&EPA
Air and Radiation EPA420-R-05-010
June 2005
United States
Environmental Protection
Agency
Analysis of Connecticut's
Request for Waiver of the
Reformulated Gasoline
Oxygen Content
Requirement for Connecticut
Covered Area:
Technical Support Document
-------
EPA420-R-05-010
June 2005
of for of the
for
Technical Support Document
Transportation and Regional Programs Division
Office of Transportation and Air Quality
U.S. Environmental Protection Agency
Docket OAR-2004-0429
-------
TABLE OF CONTENTS
EXECUTIVE SUMMARY 1
I. BASIS FOR DEP'S REQUEST FOR A WAIVER 5
II. GENERAL CONCLUSIONS EPA CAN MAKE ABOUT EMISSIONS BASED ON
DEP'S SUBMISSION AND OTHER INFORMATION AVAILABLE 6
III. INFORMATION CONNECTICUT HAS PROVIDED TO EPA 8
A. Information on VOC emissions 8
B. Information on NOx emissions 9
C. Information on CO emissions 9
D. Information on likely oxygenated RFC composition subsequent to
Connecticut's ban on MTBE 10
E. Information on likely RFC composition subsequent to a waiver of the
oxygen content requirement 11
IV. INFORMATION NEEDED TO EVALUATE NOX, VOC AND CO EMISSIONS
CHANGES RESULTING FROM AN OXYGEN CONTENT WAIVER 12
A. Background 12
B. Emissions Models 13
C. RFC Properties and Composition 16
V. EMISSIONS-RELATED CONCLUSIONS BASED ON CONNECTICUT'S
SUBMISSION
18
A. Effect of a waiver on NOx emissions based on Connecticut's submission . 18
1. Multiple gasoline properties affect NOx emissions 18
2. NOx emissions from on-road vehicles with and without a waiver 19
3. NOx emissions from non-road vehicles with and without a waiver20
B. Effect of a waiver on VOC emissions based on Connecticut's
submission 20
1. VOC on-road and non-road exhaust emissions 21
2. Evaporative VOC emissions due to RVP increases 22
3. Evaporative VOC emissions due to permeation 23
C. Effect of a waiver on CO emissions 24
D. Ethanol transport-related emissions 24
VI. CONCLUSIONS OF EPA'S ANALYSIS OF DEP'S SUBMISSION 25
-------
APPENDIX A: SUMMARY OF EPA'S ANALYSIS OF DEP'S SUBMISSION 26
11
-------
EXECUTIVE SUMMARY
Section 211 (k)(1) of the Clean Air Act (CAA) requires that the Administrator of
the Environmental Protection Agency promulgate regulations establishing requirements
for reformulated gasoline (RFC) to be used in gasoline-fueled vehicles in specified
ozone nonattainment areas. The CAA mandates RFC use in certain nonattainment
areas, based on ozone air quality and population criteria. Other ozone nonattainment
areas are allowed to opt into the RFC program. Within the state of Connecticut, RFC is
mandated in the Connecticut State portion of the Hartford-New Haven-Springfield
Consolidated Metropolitan Statistical Area (CMSA) and the Connecticut State portion of
the New York City CMSA. Parts or all of Litchfield, Hartford, Middlesex, New London,
Tolland, and Windham Counties, Connecticut are an opt-in area.
Section 211 (k)(2)(B) of the Act, 42 U.S.C. § 7545(k)(2)(B), establishes an oxygen
content requirement for federal RFC, but allows EPA to waive compliance with the
requirement under certain circumstances. Section 211 (k)(2)(B) provides that:
The oxygen content of the gasoline shall equal or exceed 2.0 percent by
weight (subject to a testing tolerance established by the Administrator)
except as otherwise required by this Act. The Administrator may waive, in
whole or in part, the application of this subparagraph for any ozone
nonattainment area upon a determination by the Administrator that
compliance with such requirement would prevent or interfere with
attainment by the area of a national primary ambient air quality standard.
In a letter dated April 22, 2002 from Connecticut Governor John G. Rowland to
then Administrator Whitman, Connecticut expressed intent to submit a waiver request
under CAA Section 211(k)(2)(B) from the federal RFC oxygen content requirement. In a
letter dated September 29, 2004 from Connecticut Department of Environmental
Protection (DEP) Commissioner ArhturJ. Rocque, Jr., to Assistant Administrator Jeff
Holmstead, Connecticut officially submitted such a request for the Connecticut RFC
area. The submission stated that because MTBE was banned in the State of
Connecticut beginning January 1, 2004, and because of the Act's oxygen requirement
for RFC, ethanol would be used as an oxygenate in the RFC areas in the State of
Connecticut. DEP asserted that the use of ethanol as a replacement for MTBE in RFC
would result in an increase in Volatile Organic Compounds (VOCs), and oxides of
nitrogen (NOx) during the summer ozone season. DEP further argued that "increases
in these pollutants will immediately interfere with Connecticut's ability to attain the
1
-------
National Ambient Air Quality Standard (NAAQS) for both one-hour and eight hour ozone
and fine particulates in its RFC areas.
In order to determine whether the RFC oxygen content requirement prevents or
interferes with Connecticut's ability to attain National Ambient Air Quality Standards
(NAAQS), it is necessary to consider the gasoline quality and related emissions impacts
that would be likely to exist with and without the RFC oxygen content requirement. To
evaluate the potential for interference with the ozone NAAQS, EPA must first consider
the emissions differences for each pollutant contributing to ozone formation (NOx, VOC
and carbon monoxide (CO)), between cases where the oxygen requirement remains in
effect ("no-waiver") and cases where the oxygen requirement is waived ("waiver"). To
evaluate the potential for interference with the particulate matter NAAQS, EPA must first
determine the emissions differences for NOx between cases where the oxygen
requirement remains in effect ("no-waiver") and cases where the oxygen requirement is
waived ("waiver") as described above for ozone. In deciding whether to grant or deny a
waiver, as a threshold matter EPA must determine what difference, if any, a waiver
would have on emissions, and what effect, if any, the difference in emissions would
have on ambient ozone and particulate matter levels.
