
predicts lower rates of forest conversion to agriculture.
o CARB revised ILUC for LCFS from 30 g CO2e/MJ to 19.8 g CO2e/MJ with the newest GTAP
results showing 7.5 g CO2e/MJ.
The FASOM and FAPRI modeling system predict effects that are not tied to ethanol use and
should be corrected.
o The latest data and science demonstrate that deforestation rates occur due to many
factors and the supply and demand of agricultural products has little effect on this
phenomenon.
Co-product credit value of distillers’ grain solubles (DGS) is higher than anticipated due to:
o Greater emissions from the displacement of soybean meal;
o Higher nitrogen (N) application rate on soybeans than originally anticipated;
o Displacement of fossil CO2 in urea feed.
A high adoption rate of corn oil extraction has led to the rapid growth in use of corn oil as
biodiesel feedstock.
o The preferred use of corn oil is biodiesel; so, the appropriate co-product treatment for
50% of the corn oil is as an energy product via allocation.
1
Soybeans and other legumes assimilate nitrogen from the atmosphere into organic compounds through a process known
as fixation.
2
The GHG intensity of urea in the GREET model represents the life cycle emissions per ton of urea. The urea molecule
includes carbon that is derived from natural gas. When urea is used as fertilizer or animal feed, the carbon is metabolized
to produce CO2. GREET counts the field emissions for urea when used as fertilizer but omits the emissions when it is used as
a co-produce animal feed.
vii |
o Corn oil when used as a biodiesel feed displaces fats such as soy oil and palm oil which
have much higher indirect land use change (ILUC) values than corn oil when treated as
DGS mass.
Ethanol plants produce lower GHG emissions than estimated in the 2010 RIA due to:
o Elimination of coal for dry mill plants with natural gas;
o Lower carbon intensity for electric power used by ethanol plants;
o Use of biogas motivated by California low carbon fuel standard (LCFS) program;
o Ongoing efficiency improvements from many sources;
o Utilization of CO2 to displace fossil sources and CO2 sequestration.
2005 Petroleum baseline in the 2010 RIA is underestimated because the baseline fails to
adequately account for:
o Higher rates of methane venting and flaring from oil production;
o Mix of secondary oil recovery technologies and oil sands.
This study found that corn ethanol has resulted in greater GHG emission reductions compared to those
originally predicted in the 2010 RIA. The results for dry mill corn ethanol plants from this Study (Figure
S.1) are aligned with the approach in the 2010 RIA. The emissions are based on GREET calculations and
adjustments to reflect EPA’s categories with projections for energy use in 2022 developed in this study.
The emissions include allocation of half of the GHG emissions associated with corn oil to biodiesel.
Higher nitrogen application rates for soybean farming, which affect the DGS co-product credit as well
as fossil carbon displaced in urea feed are also considered in the analysis. The lower carbon intensity
of electric power compared to 2010 projections is reflected in fuel production emissions. The small
effect on rice methane and livestock emissions is based on the recent study funded by the U.S.
Department of Agriculture by ICF (Rosenfeld, 2018). These results compared with appropriate
adjustments to EPA’s 2005 baseline translate into about a 48% reduction in GHG emissions as shown in
Figure S.1.
viii |
Figure S.1. Life Cycle GHG Emissions from Dry Mill Corn Ethanol and 2005 Petroleum Gasoline.
1 |
1. INTRODUCTION
As part of EPA’s 2010 Regulatory Impact Analysis (2010 RIA) of the Renewable Fuel Standard
(RFS), it conducted a life cycle assessment (LCA) of the biofuels specified in RFS2 by accounting
for direct and indirect emissions for the year 2022. The 2010 RIA identified 11 emission sources
which capture the full life cycle GHG profile of corn ethanol and compared these emissions
with those of gasoline (Figure 1.1). The highest GHG emissions for corn ethanol correspond to
international land use change (LUC) followed by Fuel Production. International LUC
corresponds to the change in carbon associated with the growth of new crops outside the U.S.
EPA estimated that these emissions include the release of soil carbon and avoided carbon
storage from forest and pastureland when these lands are converted to cropland. The
landcover change is predicted with the FAPRI model and is combined with carbon stock factors
developed by Winrock International. Fuel production emissions include the emissions
associated with natural gas combustion as well as upstream natural gas and electric power.
International farm inputs and N2O correspond the crop farming activity required to make up
for changes in U.S. farm exports. The modeling system estimated the effect of expansion in
corn production.
Figure 1.1. EPA’s Analysis of Corn Ethanol GHG emissions. (EPA, 2010)
2 |
The objective of this study is to evaluate EPA’s analysis based on the availability of new data
and a better understanding of models and assumptions. This study focuses on emission
categories with the highest impacts such as international land use change and compares the
results of 2010 RIA with the new findings. Another key effect examined in the study is the
impact of co-product credits and different methods of allocation. The study includes the
following sections.
Sections 1.1 to 1.4 provides an introduction to corn ethanol life cycle GHG emissions.
Section 2 presents domestic and international land use change and their impacts on
corn ethanol carbon intensity.
Section 3 discusses farming inputs and the sensitivity analysis.
Section 4 presents the impact of different co-products and their allocation factors on
corn ethanol carbon intensity.
Section 5 describes technologies used in ethanol production and their advancements.
Section 6 analyzes the energy sources used in the fuel production stage.
Section 7 describes the GHG emissions related to various types of extraction of fossil
fuels and their projection.
Section 8 presents the results of this study and compares them with those of other
studies and EPA RIA.
Finally, Section 9 summarizes this Study’s conclusions.
1.1 Life Cycle GHG Analysis
The RFS2 and other biofuel policies around the world require GHG reduction targets relative to
the conventional fossil fuels. The GHG reduction is measured through life cycle assessments
(LCAs), which account for cradle-to-grave emissions (and/or other environmental impacts),
starting with raw material extraction and ending with fuel consumption. LCA is a technique
used to model the environmental impacts associated with the production of materials. LCA
models assess environmental impacts over a range of categories, including energy
consumption, GHG emissions, criteria air pollution, eutrophication, acidification, water use,
land use, and others. The analysis includes a full inventory of all the inputs and outputs involved
in a product’s life cycle. Determining life cycle emissions for all inputs requires an iterative
analysis of these components because some components of the life cycle of fuels depend on
inputs that are part of the LCA. The net GHG emissions are converted to a CO2-equivalent basis
and then normalized by the energy content of the fuel (e.g. g CO2e/MMBtu). This carbon
intensity (CI), when compared with the CI of petroleum fuels, provides a measure of the net
GHG reductions of renewable fuels.
In the case of corn ethanol, with the U.S. the largest producer of corn in the world, the
hypothesis is that diverting corn to biofuel feedstock reduces the supply of corn in food and
feed markets. This effect is realized through increases in the price of corn and other agricultural
commodities globally. In order to address the shift in corn supply, farmers across the globe
switch from other crops to corn (direct land use change) or convert grasslands, wetlands, or
3 |
forests which are carbon sinks to crop production. The conversion of land to cropland results in
indirect land use change (ILUC) emissions due to the release of carbon in the soil, above ground
biomass, and the foregone sequestration of CO2. Two different approaches covering the extent
of life cycle impacts are referred to as attributional and consequential LCA. Attributional LCA
(aLCA) focuses on the direct processes used to produce and consume a product while
consequential LCA (cLCA) examines the consequences of possible (future) changes between
alternative product systems (Brander et al., 2009). An aLCA identifies the direct energy inputs
and emissions associated with corn farming and ethanol production. A cLCA identifies the net
change in global emissions due to induced impacts of corn consumption, energy inputs for
ethanol plants, and ethanol use. The 2010 RIA is aimed at calculating cLCA emissions based on
the displacement effect of corn diverted to ethanol production.
1.2 Land Use Change
The correlation between LUC and an expansion in biofuel is typically estimated with agroeconomic models. Economic models that simulate market behavior (particularly those in the
agricultural sector) are often linked to predict the location of land cover change and the
emissions associated with conversion to crops as illustrated in Figure 1.2
Figure 1.2. Modeling Flow for Determination of Total Biofuel Lifecycle Carbon Intensity,
Including Both Direct and Indirect Effects.
1.3 Modeling Approaches
The system boundary defines the scope of activities and emissions associated with a life cycle
analysis. The inputs to the system and emission flows are counted in the analysis are defined in
a system boundary diagram (SBD). The system boundary identifies how far emissions are
tracked and the treatment of co-products.
4 |
1.3.1 Approach for Revised GHG Analysis
This study combines new data on corn ethanol production with the methods used by EPA in the
RIA to develop a revised estimate of the GHG emissions associated with corn ethanol.
Repeating the details of the modeling in the RIA is not practical due to the complexity of the
FASOM and FAPRI modeling systems. This study estimated the emission categories within the
2010 RIA methodology based on energy inputs and co-product yields thereby allowing for a
comparison with the 2010 RIA results.
The system boundary used in this study is shown in Figure 1.3. Ethanol and corn oil for biodiesel
are fuel products. Corn oil is also used as animal feed as modeled in the 2010 RIA but current
fuel policies favor the use of corn oil as a biodiesel feedstock. Fermentation CO2 is another
coproduct for many ethanol plants. This study compares data on corn production, ethanol
inputs and ethanol plant yields with those in the 2010 RIA and then estimates emissions for
each of the RIA categories based on the best available data. The effect of each of the coproducts on the net life cycle emissions is examined here.
Corn Farming Ethanol
Production
WDGS
DDGS
Syrup
Ethanol
Natural
Gas
Agriculture
Inputs
Corn
Transport
Ethanol
Transport
Corn
Soybean
Urea
Feed Fuel
Power iLUC
Corn Oil
Credit
Soybean
Oil
Corn Oil for
Biodiesel
CO2
CO2
Liquefaction
CO2
Transport
iLUC
Figure 1.3. System Boundary Diagram for Corn Ethanol Production.
1.4 Global Warming Potential
The global warming potential (GWP) represents GHG emissions based on their radiative forcing
and lifetime in the atmosphere on equivalent units of carbon dioxide (CO2). These factors are
estimated by the Intergovernmental Panel on Climate Change (IPCC) and updated in each IPCC
Assessment Report (AR). The 2010 RIA used the factors provided by the IPCC’s Second
5 |
Assessment Report (SAR), however, these factors have been updated since 2010 and the most
recent one is the Fifth Assessment Report (AR5) shown in Table 1.1 (IPCC, 2014). This study uses
the AR4 factors to calculate the CI of fuels since these values are currently adopted by the EPA
for calculations of the national GHG inventory.
Table 1.1. Global Warming Potential (100-year time horizon).
Greenhouse Gas SAR AR4 AR5
CO2 1 1 1
CH4 21 25 28
N2O 310 298 265
6 |
2. DOMESTIC AND INTERNATIONAL LAND USE CHANGE
Since 2010 when EPA conducted the RIA, new findings and data on actual deforestation across
the globe, crop prices, soil organic carbon stocks, corn and ethanol yields have shown that the
2010 RIA overestimated the contribution of LUC towards the CI of corn ethanol. The 2010 RIA’s
approach, as well as new studies on LUC, are discussed below. EPA’s approach to ILUC
modeling, improved ILUC estimates, and the estimates used in this study are discussed.
2.1 EPA RIA Approach for Land Use Change
The 2010 RIA takes into account the incremental change of diverting corn crops to biofuel
production. The modeling attempts to answer the question: what would change if U.S. ethanol
use increased to 15 billion-gallon per year3
while holding constant the consumption of food.
Both the incremental farming inputs as well as the incremental effects of land conversion on
crops were estimated through macroeconomic modeling.
2.1.1 EPA Modeling Approach
The system boundary used in 2010 RIA is shown in Figure 2.1. The analysis includes the direct
emissions associated with tailpipe emissions, fuel production, fuel and feedstock transport. The
carbon in fuel is treated on a carbon neutral basis with zero emissions associated with the short
cycle carbon in ethanol and ethanol plant fermentation emissions. The effects of the corn
feedstock are analyzed in a cLCA with estimates of the effects of an incremental increase in the
use of ethanol and consumption of corn. The modeling takes into account the direct farming
emissions in the U.S. and internationally as well as the effect on rice and livestock methane
emissions due to shifts in the production of agricultural products. The U.S. emissions are
predicted with the FASOM model and the international crop production is predicted with the
FAPRI model combined with emission factors for land cover change and agricultural inputs.
The 2010 RIA also includes the indirect farming emissions associated with new crops in addition
to LUC. This method is intended to represent the replacement crop inputs as well as land use
conversion.
3
EPA 2010 RIA, Section 1.1.1.1
7 |
Figure 2.1. System Boundary Used in EPA RIA Study. (EPA, 2010)
For the purposes of discussion in this study direct and indirect land use change are described
separately.4
Direct land use change refers to land already used for a specific purpose (e.g.
growing food) and whose future use will achieve the same result. For instance, in response to
an increase in production of corn ethanol, lands previously used for food production might be
converted to corn for fuel. On the other hand, indirect land use change refers to the land whose
ultimate purpose is essentially changed from its previous use (Farm Energy, 2019). For instance,
converting forests or grasslands to agricultural land is called indirect land use change. The 2010
RIA aggregated the impacts of direct and indirect land use change in the U.S. and called it
“domestic land use change.” Also, the RIA assumed international land use change occurred as a
result of domestic biofuel production expansion.
EPA used the Forestry and Agricultural Sector Optimization Model (FASOM), developed by
Texas A&M University and others, to estimate the changes in crop acres resulting from
increased biofuel production. FASOM is a partial equilibrium model of the forest, agriculture,
4
Some argue that all LUC is indirect since corn used for biofuel production is diverted from the overall U.S. corn
supply.
8 |
and livestock for the United States. The model tracks U.S. cropland by county and estimates
emissions associated with the conversion to cropland (i.e. domestic land use change). Within
the model, the linked agricultural and forestry sectors compete for a portion of the land within
the U.S. Prices for agricultural and forest sector commodities as well as land are endogenously
determined given demand functions and supply processes. The FASOM model maximizes the
net present value of the sum of consumers’ and producers’ surpluses (for each sector) with
producers’ surplus estimated as the net returns from forest and agricultural sector activities.
The GHG calculations are based on available data on inputs from crop budgets coupled with
estimates from EPA, the IPCC, and the DAYCENT model developed by Colorado State University.
The FASOM model also estimates the energy consumption, as well as fertilizer use, of crop
production. The projection of farm inputs by FASOM was used in 2010 RIA to calculate the GHG
emissions of corn ethanol in 2022. The model takes into account shifts among agricultural
production including changes in livestock population due to changes in corn prices. The
population provides the basis for estimating livestock methane emissions.
Since FASOM is only applicable for modeling the land use change within the U.S. (domestic
LUC), EPA employed the integrated Food and Agricultural Policy and Research Institute
international models, as maintained by the Center for Agricultural and Rural Development
(FAPRI-CARD) at Iowa State University (as summarized in CRC, 2014), to estimate the changes
in crop acres and livestock production by type and by country globally (international LUC) in
the 2010 RIA. While FAPRI-CARD models how much cropland will change, it does not predict
what type of lands such as forest or pasture will be converted. Therefore, EPA used Winrock
International’s data to estimate what land types are converted into cropland in each country
(EPA, 2010). EPA also used the GTAP model and confirmed that the GTAP results were
consistent with outputs of FASOM and FAPRI models. Since then, the GTAP model has
undergone several revisions, but EPA has not compared its findings with the new results from
the GTAP model.
FASOM also predicted that cultivation of corn increases the soil carbon storage while
conversion of cropland pasture and forestland leads to more GHG emissions. Overall, the
FASOM results showed that expanding corn cultivation resulted in carbon storage (negative
value for domestic LUC). However, the results from FAPRI showed that production of 15 billion
gallons of corn ethanol reduced the corn export from the U.S. which causes other countries to
allocate more lands to corn cultivation and subsequently convert more pasture and forestland
to corn farms which leads to more GHG emissions. Conversion of Brazilian forests to corn
farming had the highest share from total emissions associated with international LUC under
the methodology used in the 2010 RIA.
9 |
2.1.2 Challenges with 2010 RIA Land Use Change Analysis
While the direct emissions from ethanol production vary among the studies, the table below
shows the large variability in estimates which are largely due to LUC. Early studies employed
worldwide agricultural models to estimate emissions from land use change (Searchinger et
al.,2008; Searchinger, et al., 2015; Fargione et al., 2008) with higher net GHG emissions for corn
ethanol compared to gasoline.
More recent studies, (Hertel et al., 2010) found that the emissions associated with land use
change were less than one-third of those projected by Searchinger (2008) and even smaller
values of land use change effect were reported by Tyner et al. (2010). The inconsistency in
indirect land use change predictions is mainly due to the differences in methods and
assumptions. Key factors include elasticity factors that affect the selection of land cover change
and carbon stocks. Further, some argue the modeled predictions of indirect land use change
are not meaningful because there is not a causal relationship between biofuel use and land
conversion (Zilberman et al., 2010). In the 2010 RIA, conversion of Brazilian forestland to corn
farm had a significant contribution to the international LUC. However, new studies found that
agricultural intensification and governmental policies and regulations have had a great impact
on GHG emissions reduction as well as decreasing the deforestation in Brazil (Silva et al., 2018;
Garrett et al., 2018). Brazil, for example, is seeking to reduce greenhouse gas (GHG) emissions
by 37% below 2005 levels by 2025 and 43% by 2030 through its announced Nationally
Determined Contribution (NDC). The role of agricultural intensification in response to increasing
commodity prices was not fully considered in the 2010 RIA and therefore international LUC was
over-estimated (Rosenfeld et al., 2018).
10 |
Table 2.1. Addressing Uncertainties in LUC Assessments
LCA parameter Uncertainty Recommendation
New LUC studies
estimate lower
emissions
associated with
international LUC.
LUC estimates vary
greatly with model,
structure,
assumptions and
target year.
While it is true that LUC modeling is greatly
based on assumptions and model structure, we
believe that after 10 years, with the availability
of new data, we can see that most of those
assumptions were not realistic. The current
rate of deforestations, yield price elasticity,
type of land being converted, etc. are not close
to what EPA projected in 2010.
Soil C sequestration
of corn is higher
than what assumed
in 2010 RIA.
The SOC data resulted
from recent studies
are inconclusive due
to variation between
studies and
dependence on
experiment duration.
Recent long-term studies on SOC in Midwest
such as Poffenbarger et al. (2017) shows that
corn farming results in a significant increase in
SOC storage. Various practices such as notillage and optimum fertilization increases the
SOC storage and more farmers are applying
these practices now.
2.2 New Findings on Land Use Change
The emissions associated with LUC include the net accumulation of carbon, taking into account
both the carbon release from land conversion and the foregone carbon sequestration. Figure
2.2 shows a simplified breakdown of the factors that affect the LUC presented by the CARB and
modeled in GTAP. The significant differences between the GTAP modeling and the
FASOM/FAPRI modeling include the carbon stock factors for released carbon as well as the
regional detail for crop shifting. GTAP, for example, takes into account prior trade history
between countries. All agro-economic models solve for prices that result in a supply and
demand equilibrium. GTAP is a general equilibrium model that includes all sectors of the
economy. FASOM and FAPRI are models including only agriculture and, in the case of FASOM,
forestry. Those models are more detailed on individual agricultural commodities. All of the
models project changes in land cover and predict changes in carbon stock through different
carbon accounting mechanisms and carbon stock data sets. All of the modeling systems need
to allocate emissions over time as they are predicting an initial “shock” of biofuel demand that
is distributed over a period of biofuel production.
11 |
Figure 2.2. Approaches to LUC Modeling.
(CARB, 2018)
While the modeling represents the inputs to the GTAP system, the basic principles are the same
for all LUC models. Improving crop yields, production of co-products, and high carbon stocks for
converted lands reduce LUC emissions. The recent key findings for corn ethanol affecting LUC
with GTAP have been:
Low conversion of land in the U.S.;
Increase in soil carbon storage due to corn farming practices;
Overall decline in deforestation rates globally;
High substitute value of Distillers’ grain solubles (DGS) as feed;
Increased cattle stock rate with pasture intensification;
Corn oil producing biodiesel increases overall fuel output.
Since an acre of land producing corn for ethanol produces as much animal feed (i.e. DGS) as an
acre of soybeans (soybean meal), the net LUC emissions in recent studies by ANL (Dunn, 2017),
which are below 10 g CO2e/MJ appear reasonable.
2.2.1 CCLUB and GTAP
LUC models also predict changing yields, both to the biofuel crop being examined as well as
other crops grown globally. These yield improvements include both projected future
improvements due to better farming practices (some of which may have nothing to do with an
expansion in biofuels), as well as yield improvements that are due to higher prices sending a
signal to the market to incentivize better farming practices, more efficient harvest, and
technology improvements. Expanded use of crops for biofuels will also affect feed prices and
shift the use of agricultural commodities. The production of DGS from corn affects feed
markets. The removal of land from feed production will also result in market shifts due to price
mediation. Higher corn prices, for example, could result in a shift from feedlot-fed cattle to
other sources of meat that are less feed intensive. The effect of displacement by DGS as well as
12 |
shifts in crop usage may be the most significant factor. Demand mediation or a reduction in the
demand for feed and food also reduces the overall requirement for land. Another key LUC
prediction is associated with cattle stocking rates on pasture as well as the selection of forest
land, marginal land or grassland. These predictions affect the carbon stock factor for LUC.
2.2.2 Other Corn Ethanol Studies
Two studies conducted by ICF for the U.S. Department of Agriculture (USDA) examined the
2010 RIA. Each study calculated the CI of corn ethanol under different scenarios (Flugge et al.,
2017; Rosenfeld et al., 2018). The studies investigated domestic and international land use
change based on recent studies and models and concluded that both domestic and
international land-use change emissions for corn ethanol are lower than those in the 2010 RIA.
Moreover, their estimates of GHG emissions of fuel production stage as well as tailpipe were
also lower than those in the RIA.
CARB has revised its estimation of international LUC (CARB, 2015) due mainly to using a newer
version of GTAP with an updated database, re-estimating energy sector demand and supply
elasticity values, the addition of cropland pasture to the U.S. and Brazil, improved treatment of
corn ethanol co-product (DGS), improved treatment of soy meal, soy oil, and soy biodiesel,
improved estimation of crop yield across the world, improved estimation of emissions factors,
and revision of demand and yield responses to price, among other things. The reduction in
estimated forest conversion is an important factor since the GHG emissions associated with
conversion of forest is significant.
Argonne National Laboratory (ANL) and California Air Resource Board (CARB) developed GREET
and CA-GREET models, respectively, which include the LCA for corn ethanol. CARB’s estimates
of ILUC have dropped from 30 g CO2e/MJ to 19.8 g CO2e/MJ based on refinements in modeling
(Tyner, 2010) and the changing CI of ethanol in Table 2.2 reflects both the ILUC and mix of fuel
production technologies. CARB’s original modeling with GTAP assumed a 1:1 displacement of
DGS with corn, but that has since been revised. Subsequent modeling has also taken into
account the displacement of other agricultural products.
13 |
Table 2.2. Life Cycle Studies Examining Corn Ethanol.
Year Study Model/ Database ILUC CI (gCO2e/MJ)
2008 Searchinger et al. (2008) FAPRI-CARD/GREET 100
2009 CARB CA-GREET.8b/GTAP 30
2010 EPA RIA GREET/FASOM/FAPRI 28
2018 ANL CCLUB/GTAP/GREET 3.9 to 7.5
2017 Flugge et al. (2017) FASOM/ FAPRI 8 to 14
2018 Rosenfeld et al. (2018) GREET/IPCC/GTAP 7 to 14
2014 CARBa CA-GREET2/GTAP 19.8
2021 Scully (2021) Review of Models 3.9
a
Average of approved pathways.
These models however, look backward at prior data crop expansion, yield, and land use data.
Ten years of increased biofuel production in the United States allows for a revised assessment
of the assumptions and results of the 2010 RIA.
2.2.3 Empirical Data
Showing the effects of LUC is challenging since the effect occurs even absent biofuel
production. No experiment can prove the “counterfactual” effect of land use change absent
biofuel production. However, significant empirical data suggests that the relationship between
crops used for biofuel production and land use change may not be as significant as predicted in
the 2010 RIA. Deforestation rates have declined in the past decade and farming practices
continue to store carbon in the soil. In fact, the drivers for deforestation are not directly
related to crop production (Zilberman, 2017).
The international LUC effect related to the conversion of Brazil’s Amazon region was significant
in the 2010 RIA, however, this anticipated relationship was not borne out in reality. When
comparing the deforestation in Brazil and corn ethanol production in the U.S. from 2004 to
2015, we can see that not only did U.S. corn ethanol production not cause an increase in
deforestation in Brazil but annual deforestation rates in Brazil’s Amazon region actually
decreased over 75 percent over that decade (Figure 2.3). These trends in forestry loss are
decoupled from biofuel use and this lack of correlation is not, but should be, incorporated into
EPA’s analysis.
14 |
Figure 2.3. Comparison of Brazilian Deforestation and U.S. Corn Ethanol Production.
(Rosenfeld et al., 2018)
Moreover, several studies have shown that corn crops produce large amounts of high carbon
root and residue and this has a major positive impact on soil carbon stocks (ACE, 2018). Figure
2.5 implies that the organic matter content of the soil has improved over time due to corn
farming. Part of domestic LUC is the carbon stock change due to crop cultivation and based on
Figure 2.5, the carbon stock due to corn cultivation is improving which leads to more GHG
emissions saving and lower impact of domestic LUC. Clay et al. (2012) studied the impact of
corn yield on soil carbon sequestration and reported that in many regions, surface soils are
carbon sinks when seeded with corn.
The issue of soil carbon storage is illustrated in comments in the literature regarding LUC
modeling. The authors of critiques of CCLUB, which represents the newest ILUC analysis from
GTAP, (Malins, 2020) argue that the Winrock data for domestic crop conversion is more
accurate (which is an option to utilize in GTAP). This is not a defensible position. Much of the
debate around LUC estimates as presented in GTAP pertains to the use of emission factors
associated with soil carbon release. CCLUB uses the CENTURY emission factors as defaults with
Winrock data used by default for international emissions. Figure 2.4 shows the comparison of
different emission factors, which support the argument that the higher Winrock emission
factors for domestic ILUC would be an appropriate estimate; however, this argument is
inconsistent with EPA’s GHG accounting as used in the U.S. GHG inventory, which uses FASOM.
Shifting to greater corn production from other crops along with the deployment of low carbon
farming practices stores carbon, as reflected in FASOM and CCLUB. Accordingly, criticisms of
the more recent versions of GTAP are misplaced; the LUC emissions in the U.S. should be
negative as shown in the 2010 RIA (which utilizes FASOM) and in CCLUB.
15 |
Figure 2.4. Carbon loss following cropland pasture conversion using Winrock, CENTURY and
AEZ-EF emission factor models. (Malins, et al., 2020).
Figure 2.5. South Dakota Top Soil Organic Matter. (ACE, 2018)
2.2.4 Modeling Results
Since 2010, numerous studies have examined the international LUC for corn ethanol and their
results showed that the international LUC was significantly lower than the 2010 RIA’s
estimation (Figure 2.6). These emissions correspond to the land cover change outside the U.S.
induced by a change to corn ethanol. Typically, agro-economic models predict a reduction in
U.S. crop exports for both corn and soybean as either corn exports are reduced or corn-soy
rotation is converted to continuous corn. The models take into account the price effects of
16 |
agricultural commodities as well as yield improvements and predict the type of land converted
to crop production. The initial ILUC estimate from the California Air Resources Board (CARB,
2009), was a total of 30 g CO2e/MJ of which about half was international LUC (see Figure 2.6).
CARB revised its ILUC analysis with a total international component of 15 g CO2e/MJ. These
values are roughly comparable to the EPA international LUC result in Figure 2.5 though the
2010 RIA analysis includes additional categories. A series of peer-reviewed publications have
shown that the international LUC is even lower. Publications from Purdue University (Tyner et
al., 2010; Taheripour et al., 2017) are based on the GTAP model; which was employed by
Argonne National Laboratories and incorporated into GREET (the model used by CARB and
other state Low Carbon Fuel Standards, such as Oregon’s Clean Fuels Program).
As discussed earlier, several studies based on GTAP evaluated biofuels induced land use
changes and GHG emissions. Tyner et al. (2010) estimated the land use change and emissions
associated with corn ethanol production using GTAP in support of the LCFS with the newer
analysis resulting in lower ILUC emissions. A more recent study (Taheripour et al., 2017)
incorporated a newer database (2011 database instead of 2004 database), added an
intensification option to the model, and updated the yield price elasticity based on new data
from the Food and Agriculture Organization (FAO). As Taheripour et al. (2017) stated, the
previous versions of the GTAP model did not account for the intensification of pasture and
assumed that a change in the harvested area equals a change in land cover, thus overestimating
the emissions associated with ILUC.
Figure 2.6. International Land Use Change Estimated by Several Studies.
(Rosenfeld et al., 2018; ANL, 2018)
17 |
2.2.5 Summary of LUC Effects
International LUC for corn ethanol CI was overestimated in the 2010 RIA as shown by recent
studies, availability of more recent data, and more realistic assumptions. Any estimation of LUC
involves significant uncertainty with the largest uncertainties associated with the yield
predictions on new and marginal land as well as the selection of land cover type. Shifts among
agricultural commodities further complicates the analysis and adds a level of opacity to the
modeling (CRC, 2014). While the results of LUC modeling are intrinsically uncertain,
improvements in models such as those documented in recent GTAP studies indicate that EPA’s
assessment of both international LUC as well as U.S. LUC are overstated. In fact, soil carbon
storage effects from corn farming should lead to a negative LUC in the U.S.
While the study by Searchinger et al. (2008) was the basis of international LUC calculation in the
2010 RIA, Zilberman (2017) has recently evaluated the assumptions made by Searchinger et al.
(2008) and concluded that “Searchinger et al. (2008) results may now be seen as fundamentally
flawed not just because the ILUC is uncertain and estimates vary considerably, but also because
it fails to capture the basic features of agricultural industries and land resources.” Dumortier et
al. (2011) employed the same model used by Searchinger et al. (2008), but used more realistic
assumptions and obtained completely different results (lower emissions). Rosenfeld et al.
(2018) used the simulation results of the 2013 GTAP-BIO model available in ANL’s CCLUB tool to
calculate the impact of international LUC on corn ethanol CI under several scenarios and
reported that the emissions associated with international LUC ranged from 1.3 to 16.9 g
CO2e/MJ. These findings that elasticity factors and other contributors to ILUC were overstated
by the 2010 RIA were confirmed in a recent paper by Scully, et al. (2021). Finally, studies that
compare ILUC modeling place a strong emphasis on Winrock land use conversion factors where
a critical assumption is that crop land pasture emission rates are half those of pasture
conversion (Malins, 2020). These same studies criticize the overestimation of soil carbon
storage from ongoing corn farming practices predicted by CENTURY. However, the studies fail
to recognize the merits of FASOM’s analysis as used in the U.S. emission inventory that reflects
real-world soil carbon storage effects.
Modeling Approach for This Study
This study combines the elements of several approaches to provide an updated assessment of
the GHG intensity of corn ethanol. Repeating the steps in the 2010 RIA is a challenging process
and EPA acknowledges this issue in the 2021 draft RIA; however, there are reasonable ways to
update corn ethanol’s CI without undertaking the extensive modeling effort completed in 2010.
Here, domestic and international LUC were calculated based on the GREET (2021) model
adjusted for the corn oil to biodiesel yield as shown in Table 2.3. The domestic and
international ILUC emissions are multiplied by an allocation factor that assigns half of the
emissions associated with corn oil production to biodiesel. The GREET model uses CCLUB (Dunn
et al., 2017) to estimate the soil organic carbon storage as well as land conversion and
associated emissions in response to biofuel expansion. Domestic LUC is based on average tillage
practice in the U.S.; however, the more no-tillage practice is used by corn farmers, the more
carbon will be stored in the soil and thus the impact of LUC will reduce.
18 |
Table 2.3. Change in GHG Emissions Due to Land Use Change (g CO2e/MMBtu).
Study Domestic International
EPA 2010 RIA -4,033 31,797
Rosenfeld et al. (2018) -2,038 9,082
GREET1_2020 -2,314 6,300
GREET1_ 2020, allocated to corn oil -2,199 5,986
The following calculation approach was used in this study. It allows for the assessment of the
newest corn farming data, addition of the GTAP analysis for ILUC, and inclusion of the original
2010 RIA emission categories.
Emissions Allocated to Corn Ethanol and Corn Oil by Energy Content
Domestic ILUC: CCLUB
International ILUC: CCLUB
Domestic Rice Methane: ICF 2018
Domestic Farm Inputs: GREET minus international fertilizer
International fertilizer: ICF 2018 (to align with RFS categories, subtracted from domestic farm
International Rice Methane: ICF 2018
Emissions Assigned to Corn Ethanol
Tailpipe: ICF 2018
Fuel Production: GREET
19 |
3. CORN FARMING
The consumption of farming inputs such as fertilizers, pesticides, and energy such as diesel and
LPG affect the GHG intensity of corn or crops that are grown to make up for corn used for
biofuel production. Crop yields yield affect both the land required for crop production and LUC.
This section includes new data on corn yield as well as crop inputs. This section also reviews
recent data on farming and aligns it with the estimates in the 2010 RIA and the current GREET
model.
3.1 Corn Farming
Historical data on corn yield indicates that the yield has increased steadily over time, from 85
bu/ac in 1988 to 172 bu/ac in 2020 as shown in Figure 3.1. The adoption of double-cross hybrid
corn, continued improvement in crop genetics, adoption of N fertilizer and pesticides, and
agricultural mechanization resulted in a steady increase of corn yield in the U.S. (Nielsen, 2017).
Aside from the steady increase of corn yield, the harvested area of corn has increased over
time. Due to the continuous improvement of corn yield, the production quantity has an upward
trend (USDA NASS, 2018). The 2010 RIA estimated the corn yield for 2022 as 185 bu/ac, based
on past 30 years of corn yields from USDA database. EPA’s projection of corn yield for 2022 is
consistent with the trendline of current data in Figure 3.1.
Figure 3.1. Corn Yield Over Time. (USDA NASS, 2020)
Management practices such as tillage, and nitrogen (N) application rate affect the GHG
intensity of crops. In order to decrease the environmental footprint and lower production costs,
farmers have started using new technologies such as precision agriculture to manage their
fertilizer consumption. Reduced tillage has become a common practice across the U.S. farms,
20 |
reduces soil emissions during the farming stage (Figure 3.2). Nitrogen inhibitors reduce the
requirement for nitrogen and also reduce the formation of N2O. Precision farming and guidance
methods also allow for the more efficient application of nitrogen. The combination of all of
these methods results in increased yield per acre and reduced nitrogen per bushel.
Figure 3.2. Changes in Corn Production Practices from 2005 to 2010. (Rosenfeld et al., 2018)
The leading corn farming states in the U.S. produce most of the ethanol in the country as shown
in Figure 3.3. The location of ethanol plants is not surprisingly coincident with corn production.
This co-location reduces corn transport distance and growth in corn production is occurring in
the states with the highest yield per acre, which is shown in Figure 3.4.
Figure 3.3. Corn Ethanol Production by State. (USDA NASS, 2018)
0
500
1,000
1,500
2,000
2,500
3,000
3,500
4,000
4,500
5,000
Iowa
Illinois
Nebraska
Minnesota
Indiana
South Dakota
Kansas
Ohio
Wisconsin
Missouri
North Dakota
Michigan
Kentucky
Texas
Colorado
Pennsylvania
Tennessee
Arkansas
New York
North Carolina
Mississippi
Louisiana
Maryland
Virginia
Georgia
Alabama
Oklahoma
Sourth Carolina
Idaho
Delaware
Washington
California
Arizona
Oregon
Million Bu / Million gal/y
Corn Production
Ethanol Production
Corn grain production data from USDA
NASS Quick Stats – 2018 Survey Data
Ethanol production capacity by state
from Ethanol Producer Magazine
14 States Produce
91% of Corn Grain
92% of Corn Ethanol
21 |
Figure 3.4. Average Corn Grain Yield vs. Production for 14 States.
(2014-2018 Weighted Average) (USDA NASS, 2018)
Iowa, Illinois, and Nebraska are the three states with the highest corn production in the U.S.
An analysis of NASS data for applied nitrogen and corn yield shows consistent reduction in the
nitrogen application rate per bushel of corn (Figure 3.5). The reduction in nitrogen application
rate is consistent with the 2010 RIA estimate discussed below.
Figure 3.5. Nitrogen Fertilizer Use Rate in the Three Largest Corn Producer States.
(USDA NASS, 2018)
22 |
Domestic agricultural use of fertilizers, pesticides, and energy was projected by FASOM in the
2010 RIA. The 2022 projections are compared to several evaluations of NASS data in Table 3.1.
The 2010 RIA used the GREET interim emission results to calculate the upstream emissions
associated with agricultural inputs. The 2022 projections for farming inputs in the RIA reflect
improved yields and advancements in farming techniques, which, in some cases, may not have
yet been achieved. Overall, this comprises a small portion of ethanol’s CI relative to the LUC
portion discussed above.
Table 3.1. Farming Inputs of Corn in the U.S.
Input Unit GREET
(2021)
Rosenfeld et
al. (2018)
USDA NASS
(2018)
EPA
RIAd
Analysis Year 2020 2015 2016 2022
N g/bu 401.5 373 380 344
P2O5 g/bu 150.6 128 165 79
K2O g/bu 152.3 130 193 98
Lime g/bu 1,457 1,150 N/Ac 260
Herbicide g/bu 6 6 3 5
Pesticide g/bu 0.01 0.1 N/A 1
Diesel Btu/bu 5,200 4,730b 6,388 9967
Gasoline Btu/bu 802 1,413 774 1042
Electricity Btu/bu 1,326 441 1,089 19
Natural Gas Btu/bu 479 1,301 1,212 1283
LPGa Btu/bu 1,026 1,723 1,297 –
a
Liquified Petroleum Gas
b The energy usage of corn ethanol was not mentioned in Rosenfeld et al. (2018), however they mentioned that
they obtained the data from GREET (2015). To make it comparable, the energy usage data for Rosenfeld et al.
(2018) were obtained directly from GREET (2015).
C
Data was not available.
d
From EPA RIA, Table 2.4-5. The values are listed per MMBtu of ethanol which appear to incorrectly labeled and
not possible. If for example, the N fertilizer of 138.8 lb/MMBtu are taken as lb/acre yield and combined with a corn
yield of 183 bu/ac from the RIA the N rate is 344 g/bu.
GREET1_2021 is the study input
3.2 Sensitivity Analysis of Farm Inputs
A sensitivity analysis was conducted to investigate the impact of each input on overall CI of corn
and the results are shown in Figure 3.6. Fertilizer application rates, farm yields, transport
distances to ethanol plants, and N2O production rates were examined for 12 corn farming
states using the GNOC model,5
which provides an easy-to-use assessment tool with global
applicability. Uncertainty distribution functions were developed based on the standard
deviation of historical data and other variability factors to provide inputs for a Crystal Ball™
simulation of the GHG intensity of corn. The analysis shows that nitrogen fertilizer and N2O
5 http://gnoc.jrc.ec.europa.eu/
23 |
emission are the most sensitive inputs, implying that a reduction in nitrogen fertilizer
application rate significantly decreases the GHG intensity of corn and the CI of corn ethanol.
Figure 3.6. Sensitivity Analysis of Farm Inputs.
24 |
4. IMPACT OF CO-PRODUCTS ON CORN ETHANOL CI
The corn farming system and ethanol production generate several co-products that were
considered in the 2010 RIA. These include DGS, Corn Distillers’ Oil (CDO), and stover that is
harvested with corn. Stover was considered as a fuel feedstock and not animal feed co-product.
The effect of these co-products on GHG emissions is discussed in the following sections. Some
ethanol plants also capture fermentation CO2.
The 2010 RIA also used the study published by Argonne National Laboratory to estimate the
DGS replacement rates for corn and soybean meal in animal feed. Production of DGS effectively
results in a credit since DGS is a suitable source of animal feed and displaces agricultural crops
like corn and soybean meal. However, since FASOM takes the production and use of DGS into
account, no further allocation (displacement) was conducted in the 2010 RIA.
4.1 DGS Co-Product
Distiller’s grains are the nutrient-rich co-product of the ethanol production process and provide
an alternative to corn and soybean meal feed. Wet distiller’s grains are sold to local markets
due to their high moisture content and low shelf life. But generally, the distiller’s grains are
dried to increase the shelf life and facilitate transportation over longer distances. The product is
referred to as Distiller’s Dried Grains with Solubles (DDGS) (Iowa Corn, 2019). In the U.S.,
ethanol plants have the capacity to produce substantially more than 15 billion gallons of
ethanol and 44 million metric tonnes of DDGS (U.S. Grain Council, 2018). This effect is
significant since an acre of land producing ethanol for corn produces as much feed as an acre of
soybeans. Due to its nutritional value, DDGS is considered a good substitute for soybean and
canola meal. A recent study has investigated the effect of DDGS vs. soybean meal and canola
cake on feed intake, milk production, and milk quality in dairy cows and concluded that DDGS
can substitute for a soybean-canola mixture without affecting feed intake, milk yield, and
quality, or sensory quality (Gaillard et al., 2017).
Figure 4.1 shows the prices of DDGS and soybean meal over time with a correlation in price
activity. Rises in soybean meal prices are followed by rises with DDGS prices supporting the
substitution effect. The replacement value of DGS was less well-understood in 2010 when corn
ethanol was a less mature technology. While the overall substitution effects are more
complicated, DDGS that displaces soybean meal results in the avoidance of emissions from
soybean farming.
25 |
Figure 4.1. Historical Prices of DDGS and Soybean Meal. (USDA ERS, 2018; World Bank, 2018)
Soybeans as legumes fix nitrogen in the soil, which provides nitrogen for soybean crop and the
following crop which is typically corn. Thus, the application of nitrogen fertilizer is not required
for soybean farming. However, without N fertilizer, the soybean yield is limited to 50 to 60
bu/ac. In order to achieve higher yields, 30 to 60 lb/ac of nitrogen fertilizer is required (Schmidt,
2016). In recent years, more fertilizers, especially nitrogen fertilizer, have been used in soybean
farming to increase yields (McGrath et al., 2013) (Schmidt, 2016). The GREET model input for
soybean farming (ANL, 2018) is 48 g/bu of nitrogen fertilizer, which is based on a 2008 study
(Huo et al., 2008). However, recent USDA data indicates that the consumption of nitrogen
fertilizer in soybean is 18 lb/ac which translates to 166 g/bu (USDA NASS, 2018). The application
of nitrogen fertilizer on soybean crop is triple the GREET input, which directly affects the
emissions related to soybean production.
Since DDGS is a substitute for soybean meal, the avoided emissions are substantially higher
than originally anticipated. Correcting the nitrogen fertilizer use for soybeans allows for a better
estimate of the displacement value of DDGS with corn ethanol production. The FASOM model
estimate for nitrogen usage in soybean farming in 2022 in the 2010 RIA appears to be less than
10 lb/ac (Figure A.1) with a projected soybean yield as 50 bu/ac in 2022. These parameters
correspond to a nitrogen application rate of 64 g/bu, which is much lower than current nitrogen
fertilizer use rate reported by USDA NASS (2018) (166 g/bu). (See Appendix A for a discussion of
nitrogen application)
By comparing the nutritional value and moisture content, one lb of DGS is equivalent to 0.781
lb and 0.307 lb of feed corn and soybean meal, respectively. Therefore, one lb of DGS
production results in the displacement of 118 g CO2e plus 96 g CO2e if replaced for soybean
meal and corn (Table 4.1).
26 |
Table 4.1. The CI of DGS Using Displacement Method.
Feed Material Soybean
Meala
Corn Total
CI (g CO2e/g)
Production 0.53 0.24
ILUC 0.32 0.03
Total 0.85 0.27
Displacement Ratio 0.307 0.781
g CO2e/lb DGS 118.4 95.7 214.1
g CO2e/MMBtu EtOH 7,694 6,216 13,910
aThe co-product credit for DGS depends on the crops that it displaces. In order to assess ILUC based
on Figure 1.3, the displacement effect of corn to DGS is already taking into account in ILUC modeling
in GREET with 5.0 lb DGS, dry basis per gal ethanol. However, the higher ILUC of soybean meal has
not been fully taking into account due to the new market introduction of DGS. The displacement
effect of urea feed is now shown here.
4.2 Corn Distillers Oil
Another important co-product of the ethanol plant is corn distillers’ oil (CDO). Since 2010, corn
oil extraction has become a common practice in bioethanol plants due to technological
advancements, although it requires additional investment (Batres-Marquez, 2018). In the U.S.,
almost 85% of dry grind ethanol plants extracted corn oil in 2015, producing about 1.22 million
metric tons of CDO (Veljković et al., 2018), and the extraction of CDO has continued to grow
(Figure 4.2), which is consistent with the projections in the 2010 RIA. Several studies have
shown that CDO has comparable properties to diesel and is used for biodiesel production
(Balamurugan et al., 2018; Kumar and Kumar, 2013).
In the U.S., CDO represented the fastest expanding oily feedstock for biodiesel production in
2013 (Grooms, 2014). The California LCFS originally had a very favorable CI for biodiesel
produced using CDO as feedstock.6
This drove increased use of CDO as feedstock. In 2018,
about 2,060 million lb of CDO, or 50% or production, was used from biodiesel production based
on EIA statistics.
6
The LCFS CI was 4 g CO2e/MJ of biodiesel for several years. This value has since been raised to about 22 g
CO2e/MJ, but the low initial value provided an incentive to use CDO as a biodiesel feedstock.
27 |
Figure 4.2. Corn Distillers’ Oil Production in the U.S. (USDA NASS, 2018; RFA, 2019)
4.2.1 Corn Oil as Coproduct of Ethanol Production in EPA RIA
EPA estimated that by 2022, 70% of dry mill ethanol plants will conduct extraction, 20% will
conduct fractionation, and 10% will not extract CDO. These estimates were incorporated into
the FASOM and FAPRI/CARD models to account for extracted corn oil as biodiesel feedstock.
The 2010 RIA projected that by 2022, 680 Mgal or 4000 million lb of CDO is produced as a byproduct of corn ethanol production and used to produce biodiesel. The RIA analyzed the
displacement of CDO with other agricultural products such as soy oil in the FASOM model. If
CDO were treated as a fuel product, it would receive a greater share of the ethanol plant
emissions and the ethanol plant emissions would be reduced. In practice, about half the CDO is
used as biodiesel; which means that a corn ethanol biorefinery produces two energy products
and the emissions and ILUC should be allocated between ethanol and CDO for biodiesel.
4.2.2 CDO Under Various Allocation Methods
Since CDO is a co-product of ethanol production, emissions from corn farming and ethanol
production should be allocated to CDO or treated as a displacement credit. Several allocation
methods allow for the treatment of CDO including displacement with soybean oil, and diesel, or
energy allocation with ethanol and DGS. Each allocation method results in a different effect on
the CI of corn ethanol shown in Table 4.2 as the estimated reduction in ethanol CI due to CDO
production. Although the RIA accounted for CDO using the FASOM model, which focuses on the
displacement of agricultural products, the energy allocation method is a better choice since
corn oil us for biodiesel production has expanded in recent years. The effect of the different
allocation approaches is shown in Table 4.2, energy allocation method results in more
reduction in CI of corn ethanol than displacing with soybean oil. While displacing CDO with
diesel is an extreme case, biodiesel from corn oil is an alternative for diesel fuel, so displacing
with diesel is an option. EPA should factor into its analysis the fuel value of CDO. Energy inputs
28 |
and emissions for ethanol plants as well as ILUC associated with corn usage should be assigned
to both ethanol and CDO.
Table 4.2. The Effect of Displacement Method of CDO on CI of Corn Ethanol.
Modeling Approach CI (g CO2e/MJ Ethanol)
EPA RIA ~-1.14
CDO displacing with soybean oila
-1.20
CDO displacing with diesel – 4.94
Energy Allocation -2.12
a
Based on 166 g/bu of nitrogen fertilizer.
4.3 Replacement Feed
Corn stover (cobs and residue) is an important part of the life cycle of corn, either as fuel or as
animal feed, but most LCA models treat them separately from starch ethanol (Welshans, 2014;
Mueller, 2015). Corn stover is used as a cellulosic feedstock for ethanol production. Corn stover
can also be used as a replacement for corn and hay or corn silage in animal feed. Mueller et al.
(2015) conducted a study to investigate the effect of corn stover removal on overall emissions
of ethanol. The analysis included a displacement credit for the 30% corn stover used as corn
replacement feed (CRF) as well as the DGS produced from the grain corn. The displacement
credit for CRF is based on a substitution ratio of 0.5 kg corn and 0.5 kg hay being equivalent to
1.0 kg of CRF on a dry matter basis. Although CRF is a suitable substitute for feed ingredients
such as corn and hay, it requires pretreatment which involves consumption of chemicals such
as calcium hydroxide. On the other hand, CRF has a feed and LUC credit. The results showed
that using corn stover as animal feed has a co-product credit of -6.6 g CO2e/MJ which
potentially reduced the corn ethanol CI. The extent of CRF was not explicitly modeled by EPA in
the 2010 RIA, but should be considered by EPA in reassessing the CI of corn ethanol.
29 |
5. BIOREFINERY TECHNOLOGIES
The performance of biorefineries affects life cycle GHG emissions due to the use of feedstock
and fuel resources as well as chemical inputs. The key factors affecting GHG emissions for dry
mill ethanol plants are shown in Table 5.1. The future energy inputs and yield for ethanol plants
were examined in the 2010 RIA. Many of the technologies that affect dry mill ethanol plants
were identified. The factors that affect energy inputs and yields, as well as the differences
between the performance projected in the RIA and actual performance are examined here.
Table 5.1. Ethanol Plant Performance Parameters.
Performance Trend Key Drivers Effect on LCA
Increased Yield
Starch hydrolysis and
fermentation efficiency
Cellulosic conversion
Higher yield reduces corn upstream
emissions and ILUC as well as DGS
mass and co-product credit.
Reduced Natural Gas
Consumption
Reduced drying energy, plant heat
integration, corn oil extraction,
advanced separation processes
Natural gas combustion and
upstream emissions are
proportional to use rate.
Reduced Electric Power
Consumption
Ongoing improvements in
efficiency and yield and
cogeneration reduce power
requirement. Corn oil separation
requires additional electrical
power.
Power generation and upstream
emissions are proportional to use
rate.
Increased Corn Oil
Production
Corn oil in DGS is extracted by
centrifuge or with solvents.
Several approaches. Substitution
for agricultural products or
allocation.
Reduced DGS Mass
Increased ethanol and corn oil
yield reduce starch and oil
component of DGS without
changing protein output.
Affects co-product credit. Protein
content is not affected. Only
carbohydrate and fat fractions are
affected by yield improvements.
Reduced Chemical
Consumption
Increased yield and improved
monitoring.
Reduced upstream life cycle for
chemical production.
CO2 Capture
Growth in CO2 capture from
ethanol plants which have a pure
CO2 stream. Avoids CO2
production from other sources.
Several possible approaches, none
used in RIA. Credit or allocation for
CO2 storage/ productive use.
The efficiency of corn ethanol biorefineries has improved (see following sections) in the past
decade resulting in the use of less corn per gallon of ethanol and lower energy inputs. Corn
ethanol plants also produce about 5% of their energy output as corn oil.7
The primary factors
affecting ethanol plant performance are discussed below.
7
0.25 lb/gal ethanol × 15,993 Btu/lb (GREET soy and canola LHV) /77,000 Btu/gal denatured ethanol = 5.2%.
30 |
5.1 Corn Ethanol Yield
Several technologies have contributed to improvements in the ethanol yield per bushel of corn.
Increased ethanol yield results in less corn used per gallon of ethanol which results in lower
farming emissions, lower land use, and LUC per gallon of ethanol. Figure 5.1 shows trends in
historical yield data as well as projections. Data from the GREET model that was available at the
time of the 2010 RIA (Version 1.8c) is compared with industry data. These values are consistent
with EPA’s projections in the RIA with the trend line from the industry data slightly under the
2022 RIA projection. However, the input to the FASOM and FAPRI modeling system is 5% lower
than the yield projected by EPA8
for dry mill ethanol plants.
Figure 5.1. Dry Mill Corn Ethanol Yield Data and Projections.
5.1.1 Ethanol Yield in EPA 2010 RIA
EPA assumed ethanol yields of 2.71 gallons per bushel for dry mill plants and 2.5 gallons per
bushel for wet mill plants and FASOM and FAPRI-CARD models used these yield assumptions.
With the growth of dry mill plants, the aggregate yield should be higher than the values in the
2010 RIA. A higher yield would result in lower fertilizer use and ILUC. A first-order
approximation is that corn farming and LUC related emissions should be 10% lower than those
predicted by EPA due to actual yield improvements.
8
2010 RIA Section 2.4.7.1 EPA states the FASOM assumption
31 |
EPA identified yield projections that are consistent with industry data.9
The discrepancy may be
due to the use of the modeling systems for other programs or challenges associated with
changing a modeling assumption. In any event, the lower corn ethanol yield overestimates the
corn feedstock requirement for ethanol production. An offsetting factor would be that the
model predicted higher production of DGS and greater co-product displacement but the net
effect would still be an overestimate of corn farming emission and land use effects.
5.1.2 Plant Debottlenecking
The debottlenecking process helps to increase the yield and reduce energy consumption in corn
ethanol plants. New technologies and reviews of material and steam flows optimize the
utilization of critical processes to boost overall throughput, increase yield from base
throughput, or both. Membrane dehydration technology is one such technology which helps in
energy reduction, purity flexibility, and debottlenecking distillation capacity and dehydration.
These improvements have contributed to the overall improvement in U.S. ethanol plants.
5.1.3 Enzymes and Chemicals
Enzymes are among energy-intensive inputs for corn ethanol production. Companies like
Syngenta and DuPont are providing enzymes that are more efficient in terms of increasing the
ethanol yield and simultaneously reducing the enzyme consumption. In a new study by Kumar
and Singh (2016) that investigates using amylase corn and superior yeast in corn ethanol
production, the authors concluded that use of amylase corn and superior yeast in the dry-grind
processing industry can reduce the total external enzyme usage by more than 80%. Combining
their use with in situ removal of ethanol during fermentation allows efficient high-solid
fermentation. Also, their study showed that the ethanol yield in their process is 4.1% higher
than the conventional process of corn ethanol production.
5.2 Energy Consumption
Ethanol plants have reduced natural gas and power consumption through numerous factors
such as heat integration, combined heat and power technologies, variable frequency drives,
advanced grinding technologies, various combinations of front and back end oil separation, and
innovative ethanol and dried distillers’ grains (DDG) recovery (Mueller, 2016). These
technologies directly affect the CI of corn ethanol. These energy-saving technologies were
identified in the 2010 RIA and EPA modeled the natural gas and electric power consumption for
corn ethanol plants that EPA projected would be built with wet and dry DGS (Figure 5.4).
9
RIA Section 1.1.1.1
32 |
Plant configurations modeled by EPA.
Baseline plant
Combined heat and power (CHP)
CHP with corn oil fractionation
CHP with corn oil fractionation and membrane separation
CHP with corn oil fractionation, membrane separation, and raw starch hydrolysis
EPA placed considerable emphasis on modeling CHP. This technology has proven borderline
economical with the lower costs of natural gas as well as lower costs of electric power. EPA
projected that 70% of dry mill plants would adopt corn oil fractionation and this adoption rate
has been exceeded.
Ten years of experience has provided insight on the actual energy use for dry mill ethanol
plants. Data from ethanol plant operation has become available from industry surveys as well
as pathway registrations under the California LCFS (Cooper, 2008; ACE, 2018; CARB, 2018 list of
plants).
The GHG intensity of dry mill ethanol plants that were registered under the LCFS in 2016 is
shown in Figure 5.2. These data are based on the CA-GREET2 model and the current CI values
for these facilities with the CA-GREET3 model would be lower. However, the broader data set
was available for more facilities in 2016. These ethanol plants that register under the LCFS tend
to be closer to California and the lower CI ethanol plants are also represented here. The lower
CI of advanced corn ethanol is attributed to the use of biomass or biogas from anaerobic
digester as sources of energy. The CI values combined with LCFS applications allows for an
estimation of the distribution of natural gas usage among these facilities. The range of natural
gas usage was distributed equally among six bins and the range of each bin is shown in Figure
5.3. The average natural gas usage is 20,706 Btu/gal, LHV. These energy use rates and trend in
reduced energy consumption over time are consistent with a survey of dry mill ethanol plants
shown in Figure 5.4. These data are consistent with an industry average natural gas use rate of
22,500 Btu/gal by 2022, which is used in the assessment of GHG emissions in Section 8.
33 |
Figure 5.2. CI of Corn Ethanol with Various Technologies Registered under CARB
(CA-GREET2 model) (CARB LCFS Pathway List)
Figure 5.3. Distribution of Natural Gas Usage Among Ethanol Production Facilities.
34 |
Figure 5.4. Decrease in Natural Gas Usage Since 2004 (EPA RIA combination denotes dry mill
plant with only natural gas which produces 63% dry DGD and 37% wet DGS).
While the electricity consumption has not decreased significantly since 2010 (Figure 5.5), it has
a decreasing trendline which implies lower electricity is being consumed by ethanol plant due
to employing newer technologies. The overall impact of electric power should be examined as
described in Section 6.6.
Figure 5.5. Electricity Conusmption in Corn Ethanol. (ACE, 2018)
35 |
5.3 CO2 from Corn Ethanol
Many corn ethanol plants provide CO2 for beverage and industrial purposes. The CO2 generated
in the fermentation process of corn-ethanol plants has a high market share such that it is the
largest single-sector CO2 source for the U.S. merchant gas markets. As a valuable product for
the food industry, not only is the CO2 not a waste product, but it also generates GHG savings
credit which lowers the final CI of corn ethanol (Mueller, 2017). Absent ethanol plants, other
sources of CO2 would need to be utilized for refrigeration, beverages, and other applications
(Mueller, 2019). Carbon in the fermentation CO2 corresponds to half of the carbon in ethanol or
about 37,000 g CO2/MMBtu. After electric power for capture and liquefaction the GHG savings
are over 30,000 g CO2/MMBtu for ethanol plants that capture CO2. In addition, at least 4
different ethanol plants are deploying carbon capture and EPA did not take into account the
benefits of CO2 capture or utilization in the 2010 RIA. The effect of these technologies is not
included in the analysis in Section 8.
36 |
6. PROCESS FUELS
6.1 EPA RIA Fuel Production
In 2010, EPA considered several process fuels and different ethanol production practices (dry
mill and wet mill) and came up with a combination of use rates for process fuels. EPA used the
ASPEN models developed by the USDA to estimate the energy use at dry mill plants. The use
rates are for a new dry mill corn ethanol refinery in 2022 that uses natural gas as its process
fuel. The plant has a fractionation technology to extract corn oil and will produce a composite
DGS coproduct that is 63% dry and 37% wet. Fuel Production emissions for this refinery were
estimated as ~28,000 g CO2e/MMBtu in 2022. The 2010 RIA used the GREET model to estimate
the GHG CI of natural gas and electricity. These data have evolved and more recent estimates
are included in the analysis in Section 8.
6.2 Phase Out of Coal
The use of coal as a fuel for ethanol plants has declined since 2010. The majority of ethanol
plants are using natural gas as process fuel and only a small portion of the energy used in
ethanol plants is coming from coal. According to corn ethanol pathways in the 2015 GREET
model, on average, only 8 percent of the energy for steam production at U.S. ethanol plants
comes from coal (ANL, 2018). EPA’s projection of reduction in coal use were consistent with
actual experience.
6.3 Natural Gas Production and Methane Emissions
Further refinements of the LCA of natural gas have led to many publications addressing the
issue of energy inputs and methane emissions from natural gas production and distribution.
GHG emissions associated with natural gas extraction have resulted in an increase in the GHG
intensity of natural gas process fuel, which is taken into account in this study. As can be seen
from Figure 6.1, the CI used for natural gas in this study was slightly higher than the CI used in
the 2010 RIA.
37 |
Figure 6.1. Well to Wheel (WTW) Carbon Intensity of Natural Gas plus Boiler Emission Factor in
GREET. (ANL, 2018, GREET versions from 1.8b to 2021)
6.4 Biogas and Biomass Process Fuel
Landfill gas and biogas are potential process fuels for biorefineries which help to reduce the CI
of biofuel (Table 6.1). The introduction of low GHG process fuel at biorefineries has been
motivated by the RFS2 as well as the California LCFS. Below are several strategies employed by
biorefineries to reduce the CI. All of these technology improvements lead to low CI ethanol that
could be analyzed by EPA in the current rulemaking.
Landfills collocated with ethanol plants;
On-site anaerobic digestions of manure with avoided methane emissions;
Anaerobic digestion of stillage;
Electricity cogeneration;
Solid fuel biomass combustion.
Table 6.1. Effect of Biogas on Carbon Intensity of Corn Ethanol.
Process Fuel
Biogas Fraction CI (g CO2/MJ), LHV
NG/Biogas Ethanol
Natural Gas 100% 69 50
On-site Landfill 50% 1 40
Dairy Anaerobic Digester 15 to 25% -250 0
38 |
6.5 Electric Power
Corn ethanol plants use electric pumps, hammer mills, and other electrical equipment. The
electrical load has steadily declined over time from over 1 kWh per gallon of ethanol to an
average of 0.65 kWh per gallon (ACE, 2018) over a 10-year period. Over the same time period,
the GHG intensity of the U.S. grid has declined from 750 to 505 g CO2e/kWh on a life cycle
basis. On the other hand, the 2010 RIA projected power use of 1.09 kWh/gal with projects of
reduced power consumption. Actual power use had dropped to about 30% less than the
projected value.