Certain underlying information is required in order to make a quantitative
estimate, or even a reasonably certain qualitative directional estimate of the emissions
changes that might occur from a waiver. This information includes knowledge of certain
emission-related fuel properties of the RFC that would be supplied to Connecticut with
and without a waiver. Models relating these fuel properties to vehicle emissions would
then be used to estimate percent differences in emissions between the "no waiver" and
"waiver" conditions. Additionally, area-specific on-road and off-road gasoline emission
inventory data are needed in order to convert relative (%) changes to absolute
(tons/day) changes.
Connecticut DEP's submissions included essentially no information or analysis
of the expected fuel properties of RFC with and without a waiver. Connecticut did not
provide sufficient information or analysis to show either quantitative or directional
estimates of the emissions differences between "no waiver" and a "waiver," for the
pollutants contributing to ozone formation (NOx, VOCs and CO) or particulate matter
formation (NOx). Changes in VOC, NOx, and CO emissions were not quantified, and
the lack of adequate information and analysis means that even the direction of any
change in NOx, and VOC is not clear. Changes in CO emissions were not addressed
in Connecticut's submission.
If Connecticut had provided quantitative estimates of the emission differences
between the "no waiver" and "waiver" situations, EPA's evaluation of its waiver
submission would have included a thorough review of the basis for these estimates (fuel
property estimation and emission modeling methodology), and subsequently, a
thorough review of whether these emission differences indicate that ambient ozone or
particulate matter are likely to be higher with the oxygen requirement than without.
-------
However, EPA's review of the information contained in DEP's waiver submissions
determined that it lacked not only the quantitative estimates of the "no waiver" to
"waiver" emission changes, but the underlying information necessary to make such
estimates. Since this information has not been submitted, EPA is unable to determine
whether implementation of the oxygen content requirement in the Connecticut RFC
area will prevent or interfere with the attainment of a NAAQS.
EPA has considered the information that Connecticut has provided which may
be relevant to an analysis of Connecticut's waiver request. EPA has determined that
the relevant no-waiver to waiver comparison cannot be made either qualitatively or
quantitatively. In making this evaluation, EPA has considered the information provided
by Connecticut with regard to the potential effect of a waiver on each of the pollutants,
NOx, VOC and CO, for both on-road and off-road gasoline vehicles and engines.
In order to evaluate the information submitted by Connecticut, we have identified
the fuel properties, vehicle fleets (e.g., on-road versus off-road, older technology
vehicles versus newer technology, etc.), and emission sources (e.g. exhaust, "as
blended" evaporative, commingling and permeation-related) that would need to be
considered in a waiver/no-waiver analysis. We have also identified emissions models
and other information necessary to make the relevant emission estimates that could be
utilized to make the waiver/no-waiver comparison.
The information that DEP has provided fails to clearly demonstrate what effect a
waiver would have an on ozone or particulate matter levels in Connecticut. This is
because: 1) there are three pollutants whose emission rates could be altered by a
waiver (NOx, CO and VOC) and all three affect ozone formation to varying degrees; 2)
the lack of information on fuel qualities with and without a waiver and the lack of other
relevant and necessary information precludes even a directional estimate of the impact
of a waiver on NOx and VOC emissions; 3) the best estimate of the net impact of a
waiver on CO emissions is that CO emissions would be greater with a waiver than
without, but the difference cannot be quantified; 4) no analysis has been provided or
performed, and the information before the agency does not allow an analysis to be
performed, on the combined effect of these emissions changes on ozone.1
In addition, (1) NOx emissions affect PM, (2) the lack of information on fuel
qualities with and without a waiver and the lack of other relevant and necessary
information precludes even a directional estimate of the impact of a waiver on NOx, (3)
no analysis has been provided or performed, and the information before the agency
does not allow an analysis to be performed, on the effect of any NOx emissions
changes on particulate matter.
CO plays a far less important role in ozone formation that NOx or VOC. Thus, even
though a gross directional determination can be made for CO, such a determination
provides little information in making an overall judgement about ozone formation.
-------
Since no determination can be made regarding the overall effect of a waiver on
emissions related to ozone and participate matter, the information that DEP has
provided fails to clearly demonstrate what effect a waiver would have on ozone or
particulate matter. Since this threshold demonstration has not been made, EPA is not
able to determine whether the oxygen requirement for RFC interferes with attainment of
the ozone or particulate NAAQS in the Connecticut RFC area. EPA concludes that
Connecticut's request therefore should be denied.
-------
I. BASIS FOR DEC'S REQUEST FOR A WAIVER
As EPA interprets it, Connecticut's request for a waiver is based primarily on
DEP's contentions that use of ethanol-oxygenated gasoline will interfere with attainment
of the ozone and particulate NAAQS because:
The NOx and VOC emissions performance of ethanol-oxygenated RFC will be
worse than the emissions performance of the MTBE-oxygenated RFC supplied
to Connecticut prior to its MTBE ban or to non-oxygenated fuel.
Commingling and permeation resulting from ethanol use will increase VOC
emissions compared to MTBE-oxygenated RFC or non-oxygenated fuel.
Ethanol-oxygenated RFC will not provide the NOx reduction benefits expected
under the RFC program because the Complex Model does not fully capture the
effects of oxygenates on vehicle emissions of Nox.
There would be emissions associated with ships and barges transporting ethanol
into the Connecticut area for adding to RFC at terminals.
Because the NOx emission performance will be worse for ethanol blended RFC
as compared to fuel under an oxygen waiver, fine particulate matter will increase.
-------
II. GENERAL CONCLUSIONS EPA CAN MAKE ABOUT EMISSIONS BASED ON
PEP'S SUBMISSION AND OTHER INFORMATION AVAILABLE
DEP submitted no emissions modeling for Connecticut in its September 29, 2004
waiver request. DEP provided no information on the likely properties of the oxygenated
or non-oxygenated RFC that would be supplied to Connecticut with a waiver, nor has
DEP discussed the market share of non-oxygenated RFC market share should a waiver
be granted.2 The only information supplied by DEP on oxygenated RFC without a
waiver is that it is likely to contain ethanol at 10% by volume. As discussed in Section
IV of this document, information on likely fuel properties would be critical to an
estimation of the emissions differences between the waiver and non-waiver scenarios in
Connecticut, as is evidenced by EPA's evaluation of California's waiver request.
In other words, DEP has asked EPA to grant a waiver of the RFC oxygen
requirement, based solely on its assertion that NOx and VOC emissions will increase
with RFC using ethanol when compared to the use of RFC oxygenated with MTBE or
non-oxygenated fuel. In light of Connecticut's ban on the use of MTBE as an
oxygenate, EPA must compare the base case of ethanol-oxygenated RFC without a
waiver to the RFC that would be supplied if an oxygen content waiver were granted.