6.5.1 Grid Carbon Intensity
The carbon intensity of electric power has declined with the expansion of natural gas
production and the declining price of natural gas (Figure 6.2). Carbon intensity of electric power
based on GREET has declined by 34% from 2010 to 2021 due to reduction in coal use and
growth in renewable power generation. The decrease in grid electricity CI directionally reduces
the corn ethanol CI since electricity is used in different stages of corn ethanol production, which
was not anticipated in the 2010 RIA with an overstatement of about 1000 g CO2e/MJ ethanol.
Figure 6.2. Carbon Intensity of Electric Power (U.S. Average).
(Power plant emissions do not include transmission losses, Source GREET)
A study at Carnegie Mellon University (CMU) examined the direct GHG emissions from the
power sector in the U.S. and found that between 2001 and 2017 the average annual carbon
intensity of electricity production in the U.S. decreased by 30%, from 630 g CO2e/kWh to 439
gCO2e/kWh (Schivley et al., 2018; EIA 2021). A similar proportional reduction in emissions
occurred for power plants in the corn belt states where most ethanol plants are located (Figure
6.3). Schivley et al. (2018) used the U.S. Energy Information Administration (EIA) database to
39 |
calculate aggregate GHG emissions and reports only power plant emissions10. The power plant
emissions are consistent with the power plant component GREET. Based on both EIA and
GREET, the CI of electricity is dropping. Note that the more recent EIA data shows a continuous
downtrend in the GHG intensity of U.S. electric power.
Figure 6.3. Change in Carbon Intensity of Electricity. (Schivley et al., 2018; EIA, 2021)
6.5.2 Renewable Power
Ethanol plants also have the opportunity to obtain lower GHG sources of electric power. Under
current fuel policies, such as California’s Low Carbon Fuel Standard, ethanol plants must use
renewable power that is directly connected to the generation source. However, renewable
power had contributed to the overall reduction in GHG emissions from the grid in the U.S.
10 Power plant emissions at the plant from GREET correspond to the “fuel” phase × (1 – loss factor)
40 |
6.6 Summary of Ethanol GHG Analysis Issues
Many factors affect the CI of corn ethanol. A summary of the issues and recommended analysis
method is shown in Table 6.2.
Table 6.2. Evaluation Issues related to GHG Analysis.
LCA parameter Analysis Issue Recommendation
Ethanol refinery
energy efficiency
has increased.
The energy efficiency
has increased in a few
refineries and it does
not reflect the
average.
Based on our analysis, the current energy
usage at the fuel production stage is close to
EPA RIA’s estimate, however, both electricity
and natural gas consumption have a declining
trend which should be considered.
Electric power GHG
intensity.
The GHG intensity of
electric power has
dropped faster than
projected in the 2010
RIA.
Update electricity mix for electric power
generation.
Emissions
associated with
gasoline is under
estimated.
EISA requires that the
EPA compare biofuel
emissions to a 2005
petroleum baseline.
The 2005 petroleum bassline analysis excluded
methane leakage and the thermal cracking of
petroleum which has lead to underestimation
of emissions associated with gasoline.
Co-product
allocation method
EPA RIA used the
replacement method
which results in lower
co-product credit.
Since corn oil is used as biodiesel feedstock
(energy source) energy allocation is a better
option which results in more reduction in corn
ethanol CI.
Fertilizer use rate
for soybean
EPA RIA used lower
fertilizer use rate for
soybean.
According to recent USDA statistics, the N
fertilizer use rate in soybean is almost three
times more than what EPA used. Higher
fertilizer rate for soybean results in more coproduct credit for DGS which replaces the
soybean meal.
41 |
7. PETROLEUM BASELINE EMISSIONS FOR 2005 ARE LARGER
THAN PROJECTED.
7.1 EPA 2010 RIA Approach in Estimation of Petroleum Baseline
EPA estimated the lifecycle GHG emissions associated with baseline gasoline transportation fuel
using the 2009 analysis performed by the National Energy Technology Laboratory (NETL). The
NETL analysis considers the GHG emissions associated with crude oil extraction both in the U.S.
refineries and refineries in other countries from which the U.S. imported oil. The emissions
from the 2010 RIA for 2005 gasoline fuel are shown in Table 7.1.
Table 7.1. Carbon Intensity of 2005 Gasoline from Well to Wheel (WTW).
GHG Emissions (g /MMBtu)
Life Cycle Step CO2 CH4 N2O CO2e
Fuel production 16,816 2,282 103 19,200
Tailpipe 77,278 3 5 78,891
EPA established the baseline RBOB (Reformulated gasoline Blendstock for Oxygen Blending) CI
for gasoline at 93.08 g CO2 e/MJ in the year 2005.11 EPA has not re-examined the CI of
petroleum since the 2010 RIA; however recent studies have shown that EPA underestimated
the emissions associated with 2005 gasoline. The key factors analyzed by these studies include:
Fugitive methane;
Flaring of associated gas;
Enhanced production methods including water flooding and thermal oil recovery;
Mix of oil sands;
Refinery complexity.
The key findings of recent studies which have more accurate data are discussed below.
7.2 New Findings on Petroleum Baseline
Researchers have studied the life cycle GHG emissions of petroleum fuels for several decades.
Many of these studies follow the process for LCA defined by International standards (ISO 14040,
2006). Initial studies examined the national inventory of GHG emissions from crude oil
production and refining with calculations of crude oil and fuel transport (Wang, 1999). Even
though GHG emissions from oil refineries are reported as part of most national GHG reporting
systems, the distribution of emissions among refined products has remained a challenge since
multiple refinery units produce a range of products.
11 California, in 2006, established a baseline CARBOB (California Reformulated gasoline Blendstock for Oxygen
Blending) CI of 95.86 g CO2 e/MJ. However, this value was updated to the 2012 value of 99.18 g CO2 e/MJ to reflect
the steady shift to higher intensity crude oils fed into U.S. refineries.
42 |
Aspects of crude oil production including flaring, indirect effects of road building, thermal
enhanced oil recovery, and crude production methods were identified as key aspects of the life
cycle of petroleum fuels (Unnasch et al., 2009; Keesom et al., 2009). Subsequent studies
expanded the modeling methods and detail for crude oil production in regions such as the EU
(Keesom et al., 2012; ICCT, 2014; COWI, 2015). More detailed models of crude oil production
have also been developed by Jacobs Consultancy (Keesom et al., 2012) and Stanford University
(El-Houjeiri et al., 2014). The California Air Resources Board (ARB) also publishes annual
estimates of the CI of crude oil (CARB, 2019b). Regional studies of crude oil for the U.S., China,
and globally are also part of the scientific literature (Cooney et al., 2016; Masnadi et al., 2018a;
Masnadi et al., 2018b; Gordon et al., 2015).
The GHG LCA emissions associated with gasoline have been examined in numerous studies
conducted by Jacobs Consultancy, Argonne National Laboratory, MathPro, and the University of
Calgary (Keesom et al., 2012; Elgowainy et al., 2014; Kwasniewski et al., 2016; Rosenfeld et al.,
2009, Abella and Bergerson, 2012). These studies show that a CI of 97 g/MJ would be more
accurate than the 93 g/MJ for the 2005 baseline value estimated in the EPA 2010 RIA due to
emissions associated with a range of crude oil production practices including oil sands
upgrading, venting and flaring or produced gas, and enhanced oil recovery technologies.
The quality and consistency of the raw crude fed into refineries determines the complexity of
processing required such that lower quality crude oil is more difficult to refine into
transportation fuels, thus resulting in higher CI. The total energy expended to recover crude oil
and the resulting GHG emissions vary depending upon the crude characteristics and the
recovery methods used. The carbon intensities per production method were analyzed in a
study that examined the CI of fuels under the RFS2 (Boland & Unnasch, 2014). The results for
different petroleum fuels are shown in Table 7.2.
Table 7.2. Petroleum Gasoline Carbon Intensity.
Petroleum Source Gasoline Carbon Intensity (g CO2 e/MJ)
Low High Average
Primary 84.50 94.6 89.55
Secondary 93.58 98.18 95.88
TEOR 100.58 120.00 110.29
Stripper Wells 101.95 116.44 109.20
Mining Upgrader 100.42 104.91 102.67
SAGD, Dilbit 105.00 115.36 110.18
Fracking 97.48 111.54 104.51
Oil Shale 113.00 159.00 136.00
Conventional oil includes primary and secondary sources of oil and these are the most well
defined and accessible sources of crude and hence the most drawn upon, the carbon intensity
for gasoline from these crude oils ranges from approximately 84 to 98 g CO2 e/MJ. TEOR
(Thermally Enhanced Oil Recovery) methods are generally implemented where the crude
43 |
characteristics (viscosity, API gravity) dictate and also to extend the life of a production well.
Heating water to produce the steam or other in-situ TEOR techniques require additional energy
inputs and can increase emissions by an additional 8 to 9% over conventional production.
Compared to conventional oil deposits, oil sands require production techniques that are
associated with greater environmental impacts. Shallow deposits are typically accessed using
strip-mining techniques, while deeper deposits are generally accessed using in situ techniques
whereby steam is injected into the reservoir to heat the bitumen until its viscosity decreases
sufficiently to allow it to flow out of the reservoir. On a WTW basis, the GHG emissions from oil
sands are generally between 5 to 15% higher than from most conventional oils. Heating water
to produce the steam used for in situ techniques and bitumen-sand separation uses large
amounts of energy, typically natural gas, and produces correspondingly large amounts of
emissions. In addition, bitumen produced from tar sands must go through more extensive
refining than conventional oil, producing additional emissions. Upgraded mining techniques
have led to advances in emissions reductions by approximately 2% over other oil sands ranges.
The emission ranges shown in Figure 7.1 show a range of crude oil types that were in
production in 2005 and are higher than the baseline in the 2010 RIA.
Figure 7.1. CI of Gasoline Estimated by Several Studies. (Unnasch et al., 2018)12
12 The Jacobs EU, JCE v4, GHGenuis, Jacobs NA, LCFS 2018, LCFS 2009, and EPA RFS2 2005 were presented in
Keesom et al. (2012), Edwards et al. (2012), S&T (2013), Keesom et al (2012), CARB (2018), CARB (2009), and EPA
(2010), respectively.
44 |
8. ESTIMATED GHG EMISSIONS FROM CORN ETHANOL
This study evaluated EPA’s 2010 LCA of corn ethanol and specifically focused on the emission
categories with the highest impacts. Since 2010 when the RIA was conducted, more data have
become available, LUC models have been revised several times and more realistic assumptions
have been made. Ten years of research provides a better understanding of the impact of
biofuel expansion on LUC both in the U.S. and across the globe. Also, the energy consumption
in the fuel production stage has been improved continuously since 2010 which should be
accounted for in EPA’s GHG LCA. Another important factor are the co-product credits where the
role of corn oil as biodiesel and the substitute value of soybean meal displacement was not fully
reflected in the 2010 RIA. The main factors analyzed in this study are discussed below.
1. International LUC has the highest share from total emissions of corn ethanol in the RIA.
Recent studies have estimated much lower values for international LUC compared to
EPA RIA. In this study, uses the GREET (2021)/CCLUB, to calculate both domestic and
international LUC. GREET uses the GTAP model which has undergone several rounds of
revision since 2010 and GTAP’s estimate of international LUC due to corn ethanol
production is almost five times lower than what EPA RIA estimated. GTAP includes
refinements in pasture utilization and projections of yield improvement reflected by
elasticities (Taheripour, 2017).
2. Corn ethanol yield affects both domestic and international LUC. EPA projected a yield of
2.71 gal/bu, however, recent data shows that the ethanol yield in dry mill process is 2.88
gal/bu and continues to improve (GREET, 2021).
3. Energy consumption in the fuel production stage has improved due to the application of
new technologies. EPA projected the natural gas consumption as the main source of
energy for dry mill process with corn oil fractionation as 25,854 Btu/gal. Data from LCFS
applications show a trend below 20,000 Btu/gal by 2022. Also, the CI of electricity used
as a source of energy in biorefining has a declining trend due to the consumption of
cleaner fuels in the production stage.
4. DGS, a byproduct of corn ethanol, is a partial substitute for soybean meal. Nitrogen
fertilizer use in soybean farming has increased recently and reached 166 g/bu (USDA
NASS, 2018). The RIA assumed a nitrogen fertilizer use rate for soybean of
approximately 64 g/bu. Higher nitrogen fertilizer use rates increases the GHG intensity
of soybean meal which results in a higher credit for the DGS co-product.
5. Corn oil is a co-product of corn ethanol that has achieved a high adoption rate. The 2010
RIA used the displacement method; however, the evolving use of corn oil is biomassbased diesel production (2021 Draft RIA, Figure 5.2.3-1). Therefore, energy allocation is
an appropriate option since the growing use of corn oil is as an energy product. The net
effect is a lower CI when both ethanol and biodiesel are treated as energy products.
45 |
This study uses the GREET (2021) model to calculate the CI of corn ethanol configured with
current ethanol plant and crop data. Since GREET lacks some consequential aspects of corn
ethanol LCA such as international rice methane emission and international livestock emissions,
the analysis in the ICF study (Rosenfeld et al., 2018) provides the basis for these parameters in
order to be consistent with the emissions categories in the 2010 RIA. The allocation treatment
of corn oil biodiesel is factored into the analysis also as shown in Table 8.1.
The estimated GHG emissions represent a hybrid between the GREET and consequential LCA
approach in the 2010 RIA. The allocation effect of corn oil as a biodiesel feedstock is taken into
account with emissions allocated between ethanol and corn oil-based diesel. Note that the
substitute value of corn oil is a small fraction of the DGS co-product and an acre of land that
produces corn for ethanol makes as much animal feed as an acre of soy beans.
46 |
Table 8.1. CI of Corn Ethanol for Dry Mill, Natural Gas Operation with Corn Oil Extraction.
Emission Category
2005
Gasoline
Revised
2005
Gasoline
RIA
GREET
1999
EPA 2010
RIA
GREET
2021 This Studya
ICF
Domestic Livestock -3,746 -2,202 -2,463 -2,340
Domestic Farm Inputs and Fertilizer N2O 16,000 8,281 11,548 9,065 11,023
International Farm Inputs and Fertilizer N2O 6,601 -987 -1,013 -1,013
Domestic Rice Methane -209 578 578 578
Tailpipe 79,004 79,004 880 880 2,420 2,483 2,359
International Rice Methane 2,089 3,795 3,894 3,700
International Livestock 3,458 -2,255 -2,038 -2,199
Domestic Land Use Change -4,033 1,374 3,432 1,374
Fuel and feedstock transport 5,000 4,265 2,160 2,217 2,217
International Land Use Change 30,000 31,797 6,139 9,082 5,986
Fuel Production 21,100 19,200 48,000 27,851 29,527 34,518 28,792
Net Emissions 100,104 98,204 99,880 77,233 47,468 52,096 59,755
a
95.2% allocation factor (fraction of ethanol output/ ethanol plus corn oil) applied to either GREET or ICF results as indicated in bold.
Natural gas consumption of 24,305 Btu/gal, LHV. International farming inputs are based on the ICF analysis even though the full burden of domestic corn
farming is represented with the GREET inputs. Domestic and international rice methane and livestock emissions are
based on the ICF values combined with the allocation factor. International and domestics land use change are based on the GREET result combined with the
allocation factor. This study does not investigate categories including international farm inputs and fertilizer N2O, domestic and international rice methane
emissions and international livestock emissions and relies on the ICF study estimates for these emission categories and are combined with the allocation factor
for corn oil. Livestock emissions include two major factors, enteric fermentation, and manure management. It has been shown by several studies that replacing
DGS with soybean meal reduces the enteric fermentation. The manure management emissions refer to emissions during collection, storage, transfer, and
treatment of manure. While the replacement of DGS reduced the enteric fermentation in domestic livestock, it was not included in estimating the international
livestock emissions in RIA analysis. Inclusion of reduction in enteric fermentation for international livestock would decrease the emissions associated with
international livestock.
47 |
Figure 8.1 shows the estimated CI is 50,417 g CO2e/MMBtu while 2010 RIA estimated the CI of
corn ethanol as 77,233 g CO2e/MMBtu. The GREET (2021) estimation of corn ethanol CI is the
lowest since it does not account for international livestock and rice emissions. The emission
estimates from the ICF analysis provide the basis for the analysis presented here. While in
GREET (2021) a small percentage (~7%) of energy for fuel production is coming from burning
coal, this analysis represents natural gas dry mill facilities, which are the new facilities
incentivized by the RFS2 and does not attempt to examine the entire range of ethanol
production technologies.
Figure 8.1. CI of Corn Ethanol for Dry Mill, Natural Gas Operation with Corn Oil Extraction.
Under the current situation and in the year 2022, Rosenfeld et al. (2018) calculated the CI of
corn ethanol as 59,755 g CO2e/MMBtu and 54,588 g CO2e/MMBtu, respectively. Rosenfeld et
al. (2018) also defined a scenario in which new technologies and better practices are employed
to reduce the emissions in corn and fuel production. They concluded that by employing
advanced technologies and introducing new co-products in the fuel production stage, and
efficient management practices such as reduced tillage, nutrient management and cover crops
in the farming stage the GHG emissions can be reduced to 27,852 g CO2e/MMBtu. These
estimates are consistent with ongoing trends in regenerative agriculture.
48 |
9. CONCLUSIONS
Life cycle GHG emission from the corn ethanol was analyzed over a range of production
technologies and analysis methods. The data in this study show that life cycle GHG emissions
for corn ethanol plants can range from 26 to 57 g CO2e/MJ. Typical dry mill facilities have a CI
in the 40 to 55 g CO2e/MJ range. The CI for the 2005 petroleum baseline is also higher than
originally projected; so, most of the ethanol plants in the U.S. produce fuel with a 45 to 55%
reduction in GHG emissions. The key factors that result in GHG emissions that are lower than
projected in the 2010 RIA include the following:
Reduced energy consumption;
Reduced GHG intensity for electric power;
Shift from coal to natural gas fuel;
Adoption of corn oil extraction with energy allocation;
Reduced rates of deforestation;
Improved rates of DGS use as animal feed;
Displacement of ILUC and N2O emissions from soy beans;
o Higher nitrogen application rates to soybeans than originally modelled;
Use of corn replacement feed from crop residue;
Introduction of lower CI process fuels for ethanol plants;
Higher GHG emissions from 2005 petroleum baseline fuels.
EPA overestimated international land use conversion in the 2010 RIA and has not updated the
analysis in the draft 2021 RIA. New ILUC studies that take into account pasture intensification
show a lower level of international ILUC and are represented in the CCLUB model from Argonne
National Laboratory (ANL). The CCLUB model incorporates the most recent modeling from
Purdue University’s GTAP program. EPA also analyzed negative direct and indirect land use
conversion emissions in the 2010 RIA. These results are confirmed in the CCLUB model from
ANL and are consistent with the basic factors affecting the growth of corn ethanol production.
Total agricultural land has not increased significantly in the U.S.
In addition, much of the growth in corn ethanol has come from a reduction in soybean
production. Corn farming increases soil carbon relative to soy farming with no till practices and
due to the fact that corn builds up soil carbon from its root mass. Criticisms of the CCLUB
model based on the choice of the CENTURY emission factors associated with crop activity are
misplaced as the emission factors based on Winrock and Woods Hole are simple
approximations that are unsubstantiated. The CENTURY approach is used in the development
of the U.S. emission inventory and is also consistent with regenerative agriculture practices that
generate voluntary carbon credits.
In addition, EPA did not sufficiently document advancements in corn ethanol technology.
Numerous ethanol plants are starting to use biogas and biomass fuel as well as implementing
carbon capture and sequestration.
49 |
Finally, EPA understated the 2005 petroleum baseline and has not acknowledged the revised
estimates of emissions in the 2021 draft RIA for this rulemaking. The refining of heavy oil as
well as flaring emissions from many international sources of crude oil, which occurred in 2005
contribute to higher GHG emissions associated with gasoline than those in the 2010 RIA.
50 |
10. APPENDIX A – NITROGEN APPLICATION RATES
Nitrogen application rates affect the GHG intensity of corn production. In addition, ethanol
plant DGS provides a replacement for crops with nitrogen application rates that are higher than
anticipated in the 2010 RIA.
Figure A.1. FASOM Average Nitrogen Fertilizer Use by Crop. (EPA, 2010, not updated in EPA
2021)
EPA RIA
51 |
Figure A.2. N2O emissions per acre from crop production.
Correcting the actual N fertilizer use in soybean farming, i.e., 166 g/bu, results in about a 460 g
CO2e/MMBtu of ethanol reduction in carbon intensity (CI) of corn ethanol with the soybean
meal substitution rates in the GREET model.13
In summary: EPA attributed a certain amount of N fertilizer to soy production. DGS displaces
soybeans that would otherwise be used as animal feed. Soybeans are more energy intense to
grow than considered in the 2010 RIA and this displacement credit should be taken into
account. The displacement value of DGS may be understated in the 2010 RIA also.
The literature review presented below examines the discrepancy between USDA NASS database
and GREET on nitrogen fertilizer use in soybean farming. Soybean, which is an annual legume,
requires a high amount of nitrogen (5 lb of N per each bushel). However, 50 to 60% of the
required nitrogen is supplied through the N-fixation process, which is a result of a symbiotic
relationship between the plant and soil bacteria (Nafziger, 2014). The nitrogen fixation process
consumes about 10% of the soybean’s energy in the form of sugars produced by
photosynthesis. According to Nafziger (2014), “at high yield levels, the crop might not be able
to produce enough sugars to go around, and that either yield will suffer, or N fixation will be
reduced.” One of the methods to overcome this issue is to add nitrogen fertilizer in the growing
season of soybean. Several studies have investigated the impact of nitrogen fertilizer
application rate on soybean yield (Mourtzinis et al., 2018; La Menza et al., 2017; Schmidt, 2016.
Mourtzinis et al. (2018) conducted one of the most comprehensive studies on soybean yield
13 (166 – 48) lb/bu ÷60 lb/bu. 0.307 lb SBM displaced per lb DDGS, 3.78 g CO2e/g N fertilizer, 0.0153 g N2O/g N.
52 |
response to N fertilizer in the U.S. which included 207 environments (experiment × year
combinations) for a total of 5991 N-treated soybean yields. While this study reported that the
soybean yield increased by an increase in N fertilizer application, in most individual
environments, the effect of a greater N-rate on soybean yield was not significant.
Figure A.3. Effect of Nitrogen Application Rate on Soybean Yield. (Mourtzinis et al., 2018)
While there was a large yield variability among environments within the same N rates,
Mourtzinis et al. (2018) generated a second-degree N polynomial function that was significant
(p = 0.0297), and it estimated the nitrogen rate of 340 kg ha−1 for maximization of soybean
yield. This rate translates to 1.8 kg N per bushel of soybean (Figure A.3). Similarly, Nafziger
(2014) studied the impact of nitrogen fertilizer on soybean yield over several years and
concluded that soybean yields response to N fertilizer ranged widely among the trials.
In another study, La Menza et al. (2017) tested the hypothesis that indigenous nitrogen sources
(N fixation and soil mineralization) are insufficient to meet crop N requirements for high yields.
For this purpose, they developed a protocol to ensure an ample N supply during the entire crop
season. They reported that soybean yield under ample N was 11% higher than the zero-N
condition. Based on the literature review, we can conclude that adding N fertilizer to soybeans
to achieve higher yields is gaining more attention, however, there is no clear trend between N
application rate and soybean yields. There are several other factors which can affect the
soybean yield such as planting date, N application timing, irrigation, etc. which need further
studies. The higher emissions associated with soybean meal have been included in the more
recent versions of GREET.
53 |
REFERENCES
Abella, J. P., & Bergerson, J. A. (2012). Model to investigate energy and greenhouse gas
emissions implications of refining petroleum: Impacts of crude quality and refinery
configuration. Environmental science & technology, 46(24), 13037-13047.
ACE (2018). American Coalition for Ethanol. The Case for Properly Valuing the Low Carbon
Benefits of Corn Ethanol.
https://bluetoad.com/publication/?i=519490&p=&pn=#{%22issue_id%22:519490,%22page%
22:0}. (Accessed 02/08/2019).
ANL (2018). Argonne National Laboratory. https://www.anl.gov. (Accessed 02/08/2019).
Balamurugan, T., Arun, A., & Sathishkumar, G. B. (2018). Biodiesel derived from corn oil–A fuel
substitute for diesel. Renewable and Sustainable Energy Reviews, 94, 772-778.
Batres-Marquez, S.P. (2018). Production and Revenue Trends in Corn Ethanol, DDGS, and Corn
Distillers Oil. Renewable Energy Report, Agricultural Marketing Resource Center, Iowa State
University.
Boland. S. and Unnasch. S. (2014). Carbon Intensity of Marginal Petroleum and Corn Ethanol
Fuels. Life Cycle Associates Report LCA.6075.83.2014, Prepared for Renewable Fuels
Association.
Brander, M., Tipper, R., Hutchison, C., & Davis, G. (2008). Technical Paper: Consequential and
attributional approaches to LCA: a Guide to policy makers with specific reference to
greenhouse gas LCA of biofuels.
CARB (2009). Proposed regulation to implement the low carbon fuel standard. Staff Report:
Initial statement of reasons. https://www.arb.ca.gov/regact/2009/lcfs09/lcfsisor1.pdf.
(Accessed 02/08/2019).
CARB (2010). LCFS pathway document California Modified GREET Pathway for the Production of
Biodiesel from Corn Oil at Dry Mill Ethanol Plants, Available at:
http://www.arb.ca.gov/fuels/lcfs/2a2b/internal/121410lcfs-cornoil-bd-rpt.pdf
CARB (2015). Calculating carbon intensity values from indirect land use change and crop based
biofuels. Appendix I: Detailed analysis for indirect land use change. California Air Resources
Board. https://www.arb.ca.gov/fuels/lcfs/iluc_assessment/iluc_analysis.pdf. (Accessed
02/08/2019).
CARB (2018). California Air Resources Board.
https://www.arb.ca.gov/fuels/lcfs/063008epa_lca.pdf. (Accessed 02/08/2019).
CARB (2019a). LCFS Pathway Certified Carbon Intensities
https://www.arb.ca.gov/fuels/lcfs/fuelpathways/pathwaytable.htm. (Accessed 02/08/2019).
CARB (2019b). LCFS crude oil lifecycle assessment. https://www.arb.ca.gov/fuels/lcfs/crudeoil/crude-oil.htm. (Accessed 02/08/2019).
Clay, D. E., Chang, J., Clay, S. A., Stone, J., Gelderman, R. H., Carlson, G. C., … & Schumacher, T.
(2012). Corn yields and no-tillage affects carbon sequestration and carbon footprints.
Agronomy journal, 104(3), 763-770.
54 |
Cooney, G., Jamieson, M., Marriott, J., Bergerson, J., Brandt, A., & Skone, T. J. (2016). Updating
the US life cycle GHG petroleum baseline to 2014 with projections to 2040 using open-source
engineering-based models. Environmental science & technology, 51(2), 977-987.
Cooper, G. (2008). LCFS and Corn Ethanol: Status of Issues. Renewable Fuels Association
Presentation to CARB, May 9, 2008.
COWI (2015). Study on actual GHG data for diesel, petrol, kerosene and natural gas.
https://ec.europa.eu/energy/sites/ener/files/documents/Study%20on%20Actual%20GHG%20
Data%20Oil%20Gas%20Executive%20Summary.pdf. (Accessed 02/08/2019).
CRC (2014). Unnasch, S., T. Darlington, J. Dumortier, W. Tyner, J. Pont and A. Broch, Study of
Transportation Fuel Life Cycle Analysis: Review of Economic Models Used to Assess Land Use
Effects. Prepared for Coordinating Research Council Project E-88-3.
CRC (2016). Follow-on study of transportation fuel life cycle analysis: Review of current CARB
and EPA estimates of land use change (LUC) impacts. Coordinating Research Council Report,
E-88-3b.
Dien, B. S., Nagle, N., Hicks, K. B., Singh, V., Moreau, R. A., Tucker, M. P., … & Bothast, R. J.
(2004). Fermentation of “quick fiber” produced from a modified corn-milling process into
ethanol and recovery of corn fiber oil. Applied biochemistry and biotechnology, 115(1-3), 937-
949.
Dumortier, J., Hayes, D. J., Carriquiry, M., Dong, F., Du, X., Elobeid, A., … & Tokgoz, S. (2011).
Sensitivity of carbon emission estimates from indirect land-use change. Applied Economic
Perspectives and Policy, 33(3), 428-448.
Dunn, J.B., Mueller, S., Qin, Z., & Wang, M.Q. (2014). Carbon Calculator for Land Use Change
from Biofuels Production (CCLUB 2015). Argonne National Laboratory (ANL). Argonne, IL
(United States).
Dunn, J., Mueller, S., Kwon, H. Y., & Wang, M. Q. (2013). Land‐use change and greenhouse gas
emissions from corn and cellulosic ethanol. Biotechnology for Biofuels, 6(1), 51.
Dunn, J. B., Qin, Z., Mueller, S., Kwon, H. Y., Wander, M. M., & Wang, M. (2017). Carbon
calculator for land use change from biofuels production (CCLUB) users’ manual and technical
documentation (No. ANL-/ESD/12-5 Rev. 4). Argonne National Lab. (ANL), Argonne, IL (United
States).
Edwards, R., Larive, J. F., Rickeard, D., & Weindorf, W. (2013). JEC Well to Wheels Analysis, Well
to Tank Report Version 4.0. European Commission Joint Research Centre.
Elgowainy, A., Han, J., Cai, H., Wang, M., Forman, G. S., & DiVita, V. B. (2014). Energy efficiency
and greenhouse gas emission intensity of petroleum products at US refineries. Environmental
science & technology, 48(13), 7612-7624.
El-Houjeiri, H. M., Vafi, K., Duffy, J., McNally, S., & Brandt, A. R. (2014). Oil Production
Greenhouse Gas Emissions Estimator OPGEE v1. 1 Draft D. User guide & Technical
documentation. User guide & Technical documentation.
EPA (2010). Renewable Fuel Standard Program (RFS2) Regulatory Impact Analysis. EPA Report
EPA-420-r-10-006.
55 |
Fargione, J., Hill, J., Tilman, D., Polasky, S., & Hawthorne, P. (2008). Land clearing and the
biofuel carbon debt. Science, 319(5867), 1235-1238.
Farm Energy (2019). What is direct land use, or direct land use change? https://farmenergy.extension.org/what-is-direct-land-use-or-direct-land-use-change. (Accessed
02/08/2019).
Flugge, M., Lewandrowski, J., Rosenfeld, J., Boland, C., Hendrickson, T., Jaglo, K., Kolansky, S.,
Moffroid, K., Riley-Gilbert, M., & Pape, D. (2017). A Life-Cycle Analysis of the Greenhouse Gas
Emissions of Corn- Based Ethanol. Report prepared by ICF under USDA Contract No. AG-3142-
D-16-0243.
Garrett, R. D., Koh, I., Lambin, E. F., de Waroux, Y. L. P., Kastens, J. H., & Brown, J. C. (2018).
Intensification in agriculture-forest frontiers: Land use responses to development and
conservation policies in Brazil. Global environmental change, 53, 233-243.
Gaillard, C., Sørensen, M. T., Vestergaard, M., Weisbjerg, M. R., Basar, A., Larsen, M. K.,
Martinussen, H., Kidmose, U., & Sehested, J. (2017). Effect of substituting soybean meal and
canola cake with dried distillers grains with solubles at 2 dietary crude protein levels on feed
intake, milk production, and milk quality in dairy cows. Journal of dairy science, 100(11),
8928-8938.
Gordon, D., Brandt, A. R., Bergerson, J., & Koomey, J. (2015). Know your oil: creating a global
oil-climate index. Washington, DC: Carnegie Endowment for International Peace.
GREET (2018). The Greenhouse Gases, Regulated Emissions, and Energy Use in Transportation
Model. Argonne National Laboratory.
Grooms, L. (2014). Biodiesel industry turns Corn oil, Farm Industry News. http://
www.farmindustrynews.com/biofuel/biodiesel-industry-turns-corn-oil. (Accessed
02/08/2019).
Hertel, T. W., Golub, A. A., Jones, A. D., O’Hare, M., Plevin, R. J., & Kammen, D. M. (2010).
Effects of US maize ethanol on global land use and greenhouse gas emissions: estimating
market-mediated responses. BioScience, 60(3), 223-231.
Huo, H., Wang, M., Bloyd, C., & Putsche, V. (2008). Life-cycle assessment of energy use and
greenhouse gas emissions of soybean-derived biodiesel and renewable fuels. Environmental
science & technology, 43(3), 750-756.
ICCT (2014). Crude oil greenhouse gas emissions calculation methodology for the fuel quality
directive.
https://ec.europa.eu/clima/sites/clima/files/transport/fuel/docs/icct_crude_ghg_calculation_
methodology_en.pdf. (Accessed 02/08/2019).
ICF Consulting Canada (2013). Independent Assessment of the European Commission’s Fuel
Quality Directive’s “Conventional” Default Value. Prepared for Natural Resources Canada.
Iowa Corn (2019). Distillers Grains. https://www.iowacorn.org/corn-uses/livestock/distillersgrains. (Accessed 02/08/2019).
IPCC (2014). Climate Change 2014: Synthesis Report. Contribution of Working Groups I, II and III
to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change [Core
Writing Team, R.K. Pachauri and L.A. Meyer (eds.)]. IPCC, Geneva, Switzerland.
56 |
ISO 14040 (2006). Environmental Management: Life Cycle Assessment, Principles and
Framework. International Organization for Standardization.
Keeney, R., & Hertel, T. W. (2009). The indirect land use impacts of United States biofuel
policies: the importance of acreage, yield, and bilateral trade responses. American Journal of
Agricultural Economics, 91(4), 895-909.
Keesom, W. H., Blieszner, J., & Unnasch S. (2012). EU Pathway Study: Life Cycle Assessment of
Crude Oils in a European Context. Prepared by Jacobs Engineering and Life Cycle Associates
for Alberta Petroleum Marketing Commission (APMC).
Keesom, W. H., Unnasch, S., & Moretta, J. (2009). Life Cycle Assessment Comparison of North
American and Imported Crude Oils. Prepared by Jacobs Engineering and Life Cycle Associates
for Alberta Energy Research Institute (AERI).
Kumar, D., & Singh, V. (2016). Dry-grind processing using amylase corn and superior yeast to
reduce the exogenous enzyme requirements in bioethanol production. Biotechnology for
biofuels, 9(1), 228.
Kumar, U. S., & Kumar, K. R. (2013). Performance, Combustion and emission Characteristics of
Corn oil blended with Diesel. Carbon, 8, 3.
Kwasniewski, V., Blieszner, J., & Nelson, R. (2016). Petroleum refinery greenhouse gas emission
variations related to higher ethanol blends at different gasoline octane rating and pool
volume levels. Biofuels, Bioproducts and Biorefining, 10(1), 36-46.
La Menza, N. C., Monzon, J. P., Specht, J. E., & Grassini, P. (2017). Is soybean yield limited by
nitrogen supply? Field crops research, 213, 204-212.
Malins, C., Plevin, R., & Edwards, R. (2020). How robust are reductions in modeled estimates
from GTAP-BIO of the indirect land use change induced by conventional biofuels? Journal of
Cleaner Production, 258, 120716.
Masnadi, M. S., El-Houjeiri, H. M., Schunack, D., Li, Y., Englander, J. G., Badahdah, A., … &
Gordon, D. (2018). Global carbon intensity of crude oil production. Science, 361(6405), 851-
853.
Masnadi, M. S., El-Houjeiri, H. M., Schunack, D., Li, Y., Roberts, S. O., Przesmitzki, S., … & Wang,
M. (2018). Well-to-refinery emissions and net-energy analysis of China’s crude-oil supply.
Nature Energy, 3(3), 220.
McGrath, C., Wright, D., Mallarino, A.P., & Lenssen, A.W. (2013). Soybean Nutrient Needs.
Agriculture and Environment Extension Publications. 189.
Mourtzinis, S., Kaur, G., Orlowski, J. M., Shapiro, C. A., Lee, C. D., Wortmann, C., … & Ross, W. J.
(2018). Soybean response to nitrogen application across the United States: A synthesisanalysis. Field Crops Research, 215, 74-82.
Mueller, S. (2017). Ethanol Industry Provides Critical CO2 Supply.
http://www.ethanolproducer.com/articles/14122/. (Accessed 02/08/2019).
Mueller, S. (2016). https://www.epa.gov/sites/production/files/2016-
05/documents/16004.pdf. (Accessed 02/08/2019).
57 |
Mueller, S., Unnasch, S., Tyner, W. E., Pont, J., & Johnson, J. M. (2015). Handling of co-products
in life cycle analysis in an evolving co-product market: A case study with corn stover removal.
Adv. Appl. Agric. Sci, 3, 8-21.
Mueller, S. (2017). Ethanol Industry Provides Critical CO2 Supply, Ethanol Producer Magazine,
March 2017.
Mueller, S. and Rushing, S. (2019). The Case for a Credit. Ethanol Producer Magazine, June
2019.
Nafziger, E. (2014). Do soybeans need N fertilizer? The Bulletin, Integrated Pest Management at
the University of Illinois. http://bulletin.ipm.illinois.edu/?p=1966. (Accessed 02/08/2019).
Nielsen, R.L. (2017). Historical corn grain yields for the U.S.
http://www.kingcorn.org/news/timeless/YieldTrends.html. (Accessed 02/08/2019).
National Renewable Energy Laboratory (NREL) (2004). Getting Extra “corn squeezins.”
Technology Brief-11/1993; http://www.nrel.gov/ docs/gen/old/5639.pdf. (Accessed
02/08/2019).
Poffenbarger, H. J., Barker, D. W., Helmers, M. J., Miguez, F. E., Olk, D. C., Sawyer, J. E., … &
Castellano, M. J. (2017). Maximum soil organic carbon storage in Midwest US cropping
systems when crops are optimally nitrogen-fertilized. PLoS One, 12(3), e0172293.
RFA (2019). 2019 ethanol industry outlook. Renewable Fuels Association,
https://ethanolrfa.org/wp-content/uploads/2019/02/RFA2019Outlook.pdf.
Rosenfeld, J., Lewandrowski, J., Hendrickson, T., Jaglo, K., Moffroid, K., & Pape, D. (2018). A LifeCycle Analysis of the Greenhouse Gas Emissions from Corn-Based Ethanol. Report prepared
by ICF under USDA Contract No. AG-3142-D-17-0161.
Rosenfeld, J., Pont, J., Law, K. (2009). Comparison of North American and Imported Crude Oil
Lifecycle GHG Emissions. Prepared for Alberta Energy Research Institute.
Scully, M. J., Norris, G. A., Falconi, T. M. A., & MacIntosh, D. L. (2021). Carbon intensity of corn
ethanol in the United States: state of the science. Environmental Research Letters, 16(4),
043001.
Silva, R. D. O., Barioni, L. G., Pellegrino, G. Q., & Moran, D. (2018). The role of agricultural
intensification in Brazil’s Nationally Determined Contribution on emissions mitigation.
Agricultural Systems, 161, 102-112.
Searchinger, T., Heimlich, R., Houghton, R. A., Dong, F., Elobeid, A., Fabiosa, J., … & Yu, T. H.
(2008). Use of US croplands for biofuels increases greenhouse gases through emissions from
land-use change. Science, 319(5867), 1238-1240.
Searchinger, T., Edwards, R., Mulligan, D., Heimlich, R., & Plevin, R. (2015). Do biofuel policies
seek to cut emissions by cutting food?. Science, 347(6229), 1420-1422.
Schivley, G., Azevedo, I., & Samaras, C. (2018). Assessing the evolution of power sector carbon
intensity in the United States. Environmental Research Letters, 13(6), 064018.
Schmidt, J.P. (2016). Nitrogen Fertilizer for Soybean? DuPont Pioneer Agronomy Library.
Taheripour, T., Zhao, X., & Tyner, W. E. (2017). The impact of considering land intensification
and updated data on biofuels land use change and emissions estimates. Biotechnology for
Biofuels, 10, 191.
58 |
S&T (2013). GHGenius Model v4.03. http://www.ghgenius.ca. (Accessed 02/08/2019).
Tyner, W. E., Taheripour, F., Zhuang, Q., Birur, B., & Baldos, U. (2010). Land use changes and
consequent CO2 emissions due to U.S. corn ethanol production: A comprehensive analysis.
Washington, DC: Argonne National Laboratory.
Unnasch. S., et al. (2018). Evaluating Carbon Intensity of Baseline Gasoline for Renewable Fuel
Programs, Life Cycle Associates Report LCA. LCA.6138.193.2018, Prepared for U.S. Grains
Council.
Unnasch, S., Goyal, L. (2017). Comments on ethanol pathway in CA_GREET2, Error in Biogenic
VOC. Comments submitted to California Air Resource Board.
Unnasch, S., et al. (2009). Assessment of Life Cycle GHG Emissions Associated with Petroleum
Fuels, Life Cycle Associates Report LCA-6004-3P. 2009. Prepared for New Fuels Alliance.
USDA ERS (2018). United States Department of Agriculture, Economic Research Service.
USDA NASS (2018). United States Department of Agriculture, National Agricultural Statistics
Service.
U.S. Grain Council (2018). https://grains.org. (Accessed 02/08/2019).
Veljković, V.B., Biberdžić, M.O., Banković-Ilić, I.B., Djalović, I.G., Tasić, M.B., Nježić, Z.B., &
Stamenković, O.S. (2018). Biodiesel production from corn oil: a review. Renewable and
Sustainable Energy Reviews 91, 531–548.
Wang, M.Q. (1999). Greenhouse Gases, Regulated Emissions, and Energy Use in Transportation,
GREET model 1.5. Argonne National Laboratory, Illinois.
Welshans, K. (2014).Corn stover a practical option for cattle forage.
https://www.feedstuffs.com/story-corn-stover-a-practical-option-for-cattle-forage-60-
107493. (Accessed 02/08/2019).
World Bank (2018). https://www.worldbank.org. (Accessed 02/08/2019).
Zilberman, D. (2017). Indirect land use change: much ado about (almost) nothing. GCB
Bioenergy, 9(3), 485-488.
Zilberman, D., Hochman, G., & Rajagopal, D. (2010). Indirect land use change: a second-best
solution to a first-class problem. UC Berkeley CUDARE Working Papers.
ActiveUS 192803838v.1
Growth Energy Comments on EPA’s
Proposed Renewable Fuel Standard Program:
Renewable Fuel Standard Annual Rules
Docket # EPA-HQ-OAR-2021-0324
Exhibit 3
1 | P a g e
ANALYSIS OF EPA’S PROPOSED RULEMAKING FOR 2020, 2021, and
2022 RVOs REGARDING LAND USE CHANGE, WETLANDS,
ECOSYSTEMS, WILDLIFE HABITAT, WATER RESOURCE
AVAILABILITY, and WATER QUALITY
Prepared For: Growth Energy
Date: February 3, 2022
Author: Pieter Booth, Principal1
Net Gain Ecological Services
This memorandum provides Net Gain’s comments and observations regarding selected technical issues
associated with EPA’s Proposed Rule for the Renewable Fuel Standard (RFS) Program Rules for 2020,
2021, and 2022 Renewable Volume Obligations (RVOs) (the Proposed Rule; EPA 2021a) and the
associated Draft Regulatory Impact Analysis (RIA; EPA 2021b). The Proposed Rule and the RIA rely
heavily on EPA’s Second Triennial Report (EPA 2018). Therefore, this memo updates and builds upon
previous findings and conclusions presented in the following reports2 attached as exhibits:
• Ramboll. August 18, 2019. The RFS and ethanol production: Lack of proven impacts to land
and water. Prepared for Growth Energy. Ramboll, Seattle, WA. (Exhibit 1).
• Ramboll. November 29, 2019. Memorandum: Supplemental analysis regarding allegations of
potential impacts of the RFS on species listed under the Endangered Species Act. Prepared for
Growth Energy. Ramboll, Seattle WA. (Exhibit 2).3
These prior analyses addressed the absence of a demonstrated causal nexus between the RFS and
land use change (LUC); adverse impacts to wetlands, ecosystems, and wildlife habitat; and adverse
impacts to water resource availability and water quality. Our analyses refuted claims by other
investigators that the RFS causes quantifiable adverse impacts to environmental media. We have
evaluated more recent scientific literature on this topic and continue to find that there is no evidence
the RFS program causes these adverse environmental impacts. Based on this finding, there is no
evidence that the Proposed Rule will result in land conversion or cause adverse impacts to wetlands,
ecosystems, wildlife habitat, water availability and water quality. We encourage EPA to update its
analysis in the RIA to address these findings and correct its potentially misleading discussion of
environmental impacts of the program.
1 Mr. Booth has over 30 years of experience as an environmental scientist specializing in environmental risk
assessment and restoration, natural resource damage assessment, environmental due diligence, and policy
analysis. He has published 19 articles and presented his work 43 times at national and international conferences.
He has acted as consulting expert on over a dozen environmental damages cases in the U.S. and has been retained
as a testifying expert on three international environmental damages cases.
2 Mr. Booth was lead author on both referenced reports.
3 Submitted by Growth Energy as “ESA Comments – Attachment B,” Docket # EPA-HQ-OAR-2019-0136
Supplemental Notice of Proposed Rulemaking; Renewable Fuel Standards Program: Standards for 2020 and
Biomass-Based Diesel Volume for 2021, and Response to the Remand of the 2016 Standards.
2 | P a g e
EXECUTIVE SUMMARY
The RIA presents only a generalized discussion of the drivers for and nature of potential impacts of
biofuel feedstock production and biofuel refining on LUC; wetlands, ecosystems, and wildlife habitat;
and water availability and water quality. It fails to recognize the complex causal links between drivers
of impacts and the potential impacts described to land and water, which in turn creates the misleading
impression that there is a causal relationship between the RFS and impacts to land and water, where
no such relationship has been established.
Conversion of Wetlands, Ecosystems, and Wildlife Habitats: The Role of
Land Use Change
Crop extensification resulting in LUC is the factor that has garnered the most attention in assessing
the potential impacts to land and water from the RFS. The causal chain linking the RFS and LUC
consists of a myriad of complex interactions among economic, biophysical, and social factors that are
very challenging to model. In its Second Triennial Report and its Endangered Species Act No Effect
Finding for the 2020 Final Rule (No Effects Finding) (EPA 2019), EPA has recognized this complexity
and the large amount of uncertainty in establishing causation between the RFS and LUC, but EPA fails
to adequately consider the implications of this uncertainty and absence of evidence in its conclusions
in the RIA. The RIA cites studies such as Lark et al. (2015) and Wright et al. (2017) in its discussion of
the RFS and LUC. More recent publications have concluded the findings of Wright et al. (2017) and
Lark et al. (2015) were flawed and based on inaccurate data. In addition, EPA does not adequately
consider that these studies fail to establish a causal link between the RFS and their reported results.
Other recent publications similarly do not find a quantitative causal relationship between the RFS and
LUC, with studies confirming our prior findings that the work of Lark et al. (2015) and Wright et al.
(2017) is unreliable. The RIA should be updated to acknowledge the shortcoming of such studies and
address the more recent literature.
Further, EPA largely ignores several important factors in play that negate or mitigate potential impacts
of the RFS on land and water; these include:
• Continued improvement in crop yield satisfies increased demand for corn without the need for
extensification and LUC.
• Cropping practices and other practices at the farm level such as conservation tilling, and
vegetative buffers minimize impacts to soil, surface water, and groundwater.
• Production of Distillers Dried Grains with Solubles (DDGS) offsets a considerable amount of
demand for corn and soy as animal feed.
• Adoption of more efficient irrigation methods and advanced farming technologies minimize use
of irrigation water, pesticides, and fertilizers.
EPA should update the RIA to address the complex economic and biophysical links in the causal chain
associating the RFS with impacts to land and water. EPA should address each important link in the
causal chain, including data gaps and lack of any evidence substantiating one or more of the causal
links in the chain. EPA should also address the mitigating factors set forth above.
3 | P a g e
EPA’s consideration of impacts to wetlands, ecosystems, and wildlife habitat in the Second Triennial
Report and the RIA consists almost entirely of general descriptions of data on nationwide wetlands
losses, conversion of grassland habitat, discussions of waterfowl habitat loss, potential impacts to
aquatic habitats, and potential impacts of grassland conversion to agriculture on insect pollinators. In
both the Second Triennial Report and in the RIA, EPA acknowledges the uncertainty of efforts to
quantify a relationship between the RFS and wetland and grassland habitat losses, yet this is not
adequately reflected in its conclusions. For example, the RIA includes the following statements relative
to the proposed volumes for 2022:
• There is a possibility that the proposed volumes for 2022 may “inspire” an increase in
feedstock production which in turn may affect wetlands (page 91).
• There is a potential to “incent” additional production of biofuels that in turn, may affect
grasslands and other ecosystems (page 96).
Given the magnitude of the uncertainties described by EPA and others in establishing causation
between the RFS and such impacts, and considering the myriad of economic, biophysical, and social
links in the causal chain, the “possibility to inspire” feedstock production or “incent” biofuel production
appears to be vanishingly small. This should be given due consideration in the RIA.
Water Quantity and Water Quality
As with the discussion of LUC, the RIA presents lengthy discussions of impacts associated with
agriculture in general (including biofuel feedstocks) and water quantity and water quality; however,
the RIA fails to adequately describe the complex economic and biophysical links in the causal chain
associating these impacts with the RFS. EPA should reevaluate the discussions of generic impacts as
well as address the causal chain as set forth above. In addition, to the extent EPA discusses the
hypoxic zones in western Lake Erie and the Gulf of Mexico it must explain that there has been no
quantitative attribution of these water quality impacts to the RFS and that any such attribution is
conjecture. We also recommend that EPA enhance the discussion of technological improvements in
agriculture that reduce water and agrichemical use.
4 | P a g e
THE RFS AND LAND USE CHANGE
The Draft RIA has three critical failings in its discussion of the potential role of the RFS on LUC:
• It relies on flawed studies that underlie the Second Triennial Report without due consideration
of more recent work that shows there is no demonstrated causal link between the RFS and
LUC, as EPA itself has acknowledged since it prepared the Second Triennial Report.
• It fails to consider the high degree of uncertainty in the causal relationship between the RFS
and biofuels prices, which is a fundamental assumption underlying assessments of the RFS,
and it fails to acknowledge that such a relationship has not been adequately quantified.
• It fails to address important mitigating factors including the continuing increase in crop yield,
adoption of conservation farming practices and modern farming technology, and production of
DDGS.
These shortcomings are discussed below.
Reliance on Flawed Research and No Established Causal Relationship
Some investigators have asserted that the RFS has resulted in extensive conversion of nonagricultural land to agriculture due to increased demand for corn for ethanol. Our findings indicate that
these claims are not borne out, in part because the studies use unreliable databases, present flawed
data analysis, and/or do not attempt to establish a causal link between the RFS, increased ethanol
production, and LUC. Indeed, EPA (2019) repeatedly asserts that no causal connection has been
established between LUC associated with corn production and the RFS.
As background, in the discussion of potential impacts of the RFS on LUC in its Second Triennial Report,
EPA repeatedly cites geospatial analysis conducted by the following researchers who used the Crop
Data Layer (CDL)4 dataset:
• Lark et al. (2015) analyzed LUC nationwide during the period 2008-2012 using CDL calibrated
with ground-based data from USDA’s Farm Service Agency (FSA), and data from the National
Land Cover Database (https://www.mrlc.gov/national-land-cover-database-nlcd-2016). The
authors reported that 7.34 million acres of previously uncultivated lands became utilized in
crop production and of those 1.94 million acres (785,000 ha.) of converted lands were planted
in corn as a first crop.
• Wright et al. (2017) assessed quantitative spatial relationships between the loss of grasslands
and the locations of ethanol refineries with the intent of associating this LUC with demand for
ethanol. Wright et al. (2017) note that approximately 2 million acres of grassland was
converted to row crops within 50 miles of an ethanol refinery between 2008 and 2012.
Several investigators have shown that reliance on inadequately corrected and verified CDL data leads
to an unacceptable level of uncertainty in geospatial analysis and potentially misleading results and
conclusions from such analysis. For example, Dunn et al. (2017) examined data for 2006-2014 in 20
counties in the prairie potholes region using the CDL, a modified CDL dataset, data from the National
Agricultural Imagery Program, and ground-truthing. Dunn et al. (2017) concluded that analyses
4 The CDL data set is developed by the U.S. Department of Agriculture National Agricultural Statistics Service
(https://www.nass.usda.gov/Research_and_Science/Cropland/SARS1a.php).
5 | P a g e
relying on CDL data returned the largest amount of LUC by a wide margin. They further concluded
that errors associated with CDL-based analyses resulted in estimates where “the number of hectares
in the potential error associated with CDL-derived results is generally greater than the number of
hectares the CDL-based analysis determined had undergone a transition from grassland, forested
land, or wetland to agricultural land.” This suggests that errors in classification inherent in the CDL can
result in uncertainty bounds that are larger than the estimates themselves (thereby even predicting
negative land conversion to agriculture).
Specifically, Dunn et al. (2017) pointed out that the findings reported by Lark et al. (2015) contradict
USDA data indicating that cropland area has remained almost constant during the period 2008-2012.
Similar critiques have been published of the findings reported by Wright et al. (2017) using the CDL
data set. The RIA does not acknowledge a major shortcoming of the study by Wright et al. (2017),
namely, the authors’ admission that their study “did not consider potential effects of other explanatory
variables.” The paper also discussed the errors in the data itself, stating that the “conversion of noncropland to cropland was mapped correctly over 70% of the time,” which means that it was mapped
incorrectly 30% of the time, a considerable percentage.
Several other recent papers challenge the findings of previous authors who have implicated the RFS in
LUC. Of note are papers by Pates and Hendricks (2019), Pristola and Pearson (2019), and Shrestha et
al. (2019). Pates and Hendricks (2019) assessed the local impact of ethanol plants on cropland
transitions and concluded that ethanol plant expansions reduce the probability of cropland conversion
by 0.5% on average and that fields near ethanol plants were 10% less likely to convert nonagricultural land into cropland than fields farther away. The authors acknowledge that this result
contradicts their underlying premise that ethanol plants induce LUC through local effect on corn prices.
They speculate (without providing evidence) that this contradiction may result from bias due to
concurrent changes in Crop Reserve Program (CRP) policy that disproportionately affected areas near
ethanol plants.
Pristola and Pearson (2019) performed a critical review of literature that was relied on by EPA in its
Second Triennial Report (and again in the RIA) regarding the RFS and LUC and concluded that major
flaws in the work by Wright and Wimberly (2013), Lark et al. (2015), and Wright et al. (2017) render
the work by these authors unreliable. Their major findings are as follows:
• All three studies reviewed relied on data from the CDL which has several shortcomings
including the inability to differentiate between native prairie, CRP, grass/hay, grass/pasture,
and fallow/idle grassland types, especially in earlier years.
• Improvements in the CDL over time make it problematic to compare land cover and land use
over relatively long time periods. Thus, results reported in the three studies might be biased
due to the CDL’s ability to better identify cropland in later years than earlier years. This bias
would give the appearance that cropland expanded, as these authors assert.
• All three studies reported cropland expansion over the conterminous United States, but this is
contradicted by data from the NASS that show a contraction of cropland from 2008 to 2012,
and that by 2017, cropland acres were below 2007 levels.
Both Lark et al. (2015) and Wright et al. (2017) relied on CDL data for Iowa for 2008 and 2012. Data
from NASS revealed that during the period 2008-2012 in Iowa there was a net increase of cropland of
6 | P a g e
38,000 acres as compared to an increase of 263,468 acres as reported by Lark et al. (2015) and
295,100 acres as reported by Wright et al. (2017).
Shrestha et al. (2019) studied the relationship between biofuel demand, food prices, and LUC. One of
the authors’ objectives was to assess the accuracy of automated land use classification as performed
by previous investigators (including Lark et al. 2015) as compared to manual land use classification
techniques. For this analysis, the authors selected study areas within three counties near Moscow,
Idaho with a total land area of 664 km2
. The areas were selected to represent a range of climates and
proportions of land cover types. Their work revealed that 10.90% of non-agricultural land was
misclassified as agriculture, whereas only 2.23% of agricultural land was misclassified as nonagricultural. The automated classification showed an 8.53% increase in agricultural land from 2011 to
2015, while the manual classification showed only a 0.31% (±1.92%) increase. The result derived via
manual classification was within the margin of error suggesting that there was no significant LUC
during the period.
These recent findings further call into question EPA’s continued reliance on flawed studies in its
discussion of the RFS and LUC, and indeed call into question whether there is any quantifiable causal
link between RFS and LUC.5
Inadequate Consideration of the Drivers of Biofuels Feedstock Prices
Studies that attempt to link the RFS with impacts to land and water (especially those studies focusing
on LUC) include a foundational presumption that there is a causal link between the RFS and biofuel
feedstock prices. Econometric models used to quantify the relationship have a high degree of
uncertainty, partly because agricultural commodities are traded on international markets and their
production is affected by highly uncertain and seasonally variable weather conditions. The RIA
acknowledges this uncertainty by stating that “…models that attempt to project prices at specific times
in the future, or in reaction to specific demand perturbations, necessarily contain high levels of
uncertainty” (page 209). The RIA goes on to discuss the relationship between grain stores and futures
prices and how annual volumes of grain stores depend on current year harvests and future year
harvest projections. The RIA provides no discussion of the relationship, if any, between grain stores
and the RFS. The RIA acknowledges that, in a general sense, grain prices are influenced by “an array
of factors from worldwide weather patterns to biofuel policies to international tariffs and trade wars”
(page 209). Finally, the RIA presents results from a meta-analysis of the impact of increased biofuel
production on corn prices.6 Based on the results of the single meta-analysis by Condon et al. (2013)
conducted almost a decade ago the RIA projects that the proposed ethanol volumes for 2021-2022
will increase the price of corn 3% per billion gallons, or $0.11per bushel.
5
In addition, several publications released since August 2019 reported on LUC such as conversion of grassland to
corn and soy (e.g., Zhang et al. 2021; Lark et al. 2019b; and Arora and Wolter 2018) and cropland expansion and
potential wildlife impacts (Lark et al. 2020); however, these studies do not attempt to establish causal linkages
between increased demand for ethanol from the RFS and LUC. Rather, these articles and others reveal a trend
among researchers toward improving the accuracy of geospatial modeling to discern specific LUC which appears to
be a shift from previous efforts to associate LUC with the RFS. In addition, several authors focused on assessing
the environmental benefits of improved agricultural practices, conservation, and restoration, and policy actions to
reduce grassland losses (e.g., Fargione et al. 2018; Lark 2020; and Runge et al. 2017).
6 EPA states that Condon et al. reviewed published papers in 2015, when in fact, the working paper was released
in 2013 and reviewed papers published between 2008 and 2013.
7 | P a g e
Despite the acknowledgment of the high degree of uncertainty in econometric models attempting to
link the RFS to biofuel feedstock prices, the RIA presents no discussion of such uncertainty and rather
proposes the projected price increases estimated by Condon et al. (2013) as fact. For example, the
RIA presents no information on whether the meta-analysis conducted by Condon et al. (2013)
controlled for how and to what extent corn prices were affected by rapid economic growth in
developing countries leading to growing food demand or how corn prices were affected by a dietary
transition from cereals toward more animal protein. As a result of these market factors alone, global
consumption of agricultural commodities has been growing rapidly. Further, the temporal fluctuation
in corn prices is highly influenced by the effect of the price of oil on production inputs such as
agrichemicals and fuel for farm equipment, and this relationship is not mentioned by EPA in either the
Second Triennial Report or the RIA. Also significant is a study by Shrestha et al. (2019) that analyzed
food price inflation and land use classification and concluded that food price inflation since 1973 was
lowest during the biofuel boom years of 1991-2016 and was most highly correlated with the price of
oil.
Figures 1 and 2 show nominal prices of West Texas Intermediate crude and corn for the latest 20-year
period (the shaded areas on the graphs show period of US recessions) and demonstrates that corn
prices track very closely to the price of oil.
Figure 1. West Texas Intermediate Crude Price ($/barrel).
8 | P a g e
Figure 2. U.S. Corn Price ($/bushel).
SOURCE: Macrotrends n.d.
By contrast to the close relationship between U.S. corn prices and the price of crude oil, Figure 3
shows a plot of U.S. average annual corn prices and the implied ethanol volumes 2012-2020 showing
no apparent relationship.
Figure 3. Corn prices versus implied conventional ethanol volumes 2012-2020.
SOURCES: https://www.nass.usda.gov/Charts_and_Maps/graphics/data/pricecn.txt and Congressional Research
Service (2020).
12.00
12.50
13.00
13.50
14.00
14.50
15.00
15.50
0.00
1.00
2.00
3.00
4.00
5.00
6.00
7.00
8.00
2012 2013 2014 2015 2016 2017 2018 2019 2020
Billion Gallons
Dollas Per Bushel
Corn Prices and Implied Conventional
Biofuel Volume
Average Monthly Corn Price ($/bushel)
Implied conventional volume (billions of gallons)
9 | P a g e
Inadequate Consideration of Mitigating Factors
The RIA, like the Second Triennial Report, fails to adequately consider important factors that act to
dampen any effect of the RFS on converting non-agricultural land to agriculture for ethanol feedstock,
including:
• Increased demand for corn for all uses has mostly, if not fully been met by increased yields.