This is the proper comparison for determining whether continuing to maintain the
oxygen content requirement interferes with attainment of the ozone or particulate matter
NAAQS. The properties of MTBE-oxygenated RFC are irrelevant to this analysis. To
put it another way, maintaining the current oxygen content requirement could not
amount to interference, unless, at a minimum, a waiver is likely to produce a better
situation with respect to ozone-forming or PM forming pollutants than would exist absent
the waiver. As noted, DEP has not provided information regarding the emission-related
2 Although the program report enclosed with DEP's September 2004 submission included
analyses for the gasolines used in a New York test program discussed in sections III-C
and III-D below, DEP has provided no basis for EPA to conclude that these formulations
are likely to represent the gasoline that would be supplied were a waiver granted. All
fuels met the 30 ppm sulfur standard. However, the report notes that the ethanol RFC
and non-oxygenated equivalent RFC test fuels were specified to have Complex Model
performance similar to the study MTBE RFC. The MTBE RFC was specified, except for
sulfur, to represent "typical year 2000 NYCMA RFC". No other restrictions or
specifications were imposed on the fuel supplier. As noted, the composition and
emissions performance of actual RFC with or without a waiver will be determined by a
number of regulatory, economic and technical constraints. Consequently the
specifications DEC imposed on the test fuels are insufficient to ensure that they will be
similar to actual future RFC, with or without a waiver.
-------
properties of the RFC that would be supplied were a waiver granted, nor did it address
the issue of non-oxygenated RFC market share were a waiver granted.3 As noted, DEP
has not provided the kinds of information necessary to reach this conclusion. In
summary, with the information at hand, EPA cannot estimate no-waiver to waiver
emission changes, as it must do to ascertain whether the oxygenate requirement
interferes with NAAQS attainment.4 This general conclusion is explained in more detail
below with respect to each individual pollutant of concern.
Below are more detailed discussions of the information submitted by
Connecticut. The following sections of this TSD discuss EPA's evaluation of the
information. Our evaluation clearly shows that the information is insufficient to come to
any definitive conclusion regarding the effect of an oxygen waiver on ozone or
particulate matter.
Ethanol-oxygenated RFC could constitute a portion, potentially a substantial portion, of
Connecticut's gasoline pool even if a waiver were granted. DEP may have implicitly
assumed in their submissions that all RFC would be non-oxygenated if a waiver were
granted, but it has provided no information or analysis, such as refinery modeling, to
substantiate this position. This is an issue not only because the pool average emission-
related property values of Connecticut's "waiver" RFC would depend on the extent to
which ethanol is used, but because substantial market shares of both ethanol RFC and
non-oxygenated RFC in Connecticut could result in "worst case" commingling-related
VOC emissions.
In the California waiver analysis, EPA evaluated twelve "no waiver" to "waiver" scenarios.
Each scenario consisted of a set of "no waiver" and "waiver" fuel properties predicted by
refinery modeling with a specific set of assumptions. EPA treated the "no waiver" case as
the reference, computed percent differences in emissions between "no waiver" and
"waiver" fuels, and applied these percent differences to emission inventories to determine
tons/day differences in NOx, VOC and CO emissions. EPA assumed that the tons/day
emission inventory information that was available for the California analysis represented
the all ethanol-oxygenated "no waiver" reference condition. EPA was therefore able to
compute "no waiver" to "waiver" tons/day differences from percent differences in a single
step. While it would be possible in a waiver analysis to calculate emissions changes from
a different reference condition than the "no waiver" baseline, it still necessary to make a
"no waiver" to "waiver" emissions comparison. If a different reference condition (e.g. pre-
MTBE ban) were used, it would be necessary to first estimate both reference to "waiver"
and reference to "no waiver" percent change and tons per day emissions changes, and
subsequently combine them to estimate the total "no waiver" to "waiver" changes.
7
-------
III. INFORMATION CONNECTICUT HAS PROVIDED TO EPA
In a letter dated September 29, 2004 from Connecticut Department of
Environmental Protection (DEP) Commissioner Arthur J. Rocque, Jr., to Assistant
Administrator Jeff Holmstead, Connecticut requested a waiver from the federal oxygen
content requirement for reformulated gasoline, under Section 211(k)(2)(B).5 The
submission stated that because MTBE was banned in the State of Connecticut
beginning January 1, 2004, and because of the Act's oxygen requirement for RFC,
ethanol would be used as an oxygenate in the RFC areas in the State of Connecticut.
DEP asserted that the use of ethanol as a replacement for MTBE in RFC would result in
an increase in Volatile Organic Compounds (VOCs), and oxides of nitrogen (NOx)
during the summer ozone season. DEP further argued that "increases in these
pollutants will immediately interfere with Connecticut's ability to attain the National
Ambient Air Quality Standard (NAAQS) for both one-hour and eight hour ozone and
particulate matter in its RFC areas.
In the sections below, EPA describes in detail the information that DEP has
submitted in detail.
A. Information on VOC emissions
DEP argues that the use of ethanol as a replacement oxygenate for MTBE will
result in substantial increases in VOC emissions due to (1) increased permeation
through vehicle components that hold gasoline, and (2) commingling of ethanol-blended
fuels with fuels that do not contain.6 However, DEP has not attempted to quantify what
Filed in docket OAR-2004-0429, document number 0001.
Permeation refers to the escape of gasoline constituents through the walls of non-metallic
fuel lines and gasoline tanks. These soft components of automotive fuel systems tend to
be more permeable to ethanol than to other hydrocarbons in gasoline. Thus, ethanol-
RFG blends tend to result in an increase in evaporative emissions due to permeation
through these components. Commingling refers to the RVP increase (with resultant
emission increase) that occurs when non-ethanol and ethanol-oxygenated gasoline
blends are mixed in the gas tanks of consumer's automobiles. Since the presence of
ethanol causes an increase in the volatility of the gasoline (as measured by the Reid
Vapor Pressure or RVP), such commingling would contribute to an increase in
evaporative VOC emissions.
-------
the increase in VOC emissions might be for Connecticut. DEP has provided some
attached studies regarding permeation and commingling in California (e.g. Docket OAR-
2004-0429, items 003 and 006-008), but failed to explain their applicability to
Connecticut or develop a methodology to quantify such effects for Connecticut.