• Cropping practices such as double cropping increase production with no additional need for
land in cultivation.
• Production of dried distillers’ grains with solubles (DDGS) by ethanol refineries offset a
considerable amount of demand for land to grow corn and soy for animal feed.
Increases in Corn Crop Yield
EPA has not adequately accounted for the fact that increased demand for corn for ethanol and the
effect of this increased demand on land conversion, if any, has been offset by increases in yield over
time. In fact, the number of acres planted in corn across the United States in the last couple of
decades has remained close to or below the total acres planted in the 1930s, despite increases in
demand for corn as human food, animal feed, and biofuels over this nearly 90-year period. The
increase in demand has largely been met by an approximately 7-fold increase in yield (bushels per
acre) (Figure 4). The USDA further anticipates that changes in corn production will result in an
increase of approximately 16.1 more bushels per acre by 2028 without a substantial increase in
farmed acreage.
Figure 4. Relative Change in Acres of Corn Planted and Yield (1926-2021)
10 | P a g e
Cropping Practices
Intensification refers to increasing the production of a crop on the same acreage of land and does not
directly result in LUC. Extensification refers to increasing production of a crop by planting on land not
previously in agriculture. A farmer can intensify production of a crop by switching crop types to a more
desired crop or by double planting a single crop (double-cropping) instead of seasonally rotating
crops. EPA’s Second Triennial Report acknowledges the potential significance of cropping practices in
meeting any increased demand for corn due to the RFS. By contrast, the RIA does not mention
cropping practices in this context, rather it discusses double cropping only briefly and only as it relates
to potential effects on water quality. In its Second Triennial Report, EPA cites a study by Ren et al.
(2016) in Iowa (the state with the largest corn production) that examined changes in corn and
soybean rotations around 2017 and found that 59% of the area that had been in corn/soy rotation
prior to 2007 was in two or more years of continuous corn after 2007. However, EPA fails to
acknowledge the most important conclusion related to LUC from this study: “… it is clear that the
expansion of corn production after 2007 was realized by altering crop rotation patterns” (Ren et al.
2016).
Further, the RIA mentions the findings reported by Plourde at al. (2013) regarding farmers switching
from corn/soy rotation to double cropping of corn as a means of increasing corn production but fails to
acknowledge its significance in the context of intensification in lieu of extensification to meet demand.
Rather, Plourde et al. (2013) is mentioned in terms of potential effects on nitrogen and phosphorous
loads to surface waters.
Use of Distillers Grains and Solubles as a substitute for Corn and Soy in Animal Feed
EPA has not adequately accounted for the fact that the ethanol industry produces large amounts of
distiller’s grains with solubles (DDGS) and that this biproduct is used as animal feed where it
substitutes for traditional grains such as corn and soy. EPA’s Second Triennial Report acknowledges
the role of DDGS in offsetting overall demand for corn and soy; however, this fact is ignored in the
RIA.7 Production of DDGS and its use as a substitute for corn and soy in animal feed likely has a very
important mitigating effect on any potential contribution of the RFS to LUC. This effect can be
estimated based on the annual volume of corn grown for ethanol (bushels), the annual figures for
yield (bushels per acre) and the following assumptions:
• 17 lbs of DGGS are produced per bushel of corn processed8
• 1 lb of DGGS is equivalent to 1.22 lbs of corn/soy (Hoffman and Baker 2021)
• One bushel of corn/soy weighs on average 58 lbs
7 The Second Triennial Report states that approximately 12% of the total corn production from 2014-2016 was
returned to the feed market in the form of DDGS. The Report also acknowledges a study by Mumm et al. (2014)
that concludes that 40% of corn grown in 2011 was estimated to be utilized in ethanol production; however, when
the offsetting effect of DDGS is accounted for, the acreage devoted to corn for ethanol goes down to 25%. The
report also does not acknowledge that these same authors estimate that the percentage of land devoted to corn for
ethanol will drop further to 13% by 2026 due to technological advances increasing crop yield as well as increasing
the efficiency of the ethanol distillation process. It is curious that the RIA mentions this study as well, but only in
the context of water quality.
8 Explaining Fluctuations in DDG Prices – Center for Commercial Agriculture (purdue.edu) (accessed 1/5/2022).
11 | P a g e
Applying these assumptions to the annual amount of corn grown for ethanol and annual yield during
the period 2008 to 2020, production of DDGS is estimated to have offset corn/soy equivalent acreage
ranging from 8.7 million acres in 2008 to 13.5 million acres in 2012 with an annual average of 11.2
million acres over the period9
.
9 Figures on corn displaced by DDGS 2010-2020 from World of Corn 2021. By comparison, the U.S. Corn Growers
Association estimates 6.0-8.6 million acres per year over the period 2010-2020 with an average of 6.9 million
acres per year but it is not clear how its estimate was calculated.
12 | P a g e
The RFS AND CONVERSION OF WETLANDS, ECOSYSTEMS, AND WILDLIFE
HABITAT
The introduction to this section in the RIA simply repeats information presented and conclusions made
in the Second Triennial Report regarding land use change. Without providing any supporting
information, the introductory paragraph to this section wrongly concludes that “Evidence from
observations of land use change suggests that some of this increase in acreage and crop use is a
consequence of increased biofuel production.”
In the discussion of wetlands, the RIA presents information in reports from several federal agencies
that describe the status and trends of U.S. wetlands. These reports and sources of data merely record
changes in wetland types nationwide and do not provide any analysis of the cause of the changes that
would be useful in the context of the RIA. The RIA goes on to discuss “several regional studies” of
changes in wetland area but highlights only one study: Wright et al. (2017). The RIA concludes that
this study demonstrated a causal connection between the proximity of an ethanol refinery and loss of
wetlands. The RIA also relies on Wright et al. (2017) in its discussion of losses of shrubland and forest
ecosystems. As described above, the study by Wright et al. (2017) has been shown to be unreliable.
The RIA also discusses loss of land in the CRP and references a single study by Morefield et al. (2016)
who conclude that CRP land lost between 2010 and 2013 largely went to conversion to row crops for
corn and soy. The authors of the study do not try to attribute the loss of CRP to increased demand for
biofuels, rather they acknowledge that important drivers of extensification at the expense of grassland
and wetlands include a combination of “commodity prices, reduced land retirement options, and
diminishing interest in land retirement programs…”
In its discussion of wildlife impacts, the RIA mentions loss of wetlands and impacts to ducks, and loss
of grasslands and impacts to grassland birds and insects. The RIA acknowledges that the effects of the
RFS on wildlife have not been studied, yet presents results from a study of grassland bird diversity
and cropland that implicates LUC in reduced species diversity, and studies of impacts to pollinators,
including a discussion of the potential role of exposure to agrichemicals.10 The RIA does not infer a
causal relationship between the RFS or crops grown for biofuel feedstock as a driver for effects to
wildlife and concludes that “[a]t present it is not possible to confidently estimate the fraction of wildlife
habitat loss or of corn or soy production that is attributable to biofuel production or use. Thus, we
cannot confidently estimate the impacts to date on wildlife from biofuels generally nor from the annual
volume requirements, specifically” (pages 98-99).
The discussion of potential impacts to wetlands ecosystems, and wildlife habitat presented in the RIA
(as well as in the Second Triennial Report) is unbalanced and creates a false impression that the
generic impacts described are attributable to the RFS. EPA should reevaluate this discussion to present
a balanced perspective that accurately presents the current state of knowledge regarding the lack of a
quantitative relationship between biofuel feedstock grown specifically to meet the RFS requirements
and potential impacts.
An expanded review of literature on this topic since our work on Exhibits 1 and 2 concludes that no
publications establish a quantitative or qualitative causal link between impacts from biofuel feedstock
10 This includes a discussion of neonicotinoids which concludes that the risk of these chemicals to pollinators is
poorly understood, and EPA’s preliminary determination is that the risk is low.
13 | P a g e
production and impacts to wetlands, ecosystems, or wildlife habitat.11 Wiens et al. (2011) suggests a
linkage between biofuels and potential biodiversity impacts by implicating demand for ethanol in the
loss of CRP lands but provides no analysis of causation. Wimberly et al. (2018) implicate corn and soy
extensification in increases in grassland habitat fragmentation in eastern South Dakota and western
Minnesota. These authors state that the LUC was driven by higher corn prices driven by increasing
demand for ethanol, and they cite Lark et al. (2015), Wright et al. (2017), and Wright and Wimberly
(2013)—all studies that have been largely discredited, as described above. Hoekman and Broch
(2018) describes benefits and “dis-benefits” of LUC ostensibly driven by higher ethanol prices but
provide no quantitative or qualitative causal links to the RFS or corn grown for ethanol.
In sum, the RIA should reflect that the latest scientific literature does not establish any causal
relationship between the RFS and impacts to wetlands, ecosystems, or wildlife habitat.
11 For many publications on this topic, that is not the goal. For example, Pleasants (2017) quantifies milkweed
stem abundance in soy and corn fields in the U.S. Midwest and develops estimates of the restoration required to
increase monarch butterfly populations but does not mention the RFS or biofuels. A paper by Landis et al. (2018)
models biodiversity and ecosystem services under different biofuels cropping systems, but makes no mention of
the RFS or corn grown for ethanol.
14 | P a g e
THE RFS AND WATER RESOURCE AVAILABILITY AND WATER QUALITY
Both the Second Triennial Report and the RIA present detailed discussions of water use to grow corn
as well as potential adverse effects to water quality from agriculture in general due to soil loss and
transport of pesticides and fertilizers to surface water bodies. However, EPA provides no causal nexus
between these potential impacts and agriculture to produce biofuels in general, or more to the point,
production of biofuels for the RFS.
Further, existing studies suggest that environmentally protective goals for biofuel production are
achievable as best management practices and technological advances in farming continue to be
adopted by the farming community.
Water Resource Availability
Water Used for Growing Corn
As is true for LUC, the relationship between corn production and water resource availability and water
quality varies geographically and temporally, and studies have failed to demonstrate a quantitative
causal link between corn grown for ethanol or the RFS and impacts to water availability. The RIA,
EPA’s Second Triennial Report, and studies cited therein provide information regarding the
geographical distribution of irrigated agriculture in general and biofuel crops relative to known
stressed aquifers, but there is no evidence or analysis provided regarding the relative impact of corn
grown for ethanol production in response to RFS mandates.
In its discussion of life cycle water use for biofuel feedstock production, the RIA relies on information
and analysis presented in the Second Triennial Report, including specific reference to Lark et al.
(2015) and Wright et al. (2017) regarding these authors’ now refuted conclusions about cropland
expansion associated with biofuels and the RFS. The RIA does not present any assessment of the
actual water intensity of corn grown for ethanol, rather it presents some general statistics regarding
use of water to grow corn and soy. For example, the RIA makes the following statements:
• 90% of corn is grown in areas where corn is non-irrigated.
• If 20% of corn production was used to produce 12 billion gallons per year of ethanol, this
would represent only 4.4% of all irrigation withdrawals (citing a study by Dominguez-Faus et
al. 2013).
• Nebraska is one of the states with the largest water withdrawals for irrigation and recent
increases in irrigation withdrawals have been largely driven by the need to irrigate corn for
ethanol (citing a report by the National Academy of Sciences; NAS 2011).
These statements are of little value without providing specific context relative to corn grown for
ethanol, and this context is highly geographically variable. In particular, the statement by NAS (2011)
regarding irrigation withdrawals in Nebraska and corn grown for ethanol is unsubstantiated by the
authors of the report and should not have been cited in the RIA. Moreover, it is outdated by over a
decade. However, the RIA acknowledges that “…there have been no comprehensive studies of the
changes in irrigated acres, rates of irrigation, or changes in surface and groundwater supplies
attributed specifically to the increased production of corn grain-based ethanol and soybean-based
biodiesel” (page 123).
15 | P a g e
A review of literature not previously presented by EPA did not reveal any studies attempting to
quantify water use specifically for growing corn that was destined to produce ethanol required to meet
RFS requirements. Xie et al. (2019a), Xie and Lark (2021), and Xie et al. (2021) present findings
related to mapping of irrigated land in the U.S., but they present no nexus between water use and
the RFS or water use and corn grown for ethanol. Xie et al. (2019b) present a method for mapping
annual irrigation distribution over the period 2000-2017 and conclude that irrigation over the period
2009-2017 was greater than over the period 2000-2008 and that the greatest increase was in
Nebraska and was associated with corn and soy. However, the paper contains no mention of the RFS,
biofuels, or ethanol and therefore does not attempt to link the increase in annual irrigation to
renewable fuels policy and its conclusion regarding association of increased irrigation with corn and
soy is unfounded.
The impacts of irrigation withdrawals to grow corn for ethanol are dependent on the existing condition
of available water resources, the use of irrigation water to produce corn relative to other crops, and
the proportion of irrigated corn grown that is used in the production of ethanol. Although the RIA
acknowledges these complexities, it fails to explicitly relate them to the RFS or to corn grown for
ethanol. The RIA postulates three approaches for estimating the change in water demand that may
result from increased ethanol volumes: life cycle water requirements for ethanol as compared to
gasoline, projected LUC and crop management, and changes in crop prices and associated economic
value of irrigation. The RIA acknowledges EPA’s inability to perform such analyses yet concludes that
there is likely to be some increased irrigation pressure on water resources due to the proposed
ethanol volumes. The RIA fails to acknowledge that “increased irrigation pressure,” to the extent such
a thing may occur, does not necessarily translate to increased overall water use or strain on existing
water resources. Moreover, the RIA asserts that the changes in irrigation may result from the
proposed volumes impact on crop prices without first establishing that any such impact on crop prices
has occurred historically or is likely to occur as the result of the proposed volumes, which are similar
to the volumes in 2019.
Water Use for Ethanol Production
Ethanol refineries have made great strides in reducing water consumption. In a 2007 Renewable Fuels
Association survey of 22 ethanol production facilities (representing 37% of the 2006 volume
produced), dry mills used an average of 3.45 gallons of water per gallon of ethanol produced and wet
mills used an average of 3.92 gallons of water per gallon of ethanol produced. Muller (2008) reported
declines in water requirements at ethanol dry mills from 5.8 gallons of water per gallon of ethanol
(gal/gal) in 1998 to 2.7 gal/gal in 2012. Wu and Chiu (2011) noted that water consumption in existing
dry mill plants had, on a production-weighted average basis, dropped 48% in less than 10 years, a
reduction that is similar to that reported by Muller (2008). These previously reported improvements in
efficiency are confirmed by the latest scientific literature. For example, Wu (2019) reports a 54%
decrease in water intensity for the ethanol industry over the period 1998-2017 and a 12% decrease
over the period 2011-2017 illustrating the gains in water use efficiency at ethanol plants.
Improvements in water use efficiency at ethanol refineries are largely ignored by EPA.
16 | P a g e
Advancements in Farming Practices Reduce Agriculture’s Impacts on Water Resource
Availability
What is clear but not adequately recognized by EPA in the RIA or Second Triennial Report is that
advancements in farming practices and technology have reduced the negative impact of farming on
water resource availability. Over the past decade, there has been increased use of precision
agriculture methods as well as standard best practices which retain soil moisture. This trend is
expected to continue and is expected to reduce the need for irrigation. As an indication of the trend in
irrigation reduction, the University of Nebraska (2018) reports that in Nebraska (as a bell-weather of
other dry western states), the percentage of all corn acreage that is irrigated has declined from a high
of 72% in 1981 to 56% in 2017.
Farms are increasingly moving away from traditional, less-efficient irrigation systems and adopting
water saving irrigation systems. In 2018, 67% of cropland acres irrigated used pressurized systems
including sprinklers and low-flow micro systems (Hrozencik and Aillery 2021). The number of farms
using inefficient gravity irrigation systems decreased from 62% in 1984 to 34% in 2013, converting
mostly to pressure-sprinkler irrigation which is more efficient than gravity irrigation. Water savings
associated with advanced irrigation systems relative to typical gravity systems are summarized
below12:
• Subsurface Drip: 25-35%
• Rainwater Harvesting: 50%
• Precision Agriculture: 13%
• Conservation Structures: 18%
In terms of the adoption of precision agriculture, almost 10% of farms use soil-moisture or plantmoisture sensing devices or commercial irrigation scheduling services. Sensor technology can
optimize irrigation scheduling and hence increase water use efficiency. It is also anticipated that
additional large industrial farms (which make up a large volume of total production) will employ water
use simulation models that are based on corn growth patterns and weather conditions. Adoption of
these technologies will continue to grow in the U.S., and particularly in the west, where 72% of water
irrigation takes place and farmers have recent experience with low water supply following the 2012-
2016 drought. In addition to changes in irrigation technologies, agricultural practices regarding the
timing of irrigation have helped reduce the amount of water applied to corn. For example, Xue et al.
(2017) have shown that corn crops can forego initial irrigation without significant adverse effects to
yield.
Genetic engineering and selection for improved drought tolerant corn cultivars has resulted in corn
strains that can tolerate a 25% reduction in water application without affecting yield. The use of
drought-tolerant corn, which was commercially introduced in 2011, increased to over 22% of the total
U.S. planted corn acreage by 2016 (Mcfadden et al. 2019). More importantly, this percent of use was
greatest in the driest corn-producing states of Nebraska (42%) and Kansas (39%). Other states that
12 See for example Ailen 2013, Barton and Clark 2014, Biazin et al. 2012, Center for Urban Education about
Sustainable Agriculture (CUESA) 2014, Gowing et al. 1999, Shangguan et al. 2002, National Research Council
2008, Netafim n.d., and Qin et al. 2015.
17 | P a g e
are not as drought prone (e.g., Minnesota, Wisconsin, and Michigan) saw drought-tolerant corn
planting ranging between 14% and 20% of total acreage in 2016. Figure 5 illustrates the downward
trend in the volume of water applied to corn over the period 1994-2018.
Figure 5. Average volume of water applied to corn over the period 1994-2018.
SOURCE of Data:
ttps://www.nass.usda.gov/Publications/AgCensus/2017/Online_Resources/Farm_and_Ranch_Irrigation_Survey/fris_2_0036_0036.
pdf Accessed 1/5/2022
Water Quality
The RIA discussion of potential water quality impacts, like the Second Triennial Report, fails to
establish a causal link between corn grown for the RFS and water quality impacts. The RIA discusses
several studies addressing impacts to soil and surface water from corn and soy, including erosion, soil
carbon depletion, and nutrient runoff. These impacts are inextricably linked to LUC as well as crop
intensification, but the results of these studies are relevant only to the extent the RFS-LUC or RFSintensification link can be demonstrated and quantified. The RIA at page 101 presents a simplistic
example calculation of the increased nitrogen applied to farm fields nationwide due to corn
extensification, but this calculation is based on the work of Lark et al. (2015) which has been shown to
be unreliable. This example calculation should be removed from the text of the RIA because it is
erroneous and misleading. Similarly, the RIA discusses data from USDA NASS for percentages of
planted corn acres that received treatment using herbicides, insecticides, and fungicides, but such
information is irrelevant to an analysis of the water quality impacts of the RFS unless it can be
quantitatively tied to the program. Finally, the RIA cites work by Garcia et al. (2017) who estimate
that corn production between 2002 and 2022 would result in nitrate groundwater contamination > 5
mg/L in areas with sandy or loamy soils. Although Garcia et al. (2017) mention the RFS as a potential
driver for increased corn production, the study does not derive a quantitative causal relationship
between their conclusions and the RFS, nor is that a stated goal of the research. It is inappropriate for
EPA to present this work without adequate context.
R² = 0.4346
0
0.5
1
1.5
2
1994 1998 2003 2008 2013 2018
Average Volume of Water Applied to Corn
for Grain or Seed (Acre-feet/Ac.)
Pressure Systems (ac-ft/ac) Gravity Systems (ac-ft/ac)
18 | P a g e
In the absence of evidence of a quantitative causal relationship, the discussion in the RIA of the
relationship between agriculture (and corn growing in particular), and proximal water quality is
unbalanced and creates the unfounded impression of a direct causal relationship.
Similarly, the RIA presents a discussion of the potential downstream effects of corn and soy
cultivation. In terms of aquatic life, the RIA presents a discussion of the biological condition of the
nation’s rivers and streams and the causative factors contributing to poor conditions. As with its
Second Triennial Report, the RIA mentions large scale hypoxia in western Lake Erie and the Gulf of
Mexico and the nexus to nutrient enrichment. The RIA also discusses the potential for downstream
impacts from herbicides and pesticides applied to corn and soy. Throughout these discussions, the RIA
provides no nexus to the RFS.
Like the Second Triennial Report, the RIA fails to acknowledge that there is no established causal
connection between corn grown for ethanol and the formation, persistence, or severity of hypoxic
events in western Lake Erie or the Gulf of Mexico. The RIA ignores a rich literature describing the
complexity of these phenomena as well as characterization and modeling of nutrient loading to these
systems from various sources, but such studies fail to establish a causal relationship between the
formation and severity of the GoM dead zone and the RFS. For example, econometric modeling by
Secchi et al. (2011) predicts an increase in corn acreage due to corn intensification spurred by
increasing corn prices and extends that prediction to estimate increased nutrient loading to the Upper
Mississippi River basin; however, these authors did not attempt to assert a causal connection between
increased corn prices (the variable that drives their analysis) and the RFS.
As illustrated in Figure 6, the RIA fails to acknowledge that nitrogen loading to the Gulf of Mexico has
remained fairly stable over the past 40 years.
Figure 6. Annual Nitrate and Nitrite Loading to the Gulf of Mexico 1980-2020.
Source: USGS n.d.
19 | P a g e
There may be no dispute that excess nutrient loading from the key watersheds that discharge into
western Lake Erie and the northern Gulf of Mexico contribute to eutrophication and hypoxia; however,
these watersheds contain a complex mix of urban and rural uses that present important sources of
nutrients as well as toxic contaminants. In any case, the direct causal link to the RFS or corn grown
for ethanol production (compared to all other uses and compared to all other agricultural activities) is
not substantiated by the Second Triennial Report or the literature cited therein and should be qualified
as such to the extent discussed in the RIA. Regional hypoxic conditions in western Lake Erie and the
Gulf of Mexico were increasing in frequency and severity, long before ethanol production increased,
and this fact should also be acknowledged by EPA.
The RIA also fails to acknowledge the importance of regional weather on the occurrence and severity
of large-scale hypoxia events. The National Oceanic and Atmospheric Administration (NOAA) states
that a major factor contributing to the large Gulf of Mexico “dead zone” in 2019 was the abnormally
high amount of spring rainfall that resulted in flows in the Mississippi and Atchafalaya Rivers that were
67% above the average flows over the previous 38 years
13. Data collected by the United States
Geological Survey (USGS) indicate that because of these high flows, nitrate loads were about 18%
above the long-term average, and phosphorus loads were approximately 49% above the long-term
average (USGS 2019).
Notwithstanding the misleading discussions presented in the RIA, the RIA correctly acknowledges 1)
that the important determinants of impacts to water and soil quality are not directly determined by
the RFS; 2) there are many effective management practices that can act to counterbalance any
negative impacts from corn for ethanol; and 3) the magnitude of potential impacts due to the RFS
cannot be estimated at this time. The RIA should be edited to present a more balanced discussion of
potential water quality impacts from biofuel feedstock agriculture within the specific context of crops
grown for biofuels to meet the goals of the RFS program.
13 Courtney and Courtney (no date) reported that predictions made by NOAA and Louisiana University’s Marine
Consortium of the areal extent of the GoM dead zone were 31% higher than the actual measured hypoxic areas
from 2006 to 2014. The authors of this paper hypothesize that GoM waters are becoming less susceptible to low
dissolved oxygen over time.
20 | P a g e
REFERENCES
Allen L. 2013. Smart Irrigation Scheduling: Tom Rogers’ Almond Ranch.
Arora, G. and P. Wolter. 2018. Tracking land cover change along the western edge of the U.S. Corn
Belt from 1984 through 2016 using satellite sensor data: observed trends and contributing
factors, Journal of Land Use Science, 13:1-2, 59-80, DOI:10.1080/1747423X.2018.1466001.
Barton B., and S. Elizabeth Clark. 2014. Water & Climate Risks Facing U.C. Corn Production.
http://www.ourenergypolicy.org/wp-content/uploads/2014/06/ceres-corn.pdf.
Biazin B., G. Sterk, M. Temesgen, A. Abdulkedir, and L. Stroosnijder. 2012. Rainwater harvesting and
management in rainfed agricultural systems in sub-Saharan Africa – A review. Physics and
Chemistry of the Earth 47–48:139–151. DOI:10.1016/j.pce.2011.08.015.
Center for Urban Education about Sustainable Agriculture (CUESA). 2014. 10 Ways Farmers Are
Saving Water. CUESA Articles:4–6.
Condon, N., H. Klemick, and A. Wolverton. 2013. Impacts of ethanol policy on corn pices: A review
and meta-analysis of recent evidence. NCEE Working Paper Series, Working Paper #13-05. U.S.
Environmental Protection Agency: National Center for Environmental Economics. Washington,
D.C.
Congressional Research Service. 2020. The Renewable Fuel Standard (RFS): An overview. Updated
April 14, 2020. https://sgp.fas.org/crs/misc/R43325.pdf
Courtney, M, and J. Courtney. No Date. Predictions wrong again on dead zone area – Gulf of Mexico
gaining resistance to nutrient loading. BTG Research, 9574 Simon Lebleu Road, Lake Charles, LA,
70607. https://arxiv.org/abs/1307.8064
Dominguez-Faus, R., C. Folberth, J. Liu, A. Jaffe, and P. Alvarez. 2013. Climate change would increase
the water intensity of irrigated corn ethanol. Environmental science & technology, 47(11), 6030-
6037.
Dunn J., D. Merz, K. Copenhaver, and S. Mueller. 2017. Measured extent of agricultural expansion
depends on analysis technique. Biofuels, Bioproducts and Biorefining 11:247–257.
DOI:10.1002/bbb.1750.
EPA. 2021a. Renewable Fuel Standard (RFS) Program: RFS Annual Rules. Proposed Rule. 40 CFR Parts
80 and 1090; EPA-HQ-OAR-2021-0324; FRL-8521-02-OAR; RUN 2060-AV11. Docket ID No.
EPA–HQ–OAR–2021–0324.
EPA. 2021b. Draft Regultory Impact Analysis: RFS Annual Rules. U.S. Environmental Protection
Agency: Assessment and Standards Division, Office of Transportation and Air Quality. EPA-420-
D-21-002.
EPA. 2019. Memorandum: Endangered Species Act No Effect Finding for the 2020 Final Rule – EPA-HQOAR-2019-0136.
EPA. 2018. Biofuels and the Environment: Second Triennial Report to Congress. DOI:EPA/600/R10/183F.
Fargione, J., S. Bassett, T. Boucher, S. Bridgham, R. Conant, S. Cook-Patton, P. Ellis, A. Falucci, J.
Forqurean, and B. Griscom. 2018. Natural climate solutions for the United States. Sci. Adv.
2018; 4 : eaat1869 14 November 2018.
Garcia, V., E. Cooter, J. Crooks, B. Hinckley, M. Murphy, and X. Xing. 2017. Examining the impacts of
increased corn production on groundwater quality using a coupled modeling system. Sci. Total
Environ. 586: 16-24.
21 | P a g e
Gowing, J.W., O. Mzirai, and H. Mahoo. 1999. Performance of Maize under Micro-Catchment
Rainwater. Harvesting in Western Pare Lowlands and Morogoro, Tanzania 2:193–204.
Hoekman, S. and A. Broch. 2018. Environmental implications of higher ethanol production and use in
the U.S.: A literature review. Part II – Biodiversity, land use change, GHG emissions, and
sustainability. Renewable and Sustainable Energy Reviews Volume 81, Part 2, January 2018,
Pages 3159-3177
Hoffman, L. and A. Baker. 2021. Estimating the substitution of distiller’s grain for corn and soybean
meal in the U.S. feed complex. United States Department of Agriculture Economic Research
Service.
Hrozencik, R. and A. Aillery. December 2021. Trends in U.S. Irrigated Agriculture: Increasing
Resilience Under Water Supply Scarcity, EIB-229, U.S. Department of Agriculture, Economic
Research Service. https://www.ers.usda.gov/webdocs/publications/102928/eib-229.pdf?v=882.4
Landis, D., C. Gratton, R. Jackson, and K. Gross. 2018. Biomass and biofuel crop effects on
biodiversity and ecosystem services in the North Central US. Biomass and Bioenergy Volume
114, July 2018, Pages 18-29.
Lark, T., S. Spawn, M. Bougie, and H. Gibbs. 2020. Cropland expansion in the United States produces
marginal yields at high costs to wildlife. Nature Communications.
https://doi.org/10.1038/s41467-020-18045-z
Lark, T. 2020. Protecting our prairies: Research and policy actions for conserving America’s
grasslands. Land Use Policy Volume 97, September 2020, 104727.
Lark, T., N. Hendricks, N. Pates, A. Smith, S. Spawn, M. Bougie, E. Booth, C. Kucharik, and H. Gibbs.
2019a. Impacts of the Renewable Fuel Standard on America’s Land and Water. Washington, D.C.
Lark, T., B. Larson, I. Schelly, S. Batish, and H. Gibbs. 2019b. Accelerated conversion of native prairie
to cropland in Minnesota. Environmental Conservation 46: 155–162.
DOI:10.1017/S0376892918000437.
Lark T., J. Salmon, and H. Gibbs. 2015. Cropland expansion outpaces agricultural and biofuel policies
in the United States. Environmental Research Letters 10. DOI:10.1088/1748-9326/10/4/044003.
Mcfadden J., D. Smith, S. Wechsler, and S. Wallander. 2019. Development, Adoption, and
Management of Drought-Tolerant Corn in the United States, EIB-204. United States Department
of Agriculture.
Morefield, P., S. LeDuc, C. Clark, and R. Iovanna. 2016. Grasslands, wetlands, and agriculture: the
fate of land expiing from the Conservation Reservce Program in the Midwestern United States.
Env. Res. Lett.
Muller, S. 2010. 2008 national dry mill corn ethanol survey. Biotechnol. Lett (2010) 32:1261-1264.
Mumm R., P. Goldsmith, K. Rausch, and H. Stein. 2014. Land usage attributed to corn ethanol
production in the United States: Sensitivity to technological advances in corn grain yield, ethanol
conversion, and co-product utilization. Biotechnology for Biofuels 7:1–17. DOI:10.1186/1754-
6834-7-61.
NAS. 2011. Renewable Fuel Standard: Potential Economic and Environmental Effects of U.S. Biofuel
Policy. National Academy of Sciences. Washington, DC.
National Research Council. 2008. Water Implications of Biofuels Production in the United States. The
National Academies Press, Washington, DC. DOI:10.17226/12039.
22 | P a g e
Netafim. (n.d.). Drip Irrigation for Corn. https://www.netafimusa.com/agriculture/cropapplications/corn/
Pates, N., and N. Hendricks. 2019. Estimating the local impact of ethanol plants on cropland
transitions. Selected Poster prepared for presentation at the 2019 Agricultural & Applied
Economics Association Annual Meeting, Atlanta, GA, July 21-July 23.
Pleasants, J. 2017. Milkweed restoration in the Midwest for monarch butterfly recovery: estimates of
milkweeds lost, milkweeds remaining and milkweeds that must be added to increase the
monarch population. Insect conservation and diversity 10:1 January 2017.
Plourde J., B. Pijanowski, and B. Pekin. 2013. Evidence for increased monoculture cropping in the
Central United States. Agriculture, Ecosystems and Environment 165:50–59.
DOI:10.1016/j.agee.2012.11.011.
Pristola, J. and R. Pearson. 2019. Critical review of supporting literature on land use change in the
EPA’s Second Triennial Report to Congress. Southern Illinois University, Edwardsville.
Qin W, C Hu, and O Oenema. 2015. Soil mulching significantly enhances yields and water and nitrogen
use efficiencies of maize and wheat: a meta-analysis. Scientific reports 5:16210.
DOI:10.1038/srep16210.
Ren J., J. Campbell, and Y. Shao. 2016. Spatial and temporal dimensions of agricultural land use
changes, 2001–2012, East-Central Iowa. Agricultural Systems 148:149–158.
DOI:10.1016/j.agsy.2016.07.007.
Runge, C., A. Planting, A. Larsen, D. Naugle, K. Helmstedt, S. Polasky, J. Donnelly, J. Smith, T. Lark,
J. Lawler, S. Martinuzzi, and J. Fargione. 2017. Unintended habitat loss on private land from
grazing restrictions on public rangelands. Research Article. J. App. Ecol.
Secchi, S., P. Gassman, J. Manoj, K. Lyubov, and C. Kling. 2011. Potential water quality changes due
to corn expansion in the Upper Mississippi River Basin. Ecological Applications 21:4
Shangguan ZP, MA Shao, TW Lei, and TL Fan. 2002. Runoff water management technologies for
dryland agriculture on the Loess. International Journal of Sustainable Development and World
Ecology 9:341–350. DOI:10.1080/13504500209470129.
Shrestha, D., B. Staab, and J.Duffield. 2019. Biofuel impact on food prices index and land use change.
Biomass and Bioenergy 124: 43-53.
https://www.sciencedirect.com/science/article/abs/pii/S0961953419300911?via%3Dihub
University of Nebraska. 2018. Nebraska Irrigated and Rainfed Corn Yield and Acreage Trends.
https://cropwatch.unl.edu/corn/yieldtrends
USGS. 2019. Very large dead zone forecast for the Gulf of Mexico.
https://www.usgs.gov/news/national-news-release/very-large-dead-zone-forecast-gulf-mexico
USGS. (n.d.). Nutrient Loading for the Mississippi River and Subbasins.
https://nrtwq.usgs.gov/mississippi_loads/#/
Wiens, J., J. Fargione, and J. Hill. 2011. Biofuels and biodiversity. Ecological Applications 31:4.
Wimberly, M., D. Narema, P. Baumanb, and B. Carlson. 2018. Grassland connectivity in fragmented
agricultural landscapes of the northcentral United States. Biological Conservation 217:121-130.
Wright C., B. Larson, T. Lark, and H. Gibbs. 2017. Recent grassland losses are concentrated around
U.S. ethanol refineries. Environmental Research Letters 12. DOI:10.1088/1748-9326/aa6446.
23 | P a g e
Wright CK, and MC Wimberly. 2013. Recent land use change in the Western Corn Belt threatens
grasslands and wetlands. Proceedings of the National Academy of Sciences of the United States
of America 110:4134–9. DOI:10.1073/pnas.1215404110.
Wu, M. 2019. Energy and water sustainability in the U.S. biofuel industry. Argonne National
Laboratory Energy Systems Division. ANL/ESD-19/5. June 2019.
Wu M., and Y. Chiu. 2011. Consumptive Water Use in the Production of Ethanol and Petroleum
Gasoline 2011 update. Energy Systems Division:100.
Xie. Y., and T. Lark. 2021. Mapping annual irrigation from Landsat imagery and environmental
variables across the conterminous United States. Remote Sensing of Environment Volume 260,
July 2021, 112445
Xie, Y., T. Lark, J. Brown, and H. Gibbs. 2019a. Mapping irrigated cropland extent across the
conterminous United States at 30 m resolution using a semi-automatic training approach on
Google Earth Engine. ISPRS Journal of Photogrammetry and Remote Sensing Volume 155,
September 2019, Pages 136-149.
Xie, Y., T. Lark, and H. Gibbs. 2019b. Irrigation dynamics in the Ogallala aquifer between 2000 –
2017. Research brief, prepared 01/06/2019. University of Wisconsin.
Xue Q, T. Marek, W. Xu, and J. Bell. 2017. Irrigated Corn Production and Management in the Texas
High Plains 2000:31–41. DOI:10.1111/j.1936-704X.2017.03258.x.
Zhang, X., T. Lark, C. Clark, Y. Yuan, and S. LeDuc. 2021. Grassland-to-cropland conversion increased
soil, nutrient, and carbon losses in the US Midwest between 2008 and 2016. Environ. Res. Lett.
16 (2021) 054018.
EXHIBIT 1
RAMBOLL. AUGUST 18, 2019. THE RFS AND ETHANOL PRODUCTION: LACK
OF PROVEN IMPACTS TO LAND AND WATER. PREPARED FOR GROWTH
ENERGY. RAMBOLL, SEATTLE, WA.
Prepared at the Request of
Growth Energy
Prepared by
Ramboll
Date
August 18, 2019
THE RFS AND ETHANOL
PRODUCTION: LACK OF PROVEN
IMPACTS TO LAND AND WATER
i Ramboll
CONTENTS
1. EXECUTIVE SUMMARY 1
1.1 Total Acres Planted in Corn Has Remained at or Below Levels in the Early 1930s While
Total Production Increased 7-Fold 3
1.2 Studies Have Failed to Establish a Quantitative Link Between the RFS and Land Use
Change 4
1.3 Changes in Agricultural Practices Broadly Reduce the Likelihood of Environmental
Impacts to Water Resource Availability and Quality 6
1.4 Recent Estimates of Health Damages from Corn Production are Unreliable and
Misleading 7
1.5 Environmental Impacts Associated with Ethanol Production Cannot be Viewed in a
Vacuum, Without Consideration of Such Impacts Associated with Gasoline Production 9
2. ACRES PLANTED IN CORN HAVE REMAINED AT OR BELOW LEVELS IN THE
EARLY 1930S WHILE TOTAL PRODUCTION INCREASED 7-FOLD 11
3. STUDIES HAVE FAILED TO ESTABLISH A QUANTITATIVE RELATIONSHIP
BETWEEN THE RFS AND LUC 14
3.1 Overview of LUC and Environmental Impacts 14
3.2 The Impetus for LUC is Influenced by Complex Factors; and the Influence of the RFS is
Poorly Understood and Likely Weak 15
3.3 Studies Relied Upon by EPA (2018a) to Quantify LUC Around the Time of Enactment of
the RFS Are Based on Unreliable Data and Likely Overestimate LUC 18
3.4 Recently Released Research Purporting to Establish a Quantitative Link Between the
RFS and LUC is Poorly Documented and Flawed 21
3.5 EPA (2018a) Failed to Adequately Account for the Role of Cropping Practices and
Production of Distillers Dried Grains with Solubles (DDGS) at Ethanol Refineries as
Important LUC Offsetting Factors 23
4. CHANGES IN AGRICULTURAL PRACTICES REDUCE THE LIKELIHOOD OF
ENVIRONMENTAL IMPACTS TO WATER RESOURCE AVAILABILITY AND
QUALITY 25
4.1 The Triennial Report’s Discussion of Water Use and Water Quality 25
4.2 Agricultural Improvements in Irrigation are Reducing Water Use 28
4.3 Technological Improvements in Agriculture Translate to Reductions in Potential Water
Quality Impacts 31
4.4 Reduction in Water Usage for Ethanol Processing 33
5. RECENT ESTIMATES OF HEALTH DAMAGES FROM CORN PRODUCTION ARE
UNRELIABLE AND MISLEADING 34
6. ENVIRONMENTAL IMPACTS ASSOCIATED WITH ETHANOL PRODUCTION
CANNOT BE VIEWED IN A VACUUM, WITHOUT CONSIDERATION OF SUCH
IMPACTS ASSOCIATED WITH GASOLINE PRODUCTION. 37
6.1 Impacts of Gasoline Production Associated with Land Use Change 37
6.2 Water Quality Impacts Associated with Spills 40
6.3 Toxicity and Other Ecological Impacts of Oil and Associated Products 41
6.4 Additional Water Quality Impacts Associated with Petroleum Production 42
ii Ramboll
6.5 Additional Water Quality and Supply Impacts Associated with Exploration, Production,
and Refining 43
7. LIMITATIONS 44
8. REFERENCES 45
TABLES
Table 1: Summary of Selected Results as Reported by Dunn et al (2017).
Table 2: Technological and Methodological Improvements to Irrigation of Corn Crops.
FIGURES
Figure 1: A) Annual Yield in Bushels of Corn Per Acre and Annual Acres Planted in Corn Versus 1926.
B) Annual Acres of Corn Planted 2004-2018.
Figure 2: Illustration of the Complexity of Biophysical, Economic, and Social Factors Affecting Planting
Decisions.
Figure 3: Illustration of Habitat Fragmentation in Jonah Field, Wyoming from Oil and Gas Production.
Figure 4: Total U.S. Planted Acres of Corn Per Year (million acres).
Figure 5: A) Annual Yield in Bushels of Corn Per Acre and Annual Acres Planted in Corn Versus 1926.
B) Annual Acres of Corn Planted 2004-2018.
Figure 6: West Texas Intermediate Crude Prices ($/barrel).
Figure 7: US Corn Prices ($/bushel).
Figure 8: Annual Nitrate and Nitrite Loading to the Gulf of Mexico 1980-2017.
Figure 9: Volume of Water Applied to Irrigated Corn Crops Since 1994, by Irrigation Method.
Figure 10: While Irrigated and Unirrigated Corn Crops Have Both Experienced General Increases in
Yield, Irrigated Crops More Reliably Produce Higher Yields.
Figure 11: Both Pesticide and Fertilizer Use on U.S. Corn Crops Appear to Have Peaked in the 1980s
Figure 12: Oil and gas field in Wyoming; Areas with Suitable Resources for Future Extraction.
Figure 13: Major Habitat Types in the United States.
iii Ramboll
ACRONYMS AND ABBREVIATIONS
API American Petroleum Institute
CDL cropland data layer
CGF corn gluten feed
CRP Conservation Reserve Program
DDGS distiller’s dried grains with solubles
EISA Energy Independence and Security Act
EPA U.S. Environmental Protection Agency
FSA Farm Service Agency
ha hectare
ITRC Interstate Technology & Regulatory Council
LUC land use change
NOx nitrogen oxide
NASS National Agricultural Statistics Service
NLCD National Land Cover Database
NOAA National Oceanic and Atmospheric Administration
NRCS National Resources Conservation Service
NWI National Wetlands Inventory
RFM Reduced form model
RFS Renewable Fuel Standard
SOA secondary organic aerosols
SOx Sulphur oxide
TPH total petroleum hydrocarbons
UOG unconventional oil and gas
USDA U.S. Department of Agriculture
USGS U.S. Geological Survey
USEIA U.S. Energy Information Administration
VOC volatile organic compound
1 Ramboll
1. EXECUTIVE SUMMARY
This report was prepared by Ramboll for Growth Energy in anticipation of the United States
Environmental Protection Agency (EPA) issuing proposed rulemaking on the Renewable Fuel
Standard (RFS), commonly referred to as the “RFS Reset.” One of the factors that EPA must
consider in resetting renewable fuel volumes in the program is potential environmental
impacts.
The key conclusion of this report is that there are no proven adverse impacts to land and
water associated with increased corn ethanol production under the RFS. Accordingly, EPA
could decide to reset renewable volumes in a manner that would incentivize greater
production and consumption of conventional corn ethanol in US transportation fuel without
discernible adverse environmental impacts to land and water, to the extent any exist. The
major factors supporting this conclusion are that continued improvements in agricultural
practices and technology indicate that increased demand for corn grown for ethanol in the
United States can be met without the need for additional acres of corn planted, while at the
same time, reducing potential impacts to water quality or water supplies.
Our review focused on analyses concerning water quantity and quality; as well as
ecosystems, wetlands, and wildlife. Analyses concerning ecosystems, wetlands, and wildlife
were presented primarily as part of the body of literature addressing land use change (LUC)
and conversion of land from non-agricultural to agricultural uses in the United States. We
focused particular attention on EPA’s recent environmental review of the RFS, Biofuels and
the Environment: Second Triennial Report to Congress (EPA 2018a), and studies relied upon
by the agency therein. Ramboll also reviewed other key publications pre- and post-dating
EPA (2018a). A full list of refences cited in this report is presented in Section 8.
We also reviewed a recent paper by Hill et al. (2019) investigating the air quality-related
health impacts of growing corn. Finally, we provide a brief overview of certain environmental
impacts of oil and gas exploration and production and gasoline refining, in response to EPA’s
(2018a) acknowledgement that its assessment is not fully comprehensive because it does
not consider a comparative assessment of the impacts of biofuels relative to petroleumderived fuels.
The principal findings of our review by topic include, but are not limited to:
• Land use change—Some investigators have asserted that the RFS has resulted in
extensive conversion of non-agricultural land to agriculture due to increased demand for
corn for ethanol. Our findings indicate that these claims are not borne out, in part
because the studies do not establish a causal link between the RFS, increased ethanol
production, and LUC. Indeed, in a follow-up analysis to its Triennial Report EPA (2018b)
reached the same conclusion―that no causal connection has been established between
LUC associated with corn production and the RFS.
– The number of acres planted in corn has remained effectively constant
despite significant increases in production. Acres planted in corn across the
United States has remained close to or below the total acres planted in the early
1930s, despite increases in demand for corn as human food, animal feed, and
biofuels over this nearly 90-year period. The increase in demand has largely been
met by an approximately 7-fold increase in yield (bushels per acre).
– Most studies asserting a connection between the RFS and LUC fail to
adequately account for the myriad factors that drive farmers’ choices to
2 Ramboll
plant a given crop or to convert non-agricultural land to cropland. The price
of corn is only one of many such factors, and the literature does not support that the
RFS is the predominant driver of pricing of this global commodity. Moreover,
assertions that the RFS drives LUC, fail to adequately recognize the increased
efficiency in corn production per acre as well as the diminished demand for corn crop
acreage due to co-products of the ethanol refining process, such as distiller’s dried
grains with solubles (DDGS). Assessments of LUC and the RFS generally fail to
recognize external factors that might be driving expansion of farmland, such as the
loss of farmland near urban areas.
• Water use and water quality—EPA (2018a) and other authors raise concerns that
increased corn grown for ethanol may be overstressing water sources and resulting in
regional water quality impacts. Our findings indicate that these concerns are not borne
out primarily due to research that fails to establish a causal relationship between corn
grown for ethanol and impacts to water use and water quality. We further find that EPA
(2018a) does not adequately acknowledge the role of advances in agricultural practices
in mitigating potential water use and water quality impacts.
– A quantitative or causal relationship between the RFS and concerns over
water use has not been established. From a geographical standpoint, much of
the corn that is used for ethanol production is grown on non-irrigated land where
impacts to water availability are minimal, and while noted, this is not quantitatively
considered by EPA (2018a). In addition, the increased adoption of modern farming
practices and precision agriculture (Vuran et al. 2018) is reducing the potential
impact of agriculture in general, including increased corn production, on water
availability. EPA (2018a), in fact, noted that the increased use of these best
management practices should substantially limit impacts to water resources. While
some investigators have claimed that growth in corn production has resulted in
greater stress to water resources, those studies that focus on negative impacts fail to
acknowledge, or do not appear to emphasize, that the current focus on best
management practices and resource protection is being widely adopted by the corn
growing community and incentives to adopt such practices continue. The technical
publications we have reviewed do not establish that the RFS drives corn planting
decisions and potential associated water impacts.
– A quantitative or causal link between corn production associated with the
RFS and adverse water quality impacts has not been established. While
observed environmental impacts, such as excessive algae blooms in western Lake
Erie and low oxygen levels in the Gulf of Mexico have been documented, we found
that the literature on this issue fails to quantitatively link these regional water quality
problems to increases in corn production for ethanol. Indeed, nutrient loading to the
Gulf of Mexico, as measured by nitrates and nitrites, has remained relatively
constant since at least 1980 despite increases in corn production. In addition, very
few investigators have looked closely at agriculture trends over the past decade that
show the implementation of modern farming practices are helping to reduce potential
watershed impacts; modern farming practices include improved products such as
slow-release fertilizers, and improved practices such as precision agriculture and
better water and stormwater management. This trend is expected to continue well
into the future and provide additional benefits to other agricultural products in
addition to corn. Finally, expected future gains in corn yield (bushels produced per
acre per year) in combination with steady or even declining fertilizer and pesticide
3 Ramboll
use (in pounds per acre per year), will naturally result in a decrease in the potential
for water quality impacts.
• The RFS Reset is well-timed to coincide with ongoing improvements in
agricultural practices—Nearly all published investigations Ramboll has reviewed that
focus on the potential impact of increased corn growth for biofuel production have
focused on past practices with only passing mention of future expectations. EPA (2018a)
acknowledges the benefits of the increased use of best management practices on the
environment. Modern agricultural practices are economically beneficial to corn producers
when they result in reduced input costs associated with water and agricultural
chemicals. The timing for increasing corn production and reduced potential
environmental impacts due to precision agriculture coincides with increased biofuel
demand, and the coincidence of these trends will benefit both producers and the
environment into the future.
1.1 Total Acres Planted in Corn Has Remained at or Below Levels in the Early
1930s While Total Production Increased 7-Fold
The United States Department of Agriculture (USDA) has maintained annual statistics on
domestic crop production for decades. Corn production in the United States annually exceeds
10 billion bushels, with approximately 50% of corn currently grown for ethanol production
and 50% for grain use. Accordingly, corn is documented to be the most widely produced feed
grain in the United States (U.S.), accounting for more than 95 percent of total production
and use followed by sorghum, barley, and oats (USDA 2019). Most of the corn crop for feed
grain is used for livestock feed. Other food and industrial products include cereal, alcohol,
sweeteners, and byproduct feeds.
While the approximate share of U.S. corn (in bushels) dedicated to production of ethanol has
increased from 4% in 1986, to 38% in 2015 (USDA-ERS 2019b), and to approximately 50%
in 2018, the total corn planting (in acres) has remained relatively stable since the 1930s
(Figure 1). On a shorter time-scale, acres of corn planted each year does vary, but when
examining data between 2007 and 2018, there is no long-term upward trend. In fact, acres
of corn decreased 8.07% in 2008, the year after the enactment of the Energy Independence
and Security Act (EISA), then rebounded through 2012, then decreased again such that in
2018, acres of corn were 4.7 % lower than in 2007. These data, from the USDA Crop
Production Historical Track Record (updated in USDA, 2019) demonstrates the increased
efficiency, planting and production of the corn crop without a need to secure appreciable
additional acreage for production. Efforts in better crop management, improved fertilizer use,
and precision agriculture are all likely contributors to improved yields. The USDA further
anticipates changes in corn production to result in an increase of approximately 16.1 more
bushels per acre by 2028 without a substantial increase in farmed acres (and with a
corresponding reduction in the use of water resources and fertilizer).
4 Ramboll
Figure 1: A) Annual Yield in Bushels of Corn Per Acre and Annual Acres Planted in
Corn Versus 1926. B) Annual Acres of Corn Planted 2004-2018.
Source: USDA Crop Production Historical Track Records, 2019
1.2 Studies Have Failed to Establish a Quantitative Link Between the RFS and
Land Use Change
The decision by farmers and landholders on whether to plant a bioenergy crop such as corn
reflects complex relationships between biophysical, economic, and social factors (Figure 2).
75
80
85
90
95
100
Acres of Corn Planted for All
2004 2009 2014 2019
Purposes (Million Acres)
Year
B
Planting
EISA Enactment
0
1
2
3
4
5
6
7
1926
1936
1946
1956
1966
1976
1986
1996
2006
2016
Yearly Value Versus 1926 Baseline
By the late 1930s, harvested corn
yield was no longer directly linked to
corn planted.
Acres of total corn planted has
generally plateaued, even into the
late 2010s.
Yield
Planting
A
5 Ramboll
Figure 2: Illustration of the Complexity of Biophysical, Economic, and Social Factors
Affecting Planting Decisions.
One factor that is of paramount importance is weather and climate. Regional weather
patterns largely dictate crop patterns across the country, but this is also influenced by the
availability (and price) of water for irrigation in areas with relatively low annual precipitation
or highly variable precipitation. The probability of severe weather such as drought and flood
as well as severe storm events in any given year, may also influence planting decisions.
Government policy is another overarching factor affecting planting decisions, and these
include potential monetary incentives associated with the U.S. Department of Agriculture
Conservation Reserve Program (USDA CRP), other conservation programs such as local
conservation easements, the availability of crop insurance, and market incentives that might
affect commodity prices. Other factors include market price and the price of production
inputs, which can be strongly influenced by the price of oil, exchange rates and trade
policies. Local market prices are influenced by a wide range of factors including status of
commodity stores, distance to markets, and competition from regional and even global
markets. Input prices are also highly variable due to market prices, and volume
requirements for some inputs such as irrigation water are weather and climate dependent.
Finally, all of the above factors, plus the availability and quality of land and ecosystem
characteristics and ecological value play into decisions regarding land use—whether to plant
new acreage (extensification) or plant more of a given crop on existing acreage
(intensification).
The influence of the RFS on LUC is poorly understood and likely weak. To the extent it
suggests otherwise, EPA (2018a) inadequately assesses the range of market and nonmarket
factors influencing land use change and does not consider key studies that suggest that the
RFS likely had a small, and perhaps negligible effect on LUC, especially changes in land use
from non-agriculture to biofuels feedstock (corn and soy). In particular, EPA (2018a) does
not adequately consider the role of farm policy such as crop insurance, land characteristics,
input and output prices, and technology on growing decisions by farmers.
6 Ramboll
Studies relied upon by EPA (2018a) to quantify LUC around the time of enactment of the
RFS are based on unreliable data and likely overestimate LUC. In particular, EPA (2018a)
cites work by several authors who report findings of considerable LUC, including LUC in
ecologically sensitive areas such as the Prairie Pothole Region, but do not sufficiently
acknowledge or discuss findings by more recent research that indicates that many of the
earlier studies were flawed or substantially overstated the extent of LUC in and around the
enactment of the RFS. EPA (2018a) also does not sufficiently acknowledge that the studies it
relied on do not establish a causal relationship between the RFS and LUC. In addition, EPA
(2018a) makes no attempt to quantify, or even describe in any detail, the potential
ecological impacts of the alleged LUC, so the actual environmental harm, if any, associated
with the RFS remains nebulous. Notwithstanding these shortcomings of the report, EPA
clarifies in a subsequent discussion of environmental impacts of the RFS that it does not
view the literature it identified in the Triennial Report as supportive of a causal link between
LUC and the RFS; rather, there is a myriad of “complex regulatory and market factors that
are relevant to such a relationship” (EPA 2018b).
A recent effort by Lark et al. (2019) to develop a quantitative link between the RFS and LUC
may be the most exhaustive effort to date, but their reliance on an uncertain “business as
usual” baseline and on estimating price increases attributable to the RFS are major
weakness of the work. Most important, the entire analysis presented by Lark et al. (2019)
rests on estimating price increases attributable to RFS, yet the authors fail to adequately
acknowledge the role of important factors such as the dietary transition from cereals toward
more animal protein in developing countries resulting in rapid growth in the consumption of
agricultural commodities. Other important factors affecting corn prices over the period
include higher oil prices and the link between the U.S. dollar exchange rate and commodity
prices. In addition, the data sets and models used in their analysis are not made explicit,
and some data are not in the public domain, precluding a thorough independent review of
their work.
EPA (2018a) also failed to adequately account for the role of cropping practices and
production of DDGS at ethanol refineries as important LUC offsetting factors. Several studies
indicate that a substantial portion of increase corn production following the introduction of
the RFS was met via farmers’ cropping practices, including switching from other row crops to
corn or double cropping corn instead of rotating between corn and soy (or other crops).
These studies are not given adequate consideration by EPA (2018a). Although EPA (2018a)
acknowledges that production of DDGS may offset some demand for corn as livestock feed,
key studies estimate this offsetting effect is considerable. In addition, EPA (2018a) does not
discuss whether and to what extent this offset for demand for corn is a market driver that
provides downward pressure for LUC to corn.
1.3 Changes in Agricultural Practices Broadly Reduce the Likelihood of
Environmental Impacts to Water Resource Availability and Quality
Advancements in technology and water management techniques have continued to increase
the efficiency in water resource management by stabilizing, and potentially reducing, the
overall volume of water necessary for corn growth. Agriculture accounts for an estimated 80
percent of national consumptive water use in the US according to the USDA’s Economic
Research Service (2018) and reaffirmed by the National Academy of Science (2019).
According to the 2012 statistics from the USDA, irrigated corn acreage represented about
25% of all irrigated acreage in western states, and about 24% of all irrigated acreage in the
eastern states (USDA-ERS 2018a). Additionally, the USDA has shown that irrigation for all
crops, including corn, has decreased even as the farming acreage has essentially been stable
7 Ramboll
over the past 35 years. The USDA attributes this trend to improvements in physical irrigation
systems and water management. The USDA also notes that significant capital investments in
on-farm irrigation is continuing, particularly in the western states, where most of the
irrigated farm-land is concentrated. As an indication of a positive trend in irrigation
reduction, the University of Nebraska, Lincoln reports that in Nebraska (as a bell-weather of
other dry western states), the percentage of all corn acreage that is irrigated has declined
from a high of 72% in 1981 to 56% in 2017 (University of Nebraska 2018).
Increasing crop yield per area of farmed land is taking place on both irrigated and
unirrigated corn crops, suggesting that changes in yield are not attributed to irrigation alone.
In certain areas, more corn is now being grown on the same number of acres, which has
resulted in increases in irrigation. However, watersheds where most intensification has
occurred are mostly in Western states which account for less ethanol feedstock than the
less- or non-irrigated Midwest and Eastern States.
Trends and expectations in the biofuel refining process also show increasing water use
efficiency and lower water demand over time (upwards of 50% reductions in recent years).
This trend is anticipated to continue as ethanol refining technology advances.
Advances in sustainable farm management, including substantial improvements in nutrient
formulation and use, and technological improvements in pesticide and fertilizer application,
will continue to reduce the potential for impacts to water quality in regional watersheds near
corn growing areas regardless of the cause of historical water quality impacts. Additionally,
the EPA acknowledges that corn production for ethanol has not been reliably linked to large
scale degradation of water quality. The hypothesized causal relationship between the hypoxic
zones in the northern Gulf of Mexico and eutrophication in Western Lake Erie with corn
grown specifically for ethanol production is weak and lacks supporting data. It is recognized
that urban and agricultural runoff in the subject watersheds have likely contributed to the
conditions; but EPA (2018a) notes that attributing these water quality issues to ethanol
production is speculative and not based on specific data.
1.4 Recent Estimates of Health Damages from Corn Production are Unreliable
and Misleading
Although the primary focus of this report is on studies assessing the implications of the RFS
program and corn ethanol production for land and water, a recent report that attempts to
link corn production to adverse public health impacts from air emissions merits a brief
response. A recent publication in Nature Sustainability (Hill et al. 2019) purports to estimate
US annual health damages caused by particulate air quality degradation from all direct farm
and indirect supply chain activities and sectors associated with maize (corn) production.
Although the authors do not reference the RFS, they do mention corn grown for ethanol, and
the publication has been referenced by third parties in a manner suggesting that corn grown
for ethanol may be associated with adverse health outcomes. Ramboll’s review indicates that
the conclusions presented by Hill et al. (2019) are unsubstantiated and likely overestimate
adverse health impacts, where it is not clear any health impacts exist.
The direct and indirect activities explored by Hill et al. (2019) include air emissions from
farms and upstream processes that produce the chemical and energy inputs used in corn
crop production: fuel, electricity, agrichemical production, transportation, and distribution.
The paper focuses on particulate matter smaller than 2.5 microns in diameter (PM2.5), which
is a concern for human health because particles of this size can penetrate deep into the
lungs and enter the bloodstream, and potentially result in both acute and chronic effects to
the respiratory and cardiovascular systems. Ramboll reviewed the underlying models and
8 Ramboll
assumptions employed in the Hill et al. (2019) analysis and we present the following
findings:
• The model relied upon by the authors uses annual-average data for emissions,
meteorology, and chemical/removal rates to estimate annual-average PM2.5 impacts. Use
of annual averages is inappropriate for representing processes that operate over shorter
time scales ranging from minutes to several months (e.g., atmospheric dispersion and
chemical formation of PM2.5) and results in a high level of uncertainty. The authors
acknowledge that this weakness in their approach results in spatial errors in annual
average PM2.5 calculations. These spatial errors can significantly impact the resulting
exposure and mortality estimates. The authors, however, do not present sensitivity
analyses to assess the impact of the model assumptions, nor do they include any
plausible range of uncertainty or variability with their modeled PM2.5 concentration or
mortality estimates.
• The 2005 modeling year upon which modeling is based is not representative of more
recent chemical conditions of the atmosphere in the U.S., which may lead to an
overestimate of the PM2.5 contributions from corn production by more than a factor of 2,
and this overestimate results in overestimates of health and economic damages.
• Several major sources of uncertainty in the modeling are not acknowledged or
accounted for by the authors, including the following key uncertainties:
– Ammonia emission estimates, which are the largest driver of mortality in the Hill et
al. (2019) modeling analysis, are also the most uncertain aspects in any PM2.5 air
quality modeling, because: (1) emissions are largely from agricultural sources that
vary both spatially and temporally due to weather and farming practices; (2) many
different methods are used to estimate ammonia emissions, and each can yield very
different emission rates and exhibit a high degree of error; (3) annual average
ammonia emission inventories used in the modeling fail to account for important
seasonal variations and related complex interactions with sulfate and nitrate
chemistry; and (4) ignoring diurnal and intra-daily ammonia emission variations
have been shown in the literature to overestimate ambient ammonia concentrations
by as much as a factor of 2.