DEP also notes that there will be emission increases (for both VOC and NOx)
due to the transport of ethanol to the Connecticut RFC area from production centers in
the Midwest, but has not provided a quantitative estimate of the emission increases that
will occur within the Connecticut RFC area.
DEP has not identified or quantified any other VOC emission changes (increases
or decreases) that could occur with ethanol use, nor has it clearly stated that no other
significant VOC emission changes will occur.
B. Information on NOx emissions
DEP claims that NOx emissions will increase with the use of ethanol blended
RFC compared to MTBE-blended RFC or non-oxygenated fuel, but has not quantified
the anticipated increases. DEP stated in its April, 2002 letter that existing test data
indicate that NOx emissions from newer vehicles increase with ethanol, although it did
not explicitly define "newer vehicles".
DEP notes that Phase II RFC regulations require refineries to produce gasoline
that reduces NOx emissions by 6.8 percent and VOC emissions by 27.4 percent
compared to 1990 levels, as quantified by the EPA Complex Model. DEP states,
however, that it believes that the EPA Complex Model does not fully capture the effects
that oxygenates such as ethanol have on emissions from current fleet vehicles. It
believes that NOx emissions will increase due to the use of ethanol compared to MTBE,
and that this actual increase in NOx emissions will not be identified by, or constrained
through the use of, the Complex Model to measure RFC compliance with phase II RFC
regulations.
C. Information on CO emissions
DEP made no explicit references to the potential effect of a waiver on CO
emissions or the relationship of CO to ozone formation in its April 22, 2002 or
September 29, 2004 letters but others have done so in studies DEP attached.
In the New York study entitled "Emission Impacts of Fuels to Accommodate the
NY State Oxy-waiver Request and MTBE Ban,"7 (Docket OAR-2004-0429-0009), New
"Program Report: Emission Impacts of Fuels to Accommodate the NY State Oxy-waiver
Request and MTBE Ban", May 6 ,2003, submitted as an enclosure to DEP's September
29, 2004 letter (Docket OAR-2004-0429-0009).
-------
York claimed that it has verified the lack of need for an oxygenate through laboratory
testing of two automobiles both with and without oxygenated fuel. This study describes
a test program conducted at New York's Automotive Emissions Laboratory (AEL), in
which two vehicles, a 1998 Plymouth Breeze and a 1997 Oldsmobile Achieva, were
tested multiple times using an MTBE-oxygenated RFC, an ethanol-oxygenated RFC,
and a non-oxygenated "equivalent RFC". The program report concluded that data from
over 200 emission tests on these two vehicles do not indicate any consistent increase in
any of the regulated emission species (HC, CO, NOx) for either the ethanol-oxygenated
RFC or the non-oxygenated "equivalent RFC" relative to the MTBE-oxygenated RFC.
Conversely, in the attached report entitled, "Demonstration that the U.S.
Environmental Protection Agency must grant California a waiver from the Federal
reformulated gasoline oxygen mandate on remand from the U.S. Court of Appeals for
the Ninth Circuit," California states that "ARB has consistently acknowledged that the
2.0 wt. % minimum oxygen requirement in the federal RFC program reduces CO
emissions from the existing fleet of vehicles on the road today."8
D. Information on likely oxygenated RFG composition subsequent to
Connecticut's ban on MTBE
DEP has provided EPA with very little information on the likely emission-related
properties of its RFG subsequent to an MTBE ban.9 These properties will determine the
exhaust and evaporative emissions associated with RFG use in a non-waiver scenario.
Connecticut stated in its September 29, 2004 letter that refiners are blending 10
(volume) percent ethanol to meet the oxygen requirement as a result of their MTBE ban.
This is equivalent to approximately 3.5 weight percent oxygen. Blending at
approximately 5.7 volume percent ethanol is required to meet the 2 weight percent RFG
oxygen requirement. Beyond this, Connecticut has provided no further information on
the composition of ethanol-oxygenated RFG subsequent to its MTBE ban.
The program report for New York's vehicle testing study, cited earlier, included
See Docket OAR-2004-0429-0004.
These emission-related properties are the parameters used in emission models to
estimate emission differences resulting from fuel formulation differences. The Complex
Model inputs affecting VOC and NOx emissions are RVP, oxygen wt%, aromatics vol%,
olefin vol%, sulfur ppm, E200 and E300. (E200 and E300 are the percent of gasoline
evaporated at 200F and 300F, respectively.) The alternative Tech 4 models which EPA
developed for its California waiver analysis use RVP, oxygen wt%, aromatics vol%, olefin
vol%, sulfur ppm, T50 and T90 as inputs . (T50 and T90 are the temperatures at which
50% and 90% of the gasoline are evaporated, and are highly correlated with E200 and
E300.). As discussed later, these models should be applicable to a portion of
Connecticut's on-road fleet, as well.
10
-------
certificates of analysis showing properties for the gasolines used in the study. The
report states "AEL believes that the study fuel specifications approximate the fuel
blending situation that refiners will face upon implementation of the NY MTBE ban."
Since the ethanol-oxygenated RFC formulation in this study was not blended with 10
volume percent ethanol and was not specific to Connecticut, EPA assumes that DEP
would not consider it to be representative of Connecticut's "no waiver" ethanol-
oxygenated RFC.
E. Information on likely RFG composition subsequent to a waiver of the
oxygen content requirement
DEP did not explicitly address this issue in the text of either submission letter.
Except as noted in the previous section and in footnote 7, DEP provided no description
of the potential composition of RFG were a waiver granted.