– The health impact assessment is based on a single epidemiological study that found
associations between PM2.5 concentrations and mortality, but a clear causal link has
not been established in the scientific community. In fact, the components of PM2.5
that may be associated with adverse health effects are yet unknown, but evidence
suggests that carbonaceous particles are more toxic than inorganic particles such as
those derived from ammonia and nitrate or sulfate.
Based on our review of literature documenting the development and testing of the simplistic
model employed by Hill et al. (2019), we conclude that the model is not able to faithfully
reproduce PM2.5 impacts estimated by more complex state-of-the-science air quality models.
In fact, its performance is at its worst for the very PM2.5 component (ammonium) that the
Hill et al. (2019) model indicates is the largest contributor to PM mortality from corn
production. This renders the modeling especially unreliable for this key PM component.
Overall, the uncertainties enumerated above result in unreliable estimates of PM2.5 exposure,
mortality and related costs associated with corn production, each associated with a large
range of variability.
9 Ramboll
1.5 Environmental Impacts Associated with Ethanol Production Cannot be
Viewed in a Vacuum, Without Consideration of Such Impacts Associated
with Gasoline Production
EPA (2018a) acknowledges its Triennial Report fails to address environmental impacts
associated with gasoline production, but it is important not to view environmental impacts of
ethanol in a vacuum given the biased view this presents.
Land use for oil and gas production is extensive. In 2011, the direct footprint of oil and gas
production was approximately 1,430,000 acres (Trainor et al. 2016). By 2040, Trainor et al.
(2016) estimate the direct footprint of oil and gas production will be approximately
15,890,000 acres.
Habitat fragmentation from oil and gas production is also high and is known to decrease
biodiversity (Butt et al. 2013). For example, the fragmentation caused by the dense
placement of over 55 pads per square mile in Texas is known to cause a reduction in habitat
quality for lizards in the short term (Hibbitts et al. 2013), while in the long term, habitat
restoration after the removal of oil and gas infrastructure does not eliminate adverse effects
to biodiversity (Butt et al. 2013).
Figure 3: Illustration of Habitat Fragmentation in Jonah Field, Wyoming from Oil
and Gas Production.
SOURCE: EcoFlight (USDA 2012)
Oil and gas products, production fluids, and refinery effluent have negative impacts on soil
and water quality and flora and fauna when released in the environment (EPA 1999, Wake
2005, Pichtel 2016). The toxicity of crude oil and its individual components has been well
studied and these products are known to have negative impacts on wildlife depending on the
exposure and dose received (Interstate Technology & Regulatory Council [ITRC] 2018).
10 Ramboll
Production water, fracking fluids, and refinery effluent, though less well-studied, have also
been found to have adverse effects on plants and wildlife, resulting in decreased populations
and biodiversity (Wake 2005, Pichtel 2016).
American Petroleum Institute (API) reported approximately 10.8 million gallons of oil were
spilled into U.S. Navigable Waters from 1997-2006 with the amount spilled per year varying
from 466,000 (2005) to 2.7 million (2004). This figure clearly does not include the Exxon
Valdez spill in Alaska in 1989 or the Deepwater Horizon spill in 2010. National data suggest
that spills from unconventional oil and gas may amount to one million gallons each year
(Patterson et al. 2017). These data are exclusive of major offshore releases and incidents.
The findings and conclusions summarized above and set forth in the remainder of this report
are subject to the limitations stated in Section 7.
11 Ramboll
2. ACRES PLANTED IN CORN HAVE REMAINED AT OR
BELOW LEVELS IN THE EARLY 1930s WHILE TOTAL
PRODUCTION INCREASED 7-FOLD
The total acres of corn planted in the U.S. has remained relatively stable and in fact has
decreased slightly since the 1930s as shown in Figure 4, while the approximate share of
U.S. corn (in bushels) dedicated to production of ethanol has increased from 4% in 1986 to
38% in 2015 and currently to approximately 50% in 2018 (USDA-ERS 2019b).
Figure 4: Total U.S. Planted Acres of Corn Per Year (million acres).
(Source: USDA 2019)
Even as the total corn acreage has been relatively stable or has slightly decreased since the
early 1930s, the yield in bushels per acre during this same approximate period has increased
dramatically as illustrated by Figure 5: A) Annual Yield in Bushels of Corn Per Acre and
Annual Acres Planted in Corn Versus 1926. B) Annual Acres of Corn Planted 2004-2018.
These statistics reported by the U.S. Department of Agriculture (USDA) are a positive sign of
the ability of farming practices to become more efficient and optimized to generate more
yield without adding additional acreage. Also noticeable is that the stability of farming
acreage and continued increase in yield extends into the last decade, following the
enactment of the EISA. In 2018, 4.7% fewer acres of corn were planted for all purposes in
the U.S. as compared with 2007, even though the approximate percentage of corn for
ethanol versus other uses has increased. There was regional variation in changes in corn
planting; for example, comparing data from 2017 with 2007, approximately two million
fewer acres of corn were planted for all purposes in Illinois, with approximately 860,000
additional acres in North Dakota. Regional changes are driven by a wide range of competing
macroeconomic conditions, mostly unrelated to ethanol production, including the relative
value of crops like spring wheat and cotton, or changes in corn outputs from other countries.
Indeed, the EPA confirmed that, for a variety of reasons, even the proposed 2019 RFS
renewable volume obligation standards would not be expected to result in an increase in
farming acreage (EPA 2018b).
0
20
40
60
80
100
120
1926
1929
1932
1935
1938
1941
1944
1947
1950
1953
1956
1959
1962
1965
1968
1971
1974
1977
1980
1983
1986
1989
1992
1995
1998
2001
2004
2007
2010
2013
2016
Million Acres
12 Ramboll
Figure 5: A) Annual Yield in Bushels of Corn Per Acre and Annual Acres Planted in
Corn Versus 1926. B) Annual Acres of Corn Planted 2004-2018.
(Source: USDA Crop Production Historical Track Records, 2019)
According to 2018 USDA projections, annual U.S. corn production is anticipated to surpass
15 billion bushels by 2025, while the USDA projects a 2.1-million acre decline in planted corn
acres for 2018/19 (Capehart et al. 2018, USDA-ERS 2018b). Schnepf and Yacobucci (2013)
cite the following projections by USDA and industry for future increases in corn yield: USDA
predicts yields will reach about 240 bushels per acre by 2050 (overall increase of 55% over
the 37-year period), whereas the outlook from biotechnology seed company Monsanto is an
increase of 300 bushels per acre by 2030, (overall increase of 93% over the 17-year period).
75
80
85
90
95
100
Acres of Corn Planted for All
2004 2006 2008 2010 2012 2014 2016 2018 2020
Purposes (Million Acres)
Year
B
Planting
EISA Enactment
0
1
2
3
4
5
6
7
1926
1936
1946
1956
1966
1976
1986
1996
2006
2016
Yearly Value Versus 1926 Baseline
By the late 1930s, harvested corn
yield was no longer directly linked to
corn planted.
Acres of total corn planted has
generally plateaued, even into the
late 2010s.
Yield
Planting
A
13 Ramboll
The continued trend of decreases in farmable acres and increases in yield will likely continue
to some stable equilibrium that will be controlled by economic and general land resource
conditions. There appears to be little or no discussion in reports and documents, such as EPA
(2018a), Lark et al (2019) and others, of the significance of these trends.
14 Ramboll
3. STUDIES HAVE FAILED TO ESTABLISH A
QUANTITATIVE RELATIONSHIP BETWEEN THE RFS
AND LUC
3.1 Overview of LUC and Environmental Impacts
In this section we first present a discussion of the lack of evidence for a quantitative causal
link between increased demand for ethanol from the RFS and LUC. Second, we present a
summary of some of the largest sources of uncertainty in studies that EPA (2018a) relies on
to assert that the RFS may have resulted in considerable LUC. Third, we discuss the
information presented by EPA (2018a) on the topics of cropping practices as well as the role
of distiller’s dried grains with solubles (DDGS) in offsetting LUC potentially associated with
the RFS.
The literature attempting to relate LUC to ethanol production generally acknowledges
shortcomings in some of the major data sets, and authors such as Lark et al. (2015) and
Dunn et al. (2017) attempt to address these shortcomings by using advanced geospatial
analysis techniques and data corrections (Lark et al. 2015, Dunn et al. 2017). Importantly,
studies relied upon by EPA (2018a) to quantify LUC around the time of enactment of the RFS
are based on unreliable data and likely overestimate LUC.
Assertions made by EPA (EPA 2010, 2018a) to link LUC (including land taken out of the CRP
as well as non-agricultural land converted to agriculture) to increased demand for ethanol
due to the RFS cannot be substantiated by the underlying literature for a variety of reasons,
including, but not limited to the following:
• There are a myriad of complex, interrelated market and non-market factors affecting
farmers’ decisions on land use and a thorough assessment of the causative factors was
not undertaken in the literature cited by EPA (2018a).
• Many studies do not differentiate among crop type (e.g., corn and soy) when looking at
LUC and thus it is not possible to establish a causal linkage between LUC and demand
for ethanol versus demand for biodiesel from those studies.
• Most studies of LUC are regional or state-specific and there is substantial inconsistency
between studies regarding the geographical area of focus. This inconsistency precludes
arriving at broad regional or national conclusions. For example, several studies focus on
LUC in the Prairie Pothole Region due to this region’s environmental fragility; whereas
other studies assessed the “western corn belt”, “lake states”, or the entire continental
United States.
• Many studies focus on specific land use types prior to conversion to agriculture (e.g.,
grassland, wetlands, or land in the CRP) and thus are not inter-comparable.
• Increased demand for all uses of corn may be met via either expansion of agricultural
land onto previously uncultivated land (extensification) and by increased production
from existing land (intensification). Intensification does not result in LUC and EPA
(2018a) does not adequately represent the role of intensification in mitigating the
propensity for extensification and LUC.
• Use of corn in ethanol refining produces substantial amount of DDGS and the use of
DDGS as a substitute for corn as livestock feed reduces the demand for corn as livestock
feed. This issue is not adequately accounted for in the assessment by EPA (2018a) of
the potential role of RFS in LUC.
15 Ramboll
• The literature assessing LUC relative to the RFS generally fails to consider the
considerable loss of agricultural land in urban areas and the role this loss may have in
extensification elsewhere.
EPA (2018a) reviewed a wealth of information documenting LUC to biofuel crops and
potential environmental impacts, but the report presents no coherent arguments or
convincing lines of evidence of: (1) a quantitative relationship between ethanol production
spurred by increase demand from the RFS and the documented LUC, or (2) quantitative
impacts to ecosystems, wetlands, or wildlife. EPA (2010 and 2018a) reference numerous
studies of LUC around the time of the enactment of the EISA. Many of these studies combine
data over the period pre- and post-2007, making it difficult or impossible to confidently
associate observed LUC to the time the RFS came into effect. Many authors also simply infer
that there is a relationship between LUC and the RFS without any meaningful exploration of
the market drivers for such change. In fact, EPA (2018b) asserts that historically the annual
RFS requirements have not driven increased ethanol production and consumption. EPA
asserts that this is due to the fact that consumption of ethanol has remained fairly steady
since 2013 (when the 10% ethanol/gasoline blend became the predominant fuel), yet corn
starch ethanol production has continued to rise well beyond the volumes required by the RFS
standard, driven by favorable export markets. Ethanol exports more than doubled over the
2013-2017 period from about 0.62 billion gallons to 1.72 billion gallons (US EIA 2018).
Irrespective of market drivers, EPA (2018a) acknowledges that attributing the causes of land
use change to any one factor, including the RFS, is difficult and speculative. Interestingly,
EPA (2018a) acknowledges many of these shortcomings, especially in their concluding
statement that “we cannot quantify with precision the amount of land with increased
intensity of cultivation nor confidently estimate the portion of crop land expansion associated
with the market for biofuels”.1 EPA (2018a) acknowledges that contributing factors to LUC
include market dynamics such as crop prices and input prices (e.g., fuel, transportation
costs, costs of equipment, etc.) and nonmarket costs such as those resulting from adverse
weather and pests. EPA (2018a) further acknowledges that these and other factors influence
land use change and that these factors may be “coincident with the passage of EISA and
therefore correlated in an empirical analysis”.2 A fundamental problem with many of the
studies cited by EPA (2018a) is that they focus on establishing correlations, or simply
temporal associations between observed LUC and the RFS, and do not establish causation.
EPA (2018b) succinctly summarizes the issue of relating LUC to the RFS as follows: “…there
is no scientific consensus about how to accurately and consistently attribute land use change
in the context of biofuels”.3
3.2 The Impetus for LUC is Influenced by Complex Factors; and the Influence
of the RFS is Poorly Understood and Likely Weak
EPA (2018a) identifies LUC as one of the primary drivers of potential environmental impacts
from increased biofuels production, and they devote an entire section to the topic. However,
EPA (2018a) also acknowledges the weakness and lack of certainty in many reports that
attempt to establish a quantitative link between the RFS and LUC. For example, EPA (2018a)
points out that the U.S. Department of Agriculture National Agricultural Statistics Service
(USDA NASS) data indicate increases in corn crops but in the absence of comprehensive land
classification “it is impossible to know whether these increases came from existing
1 EPA (2018a) at page xi
2 EPA (2018a) at page 22
3 EPA (2018b) at page 16
16 Ramboll
agricultural lands or new lands that were not recently in cultivation”.4 EPA (2018a)
additionally notes weaknesses in empirical approaches in general, including difficulty in
comparing observations and differences in how measured attributes are defined.
Consequently, EPA (2018a) acknowledges that it is difficult to attribute the causes of land
use changes, including where such changes are coincident with the passage of the EISA.
Several authors have examined LUC from the standpoint of decisions made at the individual
farm level. Wang et al. (2017) conducted surveys of 3,000 randomly selected farmers in 37
counties in South Dakota and 20 counties in North Dakota to gain an understanding of the
relative importance of different factors affecting land use decisions, and how that relative
importance changes with operator and farm characteristics. The results of their survey
indicated that the importance of crop output and input prices, innovations in cropping
equipment, and weather patterns all increase closer to the economic margin. The authors
also found that highly sloped areas are more sensitive to crop prices and crop insurance
policies than less sloped land and that as farm size increases, farmers are more sensitive to
policy issues and technological innovations (Wang et al. 2017).
Claassen et al. (2011) assessed the effect of farm policy on LUC and found that crop
insurance, disaster assistance, and marketing loans contributed to a 2.9 percent increase in
cropland acreage between 1998 and 2007 in the northern plains (Claassen et al. 2011). Miao
et al. (2015) found that crop insurance reduced the effective cost of land conversion by
stabilizing crop revenues (Miao et al. 2016).
Efroymson et al. (2016) use classical causal analysis to elucidate shortcomings of existing
studies of the relationship between biofuels policy and LUC. The authors point out that such
studies are often based on assumptions that the production of feedstock for biofuels results
in the increase in demand for food crops, which in turn, results in an increase in crop prices
and expansion of the total area devoted to agriculture; and that this cascading process
results in the loss of areas of natural vegetation, including grasslands. EPA (2018a)
acknowledges the general premise by Efroymson et al. (2016), describes the methods the
authors used, but does not describe the authors’ principal conclusion that for LUC, single
lines of evidence considered individually are insufficient to demonstrate probable cause.
Many of the studies cited by EPA (2018a) in describing a putative relationship between the
RFS and LUC indeed focus on single lines of evidence such as the temporal association
between LUC and the enactment of the RFS, correlations between LUC and farm proximity to
ethanol plants, or LUC and increased production of corn.
Fausti (2015) explored the causal linkages among genetically modified corn, ethanol
production, and corn production, hypothesizing that genetically modified corn allowed for the
expansion of corn acreage, increased corn production incentivized increased ethanol
production, and the RFS allowed this economic feedback mechanism to intensity (Fausti
2015). The author examined pre-RFS data (1996-2000) as well as post-RFS data (2009-
2013) and found that the policy-induced [RFS] increase in ethanol production after 2006 had
a statistically significant and positive effect on change in corn acres planted. However,
although this relationship was statistically significant, Fausti (2015) found that the “policyinduced” change was responsible for only 0.69% to 0.88% percent of the change in corn
acres planted.
One line of evidence for a link between RFS and LUC that has been explored by several
authors is the relationship between increased acres in corn or LUC and proximity to ethanol
4 EPA (2018a) at page 21
17 Ramboll
plants. EPA (2018a) asserts that “The finding of higher rates of conversion closer to the
biorefineries is important and suggests a causal link”.5 In support of this assertion, EPA
(2018a) cites “Motamed and Williams (2016)”.6 EPA (2018a) also states that “for instance
[Motamed et al. 2016], estimated that for every 1% increase in an area’s ethanol refining
capacity, its corn acreage and total agricultural acreage increased by 1.5% and 1.7%,
respectively”.7 However, EPA (2018a) ignores the authors’ own caveats about interpretation
of this finding. In particular, the authors implicate the observed spatial linkages to food and
animal feed, as well as ethanol production, conceding that “[t]hese outcomes may reflect the
efficient response of different producers to new economic incentives, but any externalities
associated with these evolving arrangements remain unknown”.8 In other words, no causal
link to the RFS was established.
Wright et al. (2017) is cited several times by EPA (2018a) to provide evidence of the
association between land use change (loss of grasslands) and refinery location. In particular,
Wright et al. (2017) note that approximately 2 million acres of grassland was converted to
row crops within 50 miles of a refinery between 2008 and 2012. However, EPA (2018a)
again does not acknowledge a major shortcoming of the study, namely, the authors’
admission that their study “did not consider potential effects of other explanatory
variables”.9 The paper also discussed the errors in the data itself, stating that the
“conversion of non-cropland to cropland was mapped correctly over 70% of the time “ which
means that it was mapped incorrectly 30% of the time, a considerable percentage.10
Li et al. (2018) examine the determinants of change in corn acreage and aggregate crop
acreage as a function of the establishment of ethanol plants and changes in crop prices in
the United States between 2003 and 2014. In this nationwide study, the authors report that
corn acreage is fairly inelastic with respect to both changes in nearby ethanol refining
capacity as well as changes in crop prices (Li et al. 2018). Unlike previous studies of the
relationship between LUC and ethanol refinery location that have regional focus, Li et al.
(2018) base their findings on the analysis of data for 2,535 counties in the contiguous United
States. Li et al. (2018) found that a 1% increase in ethanol capacity in a county was
associated with approximately 0.03% to 0.1% increase in corn acreage in that county and a
1% increase in corn price was associated with an approximately 0.18% to 0.29% increase in
corn acreage in a county. The authors conclude that previous studies may have
overestimated the effect of the proximity of ethanol refineries on planting of corn. The
authors did find that the expansion in corn ethanol alone, all else being equal, resulted in a
2.9-million-acre increase in acres planted in corn in 2012 relative to 2008. Critically,
however, they noted that most of the increase came from conversion of other crops to corn
rather than LUC to corn from a non-agricultural land use. Li et al. (2018) also refute previous
studies that purported to show considerable and irreversible LUC to corn, and they recognize
that the overall effect of corn ethanol production on total crop acreage was negligible (Stein
2018).
5 EPA (2018a) at page 35
6 This study is mis-cited by EPA and should have been Motamed et al (2016). See Section 8 References of this
report for full citation.
7 EPA (2018a) Box 3 at page 53
8 Motamed et al. (2016) at page 741
9 Wright et al. (2017) at page 9
10 Wright et al. (2017) at page 3
18 Ramboll
A review of the above studies indicates that a causal relationship between the RFS and LUC
has not been definitively established, and to the extent there is a causal linkage, the
relationship is likely weak. These studies as well as EPA (2018a) do not consider in a
quantitative way, the potential role of agricultural land loss on extensification. Although EPA
(2018a) present some information on agricultural land loss, these studies are not discussed
in any detail nor is the potential relationship to extensification.11 American Farmland Trust
estimates that between 1992 and 2012, almost 31 million acres of agricultural land were lost
to development—an average rate of loss of 1.55 million acres/year (Sorensen et al. 2018).
By comparison, Li et al. (2018) in their nationwide study noted an increase of 2.9 million
acres in 2012 as compared to 2008 (an average increase of 725,000 acres per year). It is
clear that farmland loss is considerable and very likely affects extensification.
3.3 Studies Relied Upon by EPA (2018a) to Quantify LUC Around the Time of
Enactment of the RFS Are Based on Unreliable Data and Likely
Overestimate LUC
One of the most pervasive issues in many studies of LUC around the time of the enactment
of the RFS is reliance on data sets that have proven to be inaccurate. Some of the key
publications that present estimates of LUC post-2007 and were relied on by EPA (2018a)
include the following:
• Wright and Wimberley (2013) reported that between 2006 and 2011, based on an
analysis of USDA’s National Agricultural Statistics Service’s Cropland Data Layer (CDL),
there was a 1.0-5.4% annual increase in the rate of change of WCB grasslands to corn
and soy with total LUC of 530,000 ha (Wright and Wimberly 2013).
• Johnston (2013) assessed wetland to row-crop transition rates in the Dakotas by
geographical information system analysis of the intersection of CDL with US Fish &
Wildlife’s National Wetlands Inventory (NWI) and the U.S. Geological Survey’s National
Land Cover Database (NLCD) and reported an annualized loss rate of 0.28% (5,203
ha./yr. over a 25-32 year period for NWI data) to 0.35% (6,223 ha./yr. over a 10 year
period for NLCD data) (Johnston 2013).
• Lark et al. (2015) analyzed LUC nationwide during the period 2008-2012 using CDL,
calibrated with ground-based data from USDA’s Farm Service Agency (FSA), and further
refined using data from the NLCD. They reported that 7.34 million acres (2.97 million
ha.) of previously-uncultivated lands became utilized in crop production while during the
same period 4.36 million acres (1.76 million ha.) of existing cropland were abandoned
with most of this being land enrolled in the CRP. They also reported that 1.94 million
acres (785,000 ha.) of converted lands were planted in corn as a “first crop.”
• Morefield et al. (2016) studied LUC using the USDA’s CDL over the 12-state Midwest
Region and report that between 2010 and 2013, 530,000 ha. (1.3 million ac.) of land
formerly in the CRP were converted to row crops with the “vast majority” of these lands
converted to soy and corn (Morefield et al. 2016). Of this 530,000 ha., 360,000 ha.
(890,000 ac.) were grassland, 76,000 ha. (188,000 ac.) were wildlife habitat, and
53,000 ha. (131,000 ac.) were wetland. They further report that areas in the Dakotas,
Nebraska and southern Iowa were hotspots for LUC.
• Mladenoff et al. (2016) assessed LUC in the Lakes States (MN, WI, and MI) and
determined that during the period 2008-2013, 836,000 ha. (2,066,000 ac.) of nonagricultural open lands were converted to agricultural use, with conversion to corn
11 EPA (2018a) Figure 14 at page 33
19 Ramboll
accounting for 480,000 ha (1,186,000 ac.) (Mladenoff et al. 2016). The authors used
USDA’s CDL data but combined shrubland and grass/pasture classifications into a single
“open land” classification and combined wetland/forest into a single class.
• Wright et al (2017) assessed grassland losses as a function of proximity to ethanol
refineries over the period 2008-2012 using USDA’s CDL and found that almost 4.2
million acres (1.7 million ha.) of arable non-cropland was converted to crops within 100
miles of refinery locations, including 3.6 million ac. (1.46 million ha.) of converted
grassland. Their analysis was based on applying a bias correction factor as per Lark et
al. (2015) and making other adjustments.
A major shortcoming of these studies is that the primary data set relied on (CDL) is poor at
differentiating between non-crop land classifications. Some authors acknowledged and
attempted to correct for this problem to varying degrees. These shortcomings limit the
confidence of conclusions regarding the form of the conversion, and even whether actual
land use conversion has occurred in some areas.
An illustration of the effect of CDL data uncertainties on many studies relied upon by EPA
(2018a) is a paper by Dunn et al. (2017). These authors examined data for 2006-2014 in 20
counties in the PPR using the CDL, a modified CDL dataset, data from the National
Agricultural Imagery Program, and in-person ground-truthing, and conclude that analyses
relying on CDL returned the largest amount of LUC by a wide margin. They further conclude
that errors associated with CDL-based analyses are a major limitation of conclusions drawn
from such analyses. In fact, the authors conclude that “the amount of hectares in the
potential error associated with CDL-derived results is generally greater than the number of
hectares the CDL-based analysis determined had undergone a transition from grassland,
forested land, or wetland to agricultural land”.12 This suggests that errors in classification
inherent in the CDL can result in uncertainty bounds that are of a larger magnitude than the
estimates of LUC.
As an example, Dunn et al. (2017) point out that the findings reported by Lark et al. (2015)
contradict USDA data indicating that cropland area has remained almost constant during the
period 2008-2012. Dunn et al. (2017) is of particular interest because the study focused on
the PPR, which has received the greatest attention due to documented ecosystem impacts
from habitat loss and wildlife impacts to sensitive species, including population declines of
prairie-dependent birds. It is interesting to note that EPA (2018a) acknowledges the specific
conclusions reported by Dunn et al. (2017) by stating that adjustments to data made by
Dunn et al. (2017) “led to much lower estimates of land use than either unadjusted CDL and
the NAIP for almost all counties examined [in the PPR]”.13 Despite this explicit
acknowledgment, EPA goes on to state that “Nevertheless, these earlier studies [referring to
the studies critiqued by Dunn at al. (2017)] qualitatively agree with patterns reported in
more recent national studies”.14 EPA’s use of the term “qualitatively agree with patterns” in
the context of studies that are attempting to quantify LUC after 2007 has little meaning and
is misleading to the extent it suggests agreement between studies where little to no such
agreement exists.
Table 1 presents a summary of selected results on the analysis conducted by Dunn et al.
(2017).
12 Dunn et al. (2017) at pages 8 and 9
13 EPA (2018a) at page 35
14 EPA (2018a) at page 35
20 Ramboll
Table 1: Summary of Selected Results as Reported by Dunn et al (2017).
Forest to Cropland (1000 ha.) Wetland to Cropland (1000 ha.)
Dunn et al. (2017) Lark et al. (2015) Dunn et al. (2017) Lark et al. (2015)
State NAIP (2013) CDL modified-CDL NAIP (2013) CDL modified-CDL
MNa 1.7 249 5.6 0 38 10
ND 0.83 222 0.44 0.01 25 7.4
SD 1.2 94 0.47 0 47 5.1
TOTAL 3.73 565 6.51 0.01 110 22.5
aIncludes forest and grassland that was converted to cropland.
CDL data has 30 m resolution and is tested for inaccuracy each year. The accuracy of the
CDL data varies yearly and regionally, which is why authors like Lark et al. (2015) make
modifications to the data in an attempt to make it more accurate. Dunn et al. (2017) tested
the accuracy of the modifications used by Lark et al. (2015) using NAIP data (see Table 1).
NAIP data are images that are have 1 to 2-meter resolution and allow side-by-side viewing
across years with high levels of accuracy. Dunn et al. (2017) found that even with the
corrections that Lark et al. (2015) made to the CDL data, the modifications produced “less
land flagged as undergoing LUC but the result may not be any more accurate than a result
produced without any modification”.15 These results suggest that for the areas assessed,
estimates using only uncorrected CDL data may overestimate actual LUC by a factor of 150
for forests and a factor of 11,000 for wetlands.
Further, EPA (2018a) mischaracterizes the accuracy of the CDL data16, as the Agency states
that CDL accuracies are generally > 90% for corn and soy and cites a study by Reitsma et al.
(2016) in support of that assertion; however the accuracies found in the article were actually
much lower than 90% for croplands (Reitsma et al. 2016). Reitsma et al (2016) used high
resolution imagery to distinguish between cropland, grassland, non-agricultural, habitat, and
water body land uses based on data from 2006 and 2012 in South Dakota. They found that
cropland accuracy ranged from 89.2% to 42.6% depending on whether there was more
cropland than grassland or the reverse. The authors chose data from South Dakota because
the state represents a climate transition such that row crops predominate in the eastern
portion of the state and grasslands predominate in the western portion of the state; the
change in the dominant vegetation allowed them to examine how the surrounding habitat
affected accuracy (Reitsma et al. 2016). The authors state that CDL errors that are inherent
to the data sets introduce uncertainty into land-use change calculations. EPA’s (2018a)
failure to recognize the difference in CDL accuracy is especially important since many
authors have documented that most of the observed LUC since 2007 has occurred at the
margins of cropland/grassland transition areas. While EPA (2018a) falls short of addressing
those specific data set concerns, EPA (2018b) recognizes that although satellite imagery can
provide information on the types of crops grown on a given parcel of land in a given year,
there is no nationwide system for tracking how crops from a particular parcel of land are
used, whether for domestically or internationally consumed biofuels or feed or other uses.
Thus, as EPA determined, its Triennial Report “did not purport to establish any causal link
between the RFS . . . and increased crop cultivation.”
15 Dunn et al. (2017) at page 10
16 EPA (2018a) at page 32
21 Ramboll
3.4 Recently Released Research Purporting to Establish a Quantitative Link
Between the RFS and LUC is Poorly Documented and Flawed
A recent presentation of research results by Lark et al. (Lark et al. 2019) appears to be an
ambitious effort to establish quantitative causal linkages between enactment of the RFS as a
policy to a variety of environmental outcomes using a series of interlinked models. However,
their approach rests on the assumption that the price of corn is heavily influenced by
increased demand for ethanol due to the RFS, yet the authors ignore other important factors
that could be equally or more important. Nor can they differentiate between price drivers
associated with global vs. domestic ethanol demand.
The modeling effort begins with estimates of increased demand for corn for ethanol and
effects of the increased demand on the price of corn. The authors then model the effect of
this increased demand on crop intensification and extensification and abandonment. The
authors then apply a “suite” of models, including what they describe as “causal economic
models” to evaluate the resultant land use changes as well as the following environmental
outcomes: NO2 emissions, carbon emissions, and consumptive water use.
With respect to the effect of RFS implementation in 2007 on LUC, the authors conclude that
during the period 2008-2016, the RFS resulted in an annual average increase of 6.9 million
acres of corn planted on existing cropland. In addition, the authors conclude that during the
same period, the RFS resulted in an annual average increase of 2.8 million acres of corn
planted on new cropland (i.e., cropland converted from other land cover types), or 43% of
the total increase in new cropland observed over the period. The authors attribute these
changes to a 30% increase in price of corn attributable to ethanol demand created by
the RFS.
The authors attempted to construct the counterfactual case; that is, simulate what the world
would have looked like without the RFS (called the “Business as Usual” scenario) and then
compare it to existing conditions in order to obtain and isolate the effects of the RFS.
However, when a counterfactual is posed that is too far from the real-world data, conclusions
drawn from even well-specified statistical analyses become based on speculation and
indefensible model assumptions, rather than empirical evidence. Unfortunately, standard
statistical approaches assume the veracity of the model rather than revealing the degree of
model-dependence, so this problem can be hard to detect. It is well understood that the
greater the distance from the counterfactual to the closest reasonably sized portion of
available data, the more the counterfactual depends upon model assumptions and
inferences. The seemingly large effects of the RFS reported by the authors are simply their
comparison between reality and a manufactured counterfactual situation which may or may
not reflect a realistic alternative state.
The authors’ entire analysis rests on estimating price increases attributable to RFS, and that
is the primary weakness evident in the work. The pricing model drives the rest of the
analysis. By not examining other model specifications, the inherent assumption regarding the
association of prices to the RFS remains speculative. In fact, corn prices over the period of
analysis were affected by a variety of other factors. For example, rapid economic growth in
developing countries led to growing food demand and a dietary transition from cereals
toward more animal protein. As a result, global consumption of agricultural commodities has
been growing rapidly. Further, most of the increase in corn prices has been driven by higher
oil prices. Figures 6 and 7 show nominal prices of West Texas Intermediate crude ($/bbl)
and corn ($/bu) for the latest 20-year period. The shaded areas reflect US recessions.
22 Ramboll
Figure 6: West Texas Intermediate Crude Prices ($/barrel).
(Source: Macrotrends. n.d.)
Figure 7: US Corn Prices ($/bushel).
(Source:
Macrotrends. n.d.)
Regarding the ability to “measure” land use change, Lark et al. (2019) explicitly recognize
many problems with spatial data interpretation and state that land use change was mapped
at the field level using the updated recommended practices by Lark et al (Lark et al. 2015).
However, the specific data sets used are not disclosed, and there is no description of how the
“recommended practices” were applied. The authors also do not provide an assessment of
whether and how the “recommended practices” improved estimates of LUC; rather they
simply present the results of their analysis. In addition to not presenting a full description of
23 Ramboll
the methods used, the authors rely on at least some data sets that are not publicly available,
therefore limiting the ability of a third party to replicate their work. For example, the authors
state that their analysis relies on a database built using field boundary data from the 2008
USDA Common Land Unit (CLU) among other data sources. The CLU database is compiled by
the USDA FSA and is not in the public domain.17
3.5 EPA (2018a) Failed to Adequately Account for the Role of Cropping
Practices and Production of Distillers Dried Grains with Solubles (DDGS) at
Ethanol Refineries as Important LUC Offsetting Factors
Numerous authors cited by EPA (2018a) who have researched LUC or increasing corn
production, and the relationship of these two phenomena to ethanol production have
acknowledged that much of the observed change (either LUC to agriculture or increasing
corn) may be attributable to cropping practices rather than conversion of non-agricultural
land to corn production. The primary cropping practices that may contribute to increased
production of corn, without implicating conversion of noncropland to row crops, are switching
fields to corn from other crops and double cropping of corn. The use of DDGS also reduces
the need for additional acreage of corn, which is often overlooked in analysis of LUC.
Similarly, EPA (2018a) fails to discuss the role of DDGS in potentially offsetting market
forces that may contribute to LUC occurring to meet demand for corn for ethanol.
3.5.1 Cropping Practices Have a Major Role in Meeting Increased Demand for
Corn
EPA (2018a) acknowledges the potential significance of cropping practices by citing, among
other studies, a study by Ren et al. (2016) in eastern Iowa that examined changes in corn
and soybean rotations around 2017 and found that the most common rotation over the
period 2002-2007 was corn/soy, but this rotation was not evident in 2007 and 2012 (with
59% of the area that had been in rotation prior to 2007 was in two or more years of
continuous corn after 2007). The most important conclusion reached by Ren at al. (2016) is
ignored by EPA (2018a): “From our analysis, it is clear that the expansion of corn production
after 2007 was realized by altering crop rotation patterns “ (Ren et al. 2016).18 Although this
study pertains to eastern Iowa it is of particular importance since Iowa is the largest
producer of corn in the US (17.4% in 2018; USDA-NASS, 2019).19
EPA also refers to a study by Plourde et al. (2013) when discussing intensification, but EPA
does not underscore the primary conclusion of these authors (Plourde et al. 2013). In
assessing data for two distinct time periods (2003–2006 and 2007–2010) in a nine state
“Central United States” area (states of AR, IL, IN, IA, MS, MO, NE, ND, and WS) these
authors found that the total area impacted by corn production only increased slightly
between the two periods, while there was a much greater increase in the intensity of
continuous corn rotation patterns. Similarly, in discussion about corn acres increasing mostly
on farms that were previously soy over the period 2006-2008, EPA cites Beckman (2013)
“…that increases in corn acreage from 2001-2012 resulted in a net decrease in barley, oats,
and sorghum “ (Beckman et al. 2013).20
17 In fact, the FSA website states the following:
CLU is not in the public domain. Section 1619 of the Food, Conservation, and Energy Act of 2008 (Farm Bill),
only allows the sharing of this data to individuals or organizations (governmental or non-governmental) certified
by FSA as working in cooperation with the Secretary of Agriculture. Users of the data must be providing
assistance to USDA programs, and must require access to CLU data to complete that work (USDA 2012).
18 Ren et al. (2016) at page 157
19 Calculated from p. 11 in USDA-NASS 2019
20 EPA (2018a) at page 40
24 Ramboll
Although EPA (2018a) acknowledges that changes in cropping practices “could be
significant,” they do not provide a quantitative or even qualitative assessment of how
significant cropping might be in meeting increased demand for corn for ethanol. Inadequate
accounting of the role of cropping practices in discussion of ethanol and LUC contributes to
the misperception that the increase in corn production to fulfill demand for corn for ethanol
necessarily results in adverse LUC.
3.5.2 Production of DDGS Has Offset a Substantial Amount of Demand for Corn
as Livestock Feed But this was Not Adequately Acknowledged by EPA
(2018a)
EPA (2018a) states that approximately 12% of the total corn production from 2014-2016
was returned to the feed market in the form of DDGS which is produced during the
distillation of corn for ethanol. EPA (2018a) also acknowledges a study by Mumm et al.
(2014)(Mumm et al. 2014) who conclude that although 40% of corn grown in 2011 was
estimated to be utilized in ethanol production, when the offsetting effect of DDGS is
accounted for, this acreage is reduced to 25%.21 Although EPA (2018a) cites some of the
findings reported by Mumm et al. (2014), they fail to acknowledge some very important
conclusions of these authors regarding potential future projections. Mumm et al. (2014)
evaluate four scenarios considering the impact of technological advances on corn grain
production, two scenarios focused on improved efficiencies in ethanol processing, and one
scenario reflected greater use of DDGS. For each scenario, Mumm et al. (2014) estimate the
land area attributed to corn ethanol. Assuming reasonable increases in corn grain yield with
anticipated new yield technologies coming into play between 2011 and 2026, the authors
estimate that the percentage of land devoted to corn for ethanol will be reduced from the
25% estimated for 2011 to 13% in 2026.
Irwin and Good (2013) reported that DDGS account for much of the decline in feeding of
whole corn to livestock since 2007-2008. According to the National Corn Growers
Association, between 1,013 and 1,222 million bushels of corn were displaced by DDGS and
Corn Gluten Feed (CGF; produced by wet milling at ethanol refineries) between 2009 and
2016 (National Corn Growers Association 2019). For illustration purposes, if we assume an
average yield of corn per acre per year of 125 bushels (USDA-NIFA n.d.), then over the
period 2009 to 2016 DDG/CGF may have displaced ~8.1 – 9.8 million acres of corn
production per year that otherwise would have gone for livestock feed. This offsetting factor
is more than the 6.9 million acres (yearly average) of corn planted on existing cropland and
the 2.8 million acres (yearly average) of new cropland alleged by Lark et al. (2019) to be
attributable to the RFS for the period 2008-2015.
21 Mumm et al. (2014) Box 3 at page 53
25 Ramboll
4. CHANGES IN AGRICULTURAL PRACTICES REDUCE
THE LIKELIHOOD OF ENVIRONMENTAL IMPACTS TO
WATER RESOURCE AVAILABILITY AND QUALITY
The relationship between corn production and water resource availability and water quality
varies geographically and temporally. What is clear but not quantitatively recognized by EPA
(2018a), is that advancements in farming practices and technology have reduced the
negative impact of farming on the environment. Recent technological advances have resulted
in considerable improvements in water use in agriculture in general, and for corn growing, as
well as reducing the use of agrochemicals such as fertilizers and pesticides. These
improvements have the effect of reducing the likelihood of adverse impacts to water
resource availability and quality.
There is no dispute that all agricultural production is strongly tied to the availability and
quality of fresh water. Farming practice is based on local and regional climatic and soil
conditions which determine whether crops are grown using irrigation from surface water or
groundwater sources or are non-irrigated and rely solely on precipitation. Approximately
one-quarter of US cropland is irrigated (NAS 2019). The total US irrigation withdrawals for all
crops in 2010 averaged approximately 115 billion gallons per day (NAS 2019). The
availability of sustainable water sources, more so than any other issue, poses the greatest
threat to crop productivity into the future. Corn is a water intensive crop; however, most
corn grown in the US is non-irrigated, and this is recognized by EPA (2018a). Over the past
decade, there has been increased use of modern and precision agriculture methods (for both
water use and agrochemical application) which retain soil moisture and reduce tilling. This
trend is expected to continue into the future, with increasing efficiency and effectiveness of
resource use, which will result in reducing water and fertilizer needs.
4.1 The Triennial Report’s Discussion of Water Use and Water Quality
Key conclusions in EPA (2018a) relevant to the RFS reset discussion include:
• The environmental impacts of increased biofuel production on water resource use and
water quality were likely negative in the past but limited in impact.
• A potential exists for both positive and negative impacts in the future with respect to
water resource use and availability, and impact to water quality both locally and
regionally.
• Environmental goals for biofuels production could be achieved with minimal
environmental impacts (including water and fertilizer/pesticide use) if best practices
were used and if technologies advanced to facilitate the use of second-generation
biofuels feedstocks.
These messages are consistent with our findings that the environmentally protective goals
for biofuel production are highly achievable as best management practices and technological
advances in farming continue to be adopted by the farming community. While challenges for
fully distributing and implementing these approaches will remain in certain areas (e.g., NAS
2019), the economic drivers for implementing best practices such as increased productivity
and savings derived from resource conservation, will undoubtedly continue to steer the
farming community toward greater implementation of modern approaches.
The most important statements presented by EPA (2018a) are the forward-looking
considerations that biofuel production can (and will) achieve environmental goals by using
modern practices. EPA (2018a), however, paints a picture of negative impacts from biofuels
26 Ramboll
feedstock production without using specific and conclusive data to support the claims. For
example EPA (2018a):
• Asserts that increased intensity of corn production on existing cultivated land and
expansion of crop land negatively impacts water quality but presents no direct evidence
of a causal link.
• Does not rely on direct analysis to assess the magnitude of potential water quality
impacts but instead makes general statements with no quantitative analysis that
connects the water quality impact to specific areas, land, or conditions.
• Recognizes that quantitative assessments are necessary to evaluate whether increases
in water demands can be directly attributed to feedstock production. However, EPA
(2018a) does not provide the studies or backup to support this evaluation, rather merely
speculates that negative impacts must exist.
EPA (2018a) suggests that growing corn for ethanol feedstock is a major contributor to
eutrophication and hypoxic conditions in the northern Gulf of Mexico and eutrophication in
western Lake Erie. EPA (2018a) attributes these conditions to substantial nutrient loading
from agricultural runoff. However, the impact, if any, from corn grown for ethanol production
on water quality and availability is not substantiated with data. For example, the attribution
by EPA (2018a) that biofuel feedstock production is a contributing factor to these conditions
appears to rely on models such as those presented by Michalak et al. (2013) that state corn
production “could” be a contributing factor and LaBeau, et al. (2014) that speculate biofuel
production “could” contribute to increased nutrient loading to surface water (Michalak et al.
2013, LaBeau et al. 2014).
There may be no dispute that excess nutrient loading from the key watersheds that
discharge into western Lake Erie and the northern Gulf of Mexico contribute to eutrophication
and hypoxia; however, the watersheds are composed of a complex mix of urban and rural
uses and wastewater discharges. Agricultural runoff should be considered an important
component; however, the direct causal link to corn grown for ethanol production (compared
to all other uses and compared to all other agricultural activities) is not substantiated.
Indeed, no studies reviewed by Ramboll convincingly link increases in biofuel production to
regional hypoxic conditions in surface water bodies. Such conditions have been increasing in
frequency and severity since the 1950s, long before ethanol production increased.
EPA (2018a) also fails to acknowledge the importance of regional weather on the occurrence
and severity of large-scale hypoxia events. For example, one major variable determining the
size of the hypoxic zone (colloquially known as the “dead zone”) in the Gulf of Mexico is the
rate of flow in the Mississippi River, which may be highly-variable on an annual basis. The
National Oceanic and Atmospheric Administration (NOAA) is predicting that the 2019 dead
zone in the Gulf of Mexico will cover an area of 7,829 square miles which is close to the
record size of 8,776 square miles in 2017 and more than one third larger than the 5-year
average size of 5,770 square miles (NOAA 2019). NOAA states that a major factor
contributing to the dead zone in 2019 is the abnormally high amount of spring rainfall that
has resulted in flows in the Mississippi and Atchafalaya Rivers that are 67% above the
average flows over the last 38 years. Data collected by the United States Geological Survey
(USGS) indicate that because of these high flows, nitrate loads are about 18% above the
long-term average, and phosphorus loads are approximately 49% above the long-term
average (USGS 2019).
Finally, EPA (2018a) also fails to recognize that changes in flood-control and navigation
improvements in the Mississippi River watershed during the first part of the 20th century
27 Ramboll
dramatically affected the amount of flow from the upper Midwest watersheds that would
enter the Gulf of Mexico without environmental buffering from natural tributaries (NOAA
2000). The higher flow rates allowed greater unimpeded flow of water containing nutrients to
the Gulf of Mexico than would otherwise have occurred (NOAA 2000).
It is interesting that while EPA (2018a) relies on speculation and qualitative studies to
associate corn grown for ethanol to hypoxia in western Lake Erie and the Gulf of Mexico, EPA
(2018a) also reports that there has been a reduction in total nitrogen concentrations in
surface water bodies in Iowa (the highest corn producing state and an area of corn growth
intensification). We note that nutrient loading to the Gulf of Mexico has been relatively stable
on average since at least 1980 – an important consideration as corn yield has increased
during this time period (USGS n.d.) even as farmed acreage has been stable. This indicates
that even during the increased use of corn for ethanol, there has been no net change to
nutrient loading to the Gulf of Mexico and thus there is no support for the assertion of a
direct relationship between ethanol production on the hypoxia conditions in the Gulf of
Mexico. This evidence refutes claims made to the contrary by EPA (2018a).
Figure 8: Annual Nitrate and Nitrite Loading to the Gulf of Mexico 1980-2017.
(Source: USGS n.d.)
The fact that agricultural practices in general can result in nutrient runoff is acknowledged,
although modern efficiencies and conservation methods have improved over time. Modern
practices apply technology for increased efficiency and harness continuously improving data
analysis to develop and implement best management practices. There is strong evidence
that the agricultural community, including biofuel feedstock producers, are adopting modern
agricultural practices (Vuran et al. 2018). EPA (2010 and 2018a) acknowledge and strongly
advocate for these modern practices and note that negative impacts to environmental
resources will be reduced with the use of modern approaches to tilling, fertilizer use, water
use, and precision agriculture. If these practices were not being implemented, the
expectation is that nutrient loading, and thus hypoxic conditions, should have been
increasing along with the increased yield over the past several decades. However, the data
from NOAA and the USGS show stability in nutrient loading, which would thus indicate that
the net flux of nutrients has not been increasing even while crop yields may have been
increasing.
28 Ramboll
4.2 Agricultural Improvements in Irrigation are Reducing Water Use
The trend of increasing yield per acre farmed extends to both irrigated and unirrigated corn
crops, indicating that changes in yield are not likely attributed to irrigation alone. According
to the 2012 statistics from the USDA (USDA-ERS 2018a) irrigated corn acreage represented
about 25% of all irrigated acreage in western states, and about 24% of all irrigated acreage
in the eastern states. Additionally, the USDA has shown that irrigation for all crops, including
corn, has decreased even as the farming acreage has essentially been stable over the past
35 years. The USDA attributes this trend to improvements in physical irrigation systems and
water management. The USDA also notes that significant capital investments in on-farm
irrigation is continuing, particularly in the western states, where most of the irrigated farmland is concentrated. As an indication of a positive trend in irrigation reduction, the
University of Nebraska, Lincoln reports that in Nebraska (as a bell-weather of other dry
western states), the percentage of all corn acreage that is irrigated has declined from a high
of 72% in 1981 to 56% in 2017 (University of Nebraska 2018).
USDA data indicate that there has been no substantial change in the volume of water applied
to corn crops (for grain or seed) since the 1990s (Figure 9) (USDA-NASS 2013). This
stability in the average volume of water applied to corn crops, combined with the plateau in
area of corn planted, suggests that the quantity of water applied to corn crops has not
substantially increased since at least the 1990s, despite intensification.
Figure 9: Volume of Water Applied to Irrigated Corn Crops Since 1994, by Irrigation
Method.
(Source: USDA-NASS 2013)
Because irrigation provides a stable water resource to the farmed field (assuming the water
source that supplied irrigation is also stable), crop yields on irrigated land are generally more
regular (e.g., less variable and often more substantial) than for non-irrigated land (Figure
10). Note, however, that from at least 1979 to 2013, increases in yield also have been
observed in unirrigated corn crops (USDA Farm and Ranch Irrigation Survey). Specifically, in
1979, irrigated land produced 127 bushels per acre on average, versus 85 for unirrigated
land. By 2013, irrigated land produced 196 bushels per acre on average, versus 129 bushels
per acre for unirrigated land, representing a 54% and 52% increase, respectively.
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1994 1998 2003 2008 2013
Acre-Feet per Acre
Average Volume of Water Applied per Acre of Corn Planted
by Irrigation Method
(Acre-Feet/Acre)
Average volume of water for
corn for grain or seed with
pressure systems (AcreFeet/Acre)
Average volume of water for
corn for grain or seed with
gravity systems (AcreFeet/Acre)
Corn = for grain or seed
29 Ramboll
Figure 10: While Irrigated and Unirrigated Corn Crops Have Both Experienced
General Increases in Yield, Irrigated Crops More Reliably Produce Higher Yields.
(Source: USDA-NASS 2013)
Regions of greatest corn production are moving eastward away from the regions of greatest
irrigated water use, providing further evidence that year-to-year changes in corn planting
have little to negligible impact on total U.S. water supply. For example, in 2016, the five
leading states in annual corn production (Illinois, Nebraska, Iowa, Minnesota, and South
Dakota) produced over 60% of the corn grown in the U.S. (USDA-ERS 2016). This statistic is
a change from the 2010s when the irrigation of corn crops was even more concentrated in
the drier Northern Plains (Colorado, Montana, Nebraska, Wyoming, and North and South
Dakota) and dry Southern Plains (Kansas, Oklahoma, Texas) regions. In 2007, the USDA
reported that the thirteen leading states in total irrigated acres for all crops of farmland,
accounted for nearly 80% of all U.S. irrigated land, but that they were concentrated in arid
western states (USDA-ERS 2018a). Of the top five corn-producing states, none made up
more than 15% of the total U.S. irrigated acreage. The increased growth in wetter states
such as Illinois and Minnesota eases the water supply demand for the total yield of all
irrigated corn acres.
USDA anticipates that changes in corn production will result in appreciable yield increases
(e.g., 16.1 more bushels per acre by 2028) (USDA-NASS 2017). It is therefore reasonable to
expect that technological and methodological changes to farming will continue to result in
significant reductions in water use per unit of corn production. Table 2 presents an overview
of prevailing opportunities for water savings in irrigated agriculture.
0
50
100
150
200
1970 1977 1984 1991 1998 2005 2012 2019 2026
Corn Yield (Bushels/Acre)
Real and Projected Average U.S. Corn Yield Per Acre (1970 to
2027) and Reported Irrigated/Unirrigated Corn Yield (1979 to
2013)
Irrigated
Unirrigated
Average yield until
2013 then projection
30 Ramboll
Table 2: Technological and Methodological Improvements to Irrigation of Corn
Crops.
Technological
Advancement
Approximate
water savings
factor
Baseline
scenario
Demonstrated
potential yield
increase
Notes
Subsurface
drip irrigation
25-35% vs. center
pivot system
15-33% Costs 40-50% higher
than center pivot systems
but returns on investment
can accrue within 2—5
years. In 2007, only
0.1% of irrigated corn
farms used this.
Rain water
harvesting and
storage
50+% vs. natural soil
runoff
20-52% Includes 1) harvesting of
surface runoff from
roads; 2) field microcatchment to increase
fallow efficiency in rain.
Precision
agriculture
13% vs. without
governmentrun weather
network
8% Includes use of global
positioning system,
geographical information
systems, in situ soil
testing, remote sensing
crop and soil status, realtime weather info.
Adoption rate slightly
higher in corn belt.
Conservation
structures
18% vs.
conventional
agriculture
27% Examples include grass
vegetation strips.
Adoption is higher in
areas of highly erodible
land.
(Sources: Netafim n.d., Gowing et al. 1999, Shangguan et al. 2002, National Research
Council 2008, Biazin et al. 2012, Allen 2013, Barton and Elizabeth Clark 2014, Center for
Urban Education about Sustainable Agriculture (CUESA) 2014, Qin et al. 2015)
Subsidized government programs offer farmers incentives to implement water conservation
strategies. For example, because of prolonged drought conditions, California recently
installed a network of 145 automated statewide weather stations, so that farmers could
manage their water resources more efficiently (CIMIS 2019).
With the focus on drought and long-term reductions in supplied water in some states (such
as California), more farms are moving away from “traditional, less-efficient application
systems” (USDA-ERS 2018a). For example, the number of farms using inefficient gravity
irrigation systems decreased from 62% in 1984 to 34% in 2013, converting mostly to
pressure-sprinkler irrigation which is more efficient than gravity irrigation, but which still
leaves room for improvement. Currently, almost 10% of farms use soil-moisture or plantmoisture sensing devices or commercial irrigation scheduling services. Sensor technology
can optimize irrigation scheduling and hence increase water use efficiency. Though less than
2% of farms use simulation models right now (USDA-ERS 2018a), the anticipation is that
additional large industrial farms (which make up a large volume of total yield) also will
employ water use simulation models that are based on corn growth patterns and weather
conditions. Adoption of these technologies will continue to grow in the U.S., and particularly
in the west, where 72% of water irrigation investment takes place and farmers have recent
experience with low water supply following the 2012-2016 drought.
31 Ramboll
Barriers to implementing these measures are lessening but it is recognized that issues
relating to the following are still at play: (1) farmer concerns about the impact of new
practices on yields; (2) tenant or lease issues that discourage the installation or use of new
equipment; (3) institutional issues related to Federal Crop Insurance Program; (4) irrigation
water rights laws like “use it or lose it;” and (5) cost of implementation. The Great Plains
area had traditionally been risk-averse to implementing subsurface drip irrigation techniques
because of the upfront costs and uncertain lifespan of the systems; however, there have
been improvements in the technology and irrigators are increasingly aware of the additional
incentives for water conservation and protecting water quality (Lamm and Trooien 2003).
Genetic engineering or selection for improved drought tolerant corn cultivars has also
contributed to increases in corn crop productivity. Additionally, genetic breeding has shown
that yields can be maintained with lower water requirements (nearly 25% reduction), in
addition to studies that suggest corn crops can forego the initial irrigation without significant
adverse effects to the harvest (Xue, Marek, et al., 2017). Mcfadden et al. (2019) reported
that with drought being among the most significant cause of crop yield reduction, the spike
in use of irrigation water to reduce such losses can be a major negative impact to water
resource availability particularly in the drier western states. Even though many waterintensive crops, including corn, are grown on non-irrigated land, the use of drought-tolerant
corn, which was commercially introduced in 2011, had increased to over 22 percent of the
total U.S. planted corn acreage by 2016 (Mcfadden et al. 2019). More important, this percent
of use was greatest in the driest corn-producing states of Nebraska (42 percent) and Kansas
(39 percent). Even the less severe drought-impacted though important corn-growing states
of Minnesota, Wisconsin, and Michigan saw drought-tolerant corn planting ranging between
14 and 20 percent of total acreage. There is no guarantee that drought-tolerant crops will be
effective against the most severe droughts; however, this use can be seen as similar to the
use of crop-insurance to protect farmers against loss while still providing product for use
during low-water years. The longer-term advantage is that less irrigation water would be
required even under normal water years.
Liu, et al (2018) states that best management practices for reducing agricultural non-point
source pollution are widely available even with the challenges related to the large number of
agricultural producers and the spatially variable and temporally dynamic nature of the
nutrient loading cycles. Greater adoption of the improved practices will rely on: (1) better
identification of the higher risk areas; (2) a commitment from local, state and federal
authorities to assist the farming community in applying the new approaches by allowing
innovations to be implemented without unnecessary regulatory impediments; and (3) better
financial incentives. Liu, et al (2018) also note that lack of information and misdirected
communications can negatively impact the adoption of new techniques and encourages
government, consumers, and farmers to work together to more consistently communicate
the advantages of technology adoption.
4.3 Technological Improvements in Agriculture Translate to Reductions in
Potential Water Quality Impacts
Government institutions including USDA and academic institutions such as California State
University, Fresno have promoted research into the use of precision agriculture to reduce the
need for both nutrient and pesticide use (as well as supplied water) because in addition to a
reduced environmental impact, the techniques result in cost savings for farmers by
improving yield per acre. In addition, the greater use of area-wide databases that provide
better information and awareness of water quality conditions helps to identify areas where
additional best management practices can be applied. For example, utilization of the USGS
32 Ramboll
water quality mapping reports (e.g., USGS 2017) helps provide data for surface water
chemistry trends (i.e., nutrients, pesticides, sediment, carbon, salinity) and aquatic ecology
from 1972 to the current editions.
Recent advancement in technology for fertilizers and pesticides have reduced the use of
agricultural chemicals while increases in crop yield continue. While use of fertilizer on corn
typically accounts for more than 40% of commercial fertilizer used in the U.S. since the
1980s (USDA-National Resources Conservation Service [NRCS] 2006, EPA 2018c), there has
been a plateau in the mass of fertilizer applied to corn crops (on average on a state-by-state
basis), as well as an overall decrease in the mass of pesticide applied to corn crops (see
Figure 11;(USDA-NASS 2013, Fernandez-Cornejo et al. 2014, USDA-ERS 2018c). In 1987,
the average mass of pesticide active ingredient application per area of corn planted in the
U.S. peaked at approximately 3.58 pounds per acre. In 1984, fertilizer use peaked at
approximately 290 pounds per acre. The USDA and EPA report similar trends; for example,
U.S. spending on pesticides for all crops peaked in 1998, and consumption of commercial
fertilizers peaked in 1981 (Fernandez-Cornejo et al. 2014, EPA 2018c).
Figure 11: Both Pesticide and Fertilizer Use on U.S. Corn Crops Appear to Have
Peaked in the 1980s
(Source: USDA [ibid.])
The application of slow released (or controlled) nitrogen fertilizer during peak uptake is one
key to improving nutrient efficiency and utilization (Lal, R. (Ed.), Stewart 2018). Under
optimum moisture and temperature conditions, use of slow released nitrogen fertilizer can
greatly reduce leaching of nutrients. However, further research is necessary to discern the
best slow release fertilizer for a given crop species (Rose 2002). Other advanced chemical
technologies such as use of bioreactors, can offer additional reductions in pesticide and
fertilizer in corn production. Bioreactors such as those that redirect water in farm fields
through tiles to underground woodchips where nitrate is removed by microorganisms, can
reduce nitrogen in run-off by 15% to 90% (Iowa Corn n.d., Christianson 2016).
Recent surveys and data from the use of the modern and technology-based agricultural
management systems have shown reduced resource needs and significant cost savings (NAS
0.1
1
10
100
1000
0
20
40
60
80
100
120
140
160
180
200
1950 1957 1964 1971 1978 1985 1992 1999 2006 2013
Fertilizer & Pesticide Use (Pounds/Acre)
(Log-10)
Corn Yield (Bushels/Acre)
Average U.S. Corn Yield (Left Axis) vs.
Average Fertilizer and Pesticide Use (Right Axis, Log-10)
(1950 to 2016)
Corn = for grain or
seed
Average State Fertilizer Use
Average U.S.
Pesticide Use
Average U.S.
Corn Yield
33 Ramboll
2019; Liu, et al. 2018). The USDA also has shown that a “guidance-based” system for corn
production can save thousands of dollars each year with a return of investment of two to
three years for this technology (USDA-NRCS 2006). Furthermore, the USDA reports that
“…precision agriculture reduces environmental pollution and improves water quality by
reducing nutrient runoff [while] other benefits include: improved crop yield; reduced
compaction [of fields]; labor savings; and more accurate farming records.” Finally, there are
fewer barriers to nearly all farmers in using precision technologies because of grants that are
available for purchasing equipment and free public access to the Federal Global Position
System that makes it economically possible for producers to use the new precision tools to
save energy and reduce costs by improving or implementing the following: (a) yield
monitoring, (b) grid soil sampling, (c) precision and variable-rate nutrient application; and
(d) soil moisture monitoring. Precision agriculture technologies are quickly adopted by
farmers in the United States; the rate of adoption for all precision technologies was 72.47
percent in 2010, as compared to just 17.29% in 1997 (Vuran et al. 2018). USDA found that
if guidance-based farming was used on just 10 percent of planted acres in the U.S., fuel use
would be cut by 16 million gallons, herbicide use would be reduced by 2 million quarts and
pesticide use would lower by 4 million pounds per year(USDA-NRCS 2006). The results
would be better environmental conditions and substantial increase in financial savings for the
farmer/producer.
4.4 Reduction in Water Usage for Ethanol Processing
Opportunities exist for implementing water reduction programs during biofuel production.
Excluding the non-fuel component, the primary processes that require water consumption in
ethanol production include heating and cooling. Water losses occur through: (1) evaporation,
drift, and blow down from cooling towers; and (2) blow down from boilers. Losses vary with
both the ambient temperature of the production plant, and the degree of boiler condensate
and blow down water reuse and recycling. Generally, dry mills use less water than wet mills.
In a 2007 Renewable Fuels Association survey of 22 ethanol production facilities
(representing 37% of the 2006 volume produced), dry mills used an average of 3.45 gallons
of water per gallon of ethanol produced and wet mills used an average of 3.92 gallons of
water per gallon of ethanol produced. Efforts to use recycled waste water are increasing and
will reduce the need for using supplied water during the conversion process.
Keeney and Muller (2006) report that in Minnesota, water use by dry mill ethanol refineries
ranged between approximately 3.5 and 6.0 gallons of water per gallon of ethanol in 2005
which followed a 21% reduction in water use by dry mill ethanol refineries from 1998 to
2005 (representing an annual reduction of approximately 3%). More recently, Dr. Steffen
Mueller of the University of Illinois (Chicago) Energy Resources Center notes that water
consumption by ethanol plants is continuing to decrease and dramatically so. Mueller (2016)
documents a reduction of approximately 5.8 to 2.7 gallons of water per gallon of ethanol
produced between 1998 and 2012 in dry mills.
Wu and Chiu (2011) noted additional trends that suggest decreases in the water demands of
existing and new ethanol plants. Freshwater consumption in existing dry mill plants had, in a
production-weighted average, dropped 48% in less than 10 years to water use rates that are
17% lower than typical mill values. Water use can be minimized even further through
process optimization, capture of the water vapor from dryers, and boiler condensate
recycling to reduce boiler makeup rates.
34 Ramboll
5. RECENT ESTIMATES OF HEALTH DAMAGES FROM
CORN PRODUCTION ARE UNRELIABLE AND
MISLEADING
A recent publication in Nature Sustainability (Hill et al., 2019) estimates US annual health
damages caused by particulate air quality degradation from all direct farm and indirect
supply chain activities and sectors associated with corn production. Although the authors do
not reference the RFS, they do mention corn grown for ethanol, and the publication has been
referenced by third parties in a manner suggesting that corn grown for ethanol may be
associated with adverse health outcomes. Ramboll’s review indicates that the conclusions
presented by Hill et al. (2019) are unsubstantiated and likely overestimate adverse health
impacts if any.
These “life-cycle” activities and sectors examined by Hill et al. (2019) include air emissions
from farms and upstream processes that produce the chemical and energy inputs used in
corn crop production: fuel, electricity, agrichemical production, transportation and
distribution. Downstream activities such as corn distribution and food/fuel processing are not
considered in the study. The authors develop an annual county-level emissions inventory of
air pollutants for all related sectors, then apply a specific “reduced form model” (RFM) that
converts those emissions into spatial distributions of annual fine particulate air
concentrations (or PM2.5) and resulting human exposure, premature mortality, and
monetized health damages.
PM2.5 comprises microscopic particles smaller than 2.5 microns in diameter, with chemical
constituents that include direct (primary) emissions (dust and smoke) along with the several
secondary compounds chemically formed in the atmosphere from gas precursor emissions:
nitrate from nitrogen oxide (NOx) emissions, ammonium from ammonia emissions, sulfate
from sulfur oxide (SOx) emissions, and secondary organic aerosols (SOA) from volatile
organic compound (VOC) emissions. PM2.5 is a concern for human health because particles of
this size can penetrate deep into the lungs and enter the bloodstream, which can potentially
result in both acute and chronic effects to the respiratory and cardiovascular systems.
Epidemiological studies have found associations between PM2.5 exposure and mortality and
these associations are used by Hill et al. (2019) to calculate health impacts from corn
production. The authors find that impacts to annual-average PM2.5 concentrations from corn
production are primarily driven by emissions of ammonia from nitrogen fertilizer.
Ramboll reviewed details of the specific RFM used by Hill et al. (2019), called the
Intervention Model for Air Pollution (InMAP; Tessum, Hill, et al., 2017) to calculate ambient
PM2.5 impacts from corn production. InMAP calculates atmospheric dispersion, chemistry and
removal (deposition) from direct PM2.5 and precursor gas emissions. It then converts
resulting annual PM2.5 concentrations to human exposure metrics from which premature
mortality and associated damages are determined. Hill et al. (2019) provide only an
overview of the process to develop emission inventories, which limits our capacity to review.
However, given the importance of ammonia emissions to the results reported by Hill et al.
(2019), we enumerate well-known uncertainties involved in estimating emissions from
agricultural activities. In addition, although Hill et al. (2019) did not provide explicit details
on the impact assessment, we provide a summary of the key uncertainties associated with
estimating health and associated costs from PM2.5 exposures. It is noteworthy that the
authors do not provide any uncertainty or sensitivity analyses that can provide important
context for the interpretation of the results and conclusions.
35 Ramboll
Based on our review of Hill et al. (2019) and of Tessum et al. (2017), we draw the following
conclusions:
• InMAP uses annual-average data for emissions, meteorology, and chemical/removal
rates to estimate annual-average PM2.5 impacts. Use of annual averages is inappropriate
for representing processes that operate over shorter time scales ranging from minutes
to several months (e.g., atmospheric dispersion and chemical formation of PM2.5). The
authors acknowledge that this weakness in their approach results in spatial errors in
annual average PM2.5 calculations. These spatial errors can significantly impact the
resulting exposure and mortality estimates. The authors, however, do not present
sensitivity analyses to assess the impact of the model assumptions, nor do they include
any plausible range of uncertainty or variability with their modeled PM2.5 concentration
or mortality estimates.
• The 2005 modeling year upon which InMAP is based is not representative of more recent
chemical conditions of the atmosphere in the U.S. because there have been significant
reductions in precursor emissions that directly reduce the capacity to form PM2.5. We
estimate that this leads to an overestimate of the PM2.5 contributions from corn
production by more than a factor of 2. Therefore, resulting health and economic
damages are likely overestimated.
• Ammonia emission estimates, which are the largest driver of mortality in the Hill et al.
(2019) analysis, are the most uncertain aspects in any air quality modeling exercise
because: (1) emissions are largely from agricultural sources that vary both spatially and
temporally due to weather and farming practices; (2) many different methods are used
to estimate ammonia emissions, and each can yield very different rates and exhibit a
high degree of error; (3) annual average ammonia emission inventories fail to account
for important seasonal variations and related complex interactions with sulfate and
nitrate chemistry; (4) ignoring diurnal and intra-daily ammonia emission variations have
been shown in the literature to overestimate ambient ammonia concentrations by as
much as a factor of 2. These numerous uncertainties and compounding error rates call
into question the estimates of emissions that drive the rest of the Hill analysis.
Based on our review, InMAP is not typically able to reproduce PM2.5 impacts estimated by
more complex state-of-the-science air quality models. In fact, its performance is worst for
the very PM2.5 component (ammonium) that Hill et al. (2019) model indicates is the highest
contributor to PM mortality from corn production. This renders InMAP especially unreliable
for this key PM component.
In addition to the number of significant uncertainties in all modeling aspects of the Hill et al.
(2019) analysis, including the emissions estimates and the RFM InMAP modeling, there is
also a significant amount of uncertainty associated with estimating health impacts from air
pollution concentrations and from quantifying the costs of these health impacts.
The health impact assessment is based on a single epidemiological study that found
associations between PM2.5 concentrations and mortality. While these studies suggest that
such an association exists, there remains uncertainty regarding a clear causal link. This
uncertainty stems from the limitations of epidemiological studies to establish causality
because these studies are based on inadequate exposure estimates and these studies cannot
control for many factors that could explain the associations between PM2.5 and mortality –
which, for example, may not be related to PM2.5. from the source being investigated (e.g.,
lifestyle factors like smoking). In fact, the components of PM2.5 that may be associated with
adverse health effects are yet unknown, but evidence suggests that carbonaceous particles
36 Ramboll
are more toxic, than inorganic particles such as those derived from ammonia and nitrate or
sulfate.
Overall, the uncertainties enumerated above result in unreliable estimates of PM2.5 exposure,
mortality and related costs associated with corn production, each associated with a large
range of variability.
37 Ramboll
6. ENVIRONMENTAL IMPACTS ASSOCIATED WITH
ETHANOL PRODUCTION CANNOT BE VIEWED IN A
VACUUM, WITHOUT CONSIDERATION OF SUCH
IMPACTS ASSOCIATED WITH GASOLINE
PRODUCTION.
EPA (2018a) acknowledges that it fails to address environmental impacts associated with
gasoline production. Spills of petroleum, gasoline, and a wide range of other fluids used in
the exploration, production, and refining processes as well as land use change to support
those activities all have adverse effect on water quality, ecosystems (including wetlands),
and wildlife. Additionally, both conventional and unconventional oil and gas extraction place
demands on water supply. Failure to address impacts associated with gasoline production
relative to impacts from ethanol production does not present a balanced view of alternative
energy sources and casts a negative bias on ethanol production. Parish et al. (2013)
recognize the importance of understanding differences in environmental effects of
alternative fuel production so that the relative sustainability of alternatives can be
adequately assessed in policy-making and regulatory decisions. Parish et al. (2013)
assessed negative environmental impacts through the supply chain for ethanol production
and gasoline production and found that impacts from ethanol production are more spatially
limited, are of shorter duration, and are more easily reversed than those associated with
gasoline production. It was beyond the scope of this report to expand upon the work of
Parish et al. (2013) or other comparative studies, rather this Section presents a brief
description of the wide range of potential impacts associated with petroleum production
stemming from land use changes as well water use and impacts to water quality.
6.1 Impacts of Gasoline Production Associated with Land Use Change
Oil and gas can be extracted using conventional or unconventional (i.e., hydraulic fracturing)
methods, with some resultant variability in associated land use change impacts. Both
methods require the construction and maintenance of a well pad and placement of pumping
machinery. To install any onshore well pad, the land must be cleared and leveled, which
requires the construction of access roads in most cases. A water well to provide water to the
site and a reserve pit for cuttings and used drilling mud may also be necessary. Once this
infrastructure is in place, the oil rig can be assembled on site. Diesel engines and electrical
generators provide the power for the rig. Once the oil has been reached, for a conventional
well, a pump is installed and much of the rig and other machinery can be removed and some
altered areas can be restored. However, the pad area and some access roads and pipelines
must remain throughout the life of the well. A typical lease area has many different oil wells
and pads that are connected by roads and utilities which fragment the surrounding habitat.
In Texas, well pad density may be over 55 pads per square mile (Hibbitts et al. 2013). The
typical lifespan of an oil or gas well is 20-30 years, though this varies due to geology and the
amount and type of oil present (Encana Natural Gas 2011). Once the well and pad have
reached the end of their life, they may be removed, and the area can be restored. However,
restoration does not eliminate the environmental damage the well caused; research has
shown that local biodiversity loss can have cascading effects on ecosystem productivity and
function (Butt et al. 2013).
In the United States, the land use change caused by wells is considerable due to the high
numbers of wells in many locations (Figure 12).
38 Ramboll
Figure 12: Oil and gas field in Wyoming; Areas with Suitable Resources for Future
Extraction.
In 2017, there were 990,677 onshore and offshore oil wells in the US, down from 1,038,698
in 2014 (U.S. EIA 2018a). The average size of an onshore unconventional well pad is 3.5
acres (Helmholtz Centre Potsdam GFZ German Research Centre for Geosciences n.d.), while
an onshore conventional well pad in Texas is about the same, or roughly 3.4 acres (Young et
al. 2018a). When only the direct footprint of onshore domestic wells is considered, the US
had over 1,429,999 acres of well pad infrastructure in 2011 (Trainor et al. 2016). Trainor et
al. 2016 predicted that by the year 2040, the direct footprint of oil and gas land use could
increase to 15,891,100 acres. The actual landscape impacts are almost double the footprint,
due to the spacing requirements of wells (Trainor et al. 2016). Thus, the full landscape
impact of oil and gas estimated for 2040 is roughly 31,782,200 acres. The large landscape
effects of oil and gas have implications for environmental effects.
Conventional and unconventional wells require roads and other impermeable infrastructure
that result in highly altered landscapes (Jones et al. 2015, Garman 2018). The land use
change to altered landscapes has direct effects on habitats and wildlife (Butt et al. 2013,
Garman 2018, Young et al. 2018b). Land use change for well construction increases habitat
fragmentation, pollution, noise and visual disturbance, and causes local habitat destruction;
all of which can decrease biodiversity (Butt et al. 2013, Garman 2018, Young et al. 2018b).
Some of these disturbances, such as fragmentation, are not unique to oil and gas extraction,
and research on their effects is explained in other literature (Brittingham et al. 2014). For
example, it is well known that fragmentation can split breeding populations and reduce
genetic variability within each population, potentially making them less adaptable to other
disturbances (Keller and Largiadèr 2003, Langlois et al. 2017).
Wildlife populations have been shown to decrease near areas with oil and gas production due
to habitat fragmentation, density of wells, human activity, noise and light pollution,
avoidance, and other factors (Jones et al. 2015). For example, habitat fragmentation by well
pads reduced the use of preferred habitats of lizards in Texas, which is likely to decrease the
populations of habitat specialist species (Hibbitts et al. 2013). Density of well pads has been
Both panels reproduced from Jones et al. 2015. a) Oil and
gas field in Wyoming; Photo by David Stubs. b) Current oil
and gas wells (dots) and areas with suitable resources for
future extraction (shading).
39 Ramboll
shown to decrease the population size of several species of songbirds in Wyoming (Gilbert
and Chalfoun 2011). Greater sage-grouse (Centrocercus urophasianus) in Montana and
Wyoming were found to avoid sagebrush habitats that would otherwise be high quality when
those areas are near natural gas development (Doherty et al. 2008). Threatened woodland
caribou (Rangifer tarandus caribou) avoid areas within 1000 m of oil and gas wells and 250
m of roads in northern Alberta, Canada, especially during calving season (Dyer et al. 2001).
This avoidance reduces available habitat and can decrease caribou population size (Hervieux
et al. 2005). Direct mortality from contact with infrastructure is also a problem; an average
of 8.4 birds die in each uncovered reserve pit each year (Trail 2006), thousands more birds
die due to gas flare stack emissions (Bjorge 1987), and many more may die due to the gas
flare stacks and gas compressors on well sites (Jones et al. 2015).
Development of areas for oil and gas production causes secondary land use conversion as
more people move into the production area. If the well is in a remote area the increase in
population size can cause other cascading negative effects such as illegal hunting and the
increase in introduction of exotic species of plants and animals. Both direct and cascading
environmental impacts can be especially harmful in delicate ecosystems, such as the Prairie
Pothole Region (Gleason and Tangen 2014).
The United States is composed of many different habitats that energy development affect
(McDonald et al. 2009), as shown in Figure 13. When comparing Figure 12 and Figure 13,
it is clear that oil and gas resources and well locations fall into many habitat categories,
although temperate grassland and temperate forest may be the most highly affected.
40 Ramboll
Figure 13: Major Habitat Types in the United States.
Figure reproduced from McDonald et al. 2009
6.2 Water Quality Impacts Associated with Spills
6.2.1 Unconventional Oil and Gas (UOG)
The most common UOG production method in the U.S. is hydraulic fracturing. A study of
UOG wells sites in Colorado, New Mexico, North Dakota, and Pennsylvania estimated 55
spills per 1,000 well-years (where a well-year is a unit denoting the operation of one well for
a period of one year; Patterson, Konschnik, et al., 2017). Actual spill rate varied by state,
from about 1% (Colorado) to 12% (North Dakota). Median spill size by state varied from 120
gallons (0.5 m3, Pennsylvania) to 1,302 gallons (4.9 m3, New Mexico). Total spill volume
over ten years (2005 to 2014) was estimated to range from 1,447 m3 (380 thousand
gallons; Pennsylvania) to 33,937 m3 (9 million gallons; North Dakota). The study found that
over 75 percent of UOG production sites spills occur during the first three years of a well’s
life. It also found that wells with one spill have a higher probability of future spills (Patterson
et al. 2017).
Relative to total oilfield spills, the number of spills at UOG production sites is relatively small.
EPA (2015) associates only 1% of spills (457 of 36,000 spills across nine states) with
hydraulic fracturing. Of the 457 spills assessed by EPA (2015), 300 were reported to reach
soil, surface water, or groundwater. The total reported spill volume includes an estimated
540,000 gallons released to soil, 200,000 gallons released to surface water, and 130 gallons
reaching groundwater (EPA 2015). Patterson et al. estimate of 6,648 spills associated with
41 Ramboll
all stages of UOG production covering ten years (2005 to 2014). By contrast, the estimate by
EPA (2015) focuses only on hydraulic fracturing and covered seven years (2006 to 2012).
6.2.2 Conventional Oil and Gas
The movement of raw petroleum and petroleum products consists of a complex distribution
and storage system, which has many chances for accidents, spills, leaks, and losses from
volatilization. Consistent national statistics are lacking for many stages in the overall oil
distribution and storage system. (ATSDR 1999). Statistics from the American Petroleum
Institute (API) based on U.S. Coast Guard data exist for U.S. Navigable waters, but these are
limited primarily to coastal areas and large rivers but can include lakes and estuaries.
Data were readily available for the period 1997-2006 from API (API 2009) and are presented
for illustration purposes. API reported approximately 10.8 million gallons of oil was spilled
into U.S. Navigable Waters from 1997-2006. This includes spills by vessels and facilities
(onshore and offshore). The amount spilled per year varied from 466,000 (2005) to 2.7
million (2004). Of the 10.8 million gallons of oil spilled over the period:
• 3.7 million gallons were from onshore facilities;
• Just over 620,000 gallons were from pipelines;
• 226,000 gallons were from offshore facilities;
• 36,000 gallons were from railroads, tank trucks, and passenger cars;
• And most of the remaining spills (5.7 million gallons) were from vessels.
The figures above do not include the Exxon Valdez spill in Alaska in 1989 of 10.8 million
gallons (API 1998 as cited in ATSDR 1999) or the Deepwater Horizon spill in 2010 (which
post-dated the API study) where EPA reports that 4 million barrels (approximately 168
million gallons) spilled during the 87-day period of the incident (EPA n.d.).
6.3 Toxicity and Other Ecological Impacts of Oil and Associated Products
Total petroleum hydrocarbon (TPH) toxicity to ecological receptors depends on the
hydrocarbon composition, exposure pathway, and exposure duration (i.e., acute or chronic).
Additionally, TPH in the form of product (e.g., crude oil) can cause physical and chemical
toxicity. Acute exposure typically occurs following an accidental release, which causes
immediate exposure to high concentrations of petroleum products. Chronic exposures are
typically associated with low-level releases over long periods of time, such as from a leaking
underground storage tanks and groundwater contamination. Acute exposure following a
large oil spill has both physical and chemical impacts and can have immediate ecosystem
impacts. In contrast, chronic low-level releases have more subtle impacts typically related to
chemical toxicity (Interstate Technology & Regulatory Council [ITRC] 2018).
EPA (1999) describes oil toxicity effects on wildlife according to four categories: physical
contact, chemical toxicity, reproductive problems, and destruction of food resources and
habitats. These categories of toxicity are described relative to acute and chronic exposures
below.
6.3.1 Physical Contact
Terrestrial plants, invertebrates, small animals (mammals, amphibians, reptiles) and birds
can become smothered by oil and aquatic organisms can similarly become smothered and
lose their ability to uptake oxygen. When fur or feathers of larger mammals or birds contact
oil, they get matted down, causing the fur and feathers to lose their insulating properties,
placing animals at risk of freezing to death. Additionally, in the case of birds, the complex
42 Ramboll
structure of feathers that allow birds to float or to fly can become damaged, resulting in
drowning for aquatic birds (EPA 1999).
6.3.2 Chemical Toxicity
Toxicity to the central nervous system is the major mechanism of toxicity to ecological
receptors. Early life-stage aquatic invertebrates and fish can also exhibit phototoxicity (ITRC
2018). These and other toxicological effects are summarized below. Chemical toxicity is
typically associated with chronic exposures, however, if petroleum products are present in
high enough concentrations, negative health effects, including mortality can occur from
acute exposure.
Oil vapors may be inhaled by wildlife, which can cause damage to some species’ central
nervous system, liver, and lungs. Animals are also at risk from ingesting oil, which can cause
red blood cell, intestinal tract, liver, and kidney damage. Skin and eye irritation can also
occur from direct contact with oil (EPA 1999). Fish that are exposed to oil may suffer from
changes in heart and respiratory rate, enlarged livers, reduced growth, fin erosion, a variety
of biochemical and cellular changes, and reproductive and behavioral responses. Chronic
exposure to some chemicals found in oil may cause genetic abnormalities or cancer in
sensitive species (EPA 1999).
6.3.3 Reproductive Effects
Oil can be transferred from birds’ plumage to the eggs they are hatching. Oil can smother
eggs by sealing pores in the eggs and preventing gas exchange. Also, the number of
breeding animals and the number of nesting habitats can be reduced by a spill.
Scientists have observed developmental effects in bird embryos that were exposed to oil.
Long-term reproductive problems have also been shown in some studies in animals that
have been exposed to oil (EPA 1999).
6.3.4 Destruction of Food Resources and Habitats
Species that do not directly contact oil can be harmed by a spill. Predators may refuse to eat
their prey because oil contamination gives fish and other animals unpleasant tastes and
smells, which can lead to starvation. Alternatively, a local population of prey organisms may
be destroyed, leaving no food resources for predators. Predators that consume contaminated
prey can be exposed to oil through ingestion. This causes bioaccumulation of oil compounds
in the food chain. Depending on the environmental conditions, the spilled oil may linger in
the environment for long periods of time, adding to the detrimental effects. In freshwater
lentic systems, oil that interacts with rocks or sediments can remain in the environment
indefinitely, leading to persistent ecological impacts (EPA 1999)
6.4 Additional Water Quality Impacts Associated with Petroleum Production
Production water and fluids used in conventional and unconventional oil and gas production
are an additional source of potential contaminants and may have negative impacts on the
environment. In the U.S., an estimated 21 billion barrels of produced water is generated
each year (Aqwatec n.d.). Production water can be highly saline (up to 15 times saltier than
seawater) and can contain elevated levels of chemicals and radioactive elements. This water
can kill vegetation and prevent plants from growing in contaminated soil (Miller and Pesaran
1980, Miller et al. 1980, Adams 2011, Pichtel 2016). Hydraulic fracturing fluids contain
numerous chemicals to enhance gas and oil extraction. EPA identified 1,173 chemicals
associated with hydraulic fracturing activities and chronic oral toxicity values are available for
147 of the chemicals identified (Yost et al. 2016). The potential for toxicity to wildlife and
ecosystems depends on the quality of the production water, which varies by production site.
43 Ramboll
6.5 Additional Water Quality and Supply Impacts Associated with Exploration,
Production, and Refining
Water is necessary for both conventional and unconventional oil and gas extraction as well
as refining with unconventional oil and gas exploration and production having the higher
water demand requirements. This makes oil and gas development a competitor for limited
water resources with nearby populations and agriculture, in a time when water rights are
often hotly contested (Strzepek and Boehlert 2010). High source water consumption can
alter stream flows and affect aquatic ecosystem function, including declines in specific fish
species around production sites (Dauwalter 2013, Jones et al. 2015). Additionally, produced
water, especially from unconventional oil and gas development, has high total dissolved
solids and may be contaminated with other chemicals, making it a pollutant that is expensive
and difficult to treat (Gregory et al. 2011, Gleason and Tangen 2014).
There are 135 petroleum refineries in the United States (U.S. Energy Information
Administration [USEIA] 2018b, 2019). Over time, the number of petroleum refineries has
decreased, but the capacity per refinery has increased (ATSDR, 1999; USEIA 2018b). Gross
crude oil inputs to refineries averaged 16.6 million barrels per day in 2017 (USEIA 2018c).
An estimated 2.3% of total refinery output is released to the environment through spills or
leaks (ATSDR 1999).
Petroleum refinery wastewaters are made up of many different chemicals which include oil
and greases, phenols (creosols and xylenols), sulfides, ammonia, suspended solids,
cyanides, nitrogen compounds and heavy metals. Refinery effluents tend to have fewer of
the lighter hydrocarbons than crude oil but more polycyclic aromatic hydrocarbons, which
are generally more toxic and more persistent in the environment (Anderson et al. 1974,
Wake 2005). Aquatic ecosystems around refinery discharges are often found to have low
biodiversity and a low abundance of fauna. Often the impacted area is limited to a specific
distance from the discharge point. This distance varies depending on the site and the
effluent. Studies have estimated the impacted range to be 200 m to 1.6 km from the effluent
site (Petpiroon & Dicks, 1982; Wharfe 1975 as cited in Wake, 2005). Refinery effluent has
also been attributed as the cause of lack in recruitment in some areas, that it may either kill
early life stages of aquatic organisms (e.g., settling larvae) or deter them from settling near
discharges (Wake 2005).
44 Ramboll
7. LIMITATIONS
The conclusions, opinions and recommendations presented herein represent Ramboll’s
professional judgment based upon reasonably available information and are products of and
limited by Ramboll’s assigned and agreed upon scope of work. In preparing this report,
Ramboll relied upon information provided by its client and/or third parties, and also relied
upon certain additional publicly available information. Ramboll, however, did not conduct an
exhaustive search or review/analysis of all potentially relevant information. The conclusions,
opinions and recommendations presented herein, and all other information contained in this
report, necessarily are valid only to the extent that the information reviewed by Ramboll was
accurate and complete. Ramboll reserves the right to revise this report if/when additional
relevant information is brought to its attention. In addition, Ramboll did not consider matters
outside of its limited scope of work. Accordingly, the conclusions, opinions, recommendations
and other information contained herein may not adequately address the needs of all
potential users of this report, and any reliance upon this report by anyone other than Growth
Energy, or use of a nature, or for purposes not within Ramboll’s scope of work is at the sole
risk of the person/entity so relying upon or otherwise using this report. Ramboll makes no
representations or warranties (express or implied) regarding this report beyond those made
expressly to its client, and Ramboll’s liability in relation to this report and its related scope of
work is limited under its client contract.
45 Ramboll
8. REFERENCES
Adams MB. 2011. Land application of hydrofracturing fluids damages a deciduous forest stand in West
Virginia. Journal of Environment Quality 40:1340–1344. DOI:doi:10.2134/jeq2010.0504.
Allen L. 2013. Smart Irrigation Scheduling: Tom Rogers’ Almond Ranch.
Anderson JW, JM Neff, BA Cox, HE Tatem, and GM Hightower. 1974. Characteristics of Dispersions and
Water-Soluble Extracts of Crude and Refined Oils and Their Toxicity to Estuarine Crustaceans and
Fish. Marine Biology:75–88. DOI:10.1016/B978-0-12-718250-6.50020-1.
API. 2009. Oil Spills in U.S. Navigable Waters (1997–2006). Washington D.C.
Aqwatec. (n.d.). About Produced Water (Produced Water 101).
ATSDR. 1999. Toxicological Profile for Total Petroleum Hydrocarbons (TPH). Atlanta, GA.
Barton B, and S Elizabeth Clark. 2014. Water & Climate Risks Facing U.C. Corn Production.
http://www.ourenergypolicy.org/wp-content/uploads/2014/06/ceres-corn.pdf.
Beckman J, A Borchers, and CA Jones. 2013. Agriculture’s Supply and Demand for Energy and Energy
Products. Ssrn. DOI:10.2139/ssrn.2267323.
Biazin B, G Sterk, M Temesgen, A Abdulkedir, and L Stroosnijder. 2012. Rainwater harvesting and
management in rainfed agricultural systems in sub-Saharan Africa – A review. Physics and
Chemistry of the Earth 47–48:139–151. DOI:10.1016/j.pce.2011.08.015.
Bjorge RR. 1987. Bird kill at an oil industry flare stack in northwest Alberta. Canadian field-naturalist.
Ottawa ON 101:346–350.
Brittingham MC, KO Maloney, AM Farag, DD Harper, and ZH Bowen. 2014. Ecological Risks of Shale Oil
and Gas Development to Wildlife, Aquatic Resources and their Habitats. Environmental Science &
Technology 48:11034–11047. DOI:10.1021/es5020482.
Butt N, HL Beyer, JR Bennett, D Biggs, R Maggini, M Mills, AR Renwick, LM Seabrook, and HP
Possingham. 2013. Biodiversity risks from fossil fuel extraction. Science 342:425–426.
DOI:10.1126/science.1237261.
Capehart T, O Liefert, and D Olson. 2018. Feed Outlook Corn Disappearance Down as Quarterly Stocks
Come in High.
Center for Urban Education about Sustainable Agriculture (CUESA). 2014. 10 Ways Farmers Are
Saving Water. CUESA Articles:4–6.
Christianson L. 2016. Reducing water pollution with microbes and wood chips. The Conversation.
CIMIS. 2019. CIMIS Overview. https://cimis.water.ca.gov/.
Claassen R, F Carriazo, JC Cooper, D Hellerstein, and K Ueda. 2011. Grassland to Cropland
Conversion.
Dauwalter DC. 2013. Fish assemblage associations and thresholds with existing and projected oil and
gas development. Fisheries Management and Ecology 20:289–301. DOI:10.1111/fme.12007.
Doherty KE, DE Naugle, BL Walker, and JM Graham. 2008. Greater sage‐grouse winter habitat
selection and energy development. The Journal of Wildlife Management 72:187–195.
Dunn JB, D Merz, KL Copenhaver, and S Mueller. 2017. Measured extent of agricultural expansion
depends on analysis technique. Biofuels, Bioproducts and Biorefining 11:247–257.
DOI:10.1002/bbb.1750.
Dyer SJ, JP O’Neill, SM Wasel, and S Boutin. 2001. Avoidance of industrial development by woodland
caribou. The Journal of Wildlife Management:531–542.
46 Ramboll
Efroymson RA, KL Kline, A Angelsen, PH Verburg, VH Dale, JWA Langeveld, and A McBride. 2016. A
causal analysis framework for land-use change and the potential role of bioenergy policy. Land
Use Policy 59:516–527. DOI:10.1016/j.landusepol.2016.09.009.
Encana Natural Gas. 2011. Life of the Well.
EPA. (n.d.). Deepwater Horizon – BP Gulf of Mexico Oil Spill.
EPA. 1999. Understanding oil spills and oil response. Washington, D.C.
EPA. 2010. Renewable Fuel Standard Program (RFS2) Regulatory Impact Analysis.
EPA. 2015. Review of State and Industry Spill Data: Characterization of Hydaulic Fracturing-Related
Spills. Washington, D.C. DOI:EPA/601/R-14/001.
EPA. 2018a. Biofuels and the Environment: Second Triennial Report to Congress. DOI:EPA/600/R10/183F.
EPA. 2018b. Endangered Species Act No Effect Finding and Determination on Severe Environmental
Harm under the General Waiver Authority for the 2019 Final Rule – EPA-HQ-OAR-2018-0167.
EPA. 2018c. Agricultural Fertilizer.
Fausti SW. 2015. The causes and unintended consequences of a paradigm shift in corn production
practices. Environmental Science and Policy 52:41–50. DOI:10.1016/j.envsci.2015.04.017.
Fernandez-Cornejo J, R Nehring, C Osteen, S Wechsler, A Martin, and A Vialou. 2014. Pesticide use in
U.S. agriculture: 21 selected crops,1960-2008,EIB-124. U.S. Department of Agriculture,
Economic Research Service:80. DOI:10.2139/ssrn.2502986.
Garman SL. 2018. A Simulation Framework for Assessing Physical and Wildlife Impacts of Oil and Gas
Development Scenarios in Southwestern Wyoming. Environmental Modeling and Assessment
23:39–56. DOI:10.1007/s10666-017-9559-1.
Gilbert MM, and AD Chalfoun. 2011. Energy development affects populations of sagebrush songbirds
in Wyoming. The Journal of Wildlife Management 75:816–824.
Gleason RA, and BA Tangen. 2014. Brine Contamination to Aquatic Resources from Oil and Gas
Development in the Williston Basin, United States. Page Scientific Investigations Report 2014–
5017. DOI:10.3133/sir20145017.
Gowing lW, O Mzirai, and H Mahoo. 1999. Performance of Maize under Micro-Catchment Rainwater.
Harvesting in Western Pare Lowlands and Morogoro, Tanzania 2:193–204.
Gregory KB, RD Vidic, and DA Dzombak. 2011. Water management challenges associated with the
production of shale gas by hydraulic fracturing. Elements 7:181–186.
DOI:10.2113/gselements.7.3.181.
Helmholtz Centre Potsdam GFZ German Research Centre for Geosciences. (n.d.). The Basics –
Operations. http://www.shale-gas-information-platform.org/categories/operations/the-basics/.
Hervieux D, R Anderson, S Cotterill, E Dzus, K Endresen, G Mercer, L Morgantini, J Nolan, F
Schmiegelow, AE Glen Semenchuk, Federation of Alberta Naturalists Stephen Stuckless, Devon
Resources Canada Inc. Rob Staniland, Talisman Energy Inc. Kevin Williams, and NOTE. 2005.
Alberta Woodland Caribou Recovery Plan 2004/05 – 2013/14. Alberta. DOI:DM-04-2001-27-2-
1262-3636-101019-ART14.
Hibbitts TJ, WA Ryberg, CS Adams, AM Fields, D Lay, and ME Young. 2013. Microhabitat selection by a
habitat specialist and a generalist in both fragmented and unfragmented landscapes.
Herpetological Conservation and Biology 8:104–113.
Hill J, A Goodkind, C Tessum, S Thakrar, D Tilman, S Polasky, T Smith, N Hunt, K Mullins, M Clark,
and J Marshall. 2019. Air-quality-related health damages of maize. Nature Sustainability 2:397–
403. DOI:10.1038/s41893-019-0261-y.
47 Ramboll
Interstate Technology & Regulatory Council (ITRC). 2018. Section 7 Ecological Risk Assessment:
https://tphrisk-1.itrcweb.org/7-ecological-risk-assessment/.
Iowa Corn. (n.d.). Conservation Practices.
Irwin S, and D Good. 2013. Understanding the Pricing of Distillers’ Grain Solubles 3.
Johnston CA. 2013. Wetland Losses Due to Row Crop Expansion in the Dakota Prairie Pothole Region.
Wetlands 33:175–182. DOI:10.1007/s13157-012-0365-x.
Jones NF, L Pejchar, and JM Kiesecker. 2015. The energy footprint: How oil, natural gas, and wind
energy affect land for biodiversity and the flow of ecosystem services. BioScience 65:290–301.
DOI:10.1093/biosci/biu224.
Keeney D, and M Muller. 2006. Water use by ethanol plants: Potential challenges. Institute for
Agriculture and Trade Policy Minneapolis, MN, Minneapolis, Minnesota.
Keller I, and CR Largiadèr. 2003. Recent habitat fragmentation caused by major roads leads to
reduction of gene flow and loss of genetic variability in ground beetles. Proceedings of the Royal
Society B: Biological Sciences 270:417–423. DOI:10.1098/rspb.2002.2247.
LaBeau MB, DM Robertson, AS Mayer, BC Pijanowski, and DA Saad. 2014. Effects of future urban and
biofuel crop expansions on the riverine export of phosphorus to the Laurentian Great Lakes.
Ecological Modelling 277:27–37. DOI:10.1016/j.ecolmodel.2014.01.016.
Lal, R. (Ed.), Stewart B (Ed. ). 2018. Soil Nitrogen Uses and Environmental Impacts. CRC Press.
Lamm FR, and TP Trooien. 2003. Subsurface drip irrigation for corn production: A review of 10 years
of research in Kansas. Irrigation Science 22:195–200. DOI:10.1007/s00271-003-0085-3.
Langlois LA, PJ Drohan, and MC Brittingham. 2017. Linear infrastructure drives habitat conversion and
forest fragmentation associated with Marcellus shale gas development in a forested landscape.
Journal of Environmental Management 197:167–176. DOI:10.1016/J.JENVMAN.2017.03.045.
Lark TJ, NP Hendricks, N Pates, A Smith, SA Spawn, M Bougie, E Booth, CJ Kucharik, and HK Gibbs.
2019. Impacts of the Renewable Fuel Standard on America’s Land and Water. Washington, D.C.
Lark TJ, JM Salmon, and HK Gibbs. 2015. Cropland expansion outpaces agricultural and biofuel
policies in the United States. Environmental Research Letters 10. DOI:10.1088/1748-
9326/10/4/044003.
Li Y, R Miao, and M Khanna. 2018. Effects of Ethanol Plant Proximity and Crop Prices on Land-Use
Change in the United States. American Journal of Agricultural Economics 101:467–491.
Liu T, JR Bruins, and TM Heberling. 2018. Factors Influencing Farmers’ Adoption of Best Management
Practices: A Review and Synthesis. DOI:10.3390/su10020432.
Macrotrends. (n.d.). Macrotrends. www.macrotrends.net. Accessed June 25 2019.
McDonald RI, J Fargione, J Kiesecker, WM Miller, and J Powell. 2009. Energy sprawl or energy
efficiency: Climate policy impacts on natural habitat for the United States of America. PLoS ONE
4. DOI:10.1371/journal.pone.0006802.
Mcfadden J, D Smith, S Wechsler, and S Wallander. 2019. Development, Adoption, and Management
of Drought-Tolerant Corn in the United States, EIB-204. United States Department of Agriculture.
Miao R, DA Hennessy, and H Feng. 2016. The effects of crop insurance subsidies and sodsaver on
land-use change. Journal of Agricultural and Resource Economics 41:247–265.
Michalak AM, EJ Anderson, D Beletsky, S Boland, NS Bosch, TB Bridgeman, JD Chaffin, K Cho, R
Confesor, I Daloglu, J V. DePinto, MA Evans, GL Fahnenstiel, L He, JC Ho, L Jenkins, TH
Johengen, KC Kuo, E LaPorte, X Liu, MR McWilliams, MR Moore, DJ Posselt, RP Richards, D
Scavia, AL Steiner, E Verhamme, DM Wright, and MA Zagorski. 2013. Record-setting algal bloom
in Lake Erie caused by agricultural and meteorological trends consistent with expected future
48 Ramboll
conditions. Proceedings of the National Academy of Sciences 110:6448–6452.
DOI:10.1073/pnas.1216006110.
Miller RW, S Honarvar, and B Hunsaker. 1980. Effects of Drilling Fluids on Soils and Plants: I.
Individual Fluid Components. Journal of Environment Quality 9:547–552.
DOI:10.2134/jeq1980.00472425000900040003x.
Miller RW, and P Pesaran. 1980. Effects of Drilling Fluids on Soils and Plants: II. Complete Drilling
Fluid Mixtures. Journal of Environment Quality 9:552–556.
DOI:10.2134/jeq1980.00472425000900040003x.
Mladenoff DJ, R Sahajpal, CP Johnson, and DE Rothstein. 2016. Recent land use change to agriculture
in the U.S. Lake States: Impacts on cellulosic biomass potential and natural lands. PLoS ONE
11:1–20. DOI:10.1371/journal.pone.0148566.
Morefield PE, SD Leduc, CM Clark, and R Iovanna. 2016. Grasslands, wetlands, and agriculture: The
fate of land expiring from the Conservation Reserve Program in the Midwestern United States.
Environmental Research Letters 11. DOI:10.1088/1748-9326/11/9/094005.
Motamed M, L McPhail, and R Williams. 2016. Corn area response to local ethanol markets in the
United States: A grid cell level analysis. American Journal of Agricultural Economics 98:726–743.
DOI:10.1093/ajae/aav095.
Mueller S. 2016. Request for Correction of Information. Chicago, IL.
Mumm RH, PD Goldsmith, KD Rausch, and HH Stein. 2014. Land usage attributed to corn ethanol
production in the United States: Sensitivity to technological advances in corn grain yield, ethanol
conversion, and co-product utilization. Biotechnology for Biofuels 7:1–17. DOI:10.1186/1754-
6834-7-61.
National Academies. 2019. Science Breakthroughs to Advance Food and Agricultural Research by
2030. The National Academies Press, Washington, DC. DOI:10.17226/25059.
National Corn Growers Association. 2019. Corn Displaced by DDG/CGF in Domestic Livestock Rations.
http://www.worldofcorn.com/#corn-displaced-by-ddg-cgf-domestic-livestock.
National Research Council. 2008. Water Implications of Biofuels Production in the United States. The
National Academies Press, Washington, DC. DOI:10.17226/12039.
Netafim. (n.d.). Drip Irrigation for Corn. https://www.netafimusa.com/agriculture/cropapplications/corn/.
NOAA. 2000. The Causes of Hypoxia in the Northern Gulf of Mexico.
NOAA. 2019. NOAA forecasts very large “dead zone” for Gulf of Mexico. https://www.noaa.gov/mediarelease/noaa-forecasts-very-large-dead-zone-for-gulf-of-mexico.
Parish ES, KL Kline, VH Dale, RA Efroymson, AC McBride, TL Johnson, MR Hilliard, and JM Bielicki.
2013. Comparing Scales of Environmental Effects from Gasoline and Ethanol Production.
Environmental Management 51:307–338. DOI:10.1007/s00267-012-9983-6.
Patterson LA, KE Konschnik, H Wiseman, J Fargione, KO Maloney, J Kiesecker, JP Nicot, S BaruchMordo, S Entrekin, A Trainor, and JE Saiers. 2017. Unconventional Oil and Gas Spills: Risks,
Mitigation Priorities, and State Reporting Requirements. Environmental Science and Technology
51:2563–2573. DOI:10.1021/acs.est.6b05749.
Petpiroon S, and B Dicks. 1982. Environmental effects (1969 to 1981) of a refinery effluent discharged
into Littlewick Bay, Milford Haven. Pembroke, Wales, U.K.
Pichtel J. 2016. Oil and Gas Production Wastewater : Soil Contamination and. Applied and
Environmental Science:24.
Plourde JD, BC Pijanowski, and BK Pekin. 2013. Evidence for increased monoculture cropping in the
49 Ramboll
Central United States. Agriculture, Ecosystems and Environment 165:50–59.
DOI:10.1016/j.agee.2012.11.011.
Qin W, C Hu, and O Oenema. 2015. Soil mulching significantly enhances yields and water and nitrogen
use efficiencies of maize and wheat: a meta-analysis. Scientific reports 5:16210.
DOI:10.1038/srep16210.
Reitsma KD, DE Clay, SA Clay, BH Dunn, and C Reese. 2016. Does the U.S. cropland data layer
provide an accurate benchmark for land-use change estimates? Agronomy Journal 108:266–272.
DOI:10.2134/agronj2015.0288.
Ren J, JB Campbell, and Y Shao. 2016. Spatial and temporal dimensions of agricultural land use
changes, 2001–2012, East-Central Iowa. Agricultural Systems 148:149–158.
DOI:10.1016/j.agsy.2016.07.007.
Rose R. 2002. Slow-release fertilizers 101:304–308.
Schnepf R, and BD Yacobucci. 2013. Renewable Fuel Standard (RFS): Overview and Issues.
Congressional Research Service. Washington, D.C.
Shangguan ZP, MA Shao, TW Lei, and TL Fan. 2002. Runoff water management technologies for
dryland agriculture on the Loess. International Journal of Sustainable Development and World
Ecology 9:341–350. DOI:10.1080/13504500209470129.
Sorensen AA, J Freedgood, J Dempsey, and DM Theobald. 2018. Farms Under Threat:The State of
America’s Farmland. American Farmland Trust, Washington, D.C.
Stein M. 2018. Corn ethanol production has minimal effect on cropland use, study shows | College of
Agricultural, Consumer and Environmental Sciences :: University of Illinois. Illinois Aces.
Strzepek K, and B Boehlert. 2010. Competition for water for the food system. Philosophical
Transactions of the Royal Society B: Biological Sciences 365:2927–2940.
DOI:10.1098/rstb.2010.0152.
Tessum CW, JD Hill, and JD Marshall. 2017. InMAP: A model for air pollution interventions. PLOS ONE
12:e0176131.
Trail PW. 2006. Avian mortality at oil pits in the United States: A review of the problem and efforts for
its solution. Environmental Management 38:532–544. DOI:10.1007/s00267-005-0201-7.
Trainor AM, RI McDonald, and J Fargione. 2016. Energy sprawl is the largest driver of land use change
in United States. PLoS ONE 11:1–16. DOI:10.1371/journal.pone.0162269.
U.S. EIA. 2018a. The Distribution of U.S. Oil and Natural Gas Wells by Production Rate. Washington,
D.C.
U.S. EIA. 2018b. U.S. refinery capacity virtually unchanged between 2017 and 2018.
https://www.eia.gov/todayinenergy/detail.php?id=36633.
U.S. EIA. 2019. Natural Gas Annual Respondent Query System: 757 Processing Capacity 2017.
https://www.eia.gov/naturalgas/ngqs/#?report=RP9&year1=2017&year2=2017&company=Nam
e.
University of Nebraska. 2018. Nebraska Irrigated and Rainfed Corn Yield and Acreage Trends.
https://cropwatch.unl.edu/corn/yieldtrends.
US EIA. 2018. U.S. Exports of Fuel Ethanol, Monthly.
USDA-ERS. 2016. USDA Agricultural Projections to 2025.
USDA-ERS. 2018a. Irrigation & Water Use.
USDA-ERS. 2018b. Agricultural Baseline Database. https://www.ers.usda.gov/dataproducts/agricultural-baseline-database/.
50 Ramboll
USDA-ERS. 2018c. Fertilizer Use and Price.
USDA-ERS. 2019a. Corn & Other Feedgrains.
USDA-ERS. 2019b. Alternative Fuels Data Center. www.afdc.energy.gov/data.
USDA-NASS. 2013. Farm and Ranch Irrigation Survey.
https://www.nass.usda.gov/Publications/AgCensus/2012/Online_Resources/Farm_and_Ranch_Irr
igation_Survey/.
USDA-NASS. 2017. Quick Stats. https://quickstats.nass.usda.gov/#965E2F5A-65F7-3904-9D95-
445421583137.
USDA-NASS. 2019. Crop Production 2018 Summary.
USDA-NIFA. (n.d.). Corn Breeding: Lessons From the Past.
USDA-NRCS. 2006. Conservation Practices that Save: Precision Agriculture.
USDA. 2012. Common Land Unit.
USDA. 2019. USDA Historical Track Record of Crop Production.
USGS. (n.d.). Nutrient Loading for the Mississippi River and Subbasins.
https://nrtwq.usgs.gov/mississippi_loads/#/.
USGS. 2017. Water-Quality Changes in the Nation’s Streams and Rivers.
USGS. 2019. Very large dead zone forecast for the Gulf of Mexico.
Vuran MC, A Salam, R Wong, and S Irmak. 2018. Internet of underground things: Sensing and
communications on the field for precision agriculture. Pages 586–591 2018 IEEE 4th World
Forum on Internet of Things (WF-IoT). IEEE.
Wake H. 2005. Oil refineries : a review of their ecological impacts on the aquatic environment 62:131–
140. DOI:10.1016/j.ecss.2004.08.013.
Wang T, M Luri, L Janssen, DA Hennessy, H Feng, MC Wimberly, and G Arora. 2017. Determinants of
Motives for Land Use Decisions at the Margins of the Corn Belt. Ecological Economics 134:227–
237. DOI:10.1016/j.ecolecon.2016.12.006.
Wright CK, B Larson, TJ Lark, and HK Gibbs. 2017. Recent grassland losses are concentrated around
U.S. ethanol refineries. Environmental Research Letters 12. DOI:10.1088/1748-9326/aa6446.
Wright CK, and MC Wimberly. 2013. Recent land use change in the Western Corn Belt threatens
grasslands and wetlands. Proceedings of the National Academy of Sciences of the United States
of America 110:4134–9. DOI:10.1073/pnas.1215404110.
Wu M, and Y Chiu. 2011. Consumptive Water Use in the Production of Ethanol and Petroleum Gasoline
2011 update. Energy Systems Division:100.
Xue Q, TH Marek, W Xu, and J Bell. 2017. Irrigated Corn Production and Management in the Texas
High Plains 2000:31–41. DOI:10.1111/j.1936-704X.2017.03258.x.
Yost EE, J Stanek, RS Dewoskin, and LD Burgoon. 2016. Overview of Chronic Oral Toxicity Values for
Chemicals Present in Hydraulic Fracturing Fluids, Flowback, and Produced Waters. Environmental
Science and Technology 50:4788–4797. DOI:10.1021/acs.est.5b04645.
Young ME, WA Ryberg, LA Fitzgerald, and TJ Hibbitts. 2018a. Fragmentation Alters Home Range and
Movement of the Dunes Sagebrush Lizard. DOI:10.1139/cjz-2017-0048.
Young ME, WA Ryberg, LA Fitzgerald, and TJ Hibbitts. 2018b. Fragmentation Alters Home Range and
Movement of the Dunes Sagebrush Lizard. DOI:10.1139/cjz-2017-0048.
EXHIBIT 2
RAMBOLL. NOVEMBER 29, 2019. MEMORANDUM: SUPPLEMENTAL
ANALYSIS REGARDING ALLEGATIONS OF POTENTIAL IMPACTS OF THE RFS
ON SPECIES LISTED UNDER THE ENDANGERED SPECIES ACT. PREPARED
FOR GROWTH ENERGY. RAMBOLL, SEATTLE WA.
Growth Energy
ESA Comments – Attachment B
Docket # EPA-HQ-OAR-2019-0136
Supplemental Notice of Proposed Rulemaking; Renewable Fuel Standards Program:
Standards for 2020 and Biomass-Based Diesel Volume for 2021, and Response to the
Remand of the 2016 Standards
November 29, 2019
1/38
MEMORANDUM
SUPPLEMENTAL ANALYSIS REGARDING ALLEGATIONS OF POTENTIAL
IMPACTS OF THE RFS ON SPECIES LISTED UNDER THE ENDANGERED
SPECIES ACT
Prepared for Growth Energy
Date 11/29/2019
OBJECTIVES AND SCOPE
This memorandum supplements the analysis in our August 2019 report, “The RFS and Ethanol
Production: Lack of Proven Impacts to Land and Water” (“Ramboll Report”), in which we analyzed
potential environmental impacts of the RFS program and concluded that there are no proven adverse
impacts to land and water associated with increased corn ethanol production under the RFS. The
impetus for this supplemental memorandum is a recent D.C. Circuit opinion on a petition for review of
EPA’s final rule setting the renewable fuel standards for 2018 (the “2018 RVO Rule”). Am. Fuel &
Petrochemical Mfrs. v. EPA, No. 17-1258 (D.C. Cir. Sept. 6, 2019). The Court remanded the rule back
to the agency to further consider petitioners’ claims that EPA failed to comply with the Endangered
Species Act (ESA). Specifically, the Court directed that under ESA Section 7, EPA must make an
appropriate determination as to whether the 2018 RVO Rule “may affect” a listed species or critical
habitat.
We are aware that the ESA Section 7 consultation issue is relevant not only to the remand in the above
case, but also to future EPA rulemakings with respect to the Renewable Fuel Standard Program (RFS),
including EPA’s proposed rule setting the renewable fuel standards for 2020 (the “2020 RVO
Rule”). Following on our 2019 Report, we are providing this supplemental analysis to explore further
whether there is any evidentiary basis in the record for EPA to conclude that the RFS program “may
affect” a listed species or critical habitat. This memorandum focuses on the technical aspects of the
record relied upon by the Court that were supplied by petitioners’ exhibits, including:
• Declaration of Dr. Tyler Lark (July 27, 2018; referred to herein as the Lark Declaration)
• U.S. Environmental Protection Agency, Biofuels and the Environment: Second Triennial Report to
Congress. Washington, D.C. (June 29, 2018)
• Declaration of C. Elaine Giessel (July 27, 2018)
• Declaration of Aaron Viles (July 20, 2018)
• Declaration of William A. Fontenot (July 24, 2018)
• Declaration of Katherine M. Slama (July 26, 2018)
• Declaration of Andrew E. Whitehurst (July 26, 2018).
Problem Understanding
The allegations of potential impacts to listed terrestrial species that are presented in the Lark
Declaration (and referenced in the Court opinion) center on an assumed relationship between the RFS
and habitat loss or degradation due to presumed land conversion to grow biofuel feedstock. The Lark
Declaration also references potential impacts to aquatic species due to an assumed relationship between
2/38 Ramboll
biofuel feedstock grown for ethanol production and water quality degradation due to use of
agrichemicals (e.g., fertilizers and pesticides) and the potential for increased erosion.
The relationship between the RFS and impacts to land and water, if any, would be effected via a
complex causal chain consisting of the following major relationships:
A. RFS and increased demand for ethanol
B. Increased demand for ethanol and upward pressure of this demand on the price of
ethanol feedstock (mainly corn)
C. Price of corn and an individual farmer’s decision to plant corn
D. Individual farmer’s decision to double crop corn, switch from another crop to corn,
or farm new land
E. Location, type, and magnitude of adverse impact to terrestrial and aquatic habitat
Each of the above relationships, in turn, encompasses several interrelated variables, each variable is
likely to change on an annual basis, and many of the relationships are co-dependent. The Lark
Declaration does not consider these relationships in a meaningful way, and instead relies on
unsupported assumptions and speculation.
There are several lines of evidence indicating that increased demand, if any, for ethanol resulting from
the RFS has not been a discernible driver of land use change. One of the most basic lines of evidence
has to do with the historical trend in the number of acres in the U.S. devoted to growing corn. Historical
data generated by the U.S. Department of Agriculture (USDA) shows that acres planted in corn
nationwide is currently at or below levels reported in 1926 and in the last 2 decades has generally
fluctuated between 80 million acres and 100 million acres (Figure 1).
The amount of land in the U.S. devoted to growing corn has remained at or below historical levels
despite the following trends:
• Total corn production (bushels per year) has increased about 7-fold over the period of record
• Corn produced for ethanol has increased by a factor of 12.5 since 1986 and now accounts for about
50% of corn grown.
This increase in corn production and corn production devoted to ethanol, without an apparent increase
in acres planted, is attributed to a 7-fold increase in corn yield (bushels per acre).
The 7-fold increase in corn production nationwide over the period of record has not been accompanied
by a nationwide increase in the acres of corn planted. This lack of association in itself calls into question
whether there is a causal link between increased demand for corn grown for ethanol and demand for
increased acreage of corn, which in turn calls into question the causal relationship between increased
demand for corn for ethanol and land conversion. The remainder of the report delves more deeply into
each step in the potential causal chain between the RFS and impacts to species. In particular, causal
steps B, C, and D are discussed in Section 2 below, and causal step E is discussed in Section 3 (for
terrestrial listed species mentioned in the Lark Declaration) and Section 4 (for aquatic listed species
3/38 Ramboll
mentioned in the Lark Declaration). Analysis of the effect of the RFS on increased demand for ethanol
(causal step A above) is outside the scope of this memorandum.
Summary of Findings
Our technical review of the assertions made in the Lark Declaration lead to the following overall
conclusions:
• Assertions that increased corn ethanol production under the RFS has resulted in land use change
and conversion of non-agricultural land to production of biofuel feedstock are unsubstantiated
for several reasons, including the following:
─ Acres planted in corn across the United States has remained close to or below the total
acres planted in the early 1930s despite increases in demand for corn as human food,
animal feed, and biofuels over this nearly 90-year period. This fact by itself calls into
question the relationship between the RFS and land use change.
─ The causal relationship between the RFS and the price of corn is not supported by the
evidence, and therefore, the Lark Declaration’s presumption that increased corn prices
drive land use change are unsubstantiated.
─ The Lark Declaration does not adequately consider the many disincentives to the farmer
of converting non-agricultural land to growing any given crop, and thus assertions in the
Lark Declaration that the RFS and price of corn has resulted in land conversion are also
unsubstantiated.
• Assertions that RFS-driven land use change has resulted in impacts to particular ESA listed
species are without foundation for multiple reasons, including:
─ The Lark Declaration asserts that land use change spurred by the RFS has resulted in
impacts to listed terrestrial species of birds, mammals, and insects.
─ The evidence presented in the Lark Declaration to support the alleged impacts are poorly
researched and the examples used to support many assertions instead actually refute
the assertions.
• Assertions that RFS-driven biofuels agriculture adversely impacts water quality are also
unsubstantiated for multiple reasons, including:
─ The Lark Declaration asserts that biofuels (corn and soy) agriculture has worsened the
Gulf of Mexico dead zone, imperiling Gulf sturgeon, loggerhead turtles, and sperm
whales, yet provides no supporting evidence; no studies are cited that specifically
quantify the effect of corn or soy crops as threatening these species or their habitats.
─ The Lark Declaration also asserts that biofuel (corn and soy) agriculture is associated
with impaired waters pursuant to Section 303(d) of the Clean Water Act but fails to
acknowledge cases in which such designations were made well before the RFS came into
effect. Our independent assessment of specific examples presented in the Lark
Declaration indicates that the allegations of impacts from corn or soy on impaired water
bodies is unsubstantiated.
In sum, there are least two important causal chains that must be quantified and linked together to
demonstrate a relationship between increased corn ethanol production under the RFS and impacts to
ESA-listed species: 1) a causal chain linking the RFS to land use change and water quality impacts; and
2) a causal chain linking impacts to land and water with specific impacts on the survival or reproduction
of ESA-listed species. Each of these causal chains is made up of many embedded biophysical and
economic relationships that, in turn, are influenced by a myriad of interrelated variables. The Lark
Declaration fails to consider these causal relationships in a meaningful way, relying instead on
unfounded assumptions and speculation to support its thesis.
4/38 Ramboll
1. Examination of the Causal Link Between the RFS and Impacts to Listed Species
1.1 Overview of Causal Analysis
Causal analysis is a method that is used to determine root causes for observed outcomes. It is used in
many fields such as medicine, business management, economics, ecology, and has been used to explore
the causes of land use change (Efroymson et al. 2016). The point of causal analysis is to look behind
outcomes or symptoms to determine the actual cause, instead of assuming the most obvious cause is
the root of the issue. For example, if a patient presents to a doctor with knee pain because they hurt
themselves gardening, the doctor may simply give pain medication. If the doctor looks deeper using a
more holistic causal analysis approach, the doctor may find that the patient is out of shape or that they
have arthritis. If the symptom is treated without fully understanding the root cause of the problem, the
problem will not be solved in the long term.
Causal analysis begins with creating a causal diagram that includes all causal components of an
outcome. In the next section, we use a causal diagram to examine how farmers make decisions about
crop species planted and land expansion.
1.2 The RFS/Land Conversion Causal Chain
The relationship between the RFS and the potential for land conversion is addressed in the Ramboll
report, primarily in Section 3.2. The decision to alter land from non-agricultural uses to agriculture in
general is made at the farm level and is influenced by a myriad of factors including predicted weather
conditions, crop output and input prices, innovations in cropping equipment, crop insurance, disaster
assistance, and marketing loans. The Ramboll report cites three publications in particular which address
the complexity of the causal relationship between increased production of corn ethanol and land use
change (Section 3.2 page 16-17). As one example, Efroymson et al. (2016) discusses the use of
formal causal analysis to clarify the relationship between biofuels policy and land use change and
concludes that studies relying on single lines of evidence alone are insufficient for establishing probable
cause. Many such studies are cited by EPA (2018) and indeed, many such studies rely on simple
temporal changes occurring around the time of the enactment of the Energy Independence and Security
Act (ESIA) or simple spatial associations (e.g., land use change proximity to ethanol plants) in an
attempt to link land use change and increasing corn production.
The assertion that the EISA increased the expected market price of corn and directly caused land use
change is not supported by a causal analysis. Figure 2 illustrates a simplified causal diagram including
the many components that influence planting decisions by farmers. It is clear from this diagram that the
expected market price of any given crop is not the only relevant factor in planting decisions. Farmers
often have limited freedom to change crop types or expand their farmed areas. For example, planting,
cultivating, and harvesting machinery is not interchangeable between all crop types. Farmers may only
be able to choose between two or three crops that their current machinery is capable of handling.
Additionally, farmers are locked into crop rotation schedules to maintain soil conditions and crop health.
Furthermore, all fields cannot be harvested at the same time due to limited machinery, so crops with
different harvest times must be planted to ensure high-quality output. If farmers are participating in
government subsidy or incentive programs, they may be limited in the types of crops they can plant.
Areas in the conservation reserve program cannot be planted until the term of the contract expires, and
water use restrictions, or limitations of irrigation machinery, can limit expansion of field size.
For farms, even if the species of crop or the expansion of field size were not restricted as described
above, market forces themselves affect planting decisions. Deciding what to plant is a gamble. Farmers
must consider many factors, including their own costs, resources, and market price estimates.
5/38 Ramboll
Successful farmers must bring in enough profit for both salaries and capital costs; meaning that costs
must be well below profits. Besides the obvious costs of fertilizer, water, labor, and machinery, the price
of transportation to get products to market must be considered, as well as costs of insurance given the
location, climate, and predicted weather. The decision to expand farmed areas could be a poor one if the
marginal costs exceed marginal profits. This is especially a concern when expanding farmland into areas
around currently farmed fields, which may be less suitable for farming because of steeper inclines or
poorer quality soil. Additional costs will also be incurred when expanding into natural areas where
drainage of wetlands or removal of trees and other obstructions will be required, which is a disincentive
to increase farmed acreage. These dynamics are explored in more detail in the following section.
2. Lack of Evidence of a Causal Link Between RFS and Land Use Change
A recent report by Lark et al. (2019) is a comprehensive attempt to establish quantitative causal
linkages between the enactment of the RFS and a variety of environmental outcomes using a series of
interlinked models. The fundamental premise of their work is the assumption that the price of corn is
heavily influenced by increased demand for ethanol due to the RFS, yet the authors ignore other
important factors that have a considerable effect on demand and supply conditions (Lark et al. 2019 is
discussed in detail in the Ramboll Report at Section 3.4.) Staab et al. (2017), for example, find that
there are many other contributing factors affecting demand for corn, including market speculation,
stockpiling policies, trade restrictions, macroeconomic shocks to money supplies, currency exchange
rates, and economic growth. As one example, rapid economic growth in developing countries led to
growing food demand and a dietary transition from cereals toward more animal protein and the corn
products used as cattle feed. As a result, global consumption of agricultural commodities has been
growing rapidly. In fact, it appears that most of the increase in corn prices has actually been driven by
higher oil prices (Figure 3). The U.S. Energy Information Administration estimated that of the total cost
per acre of producing corn in 2013 (approximately $350/ac.), nearly two thirds was spent on fuel,
lubricants, electricity and fertilizer1; and fertilizer is known to be closely linked to oil prices2.