11
-------
IV. INFORMATION NEEDED TO EVALUATE NOX, VOC AND CO EMISSIONS
CHANGES RESULTING FROM AN OXYGEN CONTENT WAIVER
A. Background
RFC, in order to comply with statutory and regulatory requirements, must meet
certain emissions performance and content standards. The emission performance
standards are specified as percent reductions in NOx, VOC and toxics emissions
relative to emissions with a statutory baseline gasoline having properties representative
of 1990 gasoline. These reductions are calculated by the Complex Model, which
estimates the emissions that 1990 technology vehicles would achieve when using both
the statutory baseline gasoline and any RFC formulation that is being evaluated. The
Complex Model calculates a percent reduction between the baseline gasoline and the
RFC at issue. (The properties which are inputs to the Complex Model have previously
been identified in a footnote.) RFC is also subject to oxygen and benzene content
standards. It must contain, on average, at least 2.1 weight percent oxygen and no more
than 0.95 volume percent benzene. In addition to the requirements imposed by the
RFC standards, both RFC and conventional gasoline (CG) are also affected by the
Mobile Source Air Toxics (MSAT) regulations. These regulations cap Complex Model
toxics emissions at their 1998-2000 levels on a refinery-specific basis.10 The Tier 2
gasoline sulfur requirements, when fully implemented, will require an average sulfur
level of 30 ppm or less in both RFC and CG.11
RFC producers must meet these emission-related regulatory constraints, and
also must supply RFC and CG that meets octane and other driveability-related
requirements. Refiners attempt to produce gasoline and other products in the most
economically advantageous manner, subject to these constraints. The manner in which
a given refinery meets this objective depends on the configuration of the refinery, the
cost and availability of various blending components (e.g. MTBE, ethanol, and alkylate
which have good octane characteristics), and other technical and economic factors.
Consequently:
10 Since much RFC overcomplied with toxics performance requirements, MSAT imposes a
tighter constraint on RFC composition than the RFC performance standard for many
suppliers.
11 The sulfur level in RFC is currently constrained by RFC performance standards, but the
Tier 2 sulfur content requirement imposes a tighter constraint.
12
-------
RFC emission performance and content standards alone do not determine
RFG's composition. Other factors may provide additional constraints on certain
emission-related properties.
RFC composition without a waiver is not uniform across refiners and refineries.
Complex Model emission performance varies and much RFC overcomplies,
particularly with the NOx and toxics standards. Emissions performance for
Connecticut's RFC cannot be fully evaluated without estimates of all the relevant
fuel properties which are inputs to the Complex Model.
The same factors that lead to variety in RFC composition and emissions
performance without a waiver would lead to variety in RFC composition and
emissions performance with a waiver.
Even though RFC must meet the same Complex Model performance standards
with or without an oxygen waiver, the actual emissions performance may be
different with and without a waiver because the composition of the fuel is likely be
different with and without a waiver.
B. Emissions Models
Emission performance, as measured by the Complex Model, is a function of
several fuel parameters. The relationship between emissions and gasoline properties
was determined by statistical regression analysis, using several thousand vehicle
emission tests from a number of different studies. Development of the Complex Model,
or any comparable model, such as California's Predictive Model, requires a number of
decisions, some of which are subjective. Additionally, there are choices in the statistical
techniques that can be used to develop these models. Thus, different models, showing
different relationships between fuel properties and emissions for the same pollutant, can
reasonably result from the same data set.
EPA developed updated models for NOx and exhaust VOC in order to evaluate
California's RFC oxygen content waiver request, using data which California used to
develop its Predictive Model. Much of these data were also used to develop EPA's
Complex Model. These alternative models could be applied to a Connecticut waiver
analysis. While these EPA models differ from the Complex Model, particularly with
respect to the relationship of NOx and oxygen, these models require the same basic
fuel property information as inputs.
In order to evaluate the effect of a change in fuel properties on vehicle emissions,
it is common practice to separate the vehicle fleet into several technology groups. It is
widely believed, and supporting data indicate, that different technology groups respond
differently to fuel property or composition changes. In the California waiver evaluation,
vehicles were divided into three technology groups: older technology vehicles (pre-1986
13
-------
model years), mid-age technology vehicles (model years 1986 through 1995) and newer
vehicles (model year 1996+). These older, mid-age, and newer technology vehicles
were referred to as Tech 3, Tech 4, and Tech 5 vehicles respectively.
EPA's alternative NOx and exhaust VOC models, like the Complex Model, are
designed to predict emissions from only a portion of the gasoline mobile source fleet. In
its California waiver analysis, the alternative NOx and exhaust VOC models which were
developed represented the Tech 4 portion of California's light duty vehicle fleet. Based
on California's EMFAC7G emission model, Tech 4 vehicles are expected to be the
second largest contributor of the three tech groups to California's NOx emission
inventory and the largest to its exhaust VOC inventory in 2005. (While the Tech 5
vehicles may have the largest contribution of these three tech groups to the NOx
inventory, there were insufficient data to produce Tech 5 fuel effects emission models
comparable to the Tech 4 models, and engineering reasons to assume that Tech 5
vehicles may not be substantially affected by certain fuel parameter changes).
EPA assumed in the California waiver analysis that newer "Tech 5" vehicles
(model years 1996+) would not respond to changes in oxygen content or other fuel
parameter changes. EPA assumed, as does DEP, that fuel control systems in modern
fuel injected vehicles will compensate for the additional oxygen in gasoline.12
EPA used the "Tech 3" model portion of California's phase 3 predictive model
representing model year 1981-85 technology, to estimate emissions changes that would
occur in older vehicles. EPA also had emission weighting factors, which were part of
California's phase 3 predictive model and which estimated the relative contribution to
emissions in year 2005 of these three technology groups. This allowed EPA to combine
separate percent change emission estimates for each of these technology groups into a
single percent change estimate, and apply this percent change to an on-road gasoline
NOx and exhaust VOC emission inventory (in tons per day). This allowed EPA to
estimate overall on-road gasoline NOx and exhaust VOC no-waiver to waiver emission
differences in tons per day.
EPA believes that an approach utilizing these models, with weighting factors
appropriate for Connecticut, could be used by Connecticut to perform the appropriate
emissions modeling for on-road vehicles in a waiver application.
In its analysis of the California waiver, EPA assumed that changes in emissions
from off-highway or non-road engines would be a function of oxygen content only, using
12 While the assumption regarding no effect certainly does not apply to the relationship
between gasoline sulfur content and NOx performance for these newer Tech 5 vehicles,
in its California waiver analysis EPA was comparing waiver and non-waiver fuel
formulations which had very small differences in sulfur content. Thus, the assumption that
the sulfur content had no effect on "Tech 5" emissions introduced little or no error in that
analysis.
14
-------
factors (in percent change in emissions per weight percent of fuel oxygen added)
published in EPA Report No. NR-003, as the basis for its estimate of percent changes in
non-road emissions. Barring new information, it would be reasonable to apply these
same non-road factors in a Connecticut analysis.