Moreover, Lark et al. (2019) and other authors who have attributed land use change to the RFS do not
adequately consider the wide range of factors that influence farmers’ individual planting decisions. These
factors determine the relative prices expected to be faced by farmers. That is, the futures prices of
different crops relative to each other help a farmer determine the crop planting mix (what and how
much). While relative prices may help a farmer determine the potential crop mix farmed on the land,
other supply factors influence the potential costs of production. These include weather, soil quality and
temperature conditions; pests and disease (McConnell 2018); moisture (Queck-Matzie 2019);energy and
fuel costs; interest rates; storage costs; seed and fertilizer costs; and “preventive” planting (Schnitkey
et al. 2019) programs such as COMBO (Crop Insurance), the Cropland Reserve Program (CRP), and
others.
Temporal uncertainty is something that farmers face in all their planting decisions. Farmers need to
decide today what and how much to plant in the next growing season. Farmers are responsive to crop
prices which act as a clearing house to reflect future demand and supply conditions and help alleviate
the uncertainty associated with future conditions. This means that a variety of factors described above
determine planting decisions, and these factors, coupled with the uncertainty of future prices and costs,
weakens the link between the supposed increase in price of corn due to the RFS and planting decisions.
In making crop mix decisions, farmers consider relative futures prices and expected profitability of
plantings (futures price vs. cost to produce; (Kleiber 2009, Staab et al. 2017, Hecht 2019, Springborn
1 https://www.eia.gov/todayinenergy/detail.php?id=18431#
2 https://agmanager.info/sites/default/files/pdf/2019.4.pdf
6/38 Ramboll
2019). Weather is also an important consideration in a farmer’s decision on whether to implement
“prevented” planting (Reiley 2019, Springborn 2019). Futures prices, profitability, and weather forecasts
are factors assessed by a farmer to determine where to plant and how much of each crop to plant
(Kleiber 2009, Reiley 2019, Springborn 2019). Farmers examine, among many other factors, the
relative price ratios of crops to determine an optimal planting mix, and if a farmer decides to increase
production of a certain crop, this can be accomplished by either producing more of the crop on existing
land (intensification) or putting new land into production (extensification, which may result in land use
change). All else being equal, extensification is the least preferred option as it is the option most likely
to involve additional expenditures such as land clearing and other preparation. This option will also be
dependent on the expected yield of new fields, which relative to existing fields, is most likely to be subor infra-marginal and will require more intensive inputs to achieve desired yields (Schiller 2017). Given
these considerations, farmers will typically consider switching crops and increasing yield on existing
acreage (Ling and Bextine 2017) before farming new land. Intensification efforts can include precision
farming as well as traditional techniques regarding plant spacing, pest management, etc. (Queck-Matzie
2019). (The positive environmental effects of precision farming and other technological advances in
agriculture are described at length in the Ramboll report at Section 4).
In summary, studies have shown only a modest effect on corn prices potentially associated with the RFS
(Kleiber 2009, Babcock and Fabiosa 2011, Carter et al. 2018, Renewable Fuels Association 2019). In
addition, factors affecting farmers’ planting decisions include much more than the expected market price
of the crop (Kleiber 2009, Staab et al. 2017). Other important factors include the expected yield of the
crop (Reiley 2019); and a myriad of production costs including the cost of seed, fertilizers and
pesticides, machinery, crop insurance, labor, fuel, and land rental costs (Corn Agronomy 2006, Staab et
al. 2017, Hecht 2019). The decision to expand crops onto new land entails additional hurdles and costs
beyond costs associated with changing crops or intensifying production on existing acreage. For these
reasons and those discussed more extensively in the Ramboll report (at Section 3.4), it is unreasonable
to draw a direct causal connection between the RFS and land use change.
3. Lack of Evidence of a Causal Link Between the RFS and Impacts to Terrestrial
Species
In the absence of a causal link between the RFS and land use change―and in particular land conversion
from grassland, wetland, or forest to corn and soy―there can be no causal link between the RFS and
impacts to terrestrial species due to loss or degradation of habitat. In an attempt to establish a causal
link between the RFS and impacts to terrestrial listed species, the Lark Declaration presented several
examples of quantitative analysis of land conversion from presumably natural land cover to presumably
corn and soy. These examples relied on approaches to land conversion analysis presented by Lark et al.
(2015). Lark et al. (2015) analyzed land use change nationwide during the period 2008-2012 using the
U.S. Department of Agriculture (USDA) Cropland Data Layer (CDL), calibrated with ground-based data
from USDA’s Farm Service Agency (FSA), and further refined using data from the National Land Cover
Database (NLCD). The approach used by Lark et al. (2015) purportedly included methods to “correct”
for known errors and uncertainties in the CDL database. However, the approach used by Lark et al.
(2015) has been shown to be flawed, resulting in a gross overestimate of land use change.
The Ramboll Report (Section 3.3 pages 19 and 20 and Table 1) discusses work by Dunn et al.
(2017) which examined data for 2006-2014 in 20 counties in the prairie potholes region using the CDL,
a modified CDL dataset, data from the National Agricultural Imagery Program, and in-person groundtruthing. Dunn et al. (2017) concluded that analyses relying on CDL returned the largest amount of land
7/38 Ramboll
use change by a wide margin. They further concluded that errors associated with CDL-based analyses
are a major limitation of conclusions drawn from such analyses. In fact, Dunn et al. (2017) concluded
that “the number of hectares in the potential error associated with CDL-derived results is generally
greater than the number of hectares the CDL-based analysis determined had undergone a transition
from grassland, forested land, or wetland to agricultural land”. This suggests that errors in classification
inherent in the CDL can result in uncertainty bounds that are of a larger magnitude than the estimates
themselves (thereby even predicting “negative” land conversion to agriculture within the uncertainty
bounds). Specifically, Dunn et al. (2017) pointed out that the findings reported by Lark et al. (2015)
contradict USDA data indicating that cropland area has remained almost constant during the period
2008-2012.
The Lark Declaration also cited other authors who purport to establish a quantitative link between the
RFS and land use change based on geographic associations (e.g., increased conversion of land to biofuel
feedstock in close proximity to ethanol refineries). The Ramboll report specifically identified the following
key flaws in studies that attempt to quantify land use change to biofuel feedstocks (Section 3.1 pages
14 and 15):
• Like Lark et al. (2015), many other studies of land use change to agriculture depended on unreliable
data sets such as CDL data, lacked ground-truthing, and were regional or state-specific. These
problems preclude extrapolation of results nationwide.
• The literature assessing LUC relative to the RFS generally fails to consider the considerable loss of
agricultural land due to growth in urban areas and the role this loss may have on the pressure to
expand agricultural lands elsewhere.
It is reasonable to presume that the Lark Declaration presented the best examples that could be found
to make the case for the habitat of a particular species having been impacted by land conversion to corn
or soy spurred specifically by the RFS. In the following sections, we analyze and respond to specific
examples presented in the Lark Declaration. In each case we analyzed, we found fatal flaws in the
examples presented in the Lark Declaration. These flaws are either associated with a lack of temporal
association or a lack of geographical association (or both) and a lack of potential causative mechanism.
3.1 Whooping crane (Grus americana)
The Whooping Crane is currently classified as an endangered species. Current places of residence
include Florida, Texas, central Canada, and Wisconsin. Migrating flocks reside in either Texas, Florida,
central Canada, or Wisconsin (Cornell University 2019) primarily in wetlands or muskeg (swampy woods
with lakes). In 1941, the total population had declined to 21 birds. Conservation efforts, including
protection of wintering grounds and educating hunters, has helped increase the population. As of 2019,
more than 350 whooping cranes reside in North America, including 174 migrating cranes (USFWS n.d.).
The population has been increasing over time, with no dip apparent after the RFS in 2008 (Figure 4). In
fact, after 2007, the population of whooping cranes appears to have increased even faster than it did
between 1990 and 2007 (Figure 4)
A total of three known flocks currently exist throughout North America: two migrating flocks and one
non-migrating flock. One migrating flock spends summers in the Wood Buffalo National Park in Canada
and winters in Texas at the Aransas National Wildlife Refuge. The other migrating flock nests in
Wisconsin for the summer and flies south to Florida in the winter. These flocks have been sighted taking
short rests in Kansas at either Cheyenne Bottoms or Quivira National Wildlife Refuge (QNWR) whilst
migrating. A non-migratory flock remains in Florida year-round (USFWS n.d.).
8/38 Ramboll
The migrating flocks reside in national refuges or national parks that have protection plans in place. For
example, the QNWR prohibits hunting when whooping cranes are present to avoid accidental shootings.
The U.S. Fish & Wildlife Service reports that refuges only integrate farming for specific wildlife
conservation efforts.
Whooping Cranes spend time in marshes, shallow bays and tidal flats, with the occasional venture to
nearby farmland. Their diet varies by area but may include fish, mice, insects, berries, seeds, crabs and
snakes. The Whooping Crane’s wide variety of food preferences opens opportunity to scavenge in
several locations, including corn fields (USFWS n.d.).
The Lark Declaration argues that conversion to cropland “adjacent to its critical habitat and wintering
grounds” may negatively impact the livelihood of the Whooping Cranes. Lark does not discuss the
landscape of the adjacent land at issue nor the distance of these adjacent habitats from the whooping
crane’s current nesting grounds. The images Lark refers to in support of this claim are in Appendix 7 to
the Lark Declaration. The image in Appendix 7 includes boundary locations of the critical habitat
(Cheyenne Bottoms and QNWR) briefly visited by the Whooping Cranes (Bloomberg n.d.), as well as the
nearest ethanol refinery. Ramboll further investigated these images and found that they did not support
Lark’s claims.
Multiple areas on the Lark Declaration’s maps that show corn or other crops growing in or near
Cheyenne Bottoms and QNWR were errors in the USDA Cropland Data Layer (CDL). One particularly
egregious error shows corn growing in the southeast corner of Cheyenne Bottoms Pool 2 (Figure 5). It
is clear from aerial imagery in Google Earth going back to 1992 that no corn is growing in Pool 2 (Figure
5). Ramboll further confirmed the lack of corn by contacting staff at the reserve on November 18 and
19 of 2019. Reserve staff confirmed that Pool 2 was usually under water, and although they had
planted a cover crop for the benefit of wildlife in some dry years, the cover crop had never been corn3.
Ramboll investigated multiple years of Google Earth aerial imagery for areas near to Cheyenne Bottoms
and QNWR and also within QNWR that Lark showed as converted to corn; these images failed to show
any new crop cultivation after 2007.
In summary, the data presented in Lark’s declaration does not support his assertion that the RFS
spurred land use change to biofuels in or near Cheyenne Bottoms and the QNWR. To the contrary, it
appears that there is little or no land use change to agriculture near either reserve that supports
whooping crane migrations in Kansas, and that such land use change, if any, has not been attributed to
the RFS. Further, the population of whooping cranes in the United States has risen and continues to rise
since the RFS, suggesting that even if the RFS has resulted in some land conversion in areas potentially
used by the whooping crane, this conversion has not resulted in any discernible adverse impact on
whooping crane populations. Due to the lack of evidence of land use change, any assertion that the
recovery of whooping crane populations would have been more rapid had it not been for the RFS would
be purely speculative4.
3.2 Piping Plover (Charadrius melodus)
There are three distinct populations of piping plover in the U.S.: Great Lakes, Northern Great Plains, and
Atlantic Coast. Piping plover populations on the Great Lakes are listed as endangered, whereas
populations in the Northern Great Plains and Atlantic coast are listed as threatened. Piping Plover
population declines have been attributed to human disturbance, habitat loss and predation. Piping
3 Phone communication between Ramboll and Cheyenne Bottoms Ranger Station (620-793-3066) on November 17th and 19th, 2019
4 This point applies as well to other species discussed below, where the data show recovery of the species during the time frame in
which the RFS program has been implemented.
9/38 Ramboll
plover management strategies are targeted at limiting access to beachgoers and off-road vehicles, pet
restrictions, and public education5.
The Lark Declaration implicates land conversion for crop production (and presumably by extension, land
conversion to corn or soy as a result of the RFS) as a potential impact to piping plover populations,
citing Cohen (2009)6 as documenting “disruption of plover habitat” in the Great Lakes endangered
population. In fact, Cohen et al. (2009) studied two Atlantic Coast piping plover breeding areas in West
Hampton Dunes, a barrier island in New York State. The only mention of land conversion made by the
authors was in reference to urban development. In addition, the authors cite predation management
(domestic cats and fox) as key to the recovery of the populations at the sites they studied. Thus, this
study cited in the Lark Declaration is not relevant to the premise that land conversion spurred by the
RFS results in impacts to piping plover, and in fact specifically points to urban development and
predation as the primary stressors to these populations.
The U.S. FWS Midwest Region fact sheet describes the following threats to Great Lakes piping plover
populations: Coastal beach habitat loss due to commercial, residential, and recreational developments;
and effect of water control structures on nesting habitat; vehicle and pedestrian use of beaches;
harassment or mortality of birds by dogs and cats; and predation by fox, gulls, and crows7. Habitat
protection measures include controlling access to nesting areas, nest monitoring and protection, limiting
residential and industrial development, and properly managing water flow8. Thus, like the Atlantic Coast
populations, land use change due to agriculture is not a recognized threat to the Great Lakes
populations.
In the Northern Great Plains, piping plover breed on river sandbars, along reservoir shorelines, and in
manmade habitat such as commercial sand mines. Similar to the Atlantic Coast and Great Lakes
populations, declines of this population are attributed mainly to harassment of birds and nests by
people, domesticated animals, and vehicles; shoreline habitat loss due to development projects; humaninduced increased predation; and water-level regulation policies that disrupt nesting behavior or destroy
nesting habitat (NRC 2005). Appendix 8 to the Lark Declaration provides an example of conversion of
some riparian forest habitat adjacent to a farm field along the Missouri River sometime between spring
2012 and late winter 2015. This example, however, does not portray any loss of critical habitat for this
species (i.e., critical habitat for the piping plover in the Missouri River is sand bar or sandy shoreline
habitat and not forest), and therefore does not support the premise in the Lark Declaration that land
conversion spurred by the RFS results in impacts to piping plover.
Thus, based on a review of the specific citations relied upon by the Lark Declaration as well as
publications by the U.S. Fish and Wildlife Service (USFWS) regarding endangered and threatened
populations of piping plover, we find no evidence that agriculture in general, or land conversion to corn
and soy due to the RFS in particular, results in impacts to piping plover. Such claims in the Lark
Declaration are unsubstantiated.
It is worth noting that, in addition to discussing land conversion, the Lark Declaration cites a study by
Fannin (1993)9 when suggesting that pesticides or other contaminants from agricultural practices (and
by extension, presumably agriculture for biofuels feedstock spurred by the RFS) could jeopardize piping
plover egg survival. Fannin and Eamoil (1993) collected 16 piping plover addled (unhatched) eggs in
1989 and 3 piping plover addled eggs in 1990 and analyzed the contents for a wide range of metals and
5 https://naturalhistory2.si.edu/smsfp/irlspec/Charad_melodu.htm
6 The Lark Declaration incorrectly cites Cohen et al. (2009)
7 https://www.fws.gov/midwest/endangered/pipingplover/pipingpl.html
8 https://www.fws.gov/midwest/endangered/pipingplover/pipingpl.html
9 We believe that the Lark Declaration incorrectly cites Fannin and Eamoil (1993)
10/38 Ramboll
several organochlorine pesticides, including DDT and its breakdown products. DDT and to a lesser
extent, other organochlorine pesticides are known to cause eggshell thinning and reproductive failure,
principally in raptors and fish-eating birds. DDT was banned from use in the United States in 1972, and
chlordane was banned in 1988. It is also widely known that many species made dramatic recoveries in
the years following the ban of DDT, most notably the bald eagle. Use of these and other organochlorine
pesticides in agriculture were terminated decades prior to the enactment of the EISA and
implementation of the RFS. Thus, any suggestion in the Lark Declaration that the use of pesticides on
biofuel crops may be resulting in eggshell thinning in piping plover lacks foundation.
3.3 Yellow-Billed Cuckoo (Coccyzus americanus)
The Western U.S. Distinct Population Segment of C. americanus (western yellow-billed cuckoo) was
proposed as threatened on October 3, 2013 (FR 79:192, October 3, 2014; USFWS 2014). Within the last
50 years the species’ distribution west of the Rocky Mountains declined substantially mainly due to loss
of streamside habitat. USFWS (2014) reports that current impacts from agricultural activities on yellowbilled cuckoo habitat are mainly associated with livestock overgrazing in riparian areas.
Yellow billed cuckoo breed in dense willow and cottonwood stands in river floodplains. The Lark
Declaration states that their threatened status is due largely to the “destruction of these habitats from
anthropogenic activities, including agriculture,” and presumably by extension, land conversion to biofuel
feedstock (corn and soy). However, the Lark Declaration fails to acknowledge that with the exception of
Glenn County in California, there is no overlap between significant corn or soy growing areas and critical
habitat for the species. This is primarily due to the fact that most corn and soy production in the U.S.
occurs east of the Rocky Mountains.
Figures 6, 7, and 8 show areas reported to be in corn in the USDA CDL database for 2018 within the
boundaries of designated critical habitat for the Western yellow-billed cuckoo in Glenn County, along
with available Google Earth aerial images of these areas. The maps in figure 6 show that with only a
couple of exceptions there is no overlap between Western yellow-billed cuckoo habitat and counties with
corn and soy cultivation. The Google Earth images in figure 7 and 8 clearly show that these areas were
in agricultural production as early as 1998, a decade before the RFS could possibly have influenced land
conversion, and at approximately 55.5 acres, they account for only 0.036% of the total available critical
habitat for the species in California (155,635 acres). Thus, not only is there no overlap between critical
habitat for this species and significant corn growing areas, but in the two instances in California where
the CDL reports corn to be grown in critical habitat, areas were in agricultural production long before the
RFS.
As with the piping plover, the Lark Declaration also suggests that western yellow-billed cuckoo is
adversely impacted by eggshell thinning due to pesticides. For the reasons described above, any
eggshell thinning observed in this species cannot possibly be associated with the RFS and any such
implied association is unsubstantiated.
3.4 Poweshiek Skipperling (Oarisma poweshiek)
The Poweshiek skipperling, was once abundant in remnant native prairie habitat in Indiana, Illinois,
Iowa, Michigan, Minnesota, North Dakota, South Dakota, Wisconsin, and Manitoba, Canada; but is now
thought to be present only in Wisconsin, Michigan, and Manitoba. The USFWS lists several stressors that
may be acting to reduce populations of the butterfly, with loss and degradation of habitat being one of
the initial stressors for its decline. The USFWS states that other stressors are unknown but might
include disease or pesticides.
11/38 Ramboll
The Lark Declaration (paragraph 15, page 12) states “Habitat fragmentation poses a key threat to the
Poweshiek skipperling, and there are several instances where land has recently been converted to
cultivate either corn or soybeans within close proximity to its critical habitat in Minnesota, North Dakota,
and South Dakota”. Paragraph 15 refers to Appendix 6, which we presume to be their best example to
illustrate land conversion due to RFS. Appendix 6 presents a map showing Poweshiek skipperling critical
habitat in Minnesota, the location of an ethanol refinery, and polygons depicting presumed land
conversion from native tall grass prairie to corn or soy. Appendix 6 is based on a comparison of data
from 2008 to 2016 and methods documented in Lark et al. (2015; see above description of shortcoming
of these methods). The second page of Appendix 6 shows two images from Google Earth, one from May
21, 2008 and another from June 23, 2011–presumably showing conversion of two farm fields adjacent
to Poweshiek skipperling critical habitat from grassland to cropland. The refinery depicted in Appendix 6
was confirmed by Ramboll to be the Valero refinery in Aurora, North Dakota; approximately 28 miles
from the illustrative farm fields.
Several facts indicate that the assertions in the Lark Declaration regarding the Poweshiek skipperling are
flawed, and, in fact, land conversion from tall grass prairie to corn or soy due to the RFS could not have
had an impact on this listed species:
• The last confirmed sightings of O. poweshiek in Minnesota were in 2007, despite extensive annual
surveys beginning in 201310. The RFS went into effect in 2008 so could not possibly have had an
adverse effect on this species in Minnesota. Similar trends were seen in other states (Environment
Canada 2011):
− In Iowa, the species was in decline by 2003 and was last observed in 2008
− In North Dakota, the species was thought to be extirpated by 2008; with only 8 individuals seen
in a survey in 2001.
− In South Dakota, The species began to disappear from five South Dakota sites in 2002 and many
of these sites were observed to be idle with no range or grass management. At these sites, the
decline was attributed loss of floral diversity, increase in grasses and forbs, and an increase in
exotic species. The species was last observed at Hartford Beach State Park and the Waubay
National Wildlife Refuge in 2002, Pickerel Lake State Recreation Area in 2004, Wike Waterfowl
Production Area in 2006, and Scarlet Fawn Prairie in 2008. Several sites where they had been
recorded in the past were surveyed in 2010 and no adults were observed.
• The Valero Aurora refinery in North Dakota began operation in 2003 and reached a capacity of 120
million gallons per year (MGY) in 200511, three years before the enactment of EISA. Therefore, any
increased demand for corn in the vicinity of the refinery would have been met prior to any possible
effect of the RFS, and therefore there cannot be a causal relationship between the RFS and land
conversion impacting O. Poweshiek in Minnesota.
• Dana (1997) conducted a survey for the Dakota skipper (Hesperia dacotae) butterfly in several
critical habitat areas of Minnesota, including Hole-in-the-Mountain Prairie Lincoln County, Minnesota
South of the town of Lake Benton (the same area depicted as critical habitat in the Lark Declaration
Appendix 6). The Dakota skipper has very similar habitat requirements as Poweshiek skipperling.
Dana (1997) states that the principal threat to this species in this area is probably the use of
herbicides for weed and brush control in privately owned pastures as well as overgrazing and mowing
10 https://www.dnr.state.mn.us/rsg/profile.html?action=elementDetail&selectedElement=IILEP57010
11 https://www.valero.com/en-us/AboutValero/ethanol-segment/aurora
12/38 Ramboll
by County Park staff, and possibly excavation for construction materials. Dana (1997) specifically
states that conversion of additional prairie to cropland (in general) is at most, a minor threat12.
• Appendix 6 of the Lark Declaration shows satellite images from Google Earth for the years 2008 and
2011 presumably to contrast land use in the year the EISA was enacted and several years after the
RFS went into effect. However, there is no information provided to substantiate the claim that the
highlighted areas were indeed grassland in 2008. In fact, when other satellite images readily
available on Google Earth are examined, it is clear that the subject areas were in agriculture as early
as 1992. Further, upon viewing the Google Earth images available in subsequent years for the
subject areas, there is no evident expansion of cropland since 1992 into what is now designated as
critical habitat for the Poweshiek skipperling (Figure 9).
For the reasons described above, the assertion in the Lark Declaration that land conversion spurred by
the RFS has adversely impacted critical habitat of the Poweshiek skipperling are unsubstantiated.
3.5 Other Insects
The Lark Declaration also mentions the threatened Dakota skipper (Hesperia dacotae), the endangered
rusty patched bumble bee (Bombus affinis), the endangered Hine’s emerald dragonfly (Somatochlora
hineana), and the endangered Salt Creek tiger beetle (Cicindela nevadica lincolniana) as other insect
species that could “potentially be affected by biofuel feedstock production”. In no case, does the Lark
Declaration provide any evidence to support that assertion. The Dakota skipper and rusty patched
bumble bee are both prairie/grassland species. Although there has been habitat loss and fragmentation
to varying degrees across the ranges of these species, there is no evidence presented that habitat loss
occurring after 2008 was directly linked to the presumed RFS-induced land conversion.
As to Hine’s emerald dragonfly, USFWS (2013) states the following:
• The greatest current threat to this species is from invasive plants
• There are effective protections against habitat loss (wetland filling and draining)
• Past habitat loss was due to commercial and industrial development.
USFWS (2013) does not mention any impact related to agriculture. Therefore, the Lark Declaration’s
assertion of impacts to Hine’s emerald dragonfly due to the RFS is unsubstantiated.
With regards to the Salt Creek tiger beetle, the Lark Declaration also provides no evidence or discussion
of the causal relationship between the RFS and impacts to this species. The Salt Creek tiger beetle is
currently found at only three sites in Lancaster County, Nebraska occupying 15 acres in saline wetland
habitat13. The Nebraska Game and Parks Commission states that the biggest threat to the habitat of this
species is stream channel modification13. The USFWS (2013b; page 33284) cites two publications from
2003 and 2005 in its statement that “in the past 150 years, approximately 90 percent of these wetlands
have been degraded or lost due to urbanization, agriculture, and drainage” but does not mention
agriculture as a threat to habitat of this species after the implementation of the RFS in 2008. In fact, the
USFWS (2013; page 33285) shows a graph presenting results of surveys of adult Salt Creek tiger
beetles 1991 to 2012 which indicates a consistent increase in population over the period 2008-2012
with an approximate doubling in numbers over that period (Figure 10). Based on the information
presented above, we find that the Lark Declaration’s assertion of impacts to the Salt Creek tiger beetle
due to the RFS is also unsubstantiated.
12 Note that these observations, including the statement regarding habitat threats, predate the EISA by 11 years.
13 Nebraska Department of Game and Parks
13/38 Ramboll
3.6 Blackfooted Ferret (Mustela nigripes)
The black-footed ferret was listed as endangered across its entire range on March 11, 1967.14 The is no
critical habitat designated for this species. Black-footed ferret population status and distribution is
closely tied to that of prairie dogs. Prairie dogs make up more than 90% of the black-footed ferret’s diet
and prairie dog burrows provide shelter and den habitat for the species. Major threats to black-footed
ferret populations include conversion of native grasslands to agriculture, prairie dog eradication
programs that were once widespread, and disease; and much of the remaining habitat for black-footed
ferret is fragmented due to fragmentation of prairie dog towns by agriculture and human development
(USFWS 2018).
The Lark Declaration states that “Given the connection between the Renewable Fuel Standard and the
conversion of grasslands to agricultural land within the Black-footed ferret’s range, further assessment
seems warranted”, but provides no explanation or evidence to support such a “connection”. The Center
for Biological Diversity (CBD 2019) reports that the last captive black-footed ferret died in 1980, and at
that time, the animals were thought to be extinct in North America. In 1981 the species was rediscovered in a Wyoming prairie dog colony. Between 1991 and 1999, about 1,200 ferrets from that
population were released at sites in Wyoming, Montana, South Dakota, Arizona and along the
Utah/Colorado border (CBD 2019). It is estimated that about 1,410 black-footed ferrets are currently
living in the wild (CBD 2019). Figure 11illustrates the estimated population status of black-footed ferrets
in the wild, including a rapid recovery beginning in about 2000 and extending past 2008, the year the
EISA was enacted and the RFS was implemented. The continual and unabated recovery of black-footed
ferret populations after 2008 also serves to undermine the assertion in the Lark Declaration that the RFS
has had adverse impacts on black-footed ferret.
Figure 12 illustrates the locations of black-footed ferret populations (reintroduced) and acres planted in
corn and soy in 2018. With few exceptions, there is no overlap between counties with some acreage in
corn or soy and locations where black-footed ferret have been introduced. The Lark Declaration presents
no evidence of impacts from land conversion spurred by the RFS, and, in fact, evidence suggests that
impacts due to loss of habitat (for all reasons) occurred long before any potential influence of the RFS
and most of the species recovery has occurred since 2008. Therefore, we conclude that the Lark
declarations’ assertions of a causal relationship between the RFS and impacts to black-footed ferret lack
foundation.
4. Lack of Evidence of a Causal Link between the RFS and Hypoxia in the Gulf of
Mexico or the RFS and Water Quality Impacts in Streams Supporting Listed
Species
4.1 Lack of Evidence of a Causal Relationship Between the RFS and Hypoxia in the Gulf of
Mexico
The alleged link between increased corn production for ethanol and hypoxia in the Gulf of Mexico (and
western Lake Erie) is addressed in Section 4.1 of the Ramboll report. While it is not unexpected that
nutrient loading (including from agriculture in general) to the Gulf of Mexico via the Mississippi River
contributes to the formation of a seasonal hypoxic “dead zone” in the Gulf of Mexico, there is no
information demonstrating a link between increased corn ethanol production under the RFS and specific
and quantifiable causes of observed hypoxic conditions in the northern Gulf of Mexico. The consensus,
based on the vast majority of technical articles we have reviewed is that hypoxia is due to algal
14 https://www.fws.gov/mountain-prairie/es/blackFootedFerret.php
14/38 Ramboll
production driven by excess nitrogen that enters the northern Gulf of Mexico via the Mississippi River
and related watersheds together with certain hydrologic conditions, including vertical stratification and
temperature dynamics within the Gulf of Mexico water column. The hypoxia condition is not new and as
shown by the U.S. Geological Survey and other institutions, has been an ongoing phenomenon for
several decades and well before the RFS was initiated in 2008. The loading of annual nitrate plus nitrite
to the Gulf of Mexico has been relatively consistent since comprehensive monitoring began in
approximately 198015 with the three largest measured annual loading values occurring in 1993, 1983,
and 1984, respectively, and thus well before the RFS was envisioned. Bianchi et al. (2010) conclude that
understanding the complexity of this highly dynamic system or predicting flux and source areas with
high precision is not reliable by simply referring to the numerous mostly general models that are relied
on by recent authors (including Lark).
The Lark Declaration (at page 20), for example, refers to the pre-RFS study by Donner and Kucharik
(2008) that “predicts” an increase in flux of dissolved inorganic nitrogen (DIN) by the Mississippi and
Atchafalaya Rivers of between 10% and 34% using models that rely on hypothetical predictions of land
use scenarios and discharge. Although Donner and Kucharik (2008) discuss the model validation
approach, the validation results are imperfect indicating considerable overestimates in some cases, and
underestimates in others. Furthermore, the model does not appear to provide a precise fit between the
simulated results and the observed DIN discharge numbers collected from the few field stations
identified in the study. An important complication in using a model like this to make predictions is the
differentiation between urban (e.g., septic, industrial and municipal waste water plants, and residential
runoff) and agricultural sources. As noted by Alexander, et al (2007), additional complications also are
that nutrient sources typically are statistically estimated in the models, and then adjusted based on the
model calibration. Model calibration uses “trial and error” processes for simulating numerous
parameters that are themselves influenced by hydrologic and biogeochemical processes, nutrient uptake
by a wide variety of soil types, climatic (short and long-term) conditions, and (as most relevant
currently), improvements in fertilizer application and cropping and drainage patterns. Essentially, by
providing examples of failed predictions using models, Bianchi, et al. (2010) make the case to not rely
solely on numerical models.
Notably, the influence of weather is a very important condition for the formation of the Gulf of Mexico
“dead zone” and is totally independent from loading of DIN from any particular sources. The influence of
weather on the formation of the Gulf of Mexico “dead zone” is discussed in the Ramboll report in
Section 4.1, page 26, where for example, the U.S. Geological Survey16 estimated that flooding in the
spring of 2019 resulted in an increased loading of nitrate and nitrite of approximately 18% when
compared to the long-term average loading to the Gulf of Mexico.
The alleged quantitative relationship between increased corn grown for ethanol and nutrient loading to
the Gulf of Mexico is further called into question by data from the U.S. Geological Survey indicating that
annual nitrate plus nitrite loading to the Gulf of Mexico has remained relatively constant over the period
1980 to 2017 (Figure 13). This indicates that even during the period of increased use of corn for
ethanol, there has been no appreciable net change to nutrient loading to the Gulf of Mexico. For this
reason alone, there is no support for the assertion of a direct relationship between ethanol production
on the hypoxia conditions in the Gulf of Mexico. In addition, EPA (2018) reports that there has actually
been a reduction in total nitrogen concentrations in surface water bodies in Iowa which is the highest
15 https://nrtwq.usgs.gov/mississippi_loads/#/GULF
16 https://www.usgs.gov/news/very-large-dead-zone-forecast-gulf-mexico?qt-news_science_products=4#qt-news_science_products
15/38 Ramboll
corn producing state. This further refutes the broadly stated allegation that there is a link between
expanded corn production (for any reason) and increased nutrient loading to the Gulf of Mexico.
The Lark Declaration mentions the following listed species as potentially impacted by the Gulf of Mexico
dead zone: the threatened Gulf sturgeon (Acipenser oxyrinchus desotoi), the loggerhead turtle (Caretta
caretta listed as endangered and threatened depending on location), and the endangered sperm whale
(Physeter macrocephalus). With regards to Gulf sturgeon, it is instructive to look at the geographical
location of critical habitat for the species and the occurrence of the dead zone in the Gulf of Mexico. The
dead zone forms west of the Mississippi River Delta over the continental shelf (< 200 m water depth) of
Louisiana and sometimes extends to Texas17. Figure 14 depicts Gulf sturgeon critical habitat occurring
exclusively to the east of the Mississippi River delta and the hypoxic zone in 2019 (the largest recorded)
located exclusively to the west of the Mississippi River delta. NOAA’s Gulf of Mexico Hypoxia Watch site
presents results from dissolved oxygen monitoring for the period 2001 to 201918. These results show
that hypoxia rarely extends near critical habitat areas for Gulf sturgeon, and when these conditions
exist, they are limited to a relatively small area offshore of Biloxi, Mississippi. Waters to the east and
south did not exhibit hypoxic conditions in any year monitored.
Moreover, the migratory behavior of Gulf sturgeon minimizes the probability of encountering hypoxic
waters, should they occur in their critical habitat. Oxygen depletion in the Gulf of Mexico increases in
late spring, worsens over the summer, then dissipates in the fall; whereas gulf sturgeon move into
rivers in the spring and fall and spend the summer months in the riverine habitat, then subadults and
adults move into estuarine waters in the fall to feed and then move into marine waters in the winter.
Thus, the Lark Declaration provides no evidence of a relationship between Gulf sturgeon critical habitat
and potential impacts from hypoxia in the Gulf of Mexico due to nutrient inputs from the Mississippi
River basin. In addition, NOAA does not list hypoxia as a threat to the species, rather it lists
contaminants, dredging, dams, and climate changes as the threats19. For these reasons, the
presumption in the Lark Declaration that the RFS has resulted in impacts to Gulf sturgeon is
unsubstantiated.
Loggerhead turtles and sperm whales have pan-global ranges and only a limited number of individuals
over a limited portion of their life spans would be likely to encounter the Gulf of Mexico dead zone.
Because both are air-breathing animals, adverse effects to these species from hypoxia, if any, could
only be indirect (e.g., reduced prey abundance).
The loggerhead turtle is the most common sea turtle in the southeastern U.S., and they nest mainly
along the Atlantic coast of Florida, South Carolina, Georgia, and North Carolina and along the Florida
and Alabama coasts in the Gulf of Mexico20. The Lark declaration states that “The increasing frequency
of red tides and harmful algae blooms in the Gulf of Mexico as well as the increased duration and extent
of the hypoxic dead zone caused by agricultural runoff in the Mississippi River have been reported to
both directly and indirectly affect sea turtles” and cites NMFS et al. (2011) for this proposition. Yet,
NMFS (2011) makes no mention of hypoxia, and red tide is only mentioned in the context of the west
coast of Florida. The Lark Declaration also states that “Loggerheads in the near-shore northern Gulf of
Mexico waters may be exposed to hypoxia…”, citing Hart et al. (2013) for this proposition. However,
Hart et al. (2013) studied nesting sites and movement patterns only along the Alabama and Florida
coasts and reported movement patterns southward along the Florida west coast, away from the Gulf of
Mexico dead zone. As noted above, since 2001, hypoxia in the Gulf of Mexico did not extend to the west
17 https://pubs.usgs.gov/fs/2006/3005/fs-2006-3005.pdf
18 https://www.ncddc.noaa.gov/hypoxia/
19 https://www.fisheries.noaa.gov/species/gulf-sturgeon
20 https://www.fisheries.noaa.gov/species/loggerhead-turtle
16/38 Ramboll
coast of Florida. Therefore, the assertions in the Lark Declaration that the RFS has resulted in impacts to
loggerhead turtles by means of hypoxia in the Gulf of Mexico lack foundation.
Sperm whales inhabit all of the world’s oceans, having one of the widest distributions of any marine
mammal. NOAA does not list hypoxia as a threat to this species, rather vessel strikes, entanglement,
ocean noise, marine debris, climate change, oil spills, and contaminants are listed as threats.21 Several
researchers have investigated the distribution of sperm whales and other cetaceans in the Gulf of
Mexico. Davis et al. (2002) report a resident breeding population within 100 km of the Mississippi delta
and suggest that the edge of the continental slope south of the Mississippi River delta provides the
oceanographic deep-water conditions with locally enhanced primary and secondary productivity. The
Gulf of Mexico dead zone does not extend to the continental slope, rather it is oceanographically limited
to the continental shelf where water depths are less than 200 m.
In sum, attributing adverse impacts to these species to hypoxia induced by nutrient enrichment in the
Mississippi River basin is speculative. Attributing any potential for adverse effects due solely to
theoretical increases in nutrient inputs from expanded corn production spurred by the RFS is
unsupportable.
4.2 Lack of Evidence of a Causal Relationship Between the RFS and Water Quality
Impairment in Streams Supporting Listed Species
Surface water use impairment is determined under Section 303(d) of the Clean Water Act, which
predates initiation of the RFS program by several decades. The Lark Declaration at Appendix 5 provides
maps of 303(d) impaired water bodies in several geographic regions and asserts a causal relationship
between the RFS and the 303(d) listing. Figure 15 compares the 303(d) maps for 2002 (as produced by
the State of Illinois) and the 2015 map presented in the Lark Declaration. This figure clearly shows that
a major water body near Carbondale has been impaired for more than 17 years—well before the RFS
went into effect in 2008. Similar comparisons can be made for areas of North Dakota used for
illustration in the Lark Declaration at page 92 where 303(d) impairments were tracked by the State in
200422), and for areas of central Minnesota (Lark Declaration at page 94) where in 2002, the Minnesota
Pollution Control Agency (MPCA) provided a list of its 303(d) impaired lakes23 noting that nutrients were
part of the root cause in many of them. The attempt in Lark’s Declaration to tie such impairments to the
RFS using the 303(d) maps (Appendix 5) is fundamentally flawed, for the reasons described below.
The maps shown in the Lark Declaration (Appendix 5) do not show watershed hydrology that explicitly
links areas of crop production, ethanol refining, and impaired water bodies. As Bianchi, et al (2010)
observed, general maps and information, such as the geographic placement of crop production in a
regional map, is insufficient to establish a causal link between the RFS and water quality due to the
complexities of numerous factors, including timing, weather, local farming practices, soil chemistry and
physical properties, hydrology, other rural and urban release mechanisms.
Another example of the Lark Declaration’s presentation of faulty data―with respect to the alleged link
between the RFS and streams with impaired water quality―is a sub-basin in northeastern Kansas
depicted on figure 5-6 of the Lark Declaration. We selected this sub-basin for closer examination
because it appears to be a worst-case example of the purported causal relationship, based on the
relatively large proportion of area identified as land converted to corn or soy and the proximity of a
relatively large area to a 303(d) listed water body. Figure 16 presents a reproduction of figure 5-6 from
21 https://www.fisheries.noaa.gov/species/sperm-whale
22 https://deq.nd.gov/publications/WQ/3_WM/TMDL/1_IntegratedReports/2004_Final_ND_Integrated_Report.pdf
23 https://mn.gov/law-library-stat/archive/urlarchive/a042033-1.pdf
17/38 Ramboll
the Lark Declaration, together with the selected sub-basin and presumed land conversion area
immediately adjacent to the stream. Figure 17 presents a simple examination of publicly-available
Google Earth aerial images of this area over time is instructive. Google Earth aerial images of this area
clearly depict it in agricultural use as early as 1991 with no apparent expansion since that time including
the period 2008-2018.
Further, we performed a spatial analysis of land allegedly converted to biofuels feedstock cultivation
after 2008 (using figure 5-6 from the Lark Declaration) within the watershed depicted in Figure 17 (even
though we know that in at least the case illustrated by Figure 17; this allegation is incorrect). Such an
analysis indicates that in the watershed area of 55,840 acres, the total area devoted to crops (exclusive
of grassland) in 2015 based on LCD NASS data, was approximately 18,940 acres (or 34% of the total
watershed area). Of the total acres in crops, approximately 880 acres (or 1.6% of the watershed) was in
corn or soy in 2015. Of the allegedly “converted” fields identified in the watershed in figure 5-6 of the
Lark Declaration, the closest field to the impaired water body is approximately 390 feet (the field shown
in Figure 17) and the average distance of all presumably converted fields to the impaired stream is
approximately 4,860 feet. Barring mass wasting of agricultural soils, very poor practices, or spills of
fertilizers, loading of nutrients to water bodies from agricultural fields (e.g., in pounds per acre per year)
is expected to decrease with distance; even at a distance of 390 feet, an appropriately managed farm
field would be expected to have very little transport of nutrients over that distance. Even if one assumed
that all of the presumably “converted” areas were indeed converted, the total loading of nutrients from
these fields (all else being equal) compared to all other agriculture would be expected to be vanishingly
small (e.g., the presumably “converted” soy and corn area is only about 4.6% of the total crop area).
As an additional example, we performed a spatial analysis of the watershed associated with critical
habitat for the Arkansas shiner (Notropis girardi) as depicted in the Lark Declaration at page 105. For
this watershed area of 471,400 acres, the total area devoted to crops (exclusive of grassland) in 2016
based on LCD NASS data, was approximately 175,500 acres (or 37% of the total watershed area). Of
the total acres in crops, approximately 590 acres (or 0.13% of the watershed) was in corn or soy in
2016. Of the allegedly “converted” fields identified in the watershed in the figure at page 105 of the Lark
Declaration, the closest field to the impaired water body is approximately 2.5 miles and the average
distance for all fields to the critical habitat is approximately 10 miles. For this example, even if one
assumes that all the area devoted to corn or soy in 2016 was the direct result of the RFS, the proportion
of total crop area and the distance between the corn or soy fields is so vanishingly small as to
undermine any claims of impact to the Arkansas shiner.
These quality control checks on the evidence presented in the Lark Declaration demonstrate the flawed
nature of the assertions presented therein. This analysis, along with the fact that the 303(d)
designations predate the RFS, undermines the assertion that there is a causal relationship between the
RFS, reduced water quality in Section 303(d) impaired streams, and potential adverse impacts to listed
aquatic species.
5. Conclusions
Our conclusions follow from the technical review of the assertions made in the Lark Declaration including
an evaluation of the literature cited and an independent check of the geographical information presented
in the Declaration. Our conclusions include the following:
• Assertions that increased corn ethanol production under the RFS has resulted in land
use change and conversion of non-agricultural land to production of biofuel feedstock
are unsubstantiated
18/38 Ramboll
─ Acres planted in corn across the United States has remained close to or below the total
acres planted in the early 1930s despite increases in demand for corn as human food,
animal feed, and biofuels over this nearly 90-year period. The increase in demand has
largely been met by an approximately 7-fold increase in yield (bushels per acre). The
lack of causal relationship between demand for corn and acres planted in corn calls into
question the causal relationship between increased demand for corn for ethanol and land
conversion, and, in turn, potential impacts of land conversion on endangered species.
─ The causal relationship between the RFS and the price of corn is unsupported by the
evidence. Recent efforts to quantify the relationship ignore the multiple domestic and
international economic factors affecting the price of corn. These factors include the
overall increase in global consumption of agricultural commodities in general, due to
expanding economies. In addition, most of the increase in the price of corn (as well as
other crops like soy and wheat) since 2005 has been attributed to higher oil prices.
─ The Lark Declaration (and the literature relied upon therein) does not adequately
consider the myriad factors that influence a farmer’s decision to convert non-agricultural
land to growing any given crop. In addition, the Lack Declaration fails to consider that
converting new land is likely the least preferred option a farmer has for increasing
production because it most likely involves additional expenditures such as land clearing
and other preparation. Nor does it consider that the potential yield that can be expected
of new fields, which, relative to existing fields, may be sub- or infra-marginal and may
require more intensive inputs to achieve desired yields. For these and other reasons,
assertions in the Lark Declaration that the RFS has resulted in land conversion are
unsubstantiated.
• Assertions that RFS-driven land use change has resulted in impacts to particular ESA
listed species are without foundation —The Lark Declaration asserts that land use change
spurred by the RFS has resulted in impacts to listed terrestrial species of birds, mammals, and
insects. However, the evidence presented is poorly researched (including citations to irrelevant
documents and misinterpretation of data) and the examples used to support many assertions
instead actually refute the assertions. For example, eggshell thinning in birds is mentioned as a
potential impact of biofuels production, yet the chemicals responsible for this adverse effect
were banned decades before the RFS took effect. In addition, several examples of supposed land
use change were presented using approaches that are shown to be flawed, among other things,
by testing the assertions against images from Google Earth. Specifically, we checked several
claims of land conversion that are based on methods by Lark et al. (2015) against historical
Google Earth Images that clearly show fields had been converted long before the RFS went into
effect (e.g., in areas allegedly impacting the whooping crane, Poweshiek skipperling, and yellowbilled cuckoo).
• Assertions that RFS-driven biofuels agriculture adversely impacts water quality are
unsubstantiated—The Lark Declaration asserts that biofuels (corn and soy) agriculture has
worsened the Gulf of Mexico dead zone, imperiling Gulf sturgeon, loggerhead turtles, and sperm
whales, yet provides no supporting evidence. The Lark Declaration fails to cite any studies that
associate corn or soy crops (let alone corn or soy crops directly traced to the RFS program) to
any impacts to these species or their habitats. In fact, information related to the life histories of
all three species indicates that the area within which the dead zone forms each summer does
not overlap geographically (or temporally, in the case of the Gulf sturgeon) with critical or
important habitat of any of the species. The Lark Declaration also fails to consider that the Gulf
of Mexico dead zone had been forming on a regular basis for decades before the RFS went into
effect. The Lark Declaration also asserts that biofuel (corn and soy) agriculture is associated
with state designation of impaired waters pursuant to Section 303(d) of the Clean Water Act but
19/38 Ramboll
fails to acknowledge cases in which such designations were made well before the RFS came into
effect. it also presents no assessment of the potential loading of nutrients to impaired water
bodies. Our independent assessment of specific examples indicates that an assertion of impacts
from corn or soy on impaired water bodies is unsubstantiated.
In sum, there are at least two important causal chains that must be quantified and linked together to
demonstrate a relationship between increased corn ethanol production under the RFS and impacts to
ESA-listed species: 1) a causal chain linking the RFS to land use change and water quality impacts; and
2) a causal chain linking these impacts to land and water with specific impacts on the survival or
reproduction of ESA-listed species. Each of these causal chains is made up of many embedded
biophysical and economic relationships that, in turn, are influenced by a myriad of interrelated variables.
The Lark Declaration fails to consider these causal relationships in a meaningful way, relying instead on
unfounded assumptions and speculation to support its thesis.
6. References
Alexander RB, RA Smith, GE Schwarz, EW Boyer, J V Nolan, and JW Brakebill. 2007. Differences in
phosphorus and nitrogen delivery to the Gulf of Mexico from the Mississippi River Basin.
Environmental science & technology 42:822–830.
Babcock BA, and JF Fabiosa. 2011. The impact of ethanol and ethanol subsidies on corn prices:
revisiting history.
Bianchi TS, SF DiMarco, JH Cowan Jr, RD Hetland, P Chapman, JW Day, and MA Allison. 2010. The
science of hypoxia in the Northern Gulf of Mexico: a review. Science of the Total Environment
408:1471–1484.
Bloomberg. (n.d.). Kansas Ethanol LLC. https://www.bloomberg.com/profile/company/0573350D:US.
Carter C, G Rausser, and A Smith. 2018. The Effect of the US Ethanol Mandate on Corn Prices.
CBD. 2019. The black-footed ferret: An Endangered Species Act Success.
https://www.biologicaldiversity.org/species/mammals/black-footed_ferret/.
Corn Agronomy. 2006. Cost of Production. http://corn.agronomy.wisc.edu/Management/L009.aspx.
Cornell University. 2019. All About Birds: Whooping Crane. The Cornell Lab.
Davis RW, JG Ortega-Ortiz, CA Ribic, WE Evans, DC Biggs, PH Ressler, RB Cady, RR Leben, KD Mullin,
and B Würsig. 2002. Cetacean habitat in the northern oceanic Gulf of Mexico. Deep Sea Research
Part I: Oceanographic Research Papers 49:121–142.
Donner SD, and CJ Kucharik. 2008. Corn-based ethanol production compromises goal of reducing
nitrogen export by the Mississippi River. Proceedings of the National Academy of Sciences
105:4513 LP – 4518.
Dunn JB, D Merz, KL Copenhaver, and S Mueller. 2017. Measured extent of agricultural expansion
depends on analysis technique. Biofuels, Bioproducts and Biorefining 11:247–257.
DOI:10.1002/bbb.1750.
Efroymson RA, KL Kline, A Angelsen, PH Verburg, VH Dale, JWA Langeveld, and A McBride. 2016. A
causal analysis framework for land-use change and the potential role of bioenergy policy. Land Use
Policy 59:516–527. DOI:10.1016/j.landusepol.2016.09.009.
EPA. 2018. Biofuels and the Environment: Second Triennial Report to Congress. DOI:EPA/600/R10/183F.
Fannin TE, and BJ Eamoil. 1993. Metal and organic residues in addled eggs of least terns and piping
20/38 Ramboll
plovers in the Platte Valley of Nebraska.
Hart KM, MM Lamont, AR Sartain, I Fujisaki, and BS Stephens. 2013. Movements and habitat-use of
loggerhead sea turtles in the northern Gulf of Mexico during the reproductive period. PLoS One
8:e66921.
Hecht A. 2019. Corn vs. Soybeans: The Farmer’s Choice. https://www.thebalance.com/corn-vssoybeans-808899.
Kleiber K. 2009. The effect of ethanol-driven corn demand on crop choice.
Lark TJ, NP Hendricks, N Pates, A Smith, SA Spawn, M Bougie, E Booth, CJ Kucharik, and HK Gibbs.
2019. Impacts of the Renewable Fuel Standard on America’s Land and Water. Washington, D.C.
Lark TJ, JM Salmon, and HK Gibbs. 2015. Cropland expansion outpaces agricultural and biofuel policies
in the United States. Environmental Research Letters 10. DOI:10.1088/1748-9326/10/4/044003.
Ling G, and B Bextine. 2017. Precision Farming Increases Crop Yields.
https://www.scientificamerican.com/article/precision-farming/.
McConnell A. 2018. Corn Growing 101. https://www.agriculture.com/crops/corn/corn-growing-101.
NMFS. 2011. Endangered and threatened species; determination of nine distinct population segments of
loggerhead sea turtles as endangered or threatened; Final Rule. Department of Commerce,
National Marine Fisheries Service and National Oceanic and Atmospheric Administration;
Department of the Interior, United States Fish and Wildlife Service; Federal Register 76:58868–
58952.
Queck-Matzie T. 2019. Farming 101: How to Plant Corn.
https://www.agriculture.com/crops/corn/farming-101-how-to-plant-corn.
Reiley L. 2019. Weather woes cause American corn farmers to throw in the towel.
https://www.washingtonpost.com/business/2019/06/18/weather-woes-cause-american-cornfarmers-throw-towel/.
Renewable Fuels Association. 2019. The impact of the idling and closure of ethanol production facilities
on local corn prices. Ellisville, MO.
Schiller B. 2017. What Does It Cost to Start a New Farm?
https://www.fastcompany.com/40458330/what-does-it-cost-to-start-a-new-farm.
Schnitkey G, C Zulauf, K Swanson, and R Batts. 2019. Prevented Planting Decision for Corn in the
Midwest.” farmdoc daily (9): 88. Department of Agricultural and Consumer Economics, University
of Illinois at Urbana-Champaign.
Springborn F. 2019. To plant? Or not to plant? https://www.canr.msu.edu/news/to-plant-or-not-toplant.
Staab BD, DS Shrestha, and JA Duffield. 2017. Biofuel impact on food prices index and land use change.
Page 1 2017 ASABE Annual International Meeting. American Society of Agricultural and Biological
Engineers.
USFWS. (n.d.). Reintroduction of a Migratory Flock of Whooping Cranes in the Eastern United States.
https://www.fws.gov/midwest/whoopingcrane/wcraneqanda.html.
USFWS. 2013a. Hine’s Emerald Dragonfly, Somatochlora hineana (Odonata: Corduliidae). 5-Year
Review: Summary and Evaluation. Chicago Ecological Services Field Office, Barrington, Illinois.
USFWS. 2013b. Endangered and threatened wildlife and plants; revision of critical habitat for Salt Creek
tiger beetle; Proposed Rule.
21/38 Ramboll
USFWS. 2014. Endangered and threatened wildlife and plants; determination of threatened status for
the western distinct population segment of the yellow-billed cuckoo (Coccyzus americanus); Final
rule. Department of the Interior, United States Fish and Wildlife Service Federal Register
79:59992–60038.
USFWS. 2018. Black-footed ferret. https://www.fws.gov/mountain-prairie/es/blackFootedFerret.php.
22/38
7. Figures
Figure 1. Total U.S. Planted Acres of Corn Per Year
0
20
40
60
80
100
120
1926
1929
1932
1935
1938
1941
1944
1947
1950
1953
1956
1959
1962
1965
1968
1971
1974
1977
1980
1983
1986
1989
1992
1995
1998
2001
2004
2007
2010
2013
2016
Million Acres
23/38 Ramboll
Figure 2. The decision about which crop to plant is made at the farm level, and takes many different
components into account
24/38 Ramboll
Figure 3. U.S. crude oil prices compared to crop prices, 2005 to 2015. From Staab, et. al. 2017
25/38 Ramboll
Figure 4. United States whooping crane population 1990 to present. Data are from the Audubon
Society’s Christmas Bird Count Database24 and are shown here by the number of cranes per person hour
of observation time.
24 http://netapp.audubon.org/CBCObservation/
0
0.0002
0.0004
0.0006
0.0008
0.001
0.0012
0.0014
0.0016
0.0018
0.002
Number of Whooping Cranes Per Person Hours
1990 1995 2000 2005 2010 2015
of Sighting
Year
Whooping Crane Population in the United States
26/38
Figure 5. Example of error in USDA Cropland Data Layer upon which Lark’s argument rests. An area within the Cheyenne Bottoms Reserve
was identified as corn by the USDA CDL. Examination of aerial imagery showed no corn, and conversations with staff at the reserve
confirmed that corn was never planted there.
27/38 Ramboll
Figure 6. Western yellow-billed cuckoo critical habitat and corn and soy production by county
28/38 Ramboll
Figure 7. Area near Los Molinos, CA where 2018 CDL show corn within the boundaries of the critical habitat for Western yellow-billed
cuckoo and Google Earth images from 1998 2014 document no conversion after 2008.
29/38 Ramboll
Figure 8. Area near Butte, CA where 2018 CDL show corn within the boundaries of the critical habitat for Western yellow-billed cuckoo and
Google Earth images from 1998 2014 document no conversion after 2008.
30/38 Ramboll
Figure 9. Aerial images from Google Earth demonstrating that the area highlighted in the Lark Declaration Appendix 6 was clearly in
agriculture as early as 1991, and there was no evident expansion of the area into what is now designated as critical habitat for Poweshiek
skipperling after 2008
31/38 Ramboll
Figure 10. Adult Salt Creek tiger beetles counted during visual surveys 1991-2012 (excerpted from Federal Register 2013)
32/38 Ramboll
Figure 11. Wild black footed ferret population status 1964 to 2012
SOURCE: https://www.biologicaldiversity.org/species/mammals/black-footed_ferret/
33/38 Ramboll
Figure 12. Location of black-footed ferret populations and counties with corn and soy planted 2018
34/38 Ramboll
Figure 13. Annual nitrate plus nitrite loading to the Gulf of Mexico 1980 to 2017
Source: USGS n.d.
35/38 Ramboll
Figure 14. Gulf sturgeon critical habitat and the Gulf of Mexico dead zone in 2019; the largest dead zone recorded
SOURCE: https://www.noaa.gov/media-release/gulf-of-mexico-dead-zone-is-largest-ever-measured by (Courtesy of N. Rabalais, LSU/LUMCON)
36/38 Ramboll
Figure 15. 303(d) maps for 2002 (as produced by the State of Illinois) and the 2015 map presented in the Lark Declaration showing that a
major water body near Carbondale has been impaired for more than 17 years—well before the RFS went into effect in 2008.
37/38 Ramboll
Figure 16. Watershed area selected for spatial analysis of presumed land conversion relative to 303(d) designated streams as identified in
Figure 5-6 of the Lark Declaration
38/38 Ramboll
Figure 17. Google Earth Images for the period 1991 through 2018 for fields adjacent to a 303(d) impaired water body identified in Figure
5-6 from the Lark Declaration as having been converted from grassland to corn or soy after 2008
Exhibit List
Growth Energy Comments on EPA’s
Proposed Renewable Fuel Standard Program:
Renewable Fuel Standard Annual Rules
Docket # EPA-HQ-OAR-2021-0324
Volume 2
Exhibit
Number
Title of Exhibit
1 Environmental Health & Engineering, Inc., Response to 2020, 2021, and 2022
Renewable Fuel Standard (RFS) Proposed Volume Standards (Feb. 3, 2022)
2 Life Cycle Associates, LLC, Review of GHG Emissions of Corn Ethanol under
the EPA RFS2 (Feb. 4, 2022)
3 Net Gain, Analysis of EPA’s Proposed Rulemaking for 2020, 2021, and 2022
RVOs, Regarding Land Use Change, Wetlands, Ecosystems, Wildlife Habitat,
Water Resource Availability, and Water Quality (Feb. 3, 2022)
4 Comments of Drs. Fatemeh Kazemiparkouhi, David MacIntosh, Helen Suh,
EPA-HQ-OAR-2021-0324 (Feb. 3, 2022)
5 Stillwater Associates, LLC, Comments to EPA on 2020-2022 RFS Rule,
Prepared for Growth Energy (Feb. 4, 2022)
6 Stillwater Associates LLC, Infrastructure Changes and Cost to Increase
Consumption of E85 and E15 in 2017 (July 11, 2016)
7 Oak Ridge National Laboratory, Analysis of Underground Storage Tank System
Materials to Increased Leak Potential Associated with E15 Fuel, ORNL/TM2012/182 (Jul. 2012)
8 Growth Energy, Comments on EPA’s Proposed E15 Fuel Dispenser Labeling
and Compatibility with Underground Storage Tanks Regulations, Docket #
EPA–HQ–OAR–2020–0448 (Apr. 19, 2021)
9 Petroleum Equipment Institute, UST Component Compatibility Library
10 Association of State and Territorial Solid Waste Management Officials,
Compatibility Tool
11 Air Improvement Resource, Inc., Analysis of Ethanol-Compatible Fleet for
Calendar Year 2022 (Nov. 16, 2021)
12 Renewable Fuels Association, Contribution of the Ethanol Industry to the
Economy of the United States in 2020 (Feb. 2, 2021)
13 ABF Economics, Economic Impact of Nationwide E15 Use (June 2021)
14 Jarrett Renshaw & Chris Prentice, Exclusive: Chevron, Exxon seek ‘small
refinery’ waivers from U.S. biofuels law, Reuters (Apr. 12, 2018)
15 Stillwater Associates LLC, Potential Increased Ethanol Sales through E85 for
the 2019 RFS (Aug. 17, 2018)
ActiveUS 192803838v.1
Growth Energy Comments on EPA’s
Proposed Renewable Fuel Standard Program:
Renewable Fuel Standard Annual Rules
Docket # EPA-HQ-OAR-2021-0324
Exhibit 4
200 College Ave – 113 Anderson Hall, Medford, MA 02155 | TEL: 617.627.3211 | FAX: 617.627.3994 |
www.engineering.tufts.edu/cee
1
Department of Civil and Environmental Engineering
February 3, 2022
Docket Number: EPA-HQ-OAR-2021-0324
Comments of Drs. Fatemeh Kazemiparkouhi,
1 David MacIntosh,
2 Helen Suh3
1 Environmental Health & Engineering, Inc., Newton, MA
2 Environmental Health & Engineering, Inc., Newton, MA and the Harvard T.H. Chan
School of Public Health, Boston, MA
3 Tufts University, Medford, MA
We are writing to comment on issues raised by the proposed RFS annual rule, the Draft
Regulatory Impact Analysis (December 2021; EPA-420-D-21-002), and the supporting
Health Effects Docket Memo (September 21, 2021; EPA-HQ-OAR-2021-0324-0124),
specifically regarding the impact of ethanol-blended fuels on air quality and public
health. We provide evidence of the air quality and public health benefits provided by
higher ethanol blends, as shown in our recently published study1 by Kazemiparkouhi et
al. (2021), which characterized emissions from light duty vehicles for market-based
fuels. Findings from our study demonstrate ethanol-associated reductions in emissions
of primary particulate matter (PM), nitrogen oxides (NOx), carbon monoxide (CO), and
to a lesser extent total hydrocarbons (THC). Our results provide further evidence of the
potential for ethanol-blended fuels to improve air quality and public health, particularly
for environmental justice communities. Below we present RFS-pertinent findings from
Kazemiparkouhi et al. (2021), followed by their implications for air quality, health, and
environmental justice.
Summary of Kazemiparkouhi et al. (2021)
Our paper is the first large-scale analysis of data from light-duty vehicle emissions
studies to examine real-world impacts of ethanol-blended fuels on regulated air pollutant
emissions, including PM, NOx, CO, and THC. To do so, we extracted data from a
comprehensive set of emissions and market fuel studies conducted in the US. Using
these data, we (1) estimated composition of market fuels for different ethanol volumes
and (2) developed regression models to estimate the impact of changes in ethanol
volumes in market fuels on air pollutant emissions for different engine types and
operating conditions. Importantly, our models estimated these changes accounting for
not only ethanol volume fraction, but also aromatics volume fraction, 90% volume
distillation temperature (T90) and Reid Vapor Pressure (RVP). Further, they did so
1 https://doi.org/10.1016/j.scitotenv.2021.151426
200 College Ave – 113 Anderson Hall, Medford, MA 02155 | TEL: 617.627.3211 | FAX: 617.627.3994 |
www.engineering.tufts.edu/cee
2
under both cold start and hot stabilized running conditions and for gasoline-direct
injection engines (GDI) and port-fuel injection (PFI) engine types. Key highlights from
our paper include:
• Aromatic levels in market fuels decreased by approximately 7% by volume for
each 10% by volume increase in ethanol content (Table 1). Our findings of lower
aromatic content with increasing ethanol content is consistent with market fuel
studies by EPA and others (Eastern Research Group, 2017, Eastern Research
Group, 2020, US EPA, 2017). As discussed in EPA’s Fuel Trends Report, for
example, ethanol volume in market fuels increased by approximately 9.4% between
2006 and 2016, while aromatics over the same time period were found to drop by
5.7% (US EPA, 2017).
We note that our estimated market fuel properties differ from those used in the
recent US EPA Anti-Backsliding Study (ABS), which examined the impacts of
changes in vehicle and engine emissions from ethanol-blended fuels on air quality
(US EPA, 2020). Contrary to our study, ABS was based on hypothetical fuels that
were intended to satisfy experimental considerations rather than mimic real-world
fuels. It did not consider published fuel trends; rather, the ABS used inaccurate fuel
property adjustment factors in its modeling, reducing aromatics by only 2% (Table
5.3 of ABS 2020), substantially lower than the reductions found in our paper and in
fuel survey data (Kazemiparkouhi et al., 2021, US EPA, 2017). As a result, the
ABS’s findings and their extension to public health impacts are not generalizable to
real world conditions.
Table 1. Estimated market fuel properties
Fuel ID EtOH
Vol (%) T50 (oF) T90 (oF) Aromatics
Vol (%) AKI RVP
(psi)
E0 0 219 325 30 87 8.6
E10 10 192 320 22 87 8.6
E15 15 162 316 19 87 8.6
E20 20 165 314 15 87 8.6
E30 30 167 310 8 87 8.6
Abbreviations: EtOH = ethanol volume; T50 = 50% volume distillation temperature; T90 = 90%
volume distillation temperature; Aromatics=aromatic volume; AKI = Anti-knock Index; RVP = Reid
Vapor Pressure.
• PM emissions decreased with increasing ethanol content under cold-start
conditions. Primary PM emissions decreased by 15-19% on average for each 10%
increase in ethanol content under cold-start conditions (Figure 1). While statistically
significant for both engine types, PM emission reductions were larger for GDI as
compared to PFI engines, with 53% and 29% lower PM emissions, respectively,
when these engines burned E30 as compared to E10. In contrast, ethanol content
in market fuels had no association with PM emissions during hot-running conditions.
Importantly, our findings are consistent with recent studies that examined the effect
of ethanol blending on light duty vehicle PM emissions. Karavalakis et al. (2014),
200 College Ave – 113 Anderson Hall, Medford, MA 02155 | TEL: 617.627.3211 | FAX: 617.627.3994 |
www.engineering.tufts.edu/cee
3
(2015), Yang et al. (2019a), (2019b), Schuchmann and Crawford (2019), for
example, assessed the influence of different mid-level ethanol blends – with proper
adjustment for aromatics – on the PM emissions from GDI engines and Jimenez and
Buckingham (2014) from PFI engines. As in our study, which also adjusted for
aromatics, each of these recent studies found higher ethanol blends to emit lower
PM as compared to lower or zero ethanol fuels.
Together with these previous studies, our findings support the ability of ethanolblended fuels to offer important PM emission reduction opportunities. Cold start PM
emissions have consistently been shown to account for a substantial portion
of all direct tailpipe PM emissions from motor vehicles, with data from the EPAct
study estimating this portion to equal 42% (Darlington et al., 2016, US EPA, 2013).
The cold start contribution to total PM vehicle emissions, together with our findings
of emission reductions during cold starts, suggest that a 10% increase in ethanol
fuel content from E10 to E20 would reduce total tailpipe PM emissions from
motor vehicles by 6-8%.
Figure 1. Change (%) in cold-start emissions for comparisons of different ethanolcontent market fuelsa
a Emissions were predicted from regression models that included ethanol and aromatics volume
fraction, T90, and RVP as independent variables
200 College Ave – 113 Anderson Hall, Medford, MA 02155 | TEL: 617.627.3211 | FAX: 617.627.3994 |
www.engineering.tufts.edu/cee
4
• NOx, CO and THC emissions were significantly lower for higher ethanol fuels
for PFI engines under cold-start conditions, but showed no association for GDI
engines (Figure 1). CO and THC emissions also decreased under hot running
conditions for PFI and for CO also for GDI engines (results not shown). [Note that
NOx emissions for both PFI and GDI engines were statistically similar for
comparisons of all ethanol fuels, as were THC emissions for GDI engines.] These
findings add to the scientific evidence demonstrating emission reduction benefits of
ethanol fuels for PM and other key motor vehicle-related gaseous pollutants.
Implications for Public Health and Environmental Justice Communities
The estimated reductions in air pollutant emissions, particularly of PM and NOx,
indicate that increasing ethanol content offers opportunities to improve air
quality and public health. As has been shown in numerous studies, lower PM
emissions result in lower ambient PM concentrations and exposures (Kheirbek et al.,
2016, Pan et al., 2019), which, in turn, are causally associated with lower risks of total
mortality and cardiovascular effects (Laden et al., 2006, Pun et al., 2017, US EPA,
2019, Wang et al., 2020).
The above benefits to air quality and public health associated with higher ethanol
fuels may be particularly great for environmental justice (EJ) communities. EJ
communities are predominantly located in urban neighborhoods with high traffic density
and congestion and are thus exposed to disproportionately higher concentrations of PM
emitted from motor vehicle tailpipes (Bell and Ebisu, 2012, Clark et al., 2014, Tian et al.,
2013). Further, vehicle trips within urban EJ communities tend to be short in duration
and distance, with approximately 50% of all trips in dense urban communities under
three miles long (de Nazelle et al., 2010, Reiter and Kockelman, 2016, US DOT, 2010).
As a result, a large proportion of urban vehicle trips occur under cold start conditions
(de Nazelle et al., 2010), when PM emissions are highest. Given the evidence that
ethanol-blended fuels substantially reduce PM, NOx, CO, and THC emissions during
cold-start conditions, it follows that ethanol-blended fuels may represent an effective
method to reduce PM health risks for EJ communities.
Summary
Findings from Kazemiparkouhi et al. (2021) provide important, new evidence of ethanolrelated reductions in vehicular emissions of PM, NOx, CO, and THC based on realworld fuels and cold-start conditions. Given the substantial magnitude of these
reductions and their potential to improve air quality and through this public health, our
findings warrant careful consideration. Policies that encourage higher concentrations of
ethanol in gasoline would provide this additional benefit. These policies are especially
needed to protect the health of EJ communities, who experience higher exposures to
motor vehicle pollution, likely including emissions from cold starts in particular, and are
at greatest risk from their effects.
200 College Ave – 113 Anderson Hall, Medford, MA 02155 | TEL: 617.627.3211 | FAX: 617.627.3994 |
www.engineering.tufts.edu/cee
5
References
BELL, M. L. & EBISU, K. 2012. Environmental inequality in exposures to airborne particulate
matter components in the United States. Environmental health perspectives, 120, 1699-
1704.
CLARK, L. P., MILLET, D. B. & MARSHALL, J. D. 2014. National patterns in environmental
injustice and inequality: outdoor NO2 air pollution in the United States. PLoS One, 9,
e94431.
DARLINGTON, T. L., KAHLBAUM, D., VAN HULZEN, S. & FUREY, R. L. 2016. Analysis
of EPAct Emission Data Using T70 as an Additional Predictor of PM Emissions from
Tier 2 Gasoline Vehicles. SAE Technical Paper.
DE NAZELLE, A., MORTON, B. J., JERRETT, M. & CRAWFORD-BROWN, D. 2010. Short
trips: An opportunity for reducing mobile-source emissions? Transportation Research
Part D: Transport and Environment, 15, 451-457.
EASTERN RESEARCH GROUP 2017. Summer Fuel Field Study (prepared for Texas
Commission on Environmental Quality by Eastern Research Group, Inc.).
EASTERN RESEARCH GROUP 2020. Summer Field Study (prepared for Texas Commission
on Environmental Quality by Eastern Research Group, Inc.).
JIMENEZ, E. & BUCKINGHAM, J. P. 2014. Exhaust Emissions of Average Fuel Composition.
Alpharetta, GA.
KARAVALAKIS, G., SHORT, D., VU, D., RUSSELL, R. L., ASA-AWUKU, A., JUNG, H.,
JOHNSON, K. C. & DURBIN, T. D. 2015. The impact of ethanol and iso-butanol blends
on gaseous and particulate emissions from two passenger cars equipped with sprayguided and wall-guided direct injection SI (spark ignition) engines. Energy, 82, 168-179.
KARAVALAKIS, G., SHORT, D., VU, D., VILLELA, M., ASA-AWUKU, A. & DURBIN, T.
D. 2014. Evaluating the regulated emissions, air toxics, ultrafine particles, and black
carbon from SI-PFI and SI-DI vehicles operating on different ethanol and iso-butanol
blends. Fuel, 128, 410-421.
KAZEMIPARKOUHI, F., ALARCON FALCONI, T. M., MACINTOSH, D. L. & CLARK, N.
2021. Comprehensive US database and model for ethanol blend effects on regulated
tailpipe emissions. Sci Total Environ, 151426.
KHEIRBEK, I., HANEY, J., DOUGLAS, S., ITO, K. & MATTE, T. 2016. The contribution of
motor vehicle emissions to ambient fine particulate matter public health impacts in New
York City: a health burden assessment. Environmental Health, 15, 89.
LADEN, F., SCHWARTZ, J., SPEIZER, F. E. & DOCKERY, D. W. 2006. Reduction in fine
particulate air pollution and mortality: Extended follow-up of the Harvard Six Cities
study. American journal of respiratory and critical care medicine, 173, 667-672.
PAN, S., ROY, A., CHOI, Y., ESLAMI, E., THOMAS, S., JIANG, X. & GAO, H. O. 2019.
Potential impacts of electric vehicles on air quality and health endpoints in the Greater
Houston Area in 2040. Atmospheric Environment, 207, 38-51.
PUN, V. C., KAZEMIPARKOUHI, F., MANJOURIDES, J. & SUH, H. H. 2017. Long-Term
PM2.5 Exposure and Respiratory, Cancer, and Cardiovascular Mortality in Older US
Adults. American Journal of Epidemiology, 186, 961-969.
REITER, M. S. & KOCKELMAN, K. M. 2016. The problem of cold starts: A closer look at
mobile source emissions levels. Transportation Research Part D: Transport and
Environment, 43, 123-132.
200 College Ave – 113 Anderson Hall, Medford, MA 02155 | TEL: 617.627.3211 | FAX: 617.627.3994 |
www.engineering.tufts.edu/cee
6
SCHUCHMANN, B. & CRAWFORD, R. 2019. Alternative Oxygenate Effects on Emissions.
Alpharetta, GA (United States).
TIAN, N., XUE, J. & BARZYK, T. M. 2013. Evaluating socioeconomic and racial differences in
traffic-related metrics in the United States using a GIS approach. J Expo Sci Environ
Epidemiol, 23, 215-22.
US DOT 2010. National Transportation Statistics. Research and Innovative Technology
Administration: Bureau of Transportation Statistics.
US EPA 2013. Assessing the Effect of Five Gasoline Properties on Exhaust Emissions from
Light-Duty Vehicles Certified to Tier 2 Standards: Analysis of Data from EPAct Phase 3
(EPAct/V2/E-89): Final Report. EPA-420-R-13-002 ed.: Assessment and Standards
Division Office of Transportation and Air Quality U.S. Environmental Protection
Agency.
US EPA 2017. Fuel Trends Report: Gasoline 2006-2016.
US EPA 2019. Integrated Science Assessment for Particulate Matter. Center for Public Health
and Environmental Assessment.
US EPA 2020. Clean Air Act Section 211(v)(1) Anti-backsliding Study. Assessment and
Standards Division Office of Transportation and Air Quality U.S. Environmental
Protection Agency.
WANG, B., EUM, K. D., KAZEMIPARKOUHI, F., LI, C., MANJOURIDES, J., PAVLU, V. &
SUH, H. 2020. The impact of long-term PM2.5 exposure on specific causes of death:
exposure-response curves and effect modification among 53 million U.S. Medicare
beneficiaries. Environ Health, 19, 20.
YANG, J., ROTH, P., DURBIN, T. D., JOHNSON, K. C., ASA-AWUKU, A., COCKER, D. R.
& KARAVALAKIS, G. 2019a. Investigation of the Effect of Mid- And High-Level
Ethanol Blends on the Particulate and the Mobile Source Air Toxic Emissions from a
Gasoline Direct Injection Flex Fuel Vehicle. Energy & Fuels, 33, 429-440.
YANG, J., ROTH, P., ZHU, H., DURBIN, T. D. & KARAVALAKIS, G. 2019b. Impacts of
gasoline aromatic and ethanol levels on the emissions from GDI vehicles: Part 2.
Influence on particulate matter, black carbon, and nanoparticle emissions. Fuel, 252, 812-
820.
ActiveUS 192803838v.1
Growth Energy Comments on EPA’s
Proposed Renewable Fuel Standard Program:
Renewable Fuel Standard Annual Rules
Docket # EPA-HQ-OAR-2021-0324
Exhibit 5
Comments to EPA on 2020-2022
RFS Rule
Prepared for
Growth Energy
By
Stillwater Associates LLC
Irvine, California, USA
February 4, 2022
3 Rainstar Irvine, CA 92614 – Tel (888) 643 0197 – www.stillwaterassociates.com
Stillwater Associates
Comments to EPA on 2020-2022 RFS Rule
i
Table of Contents
Executive Summary …………………………………………………………………………………………………………………….. 1
1 2022 Potential Ethanol Production ………………………………………………………………………………………….. 1
1.1 Corn Supply ……………………………………………………………………………………………………………………. 1
1.2 Ethanol Production …………………………………………………………………………………………………………… 3
1.3 Computation of Achievable Ethanol Supply …………………………………………………………………………. 4
2 E85 and E15 Consumption Capacity ……………………………………………………………………………………….. 8
2.1 E85 ………………………………………………………………………………………………………………………………… 8
2.1.1 E85 Dispenser Capability ………………………………………………………………………………………….. 8
2.2 E85-compatible vehicles …………………………………………………………………………………………………… 9
2.2.1 FFV ………………………………………………………………………………………………………………………. 10
2.3 Combining E85 Infrastructure and FFVs ……………………………………………………………………………. 10
2.4 E15 ………………………………………………………………………………………………………………………………. 10
2.4.1 E15 Dispensers ……………………………………………………………………………………………………… 10
2.4.2 E15 Infrastructure …………………………………………………………………………………………………… 11
2.4.3 E15 Compatible Vehicles ………………………………………………………………………………………… 11
2.5 Combining E85 and E15 Capabilities ……………………………………………………………………………….. 12
3 E85 Pricing …………………………………………………………………………………………………………………………. 13
4 Encouraging Increased Use of Higher-Ethanol Blends …………………………………………………………….. 15
5 Market Challenge in Meeting Original 2020 Percentage Standards …………………………………………… 16
5.1 RFS Impacts on Imports and Exports of Petroleum and Ethanol ………………………………………….. 19
6 EPA Cost Analysis of Ethanol Costs ……………………………………………………………………………………… 22
List of Tables
Table 1. Potential 2022 Corn Harvest using 2007/08 Planted Acres and Current Yields ………………………. 5
Table 2. U.S. Population as Estimated by the United Nations ……………………………………………………………. 5
Table 3. Calculation of Maximum Ethanol Production in 2022/23 ………………………………………………………. 7
Table 4. Summary of Corn Supply and Demand Calculations …………………………………………………………… 8
Table 5. E85 and Gasoline Average Prices by Region October 2021 ……………………………………………….. 13
Table 6. E85 and Gasoline Average Prices by Region (GGE) October 2021 …………………………………….. 14
Table 7. E85 and Gasoline Average Prices by Region for October 2019 ………………………………………….. 14
Table 8. E85 and Gasoline Average Prices by Region (GGE) October 2021 …………………………………….. 15
Table 9. RFS Final 2020 Volume Standards and Percentage Standards, February 2020 ……………………. 16
Table 10. Values for Terms in Calculation of the Final 2020 Standards (billion gallons) ……………………… 17
Table 11. Volume Obligations (2020 Final Rule) Compared to the Actual Volume Obligations (with and
without EPA’s estimate of SREs) and RIN Generation ……………………………………………………………………. 18
Table 12. Recalculation of 2020 Percentage Standard with no SREs ………………………………………………. 18
Table 13. 2020 Actual Volume Obligations calculated with revised percentage standards (with and without
EPAs current estimate of SREs) ………………………………………………………………………………………………….. 19
Table 14. U.S. Annual Ethanol Production and Capacity ………………………………………………………………… 19
Comments to EPA on 2020-2022 RFS Rule
ii
Table of Figures
Figure 1. U.S. Corn Acres and Yield (1936-2021) ……………………………………………………………………………. 2
Figure 2. U.S. Corn Acres Planted and Yield (2006-2021) ………………………………………………………………… 3
Figure 3. Improvement in Yields at U.S. Corn Ethanol Plants ……………………………………………………………. 4
Figure 4. U.S. Ethanol Production and Corn-based Feed Availability …………………………………………………. 6
Figure 5. E85 Consumption projected in AEO 2021 …………………………………………………………………………. 9
Figure 6. MY2001 Or Later Fraction of In-Use Vehicle Fleet and MY 2001 or Later Fraction of In-Use
Gasoline Consumption ……………………………………………………………………………………………………………….. 12
Figure 7. Monthly U.S. Ethanol Supply, Demand and RFS Volume Standards (2018-2021) ……………….. 20
Figure 8. Monthly U.S. Exports of Crude Oil, Petroleum Products, and Ethanol (2018-2021) ………………. 21
Figure 9. Monthly U.S. Imports of Crude Oil, Gasoline, Diesel, and Ethanol (2018-2021) …………………… 22
Comments to EPA on 2020-2022 RFS Rule
iii
Disclaimer
Stillwater Associates LLC prepared this report for the sole benefit of Growth Energy and no other party.
Stillwater Associates LLC conducted the analysis and prepared this report using reasonable care and skill
in applying methods of analysis consistent with normal industry practice. All results are based on
information available at the time of preparation. Changes in factors upon which the report is based could
affect the results. Forecasts are inherently uncertain because of events that cannot be foreseen, including
the actions of governments, individuals, third parties, and competitors. Nothing contained in this report is
intended as a recommendation in favor of or against any particular action or conclusion. Any particular
action or conclusion based on this report shall be solely that of Growth Energy. NO IMPLIED WARRANTY
OF MERCHANTABILITY SHALL APPLY. NOR SHALL ANY IMPLIED WARRANTY OF FITNESS FOR
ANY PARTICULAR PURPOSE.
Comments to EPA on 2020-2022 RFS Rule
1
1 2022 Potential Ethanol Production
EIA lists the U.S. ethanol nameplate production capacity for 2020 at 17.38 billion gallons per year1. How
much of this ethanol production capacity can be used is primarily a function of the available feedstock, corn,
and the conversion capacity of ethanol plants. We consider three different approaches to determine the
real world maximum potential ethanol production in 2022: historical maximum, previous year, and potential
expansion.
The highest year of ethanol production was 2018, when 16.061 billion gallons of ethanol were produced
domestically.2 In 2021, it is estimated that 14.87 billion gallons of ethanol were produced.3 We believe that
both of these figures represent conservative estimates of how much ethanol could reasonably be produced
in 2022. The 2021 volume was suppressed substantially by low demand for transportation fuel in response
to the Covid-19 pandemic. And neither figure accounts for the continuing growth in the productivity of U.S.
corn growers or the steady improvements in the efficiency of U.S. corn ethanol plants. As explained below
in greater detail, these developments have allowed U.S. ethanol production to continuously increase their
production capability without requiring increasing corn acreage or adversely impacting the supply of corn
available for other domestic non-ethanol demands or export markets. In fact, we conclude that, accounting
for these developments, 15.565 billion gallons could be produced domestically in 2022.
While the 15.565 billion gallons of ethanol for 2022 in Table 4 seems like an upper limit on ethanol
production in 2022, it is in fact limited by the decision to keep the planted acres constant the decision to
keep the portion of corn used for ethanol constant, and the representation of new technology
implementation as a straight line. The reality is that market forces are always in play. A positive future
market outlook may cause more acres to be planted in corn that year. It may cause plant maintenance to
be delayed until next year. A very promising technology may be implemented earlier and to a larger extent
than typical technology is implemented. Table 4 and the other tables in that section represent average
conditions which can be increased or decreased by each farmer or ethanol production facility’s market
outlook. Indeed, as noted, the actual ethanol production in 2018 exceeded our projection for 2022 by a
substantial margin, at a time when yield rates and conversion efficiency were lower than they are now.
1.1 Corn Supply
Review of U.S. corn production data through the Fall 2021 harvest shows that yields per harvested acre
have continued their long-term growth trend. Figures 1 and 2 below illustrates annual per-harvested-acre
yields and annual planted corn acres as reported by USDA4; Figure 1 presents the long-term trend since
1936 and Figure 2 focuses on 2006 through 2021. The dashed red line in Figure 2 indicates the average
annual corn plantings from 2008 through 2021 was 90.8 million acres; this is below the 93.5 million corn
acres planted in 2007, the last crop planted prior to enactment of EISA 2007.
1 https://afdc.energy.gov/data/10342, U.S. Ethanol Plants Capacity and Production 2 https://afdc.energy.gov/data/10342, U.S. Ethanol Plants Capacity and Production 3 http://ethanolproducer.com/articles/18648/eia-reduces-2021-2022-ethanol-production-forecasts 4
USDA QuickStats, https://quickstats.nass.usda.gov/.
Comments to EPA on 2020-2022 RFS Rule
2
Figure 1. U.S. Corn Acres and Yield (1936-2021)
Source: USDA, Stillwater analysis
Comments to EPA on 2020-2022 RFS Rule
3
Figure 2. U.S. Corn Acres Planted and Yield (2006-2021)
Source: USDA, Stillwater analysis
Since 2008, U.S. corn yields have grown from 153.3 bushel per harvested acre (“bu/ac”) to 177.0 bu/ac in
2021, an increase of 1.8 bu/ac each year. This is nearly the same pace as the 1.9 bu/ac each year in the
85 years since 1936. Normally to be consistent with this durable trend, we would estimate that in 2022, the
corn yield will be 178.8 bu/ac. However, this value is based on 177.0, which is itself a projection. To be
conservative we have decided to keep the 2022 projection of corn yield at 177.0.
1.2 Ethanol Production
In addition to this steadily increasing trend in corn yields, the yield of ethanol from corn processed at U.S.
ethanol plants has also steadily increased (according to data from USDA). These data, illustrated in Figure
3 below, indicate that yields have increased at a rate of over 0.010 gallons of denatured fuel ethanol (DFE)
per bushel of corn each year from 1982 through 2020 and this rate has accelerated to over 0.012 gallons
of DFE per bushel of corn each year from 2006 through 2020. These increases can be attributed to
innovation enabled by growing industry operating experience and steady improvements in both the
engineering designs of ethanol plants and the efficiency of the yeasts used in the fermentation process.
Extrapolating the long-term trend (an average yield increase of 0.0101 gallons per bushel per year since
1982) illustrated in Figure 3 allows us to estimate that the reported industry-average ethanol yield of 2,894
gallons of ethanol per bushel of corn in 2020 would increase to 2.904 gallons of ethanol per bushel in 2021
and 2.914 gallons per bushel in 2022. As a result, the 3,049 million bushels of corn which produced 9,309
million gallons of DFE in 2008 would yield 9,919 million gallons of DFE at current yields, a 7% increase.
Comments to EPA on 2020-2022 RFS Rule
4
Figure 3. Improvement in Yields at U.S. Corn Ethanol Plants
Source: USDA QuickStats, Stillwater analysis
1.3 Computation of Achievable Ethanol Supply
Combining each of the elements above, it is possible to estimate how much corn could be used for ethanol
production in 2022—and hence how much ethanol could be produced—while continuing to supply the
growing domestic market demands for corn required for all other uses (estimated based on the 10.9%
growth in U.S. population since 2007) and maintaining corn exports at the same volume as 2007.
As a first step in this analysis, we can estimate the amount of corn which can be produced on the same
number of planted acres as used in the 2007 market year.5 This analysis is presented in Table 1 below. For
purposes of this analysis, we assume that U.S. farmers plant 93.5 million acres of corn in the Spring of
2022, which is equal to the acreage planted in 2007. For the 2007 market year, we use USDA data reported
in their World Agriculture Supply Demand Estimate (WASDE) report for January 2010. Importantly, the ratio
of harvested acres to planted acres was about 92.5% in 2007, higher than 91.3% average for the most
recent 10 years. We also assume that U.S. farmers will harvest 91.3% of the planted acreage, which is the
average harvest rate over the past decade. Accordingly, we estimate that 85.4 million acres could be
harvested in Fall 2022. Applying the yield of 177.0 bu/ac, we estimate an achievable 2022 corn crop of
15,116 million bushels.
Then to assess how much of that corn would be available for domestic use, we add corn imports (USDA
estimates 25 million bushels for 2022) and subtract corn exports (using the most recent USDA figure of
2,437 million bushels in 2020/2021.) The net result is that the U.S. could have 12,704 million bushels of
corn available for all domestic uses in 2022.
The other major demands for corn are for feed, food, seed, and non-ethanol industrial uses. Accordingly,
assessment of how much corn is potentially available for ethanol production needs to also consider
domestic demand for these other markets. Many factors influence corn demand in each of these markets.
5
The market year for corn runs from September 1st through August 31st. Thus, the 2007/08 market year begins with harvesting the
corn planted in the Spring of 2007 (before EISA was enacted in December 2007) and ends prior to the harvest of the corn crop planted
in the Spring of 2008.
Comments to EPA on 2020-2022 RFS Rule
5
Table 1. Potential 2022 Corn Harvest using 2007/08 Planted Acres and Current Yields
Market Year 2007/08
Estimated 2022/23 with
2007/08 Acres
Area Planted (million acres) 93.5 93.5
Area Harvested (million acres) 86.5 85.4
Yield (bushels per acre harvested) 150.7 177.0
Production (million bushels) 13,038 15,116
Corn Imports (million bushels) 20 25
Less Corn Exports (million bushels) (2,437) (2,437)
Available Corn Supply (million bushels) 10,621 12,704
Source: USDA, Stillwater analysis
For the purposes of this analysis, we will assume that growth in U.S. population, which is projected to be
10.9% from 2008 to 2022, can be used as a proxy for overall demand growth.6 Annual data on U.S.
population as reported by the U.S. Census Bureau for 2007 through 2021 and U.S. population for 2022 as
estimated by the United Nations7 is summarized in Table 2.
Table 2. U.S. Population as Estimated by the United Nations
Year Population as of
December 31st
2007 301,903,167
2008 304,718,000
2009 307,373,750
2010 309,731,983
2011 311,918,250
2012 314,120,641
2013 316,266,088
2014 318,534,859
2015 320,822,902
2016 323,095,500
2017 325,142,676
2018 326,882,088
2019 328,460,928
2020 331,236,261
2021 332,182,892
2022 (forecast) * 334,805,269
*U.N. Forecast
Source: U.S. Census Bureau, Macrotrends.net
In estimating supply of corn for feed, it is also necessary to consider the feed co-products produced at
ethanol plants (both wet mills and dry mills). The ethanol production process only utilizes the starch
contained in the corn; all the protein, fiber, and minerals, along with much of the oil8 are contained in the
co-products9 which are highly valued as feed. Production data for these coproducts is available from USDA
6
Other factors would include changing consumer dietary preferences (impacting feed demand for cattle, swine and poultry) and
economic growth (impacting consumer demand for a wide range of products). 7
Available at https://www.macrotrends.net/countries/USA/united-states/population 8
A portion of the corn oil is separated out at most corn ethanol plants for use in applications other than feed. This corn oil product is,
thus, excluded from this analysis of feed co-products.
9
These co-products include distillers’ grains and syrup produced at dry mills and corn gluten meal and corn gluten feed produced at
wet mills.
Comments to EPA on 2020-2022 RFS Rule
6
in their monthly Grain Crushings and Co-Products Production report and Annual Summary10 with annual
data released in March of the following year. Figure 4 below illustrates, for the most recent three years, the
combination of corn used for feed, as reported by USDA in their monthly World Agriculture Supply and
Demand Estimate (WASDE) reports and feed co-products as reported in their Grain Crushings and CoProducts Production reports. This is compared to the corresponding annual ethanol production data and it
can be seen that the large decrease in ethanol production in 2020 had only a minimal effect on feed
availability.
Figure 4. U.S. Ethanol Production and Corn-based Feed Availability
Source: USDA, Stillwater analysis
The next step in our analysis is to project current U.S. corn demand for uses other than ethanol production.
USDA breaks down domestic corn demand into two categories – “Feed and Residual” (F&R) and “Food,
Seed, and Industrial” (FS&I). USDA then breaks out fuel ethanol demand from the broader FS&I total. For
our analysis, we will divide domestic non-ethanol corn demand into F&R and “Other FS&I” (i.e., the FS&I
minus corn used for ethanol production). Complete analysis of the demand for feed, however, needs to
include the feed co-products of ethanol production in addition to the direct use of corn for feed; we label
this as Total Corn-based Feed.11
For this analysis, we assume that growth in domestic demand for Total Corn-based Feed and Other FS&I
since 2007 can be estimated based on the growth in U.S. population since 2007. Estimation of the maximum
ethanol production in 2022/23 which leaves sufficient corn to satisfy the U.S.’s growing demand for Total
Corn-based Feed and Other FS&I is illustrated in Table 3 below. From the calculations in Table 1, we have
projected 12,704 million bushels of corn to be available in the U.S. during 2022/23 to supply all domestic
uses. From this, we subtract the 1,476 million bushels of corn required to satisfy demands for Other FS&I
(calculated from the reported demand in 2007/08 and the growth in U.S. population). This leaves 11,228
million bushels of corn available to supply F&R plus ethanol production. Per USDA, Total Corn-based Feed
10 https://usda.library.cornell.edu/concern/publications/v979v304g?locale=en 11 E.g., DDGS and corn gluten meal.
Comments to EPA on 2020-2022 RFS Rule
7
demand in the U.S. in 2007/08 was 6,839 million bushels which included 5,913 million bushels of corn and
926 million bushels of Feed Co-Products from ethanol production.12
Adjusting for population growth since 2007/08, the U.S. is estimated to demand 7,514 million bushels of
Total Corn-based Feed in 2022/23. Allocating those 7,514 million bushels between corn and co-products
requires an iterative calculation based on 17 pounds of co-products per bushel of corn used in ethanol
production and a projected ethanol yield of 2.914 gallons per bushel in 2022/23 based on extrapolation of
the yearly industry yield trend since 1982. Using these yields, production of 15,565 million gallons of ethanol
in 2022/23 would be expected to consume 5,341 million bushels of corn and produce 1,621 million bushels
of Feed Co-Products. This leaves an estimated 5,892 million bushels of corn available for use as F&R.
These 5,892 million bushels of corn for F&R plus the 1,621 million bushels of Feed Co-Products adds up
to the 7,514 million bushels of estimated demand for Total Corn-based Feed.
Table 3. Calculation of Maximum Ethanol Production in 2022/23
Marketing Year 2007/08 Projected 2022/23
U.S. Population 300,608,429 334,805,269
Corn Available for Domestic Use (million bushels) 10,621 12,704
Other FS&I (million bushels) 1,338 1,470
Corn Available for Feed and Ethanol (million bushels) 9,283 11,234
Feed and Residual (million bushels) 5,913 —
Estimated Feed Co-Products (million bushels) 926 —
Total Corn-based Feed (million bushels) 6,839 7,514
Estimated Feed Co-Products from 15.565 billion gallons of ethanol
production
1,621
Required Corn to supply F&R Demand (million bushels) 5,892
Corn Available for ethanol production 5,341
Ethanol production at 2.914 gallons/bushel (billion gallons) 15.565
Source: USDA, Stillwater analysis
Table 4 below recaps the above allocation of corn volume in 2022/23 which produces 15.565 billion gallons
of ethanol while planting the same number of acres planted in corn in 2007, keeping U.S. corn exports even
with 2007/08, and supplying estimated growth in domestic demand for all other uses of corn.
12 Based on an average yield of 17 pounds per bushel of corn used for ethanol production, corrected to 15% moisture content.
Comments to EPA on 2020-2022 RFS Rule
8
Table 4. Summary of Corn Supply and Demand Calculations
Market Year 2007/08 Projected 2022/23
Corn Supply (million bushels)
Corn Produced 13,038 15,116
Corn Imports 20 25
Corn Exports (2,437) (2,437)
Total Domestic Corn Supply 10,621 12,704
Corn Demand (million bushels)
Feed and Residual 5,913 5,892
Food, Seed & Industrial 4,387 6,811
Ethanol for fuel 3,049 5,341
Other Food, Seed & Industrial 1,338 1,470
Total Domestic Corn Demand 10,300 12,704
Surplus/(Shortage) 321 —
Ethanol Production (billion gallons) 9.3 15.565
Ethanol Yield (gallons/bushel) 2.735 2.914
Feed Co-Products (mmillionbushels) 926 1,621
Feed Co-Product Yield (pounds per bushel)
@ 15% moisture
17 17
Source: USDA, Stillwater analysis
2 E85 and E15 Consumption Capacity
In EPA’s RFS proposal for 2022, they state: “We do not anticipate that growth in the use of higher ethanol
blends through 2022 will increase rapidly enough to result in significantly greater volumes of ethanol
consumption in the U.S.13” This is dubious since in recent years governmental efforts such as USDA’s
Blender Infrastructure Program and Higher Blends Infrastructure Incentive Program have significantly
added to E85 and E15 infrastructure. Much of this infrastructure will be underutilized if future RFS’s do not
encourage increased usage of E85 and E15.
2.1 E85
E85 consumption capacity is a function of two factors: (a) E85 dispensers and (b) E85-compatible vehicles.
2.1.1 E85 Dispenser Capability
The AFDC reports that there are currently 4,125 stations selling E85.14 A report by Stillwater in 2018
estimated that these E85 stations would average 1.8 dispensers per station in 2022 that are already in
service providing E85 and thus that are compatible with and approved for use with E85.15 Therefore, in
2022, there will be about 7,425 E85 dispensers.
Each dispenser can dispense a typical volume of 45,000 gallons per month of E85, containing 33,300
gallons of ethanol per month. Therefore, the maximum E85 throughput capacity of the 7,425 existing E85
dispensers is 4.0 billion gallons of E85 containing 3.0 billion gallons of ethanol. 16 17 EIA projects that 320
million gallons of E85 will be used in 2022 (See Figure 5, year 2022). Given 7,425 existing E85 dispensers,
there is, therefore, the existing capability to dispense an additional 3.68 billion gallons of E85 containing
2.72 billion gallons of ethanol. That would in turn contain about 2.36 billion incremental gallons of ethanol,
i.e., gallons beyond the ethanol in the E10 that would be replaced by the additional E85.
13 https://www.epa.gov/sites/default/files/2021-12/documents/rfs-2020-2021-2022-rvo-standards-nprm-2021-12-07.pdf, page 27 14 https://afdc.energy.gov/fuels/ethanol_locations.html#/analyze?country=US&fuel=E85 15 Potential Increased Ethanol Sales through E85 for the 2019 RFS, August 17, 2018, Prepared for Growth Energy by Stillwater
Associates LLC, Table 2
16 7,425 dispensers X 45,000 gallons per month X12 months = 4.0 billion gallons 17 4.0 billion gallons of E85 X .74 gallons of ethanol per gallon of E85 = 3.0 billion gallons of ethanol
Comments to EPA on 2020-2022 RFS Rule
9
Citing 2020 data, EPA states that there are only about 3,947 stations at the end of 2020.18 In an effort to
examine a more conservative case, we assume that there are only 3,947 stations and that each station
only has a single dispenser. In this case, the maximum E85 throughput capacity of these 3,947 dispensers
is 2.31 billion gallons of E85 containing 1.71 billion gallons of ethanol. 19 20 Again assuming that 320 million
gallons of E85 will be used in 2022, there would be the existing capability to dispense an additional 1.99
billion gallons of E85 containing 1.47 billion gallons of ethanol. That would in turn contain about 1.27 billion
incremental gallons of ethanol, i.e., gallons beyond the ethanol in the E10 that would be replaced by the
E85.21
Figure 5. E85 Consumption projected in AEO 202122
EPA’s own estimate of the use of E85 in 2022 implies even lower utilization of existing E85 distribution
infrastructure. EPA relies on three estimates of E85 usage from 2020: 297, 206, and 202 million gallons23.
Using the largest value of 297 million gallons implies a utilization rate of 13% if there are 3,947 dispensers.
This utilization rate would be 7.4 % if there are 7,425 dispensers.
Each 5-percentage point increase in utilization rate would provide in an additional 116 million gallons of
E85, containing 85 million incremental gallons of ethanol, if there 3,947 dispensers and would provide an
additional 200 million gallons of E85, containing 148 million incremental gallons of ethanol, if there 7,425
dispensers.
2.2 E85-compatible vehicles
Another constraint on E85 is that E85 can only be used in FFVs so FFV capacity to use E85 needs to be
examined. In 2022, there will be about 20.4 million E85-compatible vehicles based on EIA’s estimates from
AEO 2021. These vehicles could use 588 gallons per year of E85, containing 435 gallons of ethanol (based
on the projected number of vehicle miles driven).24
18 https://nepis.epa.gov/Exe/ZyPDF.cgi?Dockey=P1013KOG.pdf, page 192 19 3,947 dispensers X 45,000 gallons per month X12 months = 2.31 billion gallons 20 2.31 billion gallons of E85 X .74 gallons of ethanol per gallon of E85 = 1.71 billion gallons of ethanol 21 1.99 billion gallons of E85 X .64 gallons of ethanol per gallon of E85 = 1.27 billion gallons of ethanol 22 https://www.eia.gov/outlooks/aeo/
23 https://nepis.epa.gov/Exe/ZyPDF.cgi?Dockey=P1013KOG.pdf, page 38, Figure 1.7.1-2 24 If a typical vehicle goes 12,000 miles per year. 12,000 miles divided by 25.4 miles per gallon (US fleet average in footnote 6 above)
results in 472 gallons of E10 or 588 gallons of E85 needed each year., using the factor 1.22 to convert E10 to E85.
0.00
0.05
0.10
0.15
0.20
0.25
0.30
0.35
0.40
2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030
E85 Consumption
AEO 2021 E85 bgy
Comments to EPA on 2020-2022 RFS Rule
10
2.2.1 FFV
The latest Annual Economic Outlook from the EIA, AEO 2021, projects the number of ethanol-flex fueled
vehicles (FFVs) expected to be in use to be about 20.4 million25. Again, EIA estimates that about 320 million
gallons of E85 will be used in 2022, see Figure 5. When this is divided by the 20.4 million FFVs, it calculates
out to be 15.7 gallons of E85 per FFV per year. Given that the average FFV is expected to travel about
12,000 miles per year and could use an estimated 588 gallons per year of E85 per FFV, this shows that
there is a very large upside potential for E85 sales which can be reached with the existing E85
infrastructure.26,27 If the 20.4 million FFVs used only E85, the maximum existing consumption capacity of
E85 would be 12.65 billion gallons per year of E85 containing 9.38 billion of ethanol. Each 5% increment
of E85 usage by FFVs would use 588 million gallons of E85 containing about 376.32 million incremental
gallons of ethanol over E10.
Table 4 below shows that in 2022 15.565 bg of ethanol could be produced with no changes in the total
farmland used for growing corn. EPA states that their proposed RFS would use 13.788 bg of ethanol28.
This would leave 1.777 bg of ethanol which would be unused by the RFS. If that unused ethanol were
instead used in E85, a total of 2.401 bg of additional E85 could be produced. Dispensing those additional
gallons of E85 would increase total E85 consumption to 2.741 billion gallons, raising dispenser utilization
to 68.5%, with the 7,800 dispensers, or 137% (the dispensers would be completely utilized dispensing 1.99
bg of E85), with 3,700 dispensers. With this 2.741 bg of E85 being consumed, the FFV fleet would be
21.7% fueled with E85.
2.3 Combining E85 Infrastructure and FFVs
As noted, EIA, in Figure 5, projects E85 consumption of 320 million gallons in 2022, which calculates to an
average of 15.7 gallons of E85 use for each of the 20.4 million FFVs in operation in 2021. Making the
reasonable assumption that the FFV vehicles are distributed in proximity to the E85 stations, 320 mg for
the FFV fleet or 15.7 gallons per FFV represents only an 8.0% utilization rate of dispensers and 3%
utilization rate of vehicles. 29 30 So, there is clearly a huge opportunity to increase the use of E85 given
existing infrastructure. The main barrier is pricing of E85 relative to E10.
There is sufficient existing E85 station infrastructure to dispense 4.0 bgy of E85 (3.0 bgy of ethanol). There
are sufficient FFVs to consume this E85 volume (100% dispenser utilization) while filling FFVs with E85
only 35% of the time. The number of FFVs will not constrain major increases in E85 sales. As noted above,
even devoting all additional ethanol production to E85 use (2.741 billion gallons) would raise existingdispenser utilization to only 68.5% and FFV utilization to only 21.7%. It appears that market forces and
pricing will continue to be the key factors impeding significant E85 growth. If EPA desires increased usage
of ethanol, E85 sales offer more ethanol volume per gallon of fuel sold than any other option.
2.4 E15
2.4.1 E15 Dispensers
The Biofuels Infrastructure Program (BIP) had plans to install 4,880 blender pump dispensers capable of
handling E15/E85 in 1486 stations by the end of 201831. This calculates out to 3.3 dispensers per station.
Other programs such as Prime the Pump and Higher Blends Infrastructure Incentive Program have added
25 https://www.eia.gov/outlooks/aeo/data/browser/#/?id=49-AEO2021&cases=ref2021&sourcekey=0, Table 39 26 https://www.epa.gov/automotive-trends/highlights-automotive-trends-report 27 12,000 miles divided by 25.4 miles per gallon (US fleet average in footnote 6 above) results in 472 gallons of E10 or 588 gallons of
E85.
28 https://nepis.epa.gov/Exe/ZyPDF.cgi?Dockey=P1013KOG.pdf, page 51, Table 2.1-1 29 320 million gallons divided by 4.0 billion gallons of E85 capacity equals 8.0% utilization. 30 17 gallons divides by 588 gallons of E85 used by an FFV every year is 3% utilization 31 https://www.fsa.usda.gov/programs-and-services/energy-programs/bip/index, state table
Comments to EPA on 2020-2022 RFS Rule
11
additional stations and dispensers to reach the 2,300 stations that EPA references in the RIA.32 It will be
assumed that all these stations also average 3.3 dispensers per station.
2.4.2 E15 Infrastructure
E15 is a relatively new fuel and E15 station and sales data is not regularly reported by government
agencies. In the draft RIA for this proposal, EPA presents a figure which shows approximately 2300 E15
stations as of January 2021.33 Stillwater has estimated that that these stations will average 3.3 dispensers
per station offering E15 by 2022 (i.e., dispensers that are compatible with and approved for use with E15.34
These 7,540 dispensers can handle about 4.1 billion gallons of E15 containing 0.205 billion gallons of
ethanol can be dispensed in 2022. 35 36
Since E15 is not allowed the 1 psi waiver that applies to E10, the above station throughputs must be
adjusted for 8.5 months instead of the 12 months used above. This reduces the distribution potential for
E15 in 2022 to about 2.9 billion gallons containing 0.145 billion gallons of incremental ethanol above the
E10 it would replace. If this infrastructure handles 1.2 billion gallons of E15 in 2022, then 1.7 billion gallons
of E15 dispensing capacity is not being utilized. This represents 59% of the E15 capacity that is unutilized
and 41% utilization of E15 dispensing capacity.
EPA in the RIA expresses some concerns about E15. One concern is that all existing or new hardware
must be compatible with E15, new underground tanks must also be approved for E15, and it is expensive
for station owners to pay for upgrading their hardware. The second concern is retailer liability for E15
damage and the additional cost associated with verifying E15 compatibility. These concerns are irrelevant
to the analysis of the throughput capacity thus far, which is limited to the existing E15 infrastructure, i.e.,
infrastructure that is already acquired, compatible with E15 (by definition), and approved for use with E15.
And with respect to the expansion of E15 infrastructure, EPA’s concerns are misplaced or overstated. The
manufacturers of nearly all dispensers have warranted them for E15 for many years, and most new
dispensers made since 2012 have met the UL 87A requirements for E15. Because stations typically
upgrade their pumps at high volume stations every 12 years, the population of E15 stations will continue to
increase, and for little additional cost because the upgrade cycles would occur anyway regardless of a
desire to expand E15 infrastructure.
EPA raises concerns in the RIA page 196 that being classified as E15 compatible is not the same as being
approved for E15 use. EPA points in particular to underground tanks and pipes. This point did not bother
EPA when creating Figure 6 In any event, EPA’s concern about approval is very misleading and overstated.
First, E15 underground tanks are generally unnecessary for the expansion of E15 delivery capacity because
most E15 today is produced by blender pumps, which do not need E15-certified tanks. Indeed, EPA admits
this on page 195 of the RIA: “the majority of service stations offering E15 today do so through blender
pumps which can produce E15 on demand for consumers through the combination of E10 (or E0) and
E85…”37 Thus for most E15 stations the underground tank requirements for E15 storage do not apply. With
a blender pump there is no E15 in contact with pipes or tanks as the E15 is produced in the dispenser.
Second, although underground tanks are not upgraded as frequently as dispensers and can be more
expensive than dispensers, the additional burden of obtaining approval to use a new E15-compatible
underground system is negligible
2.4.3 E15 Compatible Vehicles
32 Figure 6.4.3-2: Number of Retail Service Stations Offering E15, page 196
33 https://nepis.epa.gov/Exe/ZyPDF.cgi?Dockey=P1013KOG.pdf, Figure 6.4.3-2: Number of Retail Service Stations Offering E15,
page 196
34 Potential Increased Ethanol Sales through E85 for the 2019 RFS, August 17, 2018, Prepared for Growth Energy by Stillwater
Associates LLC, Table 2
35 2,300 stations X 3.3 dispensers X 45,000 gal/disp/mo X 12 = 4.1 billion gallons
36 4.1 bg E15 X 0.05 gal etoh/ gal E15 = 0.205 bg etoh
37 Draft Regulatory Impact Analysis: RFS Annual Rules (EPA-420-D-21-002, December 2021), page 194
Comments to EPA on 2020-2022 RFS Rule
12
As Figure 6 shows the vehicle pool is approaching the point where EPA claims that more than 90% of the
vehicles are compatible with E15 (not counting FFVs, which can also use E15) and the noncompatible
vehicles use less than 5% of the fuel used by the vehicle fleet.38 As time moves on, these vehicles will
become a smaller and smaller problem. Even in 2022, the capability for most of the nation’s vehicles to
use E15 far exceeds the small volumes of E15 available for this use.
Figure 6. MY2001 Or Later Fraction of In-Use Vehicle Fleet and MY 2001 or Later Fraction of InUse Gasoline Consumption
Source: EPA
2.5 Combining E85 and E15 Capabilities
As can be seen from the E85 and E15 infrastructure analyses above in 2022 the additional ethanol
capabilities of the E85 infrastructure, 2.72 bgy of incremental ethanol, vastly exceeds the capabilities of the
current E15 infrastructure at 0.145 incremental bgy of ethanol per year. However, both sets of infrastructure
can contribute to the goal of using more ethanol in the form of E85 or E15 with a total of 2.865 bgy of
ethanol.
38 Draft Regulatory Impact Analysis: RFS Annual Rules (EPA-420-D-21-002, December 2021), page 199
Comments to EPA on 2020-2022 RFS Rule
13
3 E85 Pricing
In general, there is a lack of good data with information on E85 sales, production, and pricing. However,
the Clean Cities Alternative Fuels Price Report is a fair source for E85 pricing data.39 It is a quarterly report
that collects samples of E85 prices by U.S. Department of Energy (DOE) PADDs for 15 days each quarter.
These prices are averaged and then compared to E10 gasoline prices in the same PADD areas. The report
also adjusts the E85 prices for energy content relative to E10 on a gasoline gallon equivalent (GGE) basis.
The most recent report was for October 2021 and the E85 and gasoline average prices from that report are
shown in Table 5. The average E85 price nationally for October 2021 was $2.73 per gallon compared to
$3.25 for E10 gasoline. These prices and the difference in prices of -$0.52 are about the same as most of
the areas of the country except for New England and California. New England has an E85 price that is
$0.32 above gasoline and California E85 is priced nearly $1.00 below California reformulated gasoline.
This implies that in New England E85 sellers are not pricing competitively compared to gasoline on an
energy basis but in California the market pricing is very competitive to E10 gasoline.
Table 5. E85 and Gasoline Average Prices by Region October 2021
Region
E85
Prices
($/gal)
Gasoline
Prices
($/gal)
Price
Difference
New England $3.55 $3.23 $0.32
Central Atlantic $2.68 $3.16 -$0.48
Lower Atlantic $2.68 $3.08 -$0.40
Midwest $2.69 $3.08 -$0.39
Gulf Coast $2.52 $2.82 -$0.30
Rocky Mountain $3.05 $3.54 -$0.49
West Coast $3.35 $4.34 -$0.99
National Average $2.73 $3.25 -$0.52
The Clean Cities Alternative Fuels Price Report also converted the E85 prices to a gasoline equivalent price
by adjusting for the lower energy content of E85. A factor of 70% was used for adjusting the ethanol gallon
energy content to a gasoline energy content equivalent. It was also assumed that E85 had on average
70% ethanol. EIA uses 74% in its E85 calculations. These GGE prices are shown in Table 6.40 As can be
seen, in all cases the E85 fuel was priced higher than gasoline on an energy equivalent basis. California
alone seems to have E85 priced nearly equal to gasoline on an energy basis. The rest of the country has
E85 prices that are adjusted to reflect some of the reduced energy, but it appears that stations in these
areas are attempting to keep around half of the cost of the energy difference for themselves. This may
reflect that many states have mandates for FFV and E85 usage in state and government fleets and the E85
retailers adjust their prices only enough to maintain some portion of the non-mandated E85 market.
39 https://afdc.energy.gov/files/u/publication/alternative_fuel_price_report_october_2021.pdf, Table 8 40 https://afdc.energy.gov/files/u/publication/alternative_fuel_price_report_october_2021.pdf, Table 17a
Comments to EPA on 2020-2022 RFS Rule
14
Table 6. E85 and Gasoline Average Prices by Region (GGE) October 2021
Region
E85
Prices
($/GGE)
Gasoline
Prices
($/gal)
Price
Difference
New England $4.62 $3.23 $1.39
Central Atlantic $3.48 $3.16 $0.32
Lower Atlantic $3.48 $3.08 $0.40
Midwest $3.50 $3.08 $0.42
Gulf Coast $3.28 $2.82 $0.46
Rocky Mountain $3.96 $3.54 $0.43
West Coast $4.36 $4.34 $0.02
National Average $3.55 $3.25 $0.30
There is little doubt that 2020 and 2021 gasoline and E85 prices reflect the large drop in transportation fuel
demand caused by the Covid-19 epidemic, but the pandemic’s effect may have been similar with respect
to each type of fuel. In order to examine E85 pricing during a more normal time period, the same two tables
used above have been copied from the Clean Cities Alternative Fuels Price Report for October 2019.41 In
2019, the national average E85 price was $2.28 per gallon and the national average gasoline price was
$2.68 per gallon. These prices are lower than the October 2021 prices and reflect an average of $0.40 less
for E85 than for E10 gasoline. In 2019, New England had E85 priced above gasoline just as in 2021 and
California again had the lower E85 prices relative to gasoline.
Table 7. E85 and Gasoline Average Prices by Region for October 2019
Region
E85
Prices
($/gal)
Gasoline
Prices
($/gal)
Price
Difference
New England $2.85 $2.69 $0.16
Central Atlantic $2.34 $2.45 -$0.11
Lower Atlantic $2.26 $2.46 -$0.20
Midwest $2.18 $2.49 -$0.31
Gulf Coast $2.07 $2.25 -$0.18
Rocky Mountain $2.27 $2.73 -$0.46
West Coast $3.08 $3.93 -$0.85
National Average $2.28 $2.68 -$0.40
Table 8 shows the 2019 GGE adjusted prices for E85.42 The results are very similar to the results for 2021.
Again, California was the only region that came close to pricing E85 at parity with E10, but it still priced E85
above E10 on an energy-equivalent basis.
41 https://afdc.energy.gov/files/u/publication/alternative_fuel_price_report_oct_2019.pdf, Table 7 42 https://afdc.energy.gov/files/u/publication/alternative_fuel_price_report_oct_2019.pdf, Table 17a
Comments to EPA on 2020-2022 RFS Rule
15
Table 8. E85 and Gasoline Average Prices by Region (GGE) October 2021
Region
E85
Prices
($/GGE)
Gasoline
Prices
($/gal)
Price
Difference
New England $3.70 $2.69 $1.01
Central Atlantic $3.04 $2.45 $0.59
Lower Atlantic $2.94 $2.46 $0.48
Midwest $2.83 $2.49 $0.34
Gulf Coast $2.69 $2.25 $0.44
Rocky Mountain $2.95 $2.73 $0.22
West Coast $4.00 $3.93 $0.07
National Average $2.96 $2.68 $0.28
From looking at Tables 6 and 8, it appears that since 2019 and even through the Covid epidemic, E85 has
consistently been priced above E10 gasoline on an energy equivalent basis on average in all regions of the
country. If we assume that E85 purchasers are completely rational, it is obvious why utilization of E85 has
not been higher: consumers recognize that it is too expensive. By and large, only price-insensitive E85
purchasers, such as those that are mandated to use E85 in their state, or through Federal mandate, or for
fleet fueling for FFVs, purchase E85. Additionally, there are likely some consumers that are committed to
E85 for non-economic reasons, who are also relatively price-insensitive; these can include corn farmers,
those involved in ethanol production, and environmentalists. These groups of price-insensitive consumers
have likely been the primary market for E85 retailers so far. Because they are few in number and not
growing rapidly, the use of E85 has also not been growing much. In order to substantially increase the use
of E85, it must be marketed to general consumers, who are highly price sensitive when purchasing fuel.
And therefore, E85 must be priced at or below parity with E10 on an energy-equivalent basis. Put another
way, retailers have been marketing E85 as a premium niche product, seeking high margins on low volume,
but to take advantage of the infrastructure to markedly increase E85 use, they would need to market it as
a mass-market low-cost, high-volume commodity.
In sum, the only real barrier to increasing the use of E85, and thereby the concentration of ethanol in the
nation’s fuel supply, is the relative pricing of E85 to E10. E85 has consistently been priced more expensive
than E10 and most consumers are highly sensitive to this premium. The solution, then, is to lower the
relative price of E85 to E10. The RFS provides a direct mechanism to achieve this: higher RFS standards
tighten the supply of RINs, increasing their price. Higher RIN prices translate into greater discounts relative
to E10. And because E85 has such a higher concentration of ethanol, it can achieve a greater discount
through higher RIN prices.
4 Encouraging Increased Use of Higher-Ethanol Blends
The RFS is designed to allow the lowest priced renewable biofuel to meet the demand for the conventional
renewable biofuel requirement. Many parties believe that the conventional renewable biofuel category is
designated to provide a mandate for ethanol D6 RINs but this category can be met by using any renewable
biofuel allowed under the RFS. As a result, meeting this category’s requirements is based on acquiring the
least expensive marginal RINs.
The marginal RIN is defined as the type of RIN which provides the lowest cost option for RFS-Obligated
Parties to comply with the last increment of their annual obligation. This builds upon the assumption that
an obligated party will seek to comply with its obligations in the most cost-effective manner available to it;
thus, it will use the overall lowest cost option first to comply with as much of its obligation as possible and
then, successively, utilize higher cost options in order of increasing cost to the extent needed to satisfy its
total obligation for a year. As the RFS is comprised of four nested obligations, assessment of the marginal
RIN is more complex than the assessment of the marginal supplier in most other market contexts.
Comments to EPA on 2020-2022 RFS Rule
16
Identifying the marginal RIN is useful as a tool for predicting the market response to potential changes to
RFS policies.
Without the value of the D6 RIN export ethanol is more expensive than domestic RFS ethanol, so that
anytime ethanol is desirable for RFS compliance, the market will first divert from exports. Put differently,
anyone choosing to export U.S. ethanol chooses to lose the value of the D6 RIN he could get if he could
use the ethanol domestically. This is important because it means that we have 1.3-1.5 billion or more gallons
of ethanol, with existing land use and production capacity, ready to use domestically if the right price signals
are sent through the RFS43 44. There is no need to worry about marginal land use effects or marginal price
effects for corn or food.
5 Market Challenge in Meeting Original 2020 Percentage Standards
EPA originally finalized the annual volume obligations and the corresponding percentage standards for the
four RFS renewable fuel categories in February 2020.45 The volumes and corresponding percentage
standards are presented in Table 9 below. As each Obligated Party’s obligations under the RFS are set by
applying the percentage standards to the sum of their gasoline and diesel supply to the 49-state RFS
market,46 the finalization of the 2020 percentage standards enabled them to calculate their obligations and
act to acquire all RINs needed for compliance with each their year-to-date obligations every day for the rest
of 2020. Accordingly, when Obligated Parties closed their books for 2020, they were able to precisely know
their obligations and take appropriate actions to secure any remaining shortfall in their required number of
2019 and 2020 RINs.
Table 9. RFS Final 2020 Volume Standards and Percentage Standards, February 2020
Fuel Category
2020 Final Volume
Standard (billion
gallons)
2020 Final Percentage
Standards
Cellulosic biofuel 0.59 0.34%
Biomass-based diesel 2.43* 2.10%
Advanced biofuel 5.09 2.93%
Renewable fuel 20.09 11.56%
Undifferentiated Advanced Biofuels (Implied) 1.445 0.83%
Conventional Biofuels (Implied) 15.00 8.63%
*Established in the 2019 final rule (83 FR 63704, December 11, 2018)
In calculating the final percentage standards for 2020, EPA used data provided by EIA in their Short-Term
Energy Outlook (STEO) for October 2019 and estimated the volume of small refinery volumes to be granted
exemptions based on actual exemptions granted annually for 2016 through 2018. These volumes, originally
published in Table VII.C-1 of the 2020 Final Rule are reproduced in Table 10 below along with the revised
volumes in the December 2021 proposed rule.
43 https://www.fas.usda.gov/ethanol-2020-export-highlights
44 https://www.fas.usda.gov/ethanol-2019-export-highlights
45 Renewable Fuel Standard Program: Standards for 2020 and Biomass-Based Diesel Volume for 2021 and Other Changes, 85 FR
7016, February 6, 2020. 46 Gasoline and diesel supplied to Alaska do not generate RFS obligations.
Comments to EPA on 2020-2022 RFS Rule
17
Table 10. Values for Terms in Calculation of the Final 2020 Standards (billion gallons)
Value for 2020 Standards
Term Description February 2020
Final Rule
December 2021
Proposal
RFVCB Required volume of cellulosic biofuel 0.59 0.51
RFVBBD Required volume of biomass-based diesel a 2.43 2.43
RFVAB Required volume of advanced biofuel 5.09 4.63
RFVRF Required volume of renewable fuel 20.09 17.13
G Projected volume of gasoline 142.68 123.25
D Projected volume of diesel 55.30 50.49
RG Projected volume of renewables in gasoline 14.42 12.63
RD Projected volume of renewables in diesel 2.48 2.15
GS Projected volume of gasoline for opt-in areas 0 0
RGS Projected volume of renewables in gasoline for opt-in areas 0 0
DS Projected volume of diesel for opt-in areas 0 0
RDS Projected volume of renewables in diesel for opt-in areas 0 0
GE Projected volume of gasoline for exempt small refineries 4.24 0.00 – 4.80
DE Projected volume of diesel for exempt small refineries 3.02 0.00 – 3.39 a The BBD volume used in the formula represents physical gallons. The formula contains a 1.5 multiplier to convert this physical
volume to ethanol-equivalent volume.