In its California analysis, EPA had emission inventory information for both on-
road and non-road mobile sources. Thus, knowing oxygen content and other fuel
properties, and oxygenated/non-oxygenated market share for each waiver case, EPA
was able to estimate the percent difference in emissions between waiver and non-
waiver scenarios, apply this difference to the inventory to estimate a tons/day difference
and combine the on-road and non-road changes.
For the reasons described in EPA's California waiver analysis13, EPA assumes
that carbon monoxide emissions would be primarily affected by fuel oxygen content,
with CO emissions increasing as oxygen content decreases. For California, EPA
modeled on-road CO changes using "percent reduction per weight percent oxygen
increase factors" published in a California Air Resources board staff report.14 EPA used
factors from NR-003, cited earlier, to model off-road CO changes. The on-road percent
changes may be somewhat California-specific, since they were derived from California's
EMFAC emission factors model. It is probably more appropriate to quantify the CO
emission changes in Connecticut using the Mobile model.
In summary,
There are models which may be superior to the Complex Model for evaluation of
on-road vehicle emissions differences in Connecticut with and without an
oxygenate waiver.
These alternative models require essentially the same fuel property information
as the Complex Model, and emission performance for Connecticut's RFC cannot
be evaluated using these models without estimates of all the fuel property inputs.
It is reasonable to estimate off-road exhaust emission and on-road CO changes
from oxygen changes alone and the methods to estimate these emission
changes could (if the requisite information were available) be used in evaluating
a waiver request in Connecticut.
13 EPA Docket A-2000-10, Docket Number ll-B-2.
14 California Air Resources Board (GARB) Staff Report: Initial Statement of Reasons;
Proposed California Phase 3 Reformulated Gasoline Regulations; released October 22,
1999.
15
-------
C. RFG Properties and Composition
Knowledge of the likely formulation of RFG in Connecticut with and without an
oxygen content waiver is critical to EPA's evaluation of whether or not the oxygen
content mandate interfered with attainment of the ozone or particulate matter NAAQS.
Because this information has not been provided, EPA is not able to conclude what
effect a waiver would have on vehicle emissions in the Connecticut RFG area. Some
information exists regarding the composition and properties of ethanol-oxygenated RFG
15 supplied to Connecticut during the ozone season subsequent to Connecticut's MTBE
ban. This RFG does not, on average, meet the ultimate 30 ppm sulfur requirement
which the Tier 2 gasoline sulfur regulations will impose on both conventional and
reformulated gasoline, but reflects transitional sulfur requirements which become more
stringent over a multi-year period. Even though these data do not represent full
implementation of the Tier 2 sulfur requirement, they are likely to be somewhat
informative of the composition of the RFG that will be supplied to Connecticut in the
future in the absence of an oxygen content waiver.16 However, these data do not
predict the composition and properties of the RFG that would be supplied to
Connecticut if a waiver were granted. Consequently, this information, by itself, is
insufficient to evaluate the impact of an oxygen content waiver.
Connecticut's submissions provide no source of real world data or predicted data
that is informative of the composition of the RFG that would be supplied to Connecticut
if EPA waived the oxygen content requirement. In its waiver submission, California
relied on refinery modeling studies to predict the properties of California Phase 3 RFG
(CaRFGS) and, similarly, EPA relied on expansions of these same refinery modeling
studies to evaluate California's submission. These studies predicted the properties of
the CaRFGS that would be produced if an oxygen content waiver were granted, as well
as the properties of the oxygenated CaRFGS that would be produced without a waiver.
This modeling was conducted on those refineries supplying RFG to California, so is
15 RFG surveys are conducted each year by the RFG Survey Association, an association of
refiners, importers and oxygenate blenders. Gasoline samples are collected at retail
stations in RFG covered areas, under a survey plan which EPA approves, and analyzed
for emission-related properties. Independent contractors perform the sample collection
and analysis. These surveys are required by 40 CFR 80.67, for suppliers who choose to
meet RFG standards on an average basis, rather than on a "per gallon" basis. These
surveys are intended to ensure that the RFG supplied to each covered area meets
standards. Specific requirements pertaining to these surveys can be found in 40 CFR
80.68.
1R
EPA has published average property information from RFG surveys through 2003.
Although 2004 averages have not been published, EPA would have provided data from
2004 surveys to Connecticut DEP as soon as it was available, had it been requested.
Ozone-season RFG surveys in CT were conducted between June and late August. EPA
receives data approximately 30 days after each survey is completed and received all
ozone season survey data for CT prior to October, 2004.
16
-------
California-specific.17 Furthermore, California reformulated gasoline standards, which
are applicable to RFC sold in federal RFC areas in California, differ from federal RFC
standards. Thus, the predictions of the California fuel properties from this modeling are
not applicable to the analysis of the Connecticut waiver since different modeling
constraints would be applied for the California case versus the Connecticut case. For
example, modeling indicated that it was infeasible for California refiners to make RFC
oxygenated at 10% ethanol, while Connecticut is likely to receive 10% ethanol during
the ozone season.18 The California modeling does predict that substantial amounts of
ethanol-oxygenated gasoline would be used in California, even with a waiver of the
federal RFC oxygen content requirement. Although the California analysis is certainly
not directly applicable to Connecticut, the expected use of ethanol in California
demonstrates that, lacking information to the contrary, EPA cannot assume that ethanol
will not be used in Connecticut in a waiver situation. Ethanol may be used to make up,
in part, for the volume and octane which MTBE provided even if EPA waived the oxygen
content requirement. Other studies support such a conclusion.19
17 In fact, refinery modelers typically use completely different models to estimate the
composition of gasoline produced for the Connecticut area versus gasoline produced for
the California market.
1 ft
The modeling which EPA commissioned for the California waiver analysis represented the
characteristics of the California refinery system. Additionally, the modeling required that
California Phase 3 RFC meet California standards for NOx and hydrocarbons based on
California's Phase 3 predictive model, rather than the Complex Model. Thus, the RFC
modeled for the California waiver analysis was required to meet a much more stringent
NOx standard than the Federal standard. Additionally, since the predictive model, unlike
the Complex Model, indicates that NOx emissions will increase as oxygen content
increases, California's NOx performance requirements impose a practical limit on weight
percentage of oxygen that can be used in California blends. Consequently, EPA has little
confidence that the forecasts of emission-related fuel parameters which these models
produced would be applicable to Connecticut.