In their proposed rule for volume requirements for 2020, 2021, and 2022, EPA proposes to revise the final
volumes and percentage standards for 2020 on the basis that – 47
Since we promulgated those standards, several significant and unanticipated events occurred that affected
the fuels markets in 2020. The two most prominent of these events were:
The COVID–19 pandemic and the ensuing fall in transportation fuel demand, especially the
disproportionate fall in gasoline demand relative to diesel demand, which significantly reduced
the production and use of biofuels in 2020 below the volumes we anticipated could be achieved,
and
The potential that the volume of gasoline and diesel exempted from 2020 RFS obligations
through small refinery exemption (SREs) will be far lower than projected in the 2020 final rule.48
To evaluate the impact of these factors on obligated parties, we compare EPA’s February 2020 Final
Volume obligations for each of the four RFS categories with the volume obligations calculated by applying
2020 actual volumes to the finalized 2020 Percentage Standards (with and without inclusion of EPA’s
current estimates for exempted gasoline and diesel volumes) and to EMTS data for net 2020 RIN
generation and 2020 RIN Separations (excluding separation for export). 49 50 51 These comparisons, which
do not include any usage of carryover RINs, are presented in Table 11 below. This analysis shows that the
inherent flexibilities in the RFS due to the use of percentage standards, the availability of cellulosic waiver
credits (CWCs) to meet shortfalls in cellulosic biofuel availability, and the nested structure of the RFS
obligations means that actual separations of 2020 vintage RINs were sufficient to sufficient to satisfy the
cellulosic biofuel, biomass-based diesel, and advanced biofuel obligations for 2020 as previously
finalized.52 Additionally, net RIN generation for the Renewable fuels category was more than sufficient to
enable compliance with the previously finalized 2020 Renewable Fuel obligation if EPA’s estimate for small
refinery exemptions (SREs) in 2020 is accurate, thus use of carry-over RINs to cover the shortfall in 2020
conventional RIN separation is appropriate.
47 Renewable Fuel Standard (RFS) Program: RFS Annual Rules, 86 FR 72436, December 21, 2021. 48 RFS Annual Rules, Sub-Section I.B., December 21, 2021 49 4.80 billion gallons of gasoline and 3.39 billion gallons of diesel as indicated in the rightmost column of Table 6. 50 Defined by EPA as the total number of RINs generated minus the number of invalid RINs generated. 51 RINs associated with export volumes are not eligible for use in meeting RFS volume obligations. EIA data indicates that 1,317
million gallons of ethanol were exported from the U.S. in 2020. These would have been available to supply the domestic U.S. market
if required for RFS compliance.
52 While a portion of 2020 vintage RIN separations occurred in 2021, this analysis has not taken credit for the portion of 2019 vintage
RIN separations which occurred in 2020.
Comments to EPA on 2020-2022 RFS Rule
18
Table 11. Volume Obligations (2020 Final Rule) Compared to the Actual Volume Obligations (with
and without EPA’s estimate of SREs) and RIN Generation
Fuel Category
2020 Final Volume
Standard (billion
gallons)
Actual Volume
Obligations with 2020
Percentage
Standards, no SRE
Actual Volume
Obligations with
2020
Percentage
Standards, est
SRE
2020 Net
RIN
Generation
2020 RIN
Separation
Cellulosic biofuel 0.59 0.54 0.51 0.51 0.50
Biomass-based diesel 2.43 2.15 2.04 2.88 2.48
Advanced biofuel 5.09 4.66 4.42 5.33 4.66
Renewable fuel 20.09 18.38 17.43 18.26 17.06
Source: EIA, Stillwater analysis
To understand the influence of EPA’s estimate for SREs in their calculation of the 2020 Final Percentage
Standards in February 2020, we recomputed these values with the gasoline and diesel exempt volume
projections both set to zero. This effectively increases the denominator used in calculating the percentage
standards. This revised calculation (“2020 Percentage Standards, no SREs”) is compared to the February
2020 calculation (“2020 Final Percentage Standards”) in Table 12 below.
Table 12. Recalculation of 2020 Percentage Standard with no SREs
Fuel Category
2020 Final Volume
Standard (billion
gallons)
2020 Final
Percentage
Standards
2020
Percentage
Standards, no
SREs
Applicable Volume (bgal) — 173.82 181.08
Cellulosic biofuel 0.59 0.34% 0.33%
Biomass-based diesel 2.43 2.10% 1.34%
Advanced biofuel 5.09 2.93% 2.81%
Renewable fuel 20.09 11.56% 11.09%
Undifferentiated
Advanced Biofuels
(Implied)
1.445 0.83% 0.80%
Conventional Biofuels
(Implied) 15 8.63% 8.28%
Source: EPA, Stillwater analysis
These alternative percentage standards are next used to compute what the 2020 volumes would have been
had EPA accurately projected both transportation fuel use and SREs for 2020. These calculations are
presented below in Table 13 and can be compared with Table 12 above. With this alternative calculation,
it can be seen that cellulosic biofuels RIN generation (0.51 billion) and RIN separation (0.50 billion) both
fell slightly short of the actual volume obligation of 0.52 billion RINs. RIN separations in this calculation
exceed the actual volume obligations for both biomass-based diesel and advanced biofuels. Renewable
fuels separations, at 17.06 billion fall short of the actual volume obligation of 17.64 billion. However, we see
that 2020 net RIN generation, at 18.26 billion significantly exceeded the actual volume obligation – this tells
us that sufficient RINs were generated to meet the 2020 standards as originally finalized if the EPA’s 2020
estimation of SREs is removed from the analysis.
Comments to EPA on 2020-2022 RFS Rule
19
Table 13. 2020 Actual Volume Obligations calculated with revised percentage standards (with and
without EPAs current estimate of SREs)
Fuel Category
2020 Final
Volume
Standard
(billion gallons)
Actual Volume
Obligations with
2020 Percentage
Standards, no SRE
Actual Volume
Obligations with
2020 Percentage
Standards, est SRE
2020 Net
RIN
Generation
2020 RIN
Separation
Cellulosic biofuel 0.59 0.52 0.49 0.51 0.50
Biomass-based diesel 2.43 2.06 1.96 2.88 2.48
Advanced biofuel 5.09 4.47 4.24 5.33 4.66
Renewable fuel 20.09 17.64 16.73 18.26 17.06
Undifferentiated
Advanced Biofuels
(Implied)
1.445 1.27 1.20 0.29 0.31
Conventional Biofuels
(Implied) 15 13.17 12.49 12.93 12.40
Source: EIA, Stillwater analysis
5.1 RFS Impacts on Imports and Exports of Petroleum and Ethanol
To examine the impact of the RFS on imports and exports of petroleum (crude oil and products) and ethanol,
we will review data available from EIA. A key element of this analysis begins with EIA’s assessment of U.S.
ethanol production capacity and production. These data are summarized in Table 14 below. EIA reports
U.S. fuel ethanol production capacity annually as of January 1st of each year; the most recent data available
are for January 1, 2021.53 EIA also reports monthly fuel ethanol production.54 The ratio of the two gives us
the annual capacity utilization which we measure in percent; the average utilization achieved in the years
immediately preceding Covid-19, 2018 and 2019, was 95%. Assuming that the industry can regularly
operate at 95% of capacity and that U.S. ethanol capacity remains unchanged from the 17.546 billion
gallons per year cited by EIA at the beginning of 2021, this suggests that the U.S. ethanol industry has the
capability to produce at a rate of 16.669 billion gallons per year if sufficient supplies of corn are available
and if there is demand for the product.55
Table 14. U.S. Annual Ethanol Production and Capacity
Year Annual
Capacity
(million gallons)
Monthly
Capacity
(million gallons)
Annual Production
(million gallons)
Annual Utilization
(%)
2018 16,542 1,379 16,091 97%
2019 16,908 1,409 15,778 93%
2020 17,378 1,448 13,941 80%
2021 (estimated)56 17,546 1,462 14,716 84%
Potential 202257 17,546 1,462 16,669
(@ 95% utilization)
95% (assumed)
Source: EIA, Stillwater analysis
Figure 7 below illustrates the monthly supply and demand for U.S. ethanol from 2018 through October
2021. As seen in this chart, monthly production (orange line) averaged 95% of capacity (blue line) in 2018
and 2019. Capacity utilization dropped to a low of 49% in April 2020 due to the impacts of the pandemic
and operating rates have only recovered to 84% of capacity in 2021 (based on data for January through
October). Domestic demand (green bars) for March through October of 2021 has nearly recovered to preCovid levels of about 1.2 billion gallons per month while export volumes (light gray bars) continue to be
below pre-pandemic levels. The black line on the chart corresponds to the finalized implied conventional
biofuels obligation of 15 billion gallons per year (1250 million gallons per month) for 2018 through 2020;
53 https://www.eia.gov/petroleum/ethanolcapacity/
54 https://www.eia.gov/dnav/pet/pet_pnp_oxy_dc_nus_mbbl_m.htm 55 https://www.eia.gov/petroleum/ethanolcapacity/
56 Annual production estimated based on data through October 2021. 57 Assumes no change in operable capacity since January 1, 2021.
Comments to EPA on 2020-2022 RFS Rule
20
while domestic ethanol demand in 2020 was clearly below this level, pre-pandemic ethanol demand
averaged near this target.58 The dashed red line represents the 2020 and 2021 volumes in EPA’s December
2021 proposal. In summary, these data demonstrate that the U.S. ethanol industry has significant unused
capacity which could be used to supply as much as about 16.7 billion gallons per year of ethanol for the
U.S. fuels market.59
Figure 7. Monthly U.S. Ethanol Supply, Demand and RFS Volume Standards (2018-2021)
Source: EIA, Stillwater analysis
Exports continue to play a major role in U.S. petroleum and ethanol markets. Figure 8 below illustrates the
recent trends in U.S. exports of crude oil, petroleum products, and ethanol. U.S. crude oil exports prepandemic were increasing steadily from over 42 million barrels per month in January 2018 to a high of 112
million barrels in March 2020; since the onset of the pandemic, they have varied in the range of about 80
to 100 million barrels per month. U.S. exports of petroleum products (including gasoline, diesel fuel, and
other products) generally ranged around 100 million barrels per month pre-pandemic, dropped to a low of
58 million barrels in May 2020, and then rapidly recovered to a range of 80-90 million barrels per month
since mid-2020. In contrast, U.S. ethanol exports are considerably smaller and have been generally
declining since 2018 – from an average of 3,400 barrels per month in 2018 to an average of 2,300 barrels
per month in 2021 (January through October). This decline in ethanol exports since before the pandemic is
attributable to restrictive trade policies implemented by China and Brazil, two of the largest markets for U.S.
ethanol exports.
58 The actual RFS obligation scales with annual gasoline and diesel demand each year. As discussed in Section 5 page 25 of this
report, this regulatory scaling mechanism effectively eliminates any need for EPA to reduce the obligation volumes for 2020 as they
currently propose.
59 Based on current nameplate capacity of 17.5 billion gallons per year and the 95% sustained capacity utilization achieved in 2018
and 2019.
Comments to EPA on 2020-2022 RFS Rule
21
Figure 8. Monthly U.S. Exports of Crude Oil, Petroleum Products, and Ethanol (2018-2021)
Source: EIA, Stillwater analysis
Looking at U.S. Imports, as shown in Figure 9 below, shows a somewhat different pattern than U.S. exports
reviewed above. Crude oil imports are roughly double the amount of crude oil exports, with recent highs of
250 million barrels per month during the summer of 2028 followed by a gradual decline to a recent low of
155 million barrels in September 2021 because of growing U.S. production and exports and the loss in
domestic demand with the pandemic. Since September 2021, crude imports have risen to recent levels
near 200 million barrels per month. During this time period, U.S. gasoline and diesel imports have been
relatively steady, averaging nearly 67 million barrels per month and 6.5 million barrels per month,
respectively. U.S. ethanol imports during this time period have been small and irregular, comprised primarily
of opportunistic imports of Brazilian sugarcane ethanol during Brazil’s harvest season and directed primarily
towards California’s LCFS market where the low carbon intensity is highly valued. Over the time frame
shown, ethanol imports averaged less than 0.2 million barrels per month.
Comments to EPA on 2020-2022 RFS Rule
22
Figure 9. Monthly U.S. Imports of Crude Oil, Gasoline, Diesel, and Ethanol (2018-2021)
Comparing exports and imports, the U.S. remains a large net importer of crude oil and a net exporter of
petroleum products. The U.S. is also the world’s largest ethanol exporter with net exports declining in recent
years due to the combined effects of trade frictions with China and Brazil and the impacts of Covid-19. With
the U.S. ethanol industry operating well below capacity, it has the capability of significantly increasing
supply to U.S. and international markets if current regulatory and trade barriers were relaxed.
Based on the above analysis, increasing the implied RFS obligation for conventional biofuels would be
expected to increase demand for domestic ethanol production. As U.S. ethanol plants are currently
operating below demonstrated capacity, this increase can be accommodated without the need for the U.S.
to reduce its current level of exports. As U.S. ethanol imports are currently small, opportunistic, and driven
by LCFS value it is unlikely that increased RFS obligations would result in material increases of ethanol
imports. With higher levels of ethanol in the U.S. gasoline pool, petroleum refineries would be expected to
re-optimize between lower rate (reducing U.S. crude imports) and increased net exports of petroleum
products (reducing crude demand in the rest of the world) depending on short-term market conditions.
Regardless of how the U.S. refining industry were to re-optimize, global GHG emissions would be reduced.
6 EPA Cost Analysis of Ethanol Costs
In the RIA, EPA has projected an increased cost for ethanol in 2022.60 Table 7.4-1 in the RIA projects a 3%
increase in corn ethanol prices for 2022 and Table 7.5-1 sets the price increase at $0.14 per bbl of corn.
Historically, the process of producing corn and ethanol have gotten more efficient each year. If this is
coupled with a drop in gasoline demand and the resulting drop in demand for blending ethanol, it would be
expected for corn and ethanol prices to drop slightly rather than increase as EPA is proposing.
60 Draft Regulatory Impact Analysis: RFS Annual Rules (EPA-420-D-21-002, December 2021), page 214
ActiveUS 192803838v.1
Growth Energy Comments on EPA’s
Proposed Renewable Fuel Standard Program:
Renewable Fuel Standard Annual Rules
Docket # EPA-HQ-OAR-2021-0324
Exhibit 6
Infrastructure Changes and Cost to Increase
Consumption of E85 and E15 in 2017
Prepared for
Growth Energy
By
Stillwater Associates LLC
Irvine, California, USA
July 11, 2016
Stillwater Associates
3 Rainstar Irvine, CA 92614 – Tel (949) 653 5899 – Fax (949) 786 4395 – stillwaterassociates.com
Infrastructure Changes and Cost to Increase RFS Ethanol Volumes through Increased E15 and
E85 Sales in 2017
i
Disclaimer
Stillwater Associates LLC prepared this report for the sole benefit of Growth Energy.
Stillwater Associates LLC conducted the analysis and prepared this report using reasonable care
and skill in applying methods of analysis consistent with normal industry practice. All results are
based on information available at the time of presentation. Changes in factors upon which the
report is based could affect the results. Forecasts are inherently uncertain because of events that
cannot be foreseen, including the actions of governments, individuals, third parties and
competitors. NO IMPLIED WARRANTY OF MERCHANTABILITY SHALL APPLY.
Infrastructure Changes and Cost to Increase RFS Ethanol Volumes through Increased E15 and
E85 Sales in 2017
ii
Table of Contents
Executive Summary ………………………………………………………………………………………………………….. 1
1 The Objective of the Study …………………………………………………………………………………………. 2
2 E85 Analysis …………………………………………………………………………………………………………….. 2
2.1 Case 1: Incremental Ethanol Consumption Through Existing E85 Infrastructure ………… 2
2.2 Case 2: Expanding Infrastructure to Deliver E85 in 2017 …………………………………………. 4
2.3 Transporting Additional E85 …………………………………………………………………………………. 6
3 Economics of E85 Infrastructure Changes ……………………………………………………………………. 7
3.1 Single Station Single Dispenser Economics …………………………………………………………… 7
4 How to Increase Sales at Existing E85 Stations ……………………………………………………………. 9
4.1 Customer Segmentation ……………………………………………………………………………………… 9
4.2 Gross Margins ………………………………………………………………………………………………….. 12
4.3 What RIN Price is Needed for Short Term Volume Growth …………………………………….. 16
4.4 Impact of Different Demand Curves on Optimum Gross Margin ……………………………… 18
5 E15 Analysis …………………………………………………………………………………………………………… 19
5.1 Expanding Infrastructure to Deliver E15 in 2017 …………………………………………………… 19
5.2 The Time since the Last Dispenser Replacement is Important ……………………………….. 19
5.3 Station Costs to Upgrade to E15 ………………………………………………………………………… 20
5.4 Costs for the Blender Pump Option …………………………………………………………………….. 22
5.5 The Phase-In for 2017 ………………………………………………………………………………………. 22
5.5.1 E15 dispenser economics ……………………………………………………………………………. 22
5.6 The Upgrade Cost Until E15 is Available at the Terminal ……………………………………….. 24
List of Tables
Table 3.1 – Single Dispenser Economics for addition of second E85 Dispenser at Existing E85
station on replacement cycle …………………………………………………………………………………………….. 7
Table 3.2 – Single Dispenser Economics for addition of a second E85 Dispenser at Existing E85
station off replacement cycle …………………………………………………………………………………………….. 7
Table 3.3 – Single Dispenser Economics for addition of a new E85 Dispenser at Existing E10
station off replacement cycle …………………………………………………………………………………………….. 8
Table 3.4 – Single Dispenser Economics under scenario 3 where the new E85 dispenser added at
the previously E10-only station has 50% dispenser utilization ……………………………………………….. 8
Table 3.5 Single Dispenser Economics for addition of a new E85 Dispenser at Existing E10
station (off replacement cycle) with 10% dispenser utilization ……………………………………………….. 9
Table 4.1 – Estimates of Customer Breakdown …………………………………………………………………. 10
Table 5.1 – Station Costs to Upgrade to E15 – Two Gasoline Tank Station …………………………… 21
Table 5.2 – Station Costs to Upgrade to E15 – Three Gasoline Tank Station ………………………… 21
Table 5.3 – Station Costs to Upgrade to E15 – Two Gasoline Tank Station …………………………… 23
Table 5.4 – Rate of Return on Investment ………………………………………………………………………… 23
Table 5.5 – Station Costs to Upgrade to E15 – Three or More Gasoline Tank Station …………….. 23
Table 5.6 – Rate of Return on Investment ………………………………………………………………………… 24
Table of Figures
Figure 4.1 – E85 Site Volume, One Dispenser ………………………………………………………………….. 11
Figure 4.2 Comparing Stillwater and Brattle Demand Curves ……………………………………………… 12
Figure 4.3 – E85 Gross Margin Estimates ………………………………………………………………………… 13
Figure 4.4 – Calculated Combined Supplier/Dealer E85 Gross Margin …………………………………. 14
Figure 4.5 – Optimum E85 Price Point vs. RINs Price ………………………………………………………… 15
Figure 4.6 – RINS Price that Causes E85 to be Priced at 30% Discount to E10 (Ethanol Price) . 16
Figure 4.7 – RINS Price that Causes E85 to be Priced at 30% Discount to E10 (RBOB Price) … 16
Figure 4.8 – RINs Price Required to Increase E85 Sales by a Factor of 5 …………………………….. 17
Infrastructure Changes and Cost to Increase RFS Ethanol Volumes through Increased E15 and
E85 Sales in 2017
iii
Figure 4.9 – Comparison of Stillwater and Brattle Log-Log Curves ………………………………………. 18
Infrastructure Changes and Cost to Increase RFS Ethanol Volumes through Increased E15 and
E85 Sales in 2017
1
Executive Summary
Congress legislated both the 2005 Renewable Fuel Standard (RFS) and the updated 2007
Standard (RFS2) as a mechanism to mandate the phasing in of renewable biofuels into U.S.
transportation fuels. On an annual basis, the administering agency, the U.S. Environmental
Protection Agency (EPA), is expected to propose and finalize new volume obligations for the four
RFS categories of cellulosic biofuels; advanced biofuels, biomass-based diesel and total
renewable biofuels. Ethanol has become the predominant biofuel used to meet three of the four
RFS2 categories. Ethanol can be used in transportation fuel when it is blended with gasoline at
various levels. The most popular of these has been E10, which is 10 percent ethanol and 90
percent petroleum blendstocks. Ethanol can also legally be blended as E15, a blend of up to 15
percent ethanol, or as E85. E85 can contain 51 to 83 percent ethanol blended with petroleum
blendstocks or natural gasoline. E85 can only be used in Flexible Fueled Vehicles (FFVs). FFVs
comprise about eight percent of the nation’s transportation vehicle fleet.
As the RFS2 mandates for ethanol have risen, the nation has begun to approach the so-called
E10 blendwall, that point at which nearly all of the nation’s gasoline has been blended at the 10
percent ethanol level. To get around the E10 blendwall, it is necessary to find pathways to blend
more than 10 percent ethanol into ever larger portions into the nation’s gasoline pool. E15 and
E85 are the primary pathways to increase ethanol consumption beyond 10%.
In its latest RFS2 proposal for 2017, EPA has proposed standards that result in modest increases
in ethanol usage but has discounted the additional contribution from E85 and E15. Growth
Energy has requested that Stillwater Associates examine the distribution infrastructure for
pathways to potentially increase the supply of E15 and E85 at the retail station level. Stillwater
has considerable experience in the transportation fuels distribution space.
Stillwater evaluated the current state of fuels distribution, from the supply source though the
pipeline and terminal network to the service station and to the consumer. For E85, Stillwater
found that there are enough E85 stations and E85 dispensers in the U.S. to substantially increase
the volumes of ethanol used in transportation fuels. The simplest case where E85 throughput is
increased in the roughly 3,100 existing E85 stations with no new hardware required can increase
E85 sales by 1.674 billion gallons per year (bgy) and increase ethanol usage by 1.108 bgy if EPA
would only provide sufficient economic incentives to current FFV owners using E10. This is very
low hanging fruit in terms of increasing renewable fuels usage.
Stillwater analyzed the reasons for the current low consumption of E85 and found that E85 needs
to sell below its energy parity value compared to E10 in order to increase sales to price conscious
E10 consumers. Stillwater found that EPA’s recently established and currently proposed RFS
renewable standards fall short of providing a sufficient driving force to increase D6 RIN value to
the point where E85 prices can be set far enough below energy parity with E10 to establish a
tipping point where larger E85 sales volumes enable even lower E85 prices to the consumer.
Stillwater also found that ethanol volumes can be increased significantly through the use of E15
or E85 by making relatively modest investments to expand the infrastructure for delivering E85 or
E15.
Infrastructure Changes and Cost to Increase RFS Ethanol Volumes through Increased E15 and
E85 Sales in 2017
2
1 The Objective of the Study
On May 31, 2016, the EPA issued a notice of proposed rulemaking on the 2017 Renewable Fuel
Standards and the biomass-based diesel standard for 2018. For 2017, EPA is proposing
standards based on an assumption that the maximum reasonably achievable volume of ethanol
usage is approximately 0.2 bgy above the E10 blendwall most of which is E85. Growth Energy
has requested that Stillwater Associates evaluate whether more volumes of incremental ethanol
are reasonably achievable through E85 and E15 if EPA were to require additional ethanol above
the E10 blendwall through implementation of the RFS.
In this report, Stillwater assesses the ability of the fuel system to deliver greater volumes of E85
right now. We then analyze the potential pathways for expanding infrastructure for selling E85 or
E15. Stillwater prioritizes low-cost solutions for expansion. Stillwater also analyzes the financial
dynamics of the market to determine what kind of incentives are needed to spur the necessary
investment in upgraded infrastructure, and develops a market segmentation model that
illuminates what is needed from RFS volume requirements and the RIN market to create those
incentives.
Stillwater did NOT examine the actual production capacity of ethanol manufacturing facilities but
will assume that sufficient domestic production is available to fulfill the incremental supply.
Additionally, Stillwater did assume that model year 2001 and later U.S. automobile and truck
fleets are capable of using E15 and that original equipment manufacturer warranty issues will not
impede renewable fuel consumption. Support, or lack thereof, from the oil industry is assumed to
be out of scope for the purposes of this report.
2 E85 Analysis
In the E85 portion of this analysis, Stillwater first identifies the potential increases in E85 sales
volumes through existing stations. We then assess the cost of expanding E85 distribution
capacity by additional E85 dispensers at existing E85 stations and at E10-only stations, and
estimate the magnitude of the possible expansion. Next Stillwater analyzes the investment costs
in terms of rates of return and the need for increased margins from the point of adding a single
new dispenser. The margins required to achieve desirable rates of return are minimal if the new
dispenser is fully utilized but they increase if the dispenser has low E85 throughput, suggesting a
strong incentive for high RIN prices and high corresponding E85 discounts. Then Stillwater
models the behavior of several segments of E85 customers and discovers that E85 has seldom
been priced sufficiently below energy parity with E10 to attract price-sensitive E10 customers,
which constitute by far the largest segment of the market. It appears that there is a tipping point in
E85 price below which E85 sales volumes can increase rapidly. Finally, Stillwater discusses the
ethanol-E85 supply chain and how RINs and ethanol price reductions move through the supply
chain.
2.1 Case 1: Incremental Ethanol Consumption Through Existing E85 Infrastructure
Existing infrastructure is capable of delivering volumes of E85 far beyond what EPA has
proposed. The ability to deliver E85 is a function of three factors:
1. The number of E85 stations;
2. Dispenser throughput; and
3. The location of stations relative to vehicles that can use the fuel, i.e., flex-fuel vehicles
(“FFVs”).
We address each in turn.
Infrastructure Changes and Cost to Increase RFS Ethanol Volumes through Increased E15 and
E85 Sales in 2017
3
Stations. According to EPA, there were 3,126 E85 stations in the United States as of March
2016.1
By the time 2017 begins, that figure will certainly be higher as a result of additional
upgrades and various programs targeted to increase the availability of E85, such as BIP (USDA’s
Biofuel Infrastructure Partnership) and the “Prime the Pump” program. EPA notes that BIP is
expected to have added 1,486 E85 stations by the end of 2016. Therefore, we can assume that
there will be at least 4,612 E85 stations at the start of 2017. But to make our analysis extremely
conservative, we will assume that there are 3,100 E85 stations at the start of 2017. Further, that
figure will undoubtedly increase over the course of 2017. We address the potential for
infrastructure expansion in 2017 later; for now, and again to develop the most conservative
analysis, we will assume for present purposes that the number of E85 stations does not increase
during 2017 but rather remains at 3,100 for the entire year.
Dispenser throughput. In an influential study entitled “Feasibility and Cost of Increasing U.S.
Ethanol Consumption Beyond E10,” leading researchers Bruce Babcock and Sebastien Pouliot
examined an E85 service station in Minnesota and found that it sold almost 50,000 gallons in one
month.2
Accordingly, they assume that the average E85 station can deliver 45,000 gallons of E85
per month. That assumption accords with a rule of thumb in gasoline marketing that the average
station will sell two million gallons of fuel per year with four dispensers (2 hoses each). That
standard converts to just about 42,000 gallons per month per dispenser.
More careful analysis confirms Babcock and Pouliot’s finding and the rule of thumb, but further
shows that they are very conservative and reflect a model in which there is minimal customer wait
time at the pump. We assume for purposes of this discussion that the average E85 station has
one E85 dispenser, located on a fueling island allowing two vehicles access at the same time with
one fueling hose on each side of the island. For safety reasons, the EPA has established a rule
that limits the rate at which gasoline or methanol is pumped into motor vehicles—the “flow rate”—
to 10 gallons per minute.3
While every dispenser has its own self-contained pumping mechanism,
it is designed to be shared by both attached hoses, allowing one dispenser to fuel two vehicles
simultaneously. While flow rates vary service station to station, and then by dispenser, we
assume conservatively for purposes of this discussion a flow rate of just three gallons per minute,
assuming two vehicles are using the dispenser at the same time. The average volume of gasoline
purchased per transaction (which may or may not completely fill the vehicle gasoline tank) is
approximately 12 gallons, which at a flow rate of three gallons per minute would result in the
average fueling not exceeding four minutes. We then assume conservatively that it takes four
minutes for the just-fueled vehicle to leave the fueling island and the next vehicle to situate at the
dispenser after a modest time gap (though we think it could reasonably take as little as two
minutes), yielding an eight-minute fueling cycle per vehicle. At that rate, each hose on the fuel
dispenser could service 7.5 vehicles per hour, for a total of 15 vehicles per hour per dispenser.
While a typical service station is open 24 hours per day (usually set by contractual terms),
approximately 75 of its fuel sales take place over a 12-hour peak period with very little taking
place during the late evening or early morning hours. Therefore, we assume that drivers fill up at
the maximum rate during the 12 peak hours, and that that defines 75% of the daily throughput for
the dispenser. Specifically, using the typical fueling volume of 12 gallons per transaction, a single
E85 fueling dispenser with two hoses would dispense 180 gallons of fuel during each of the peak
hours of operations per day, which works out to 2,160 gallons during the entire peak window,
2,873 gallons total per day, and 86,184 gallons total per month, assuming that daily sales are
ratable through a 30-day month (i.e., that the same volume is sold daily—an assumption that
likely has only marginal effect on the results). This analysis shows that the average station with a
single E85 dispenser could deliver approximately twice the volume of E85 that Babcock and
Pouliot assumed and that the rule of thumb suggests.
1
EPA’s count is likely too low. According to e85prices.com, there are about 3,450 E85 stations today. 2
Babcock, Bruce A., Pouliot, Sebastien. Feasibility and Cost of Increasing U.S. Ethanol Consumption Beyond E10. Iowa
State University Center for Agricultural and Rural Development – CARD Policy Briefs. January 2014.
http://www.card.iastate.edu/publications/dbs/pdffiles/14pb17.pdf
3
EPA. Transportation and Air Quality. http://www.epa.gov/oms
Infrastructure Changes and Cost to Increase RFS Ethanol Volumes through Increased E15 and
E85 Sales in 2017
4
For purposes of the rest of this report, we will therefore assume (consistent with Babcock and
Pouliot’s finding) a very conservative throughput of 45,000 gallons per dispenser per month.
FFVs. According to the U.S. Department of Energy’s (DOE) Alternative Fuels and Advanced
Vehicles Data Center (AFDC), there are more than 17.4 million FFVs on U.S. roadways today. In
fact, that figure is likely higher—almost 21 million, according to a recent report by Air
Improvement Resource, Inc.
Total incremental consumption capacity. Finally, we consider how much ethanol can be
consumed through this system as E85 above the amount of ethanol that can be consumed as
E10, i.e., the existing system’s capacity to deliver and consume incremental ethanol as E85. For
purposes of this discussion, we assume, as EPA does, that E85 contains 74% ethanol but adds
the equivalent of 66.2% ethanol over the gallon of E10 that the E85 displaces (specifically, EPA
states every gallon of ethanol use in excess of E10 requires 1.51 gallons of E85).
Assuming 3,100 E85 stations each with a single E85 dispenser distributing 45,000 gallons of E85
per month, this system can distribute about 139.5 million gallons of E85 per month, or 1.674
billion gallons of E85 per year. That is far higher than the 200-300 million gallons of E85 that EPA
assumed for 2017. And it equates to about 1.108 billion gallons of incremental ethanol per year.
There is no reason to find that the fleet would be unable to consume that entire capacity of E85.
The recent report by Air Improvement Resource finds that the existing FFV fleet of 21 million can
consume about 17.13 billion gallons of E85 per year.4
Even if AFDC’s smaller fleet size is used, it
could still amply consume all the E85 that could be delivered by the existing infrastructure.
The only remaining question with respect to the capacity of the existing system to deliver and
consume E85 is whether the FFVs are proximate to E85 stations. To assess this, we return to the
Babcock and Pouliot paper. Prior to that paper, studies and papers by other authors had simply
attempted to extrapolate potential E85 sales using linear models based upon E10 consumption
rates. Such analysis is off target, since FFVs are the only vehicles that can use E85 fuel. Babcock
and Pouliot used detailed data extracts for the geographical distribution of FFVs across the U.S.
down to the zip code level, and the corresponding data of existing E85 service stations with
infrastructure already in place. Assuming that station throughput (as noted) was 45,000 gallons of
E85 per month, that there were 14.6 million FFVs on the road, that there were 3,000 E85 stations
nationwide, and that FFVs would buy E85 from stations within a 10-mile radius, Babcock and
Pouliot determined that 1.2-1.3 billion gallons of E85, containing one billion gallons of ethanol,
could be consumed in a year. Their result of 1.2-1.3 billion gallons of E85 is of course less than
the 1.674 billion gallons of E85 computed above. But since they conducted their study in 2013,
the number of E85 stations and the number of FFVs have increased, thus increasing the
likelihood that an FFV is within 10 miles of an E85 station, and so it is likely that if the Babcock
and Pouliot analysis were re-run using today’s figures, the result would be much higher and
closer to the full 1.674 billion gallons of E85 of throughput capacity. In other words, the Babcock
and Pouliot results reflect a very conservative estimate of the volume of E85 and incremental
ethanol that could be reasonably consumed in 2017.
In sum, there is no doubt that much more than one billion gallons of E85 could be consumed
nationally in 2017 using existing E85 infrastructure.
2.2 Case 2: Expanding Infrastructure to Deliver E85 in 2017
In this section, we examine low-cost ways to expand infrastructure for delivering E85 to
consumers in 2017.
4
Air Improvement Resource, Inc. Analysis of Ethanol-Compatible Fleet for Calendar Year 2017. July 11, 2016.
Infrastructure Changes and Cost to Increase RFS Ethanol Volumes through Increased E15 and
E85 Sales in 2017
5
Cost to add an E85 dispenser at an existing E10-only station. There are two principal pieces
of infrastructure needed to deliver E85: the dispenser and the underground storage tank.
There are two basic kinds of dispensers: blender pumps, which cost about $20,000; and E85
pumps, which cost about $15,000.
According to a report by the National Renewable Energy Laboratory (NREL) called “E85 Retail
Business Case,”5
there are three methods for an existing service station to obtain the necessary
tank to introduce an E85 dispenser:
1. Mid-grade conversion – The retailer cleans an existing (E10) tank and replaces or retrofits
associated non-compatible piping and other equipment. This applies to cases where
stations have a third tank for mid-grade that can be replaced by a blending valve (for
regular and premium to make mid-grade), cases where stations have an extra regular
grade tank, or cases where diesel is replaced because the sales are deemed negligible.
2. New tank – The retailer installs a new underground storage tank and retrofits or replaces
associated non-compatible piping and other equipment. In this case, the retailer retains
the sales of regular and premium fuel.
3. Premium conversion – The retailer fills the premium-grade tank with E85 after cleaning it
and replacing associated non-compatible piping and other equipment. This case applies
to stations that blend their mid-grade rather than draw it from a designated mid-grade
tank, so the retailer can no longer offer either mid-grade or premium-grade gasoline once
the tank is converted.6
With the movement to E10, most E10 stations have tanks that are capable of holding E85. As set
forth in more detail in an NREL report, all steel tank manufacturers have issued signed letters
indicating compatibility with E100, as have fiberglass tanks manufactured in the last ten years.7
The only potential issue would be older fiberglass tanks, where compatibility and manufacturer
approval for use depends on age, manufacturer, and whether the tank is single- or doublewalled.8
Particularly since EPA promulgated its recent underground storage tank rule, EPA has increased
efforts to ensure stations have documentation to show that the tank is approved. This may be a
concern for older stations. However, in the past two years, tank and equipment manufacturers
have made strides toward updating their records for older equipment design and the types of
materials used and supplying this information to the station owners. In fact, EPA’s rule has
created a cottage industry of consultants willing to help the station owner meet the documentation
requirements for EPA, fire marshal, and insurance purposes. While this service comes with a
cost, it is generally cheaper than replacing the equipment and Stillwater’s cost estimates should
cover these expenses.
5
Johnson, C. and Melendez, M. E85 Retail Business Case: When and Why to Sell E85. NREL. December 2007.
http://www.afdc.energy.gov/pdfs/41590.pdf
6
DOE EERE. Clean Cities – Building Partnerships to Reduce Petroleum Use in Transportation.
http://www1.eere.energy.gov/cleancities/
7
Moriarty, K., Yanowitz, J., E15 and Infrastructure, NREL. May 2015.
http://www.afdc.energy.gov/uploads/publication/e15_infrastructure.pdf
8
There is no specific limit on how long a tank can last until it must be replaced. Tanks now have leak detection and
corrosion monitoring, so they can be monitored and replaced before failure. Under the right conditions many tanks last 30
years or longer, but there are some locations where a tank is unlikely to last for 20 years.
Infrastructure Changes and Cost to Increase RFS Ethanol Volumes through Increased E15 and
E85 Sales in 2017
6
Method 1 is the lowest-cost path and the one we focus on. NREL has estimated the cost of the
underground work associated with Method 1 as $15,000, and thus $30,000 to complete the
conversion, i.e., including the new E85 dispenser.9
Cost to add an E85 dispenser at an existing E85 station. Adding another E85 dispenser to an
existing E85 station is cheaper because the only expense is the new dispenser—$15,000 if it is
an E85 dispenser. The station will already have the necessary tank and associated piping and
equipment.
Taking advantage of the natural replacement cycle. Whether upgrading an E10-only station
or an existing E85 station, the effective cost can be reduced by taking advantage of the typical
replacement cycle. Gasoline stations generally replace their dispensers every seven years.10
Upgrading infrastructure to support E85 in conjunction with ordinary infrastructure replacement
reduces the upgrade cost to its marginal cost over the regular replacement cost. Since the cost of
an E10 dispenser is $10,000, the marginal cost of the upgrades described above can be reduced
by this amount.
The consumption that could be supported simply by taking advantage of the ordinary replacement
cycle to upgrade to E85 is sizeable. There are about 155,000 stations in the United States, which
means that about 22,140 stations are replacing their dispensers every year. Of course, not all the
replacement occurs on January 1; it is spread over the year. Assuming that this replacement
cycle occurs ratably over the year, i.e., at a constant rate, 1,845 stations replace their dispensers
every month. If EPA sent a strong signal to the market through the RFS and even one third of
these already-upgrading stations upgraded to offer E85 with one dispenser, then that would mean
an additional 7,380 stations offering E85 at the end of 2017, or (assuming ratable installation over
the year) the equivalent of an additional 3,690 stations operating for all of 2017. Given the
throughput discussed above of 45,000 per dispenser per month those stations could deliver an
additional approximately two billlion gallons of E85 over the course of 2017.
It is reasonable to assume that the industry could hit the ground running on January 1, 2017,
because the final 2017 RFS rule would give it a one-month lead time to prepare.
Existing activity to expand E85 infrastructure. Expansion of E85 infrastructure is already
underway. As noted above, EPA expects the BIP program to add 1,486 E85 stations. Through
BIP and “Prime the Pump,” many large independent chains are working to significantly increase
the number of E85 stations, including Sheetz, Kum & Go, Murphy USA, Protec Fuel, Thorntons,
MAPCO, Minnoco, Cenex, and RaceTrac. And other chains that have worked to expand E85
capabilities significantly include Speedway, Kwik Trip, Spinx, Rebel Oil, Break Time (MFA), MFA
Oil, Meijer Gas, Super Pantry, Bosselman’s Pump & Pantry, Kroger, Petro Serve USA, and Road
Ranger.
2.3 Transporting Additional E85
While the distribution system must move four gallons of E85 for every three gallons of gasoline,
most of the E85 will move from local ethanol production facilities or ethanol tanking facilities to the
stations by truck. E85 is primarily blended at ethanol plants in the Midwest and mostly trucked to
E85 stations that are close to the ethanol production facilities. Trucking assets will require some
redeployment (from product terminals to ethanol plants or ethanol storage facilities) but this
should not be a constraint on the distribution system. Rebalancing these truck transportation
requirements results in little change to the overall number of trucks. Because the ethanol
distribution system is already handling substantial ethanol volumes through E10, significant
increases in ethanol consumption are possible without much impact on the gasoline or ethanol
distribution system.
9
Moriarty, K., Johnson, C., Sears, T. and Bergeron, P. E85 Dispenser Study. NREL. December 2009.
http://www.afdc.energy.gov/pdfs/47172.pdf
10 Stillwater estimate. See Section 5.5.1 for details.
Infrastructure Changes and Cost to Increase RFS Ethanol Volumes through Increased E15 and
E85 Sales in 2017
7
The station tankage for E85 should also not be a concern. Even for small stations, the station’s
largest tank is sized to move about 85 percent of the volume (regular gasoline) through the two
dispensers in a day. If this becomes tight, the station will simply move to twice a day deliveries of
E85.
E85 is primarily blended at ethanol plants in the Midwest and mostly trucked to E85 stations that
are close to the ethanol production facilities. Because the ethanol distribution system is already
handling substantial ethanol volumes through E10, significant increases in ethanol consumption
are possible without much impact on the gasoline or ethanol distribution system.
3 Economics of E85 Infrastructure Changes
3.1 Single Station Single Dispenser Economics
The best way to examine the economics of E85 is through the eyes of a single station adding an
E85 dispenser. This analysis will be for a station that already has three or more gasoline tanks.
By adding E85 none of the current grades are lost, so the current station economics continue with
the added margins from the new E85 to offset the added required investments.
Since dispensers are replaced about every seven years, we assume a project with a seven-year
life. We examined three scenarios, described above:
1. Adding an E85 dispenser to an existing E85 station on the replacement cycle, which has
an initial investment of $5,000;
2. Adding an E85 dispenser to an existing E85 station off the replacement cycle, which has
an initial investment of $15,000;
3. Adding an E85 dispenser to an E10-only station off the replacement cycle, which has an
initial cost of $30,000.
Using our assumption that a dispenser will move 45,000 gallons per month, we further assume
that the new E85 dispenser will move 540,000 million gallons per year and 3.78 million gallons
over the seven-year investment period.
We examine the economics with two rates of return: 10%, which is a reasonable target for
independent stations; and 15%, which is a reasonable target for a large corporation.
In Table 3.1, a simple breakeven analysis of Scenario 1 reveals that the station needs to make
0.13 cents per gallon additional margin to recover the initial investment, an additional margin of
0.33 cents per gallon to earn a 10% return, and 0.38 cents per gallon to earn a 15% return. Given
that these required margins are far less than one cent per gallon, the station owner should have
little hesitation making this investment in E85, assuming that he believes there will be reasonable
demand to fully utilize his E85 dispenser. We explain in more detail below the reasons to believe
that this throughput can be achieved in light of demand patterns and the stations’ optimal gross
margin analysis.
Table 3.1 – Single Dispenser Economics for addition of second E85 Dispenser at Existing
E85 station on replacement cycle
Table 3.2 – Single Dispenser Economics for addition of a second E85 Dispenser at Existing
E85 station off replacement cycle
FULLY UTILIZED E85 DISPENSER
INVESTMENT LIFE THROUGHPUT RATE OF RETURN BREAKEVEN
YR GALLONS CENTS PER GALLON
$ 5,000 7 3,780,000 0.13
$ 5,000 7 3,780,000 10% 0.33
$ 5,000 7 3,780,000 15% 0.38
Infrastructure Changes and Cost to Increase RFS Ethanol Volumes through Increased E15 and
E85 Sales in 2017
8
Table 3.2 shows the results under Scenario 2, where a second E85 dispenser is added at an
existing E85 station off replacement cycle, i.e., paying the full cost for the upgrade rather than just
the marginal cost. The $15,000 cost for adding a second dispenser at existing E85 stations
represents 0.40 cents per gallon on a simple breakeven basis, 0.98 cents per gallon for a 10%
rate of return, and 1.14 cents per gallon for a 15% rate of return. While these margin increases
are around one cent per gallon and slightly higher, this should still be an easy investment
decision for the station owner to make. The only assurance that a station owner would need
under these circumstances, is that there will be sufficient demand for E85.
Table 3.3 – Single Dispenser Economics for addition of a new E85 Dispenser at Existing
E10 station off replacement cycle
Table 3.3 shows the results under Scenario 3, where an E85 dispenser is added at an existing
E10-only station off replacement cycle. The $30,000 cost for adding a new E85 dispenser
represents 0.79 cents per gallon on a simple breakeven basis, 1.96 cents per gallon for a 10%
rate of return, and 2.29 cents per gallon for a 15% rate of return. These margin increases
required are above the one cent per gallon threshold used by station owners and as such would
require serious decision making by the station owner. The station owner would have to expect to
capture additional RIN value through higher E85 margins or attract additional new volumes to
make this kind of investment.
Tables 3.1, 3.2, and 3.3 demonstrate the E85 economics if the dispenser is fully utilized. What do
the economics look like at less than full utilization? Table 3.4 shows not surprisingly that the
station needs double the margin increase if the dispenser is only half utilized. Thus the station
owner deciding to add an E85 dispenser must worry not only about the additional margin needed
to pay off his investment but also about how much each dispenser is used.
Table 3.4 – Single Dispenser Economics under scenario 3 where the new E85 dispenser
added at the previously E10-only station has 50% dispenser utilization
FULLY UTILIZED E85 DISPENSER
INVESTMENT LIFE THROUGHPUT RATE OF RETURN BREAKEVEN
YR GALLONS CENTS PER GALLON
$ 15,000 7 3,780,000 0.40
$ 15,000 7 3,780,000 10% 0.98
$ 15,000 7 3,780,000 15% 1.14
FULLY UTILIZED E85 DISPENSER
INVESTMENT LIFE THROUGHPUT RATE OF RETURN BREAKEVEN
YR GALLONS CENTS PER GALLON
$ 30,000 7 3,780,000 0.79
$ 30,000 7 3,780,000 10% 1.96
$ 30,000 7 3,780,000 15% 2.29
50% UTILIZED E85 DISPENSER
INVESTMENT LIFE THROUGHPUT RATE OF RETURN BREAKEVEN
YR GALLONS CENTS PER GALLON
$ 30,000 7 1,890,000 1.59
$ 30,000 7 1,890,000 10% 3.91
$ 30,000 7 1,890,000 15% 4.58
Infrastructure Changes and Cost to Increase RFS Ethanol Volumes through Increased E15 and
E85 Sales in 2017
9
Table 3.5 below reveals some insights about existing E85 stations. Again using Scenario 3, this
table shows that at low throughputs the margin required to pay off investments is in the $0.20 per
gallon range. It could be said that station owners are not gouging the E85 customer or failing to
pass on enough of the RIN value but are simply holding on to the high E85 margin because it is
needed to pay off their investment due to the very low E85 throughput per station.
Table 3.5 Single Dispenser Economics for addition of a new E85 Dispenser at Existing E10
station (off replacement cycle) with 10% dispenser utilization
4 How to Increase Sales at Existing E85 Stations
Past characterizations of E85 consumers have assumed they are a single group that follow
standard economic rules. Here, we explore a logical segmentation of E85 customers to better
explain observed demand patterns versus price, then extend this model by estimating gross
margin in the supplier-retailer chain to explain observed retail pricing behavior. In the current
pricing situation between gasoline, ethanol, and RIN prices, dealers and retailers are pricing E85
higher than energy parity with E10 because this price level generates the largest gross margin. A
combination of higher RIN price and lower ethanol-relative-to-gasoline price can change this
optimum price point to increase E85 sales volume dramatically from current levels. The RIN price
required to increase sales volume in the short to mid-term by changing pricing behavior and over
the long term by providing incentives to build E85 fueling infrastructure is calculated below. EPA
can create the environment for this E85 growth by setting 2017+ obligations for ethanol high
enough to sustain these necessary RIN values.
4.1 Customer Segmentation
The different sloped lines obtained by Korotney in “Correlating E85 Consumption Volumes with
E85 Price” are in part due to different geographies, but are also likely to be due to different types
of potential customers who react differently to price. For example, demand in California appears
to have no response at all to price. This is inconsistent with behavior of the typical price-seeking
consumer. Also, all other states show a small but steady increase in demand for E85 when prices
are higher than energy parity with E10. (See Appendix.) A purely price-seeking consumer who is
aware of this would not purchase E85 until it was priced at or below energy parity. In fact, there
are a number of reasons to believe that such a price-seeking consumer would only start to
increase E85 consumption when the price is somewhat below parity due to the inconveniences of
refueling more often and traveling farther to find E85, which is currently only sold at about 2% of
retail sites. Variations in E85 energy content (since ethanol content varies from 51% to 81%) also
complicates the decision, so the consumer may also require a bit more of a discount.
To account for these issues, we have developed a working hypothesis based on our extensive
experience with the retail gasoline market on how to segment E85 customers in a way to better
account for the observed buying behavior. Our estimates of the customer breakdown can be
summarized as in the following table:
10% UTILIZED E85 DISPENSER
INVESTMENT LIFE THROUGHPUT RATE OF RETURN BREAKEVEN
YR GALLONS CENTS PER GALLON
$ 30,000 7 378,000 7.94
$ 30,000 7 378,000 10% 19.56
$ 30,000 7 378,000 15% 22.89
Infrastructure Changes and Cost to Increase RFS Ethanol Volumes through Increased E15 and
E85 Sales in 2017
10
Table 4.1 – Estimates of Customer Breakdown
Segment Description % of
FFV
Owners
% of
Current
E85
Demand
Total US
Vol.
Demand
Available,
(mgy)
Vol.
Per site
per mo.
Price
Point
Notes
Committed Either Brand
or
contractually
obligated to
consume.
0.5% 30% 50 1,400 Doesn’t
matter
Includes federal,
state and
municipal fleets
or businesses
who have
committed to
E85.
Believers Believe it’s
the right
thing to
do. Will
consume if
price
approaches
energy
parity.
3% 60% 300 8,400 Sliding
scale
that
increases
from 5 to
25%
discount
from E10
Supporters of
renewable fuel,
some farmers or
other corn
proponents. Also
could be car
renters who fill
up before
returning FFVs.
Mass
Consumers
Price takers
will
consume
when
economical,
including
price,
convenience
and risk.
93.5% 10% 93500 262,000 Sliding
Scale
from 25%
to 50%
discount
from E10
Most consumers
try to buy the
best fuel for the
money, but are
influenced by
other issues too.
Disbelievers Will not
consume,
regardless
of price.
3% 0% 0 0 Begins
only at
steep
discounts
of 40%
or more
No need to
consider this
group.
Some corroboration of this model is provided by EIA data which show that federal and state fleets
consumed nearly 44 million gallons of E85 in 2014, which is about 27% of estimated total
consumption11,12.
It is important to realize that there are many ways to segment fuel customers along completely
different dimensions. Also note that the distribution will vary by geography, and the number of
consumers in each segment can only be roughly estimated. However, despite these limitations,
this structure accounts for many observations of demand response to price, and enables
additional investigation of phenomena at the dealer-customer interface.
11 EIA. Federal Fleet Fuel Consumption Data. http://federalfleets.energy.gov/performance_data#waivers 12 EIA. State Fleet and Fuel Data. http://www.eia.gov/renewable/afv/users.cfm?fs=a&ufueltype=e85
Infrastructure Changes and Cost to Increase RFS Ethanol Volumes through Increased E15 and
E85 Sales in 2017
11
This model leads to a volume curve for a one-dispenser E85 site which has a customer base
representing the U.S. as a whole to look something like this:
Figure 4.1 – E85 Site Volume, One Dispenser
To test how reasonable these results are, consider how the left half of this curve looks like the
regression analyses for each of the five states analyzed in Korotney’s analysis.13 California looks
like the very far left part of the curve only because so much of the demand there is by consumers
committed to its use. The other four states look very much like the part of the curve shown for the
Believers with small positive slopes. The overall slope of this part of the curve is consistent with
Korotney’s results. Consider the demand response we attribute to Mass Consumers on the right
hand side of the curve. If E85 were discounted by 35% to E10, only 15% of the owners of FFVs
(or 16% of those we are calling Mass Consumers) would be needed to create demand ten times
larger than today’s typical demand of less than 5,000 gallons per month. Based on the work by
Babcock and Pouliot in 2013 (with lower station counts and a small FFV fleet than exist today),
more than 30% of FFVs are located within five miles of an E85 station, so attracting half of these
local FFVs with E85 discounted to only 65% of E10 price seems very reasonable if not
conservative.
Drawing from our experience in the industry, we also believe that the right-hand side of the curve
is reasonable, assuming that the discounts shown persisted in a sustained pricing environment
(e.g., as would occur if EPA meaningfully changed how it implemented the RFS). In our
experience, customers are very price-sensitive. For example, we have seen evidence of
significant customer movement when different retailers engage in price wars over gasoline.
Similarly here, once the inconvenience of E85 is compensated for below energy parity, we would
expect retailers to market the price savings and for FFV owners to take advantage of them.
Indeed, if E85 were discounted by 35% to E10, only 15% of the owners of FFVs (or 16% of those
we are calling Mass Consumers) would be needed to create demand ten times larger than
today’s typical demand of less than 5,000 gallons per month. As stated above, based on the work
by Babcock and Pouliot, more than 30% of FFVs are located within five miles of an E85 station
(and 55% of FFVs are located within ten miles).
13 “Memo to docket on Correlating E85 consumption volumes with E85 price,” memorandum from David Korotney to EPA Air Docket EPA-HQ-OAR-2015-0111.
Infrastructure Changes and Cost to Increase RFS Ethanol Volumes through Increased E15 and
E85 Sales in 2017
12
The graph below contrasts Brattle’s14 log-log curve with Stillwater’s assessment:
Figure 4.2 Comparing Stillwater and Brattle Demand Curves
Note how similar the curves are in the range below the point of E85 energy parity which occurs at
an E85 discount of about 22% to E10. These should agree at this point because there are ample
observations of consumer behavior to correlate it with price. The area where the two demand
curves diverge is where there is not enough data to discern price behavior. Accordingly, the
Brattle demand curve is a reasonable extrapolation of the existing data that show the beginning of
change near energy parity. However, Stillwater’s customer segmentation analysis predicts that
there should be a distinct change in demand response to price as the price discount to E10
increases below energy parity because price seeking customers begin to see better value, and
we believe these are the vast majority of FFV owners.
There is another difference between these curves that is important to realize. To achieve strong
demand at E85 discounts of 30% or more to E10 there are two key requirements. First, local FFV
owners will need to know where to find the E85 site. Second, FFV owners will need to know that
E85 will be consistently priced at levels that make it attractive relative to E10. We believe this
level is 25-30% below E10, but in reality it is related to other factors including general price level,
local competition for E85 sales, and local concentration of FFVs. Consumers will not drive around
looking for the single local E85 site if it is often more expensive to use than E10.
Next, we examine the incentives that fuel suppliers and retailers have for pricing E85 by looking
into the gross margin available to them. We’ll first consider the situation with recent prices with
the Stillwater demand curve and later generalize the predictions for a range of prices with the use
of Brattle’s log-log price curve.
4.2 Gross Margins
If dealers will not discount E85 by more than 25% relative to E10, how does this matter? To
explore this issue, we have created a simple model of retail pricing to estimate gross margins in
14 Peeking Over the Blendwall An Analysis of the Proposed 2017 Renewable Volume Obligations The Brattle Group July
11,2016
Infrastructure Changes and Cost to Increase RFS Ethanol Volumes through Increased E15 and
E85 Sales in 2017
13
the supply chain with the following assumptions (using 2016 average prices in Los Angeles
through June as a proxy):
Ethanol Price = $1.62/gallon
Gasoline (BOB) Price = $1.41/gallon
Ethanol RIN Price = 74 cents
Supplier E10 Margin to Retailer = 5 cents per gallon (cpg)
Retailer E10 Margin = 10 cpg
E85 volumes according to the above curve
Gross margin calculated across fuel supplier and retailer
Fuels tax = 40 cpg*15
Using these assumptions, we calculate E85 gross margins for two cases shown on the left priced
at 14% and 35% discounts to E10. On the right we do the same calculations but with the RIN
price increased from 74 to 124 cents.
Figure 4.3 – E85 Gross Margin Estimates
15 (Fuels taxes vary dramatically by state, and in many states are lower for E85 than for E10.15 Here we assume a
moderate volumetric tax of 40 cents on every gallon of fuel. This penalizes E85 relative to E10 since the 22% higher
volume of E85 needs to be purchased results in 22% higher taxes per mile driven. If the fuels tax is implemented as a
sales tax based on a percent of sales price and E85 is priced below energy parity to E10, then it actually favors E85
slightly.) The following link: http://www.ncsl.org/research/transportation/taxation‐of‐alternative‐fuels.aspx#one for the National
Council of State Legislatures lists much of the data on state taxes.
Infrastructure Changes and Cost to Increase RFS Ethanol Volumes through Increased E15 and
E85 Sales in 2017
14
The first case shows a gross margin (“GM”) of $1,000/month for pricing above energy parity at a
14% discount to E10 price. It also shows a negative margin at a much steeper discount of 35%
relative to E10 because the dealer would have to price below cost in order to attract the price
seeking consumers. Clearly, E85 cannot be economically priced below energy parity with these
price assumptions. In the second case, all of the assumptions are identical except that RINs are
priced much higher at 124 cents. In this case, the gross margin increases 170% with the deeper
discount because the increased volume more than overcomes the decreased margin per gallon
sold. Note that the optimum E85 sales price with higher RIN prices (and constant RBOB prices) is
significantly lower. With only a change in RIN price, the dealer can profitably increase gross
margin by selling more E85 at a much lower price. Also note that this increase in E85 sales can
occur without increases in retail infrastructure. Last, note that this also results in more competitive
E85 pricing without more E85 competition. This happens because E85 becomes price
competitive with E10 so that consumers with FFVs will choose to fill with E85 because it less
expensive for them.
While the E85 dealer may lose some of his E10 business to E85, because only 2% of retail sites
have E85 it is more likely that he will increase overall site volume and profitability by attracting
FFVs that were being filled at competitors’ sites. It may be pointed out that this is the gross
margin across the fuel supplier and retailer, so that the retailer may not be able to set his price at
the joint optimum. However, both supplier and marketer have incentives to find this price point
even if the margin is not shared equally. The additional benefit to the retailer from increased site
traffic further increases the chances of finding a price point that results in increased sales
volumes.
We repeated calculations like those above to estimate GM as a function of price point and the
impact of RIN prices in the 2016 price environment, as shown in the following figure:
Figure 4.4 – Calculated Combined Supplier/Dealer E85 Gross Margin
The yellow line is the volume curve derived from the customer segmentation analysis. The blue
line shows the GM curve that an E85 retailer would expect in the assumed pricing environment.
The 2016 environment, with ethanol priced above gasoline, is difficult for E85 marketing. Margins
Infrastructure Changes and Cost to Increase RFS Ethanol Volumes through Increased E15 and
E85 Sales in 2017
15
tend to be low and with RINs priced at 74 cents, the E85 price point that optimizes gross margin
is very close to the 14% below E10 observed recently. This indicates that E85 marketers are
pricing to maximize gross margin as we would expect, and gives another validation for the
structure of this model. This figure also shows that GM increases with RIN price and that optimum
GM increases even more with higher RIN prices when the sales price is discounted more heavily.
Increasing the RIN price from 74 to 100 cents results in a doubling of GM with five times the sales
volume. Increasing the RIN price further to 126 cents results in nine times both the gross margin
and sales volume with reduced E85 sales prices. The trend is shown in the graph below:
Figure 4.5 – Optimum E85 Price Point vs. RINs Price
This analysis demonstrates that high RIN prices increase E85 gross margins, providing incentives
to build E85 infrastructure. They also (interestingly) provide incentives to price more competitively
and sell substantially more volume in the short term. So E85 sales volumes can increase
substantially in both the short term and the long term if RIN prices can be maintained at a specific
level that is a function of the RIN price, and the relative price of ethanol to RBOB, and fuels tax
rate. The RIN price needed to provide the right incentives for increased E85 sales varies with
gasoline and ethanol prices as shown in the next section.
One last note on value pricing is that it has at times been very successful in the fuels market.
ARCO was very successful for decades at pricing below other majors. At one point in time, a 5
cpg discount in street price was enough to enable an average volume per site that was double
the industry as a whole. This enabled dealers to amortize fixed costs over twice the volume of
competitors and resulted in increased site traffic that improved the profitability of AM/PM brand
convenience stores located on ARCO sites. Today, an example of a successful value priced retail
site is Costco, which has an average volume many times that of an average gasoline station.
Value has been, and continues to be (along with quality, convenience, and others), one of the
dimensions of differentiation in the retail fuel space. While these examples do not indicate what
the price response to E85 will be, they do demonstrate that there are many consumers who are
price conscious.
Infrastructure Changes and Cost to Increase RFS Ethanol Volumes through Increased E15 and
E85 Sales in 2017
16
4.3 What RIN Price is Needed for Short Term Volume Growth
All of the calculations so far have been with a narrow range of fuel taxes, gasoline, ethanol, and
RIN prices. In this section a wide range of these parameters will be used to show specifically
what RIN price is needed so that the optimum retail price point is discounted by 30% relative to
E10. With the Stillwater demand curve, this results in a site sales-volume increase of five times
current average site volume when priced at a 14% discount to E10. Below are two graphs that
show ethanol RIN price levels required using this simple model (the S curve from Figure 4.5) to
increase E85 sales volumes by a factor of five from 4,800 gallons per month to 24,000 gallons
per month.
Figure 4.6 – RINS Price that Causes E85 to be Priced at 30% Discount to E10 (Ethanol
Price)
Figure 4.7 – RINS Price that Causes E85 to be Priced at 30% Discount to E10 (RBOB Price)
Infrastructure Changes and Cost to Increase RFS Ethanol Volumes through Increased E15 and
E85 Sales in 2017
17
The graph below shows the impact of Fuels Tax level on RIN price required to increase E85 sales
by a factor of five:
Figure 4.8 – RINs Price Required to Increase E85 Sales by a Factor of 5
From these analyses, we can make the following generalizations for what is needed to provide
incentives to increase E85 sales by five times at existing infrastructure from current levels while
holding the other parameters constant:
Infrastructure Changes and Cost to Increase R