19 The NESCAUM report submitted as enclosure B of NY's January 6, 2003 submission
(Docket OAR-2003-0004) states on page 9, vol.2 "If the oxygen requirement were waived
or lifted, it is highly likely that some amount of ethanol would still be blended into RFC..."
17
-------
V. EMISSIONS-RELATED CONCLUSIONS BASED ON CONNECTICUT'S
SUBMISSION
A. Effect of a waiver on NOx emissions based on Connecticut's submission
1. Multiple gasoline properties affect NOx emissions
With regard to DEP's comments regarding the complex Model, EPA does agree
that the Complex Model may over-estimate the NOx emission performance of certain 10
vol. % (about 3.5 wt% oxygen) ethanol-oxygenated RFC blends, because the Complex
Model does not show a NOx increase with increasing oxygen. According to EPA's
models developed during the California waiver analysis and readily available to
Connecticut DEP, all other properties being equal, as oxygen content increases, NOx
emissions will also increase for Tech 4 vehicles. Thus, if two RFC blends with different
oxygen content were formulated to have the exact same Complex Model NOx
performance (e.g. 6.8 percent reduction from 1990 baseline, the averaged standard),
EPA would suspect that the "true" NOx performance of the blend with the higher oxygen
content would be worse than that of the blend with the lower oxygen content. The
problem with this simplistic argument is that there are many properties and components
of RFC other than oxygen content which would likely be different in a waiver versus no-
waiver scenario, and these other properties and components also affect NOx emissions.
(This is discussed in the previous section of this TSD.) Thus, neither the magnitude nor
direction of a NOx change can be predicted using the data submitted by DEP since the
argument above is based only on oxygen content.20
20 Although oxygen increases taken alone (i.e., all other properties being equal) directionally
increase NOx emissions in EPA's alternative models for Tech 4 vehicles, real world
situations clearly show that other RFC properties can and do change the magnitude and
even the direction of this change. In EPA's evaluation of California's waiver request using
the alternative models, NOx emissions clearly go down where average oxygen content of
the gasoline pool is lower (i.e., all of the waiver scenarios). However, in other real world
situations fuels can exhibit better NOx performance even with higher oxygen content due
to differences in other properties that affect NOx emissions. To further illustrate this point,
EPA has examined "real world" VOC-controlled RFC survey data collected during the
2003 ozone season and computed two sets of average fuel properties, one from an RFC
area with lower oxygen content and one from an area with higher oxygen content. EPA
utilized averages from surveys conducted in the New York-New Jersey-Long Island-
Connecticut area to represent the lower oxygen content RFC, and averages from
Louisville, Kentucky surveys to represent the higher oxygen content RFC. EPA computed
the NOx performance for these two "average fuels" using EPA's alternative NOx models.
18
-------
2. NOx emissions from on-road vehicles with and without a waiver.
Given property estimates for Connecticut's "waiver" and "no-waiver" gasolines,
a complete waiver submission would presumably use EPA's alternative "Tech 4" NOx
models to estimate the emissions change for a portion of the on-road fleet, since these
models were developed with data that are not California-specific. As explained earlier,
the information which DEP has submitted is insufficient to even determine the expected
direction of the "Tech 4" "no waiver" to "waiver" NOx emission change.
An assumption similar to that made for the California analysis, i.e., that the "no
waiver" to "waiver" fuel property differences would have no effect on exhaust emissions
in "Tech 5" vehicles, would likely be appropriate for Connecticut as long as there was
little difference in the sulfur content of the "no waiver" and "waiver" fuels. While this is
somewhat uncertain prior to the implementation of the 30 ppm Tier 2 standard, this
should be the case with that standard in place. Consequently, it would likely be
appropriate to assume that "Tech 5" "no waiver" to "waiver" emission differences would
be zero.
It would probably be adequate to use the "Tech 3" NOx model portion of
California's phase 3 predictive model to estimate NOx emission changes for older
vehicles. This model, although it was created by California to represent a portion of
their fleet, was not based solely on California vehicles. This model requires the same
fuel parameter inputs as the alternative "Tech 4" NOx models which EPA developed.
Thus, it cannot be used to determine the probable direction of the NOx emission change
without estimates of Connecticut's "no waiver" and "waiver" RFC parameters.
EPA applied California-specific emission weighting factors for Tech 3, 4 and 5
vehicles in its California analysis. They represent the California fleet in 2005 and were
estimated using a California-specific emission factors model. Weighting factors more
applicable to the federal fleet and/or to the Connecticut fleet could be developed with
EPA's Mobile model. However, DEP did not submit the fuel parameter information to
calculate the Tech 3 or Tech 4 emission changes, nor did it develop weighting factors
for the various Tech groups. Therefore, it is not possible, given the data submitted, to
draw any reasonable conclusions about the magnitude or the direction of NOx
emissions changes in on-road vehicles for the Connecticut waiver/no-waiver
comparison.
The NOx performance of the Louisville, KY "average fuel" is better than that of the New
York "average fuel" based on the EPA alternative model, even though the oxygen content
of the Louisville "average fuel" was higher. (See Docket OAR-2003-0004 Document II-B-
1.)
19
-------
3. NOx emissions from non-road vehicles with and without a waiver
EPA assumes that granting an oxygen content waiver would result in some
decrease in the average oxygen content of Connecticut's RFC.21 In that case, there
would be some percent decrease in non-road NOx emissions with an oxygen content
waiver. However, EPA cannot estimate how much non-oxygenated RFC would be used
if a waiver were granted in Connecticut. Consequently, EPA can only qualitatively
estimate a directional decrease in NOx for non-road vehicles and engines with a waiver.
DEP has not submitted any NOx emission inventory information as part of its
waiver request. While the non-road percent change NOx factors are large, EPA
assumes that the non-road gasoline NOx emissions inventory for Connecticut is small
compared to the on-road gasoline NOx emissions inventory. Consequently, even if
EPA concludes that non-road gasoline NOx emissions will be lower with a waiver, EPA
cannot determine with confidence, based only on a qualitative assessment, whether
total mobile source gasoline NOx emissions would be lower with or without a waiver.
B. Effect of a waiver on VOC emissions based on Connecticut's submission
Evaluating the potential impact of ethanol-oxygenated gasoline on NOx
emissions between a waiver and no-waiver scenario only calls for considering changes
in exhaust emissions. Analysis of VOC impacts is considerably more complex. In
addition to the changes in on-road and non-road exhaust emissions, one must also
consider evaporative VOC emission changes that may occur as a result of "as-blended"
RVP differences between various gasoline formulations.22 There are also other
emission effects that may occur as a result of ethanol use. Commingling of ethanol and
non-ethanol gasolines in vehicle fuel tanks causes increases in RVP which can result in
higher evaporative VOC emissions. Additionally, use of ethanol-oxygenated gasoline is
believed to increase VOC permeation emissions.
21 Although without the appropriate analysis, it cannot be said with any certainty that the
waiver would result in the production of non-oxygenated RFC. At least one refiner has
strongly supported granting New York's request for a waiver (Docket OAR-2003-0004) .
The letter of support from Tosco Corporation enclosed with New York's January 6, 2003
submission clearly indicates that less oxygen would be utilized in New York RFC if a
waiver were granted. One would presume that support for a waiver from the refining
community would indicate that at least some refiners would blend RFC without oxygen or
the waiver would have no effect on RFC production.
22 "As-blended" RVP refers to the RVP resulting from blending at a refinery or terminal. It is
distinct from the RVP that may exist in commingled fuel in a vehicle fuel tank.
20
-------
1. VOC on-road and non-road exhaust emissions
As previously noted, for its California waiver analysis EPA developed alternative
exhaust hydrocarbon models, estimating changes in exhaust hydrocarbon emissions as
a function of various fuel parameters. For Tech 4 vehicles (Model Year 1986-1995),
these models predict that as oxygen content increases from zero, with other
parameters constant, exhaust hydrocarbon (HC) will first decrease and then increase.23
According to these different models, therefore, decreasing oxygen content from a 3.5%
"no waiver" level, in the absence of other parameter changes, could either reduce or
increase exhaust HC emissions from Tech 4 vehicles, depending on the "waiver"
oxygen content. However, these HC models are a function of several parameters, i.e.,
the same parameters that are inputs to the alternative NOx models identified earlier. As
discussed and demonstrated earlier with respect to NOx, knowing only that oxygen
content is likely to decrease with a waiver, and even knowing the pool average oxygen
content with a waiver, does not provide enough information to determine the net effect
on emissions, because changes in other parameters could offset emission changes due
to oxygen. Consequently, without further information, EPA cannot predict even the
direction of the exhaust VOC emission differences between "no waiver" and "waiver"
RFC in Tech 4 vehicles.
EPA believes that the assumptions discussed earlier for Tech 3 and Tech 5 NOx
emissions are valid for Tech 3 and Tech 5 exhaust HC as well; that exhaust HC will
probably not be substantially affected by oxygen content or other parameters for Tech 5
vehicles, and the "Tech 3" exhaust HC portion of California's Phase 3 Predictive Model
may be the best available model to predict fuel-related exhaust HC changes in older
vehicles. However, this Tech 3 model requires the same fuel parameter inputs as the
alternative "Tech 4" HC models which EPA developed.
In its California analysis, EPA used Tech 3, Tech 4 and Tech 5 exhaust HC
weighting factors. As with NOx, these weighting factors were derived with California's
mobile source emissions model, and incorporated into the predictive model. The
emission weighting factors which were applied in the California analysis are California-
specific; they represent the California fleet in 2005 and were estimated using a
California-specific emission factors model. Weighting factors more applicable to the
federal fleet and/or to the Connecticut fleet could be developed with EPA's Mobile
model. However, DEP did not submit the fuel parameter information to calculate the
Tech 3 or Tech 4 emission changes nor did it develop weighting factors for the various
Tech groups. Therefore, it is not possible, given the data submitted to make
conclusions about either the magnitude or the direction of exhaust VOC emissions from
on-road vehicles for the waiver/no-waiver comparison.
23 While for certain discussions, it may be useful to make a distinction between VOC
emissions and HC emissions, for purposes of the discussion here, there is no difference
between references to VOC and HC emissions. Generally speaking, we use the term HC
as the appropriate convention when referring to emissions models.
21
-------
As we did in the California analysis, EPA assumes that non-road exhaust VOC
emission changes are largely a function of oxygen content, and that exhaust VOC
emissions would increase as oxygen content decreases.24 Thus EPA would expect an
increase in non-road exhaust VOC emissions with a waiver. It would probably be
appropriate to model these emissions for Connecticut using the factors which EPA used
in the California analysis. These factors are incorporated into EPA's NONROAD2004
model, and this model could be used directly to estimate nonroad exhaust emission
levels, in tons per day, for the RFC areas in Connecticut in the absence of a waiver.
However, since DEP has not addressed the issue of non-oxygenated market share with
a waiver and has not provided an estimate of pool average oxygen content with a
waiver, it is not possible to estimate the magnitude of the "no waiver" to "waiver"
emission change in non-road VOC exhaust emissions.
2. Evaporative VOC emissions due to RVP increases
Non-exhaust emissions of VOC (except permeation) are modeled as a function
of RVP only.25 In other words, if estimates are available of RVP changes between the
waiver and non-waiver scenarios, existing models can be used to predict non-
permeation changes in evaporative emissions of VOC. The RVP level in ozone season
RFC is tightly constrained by the need to comply with Complex Model VOC
performance standards.26
As described earlier, an RVP boost and associated increase in VOC emissions
can occur within the gas tanks of cars when fuel containing ethanol is commingled with
fuel that does not contain ethanol. Connecticut asserts that commingling will increase
when ethanol is substituted for MTBE as an oxygenate in RFC. Presumably
Connecticut is referring to the commingling of ethanol-oxygenated RFC purchased
within the Connecticut RFC area with non-ethanol gasolines purchased outside of the
Connecticut RFC area. Connecticut does not attempt to quantify this effect. While this
may occur in a no-waiver scenario, Connecticut does not address the central question
24 Available data and models do not accommodate consideration of additional fuel
properties.
25 That is, RVP is the only fuel parameter input assumed to affect evaporative emissions.
The evaporative emission estimates produced by EPA's Mobile model are also affected
by ambient temperature inputs and other area-specific factors. In its California waiver
analysis, EPA used a formula, derived with California's EMFAC7G model, to estimate the
VOC emission differences resulting from RVP differences. It would be appropriate to use
a non-California model such as the Mobile model or the Complex model to model any
relevant RVP-related "no waiver" to "waiver" evaporative emissions changes in
Connecticut.
26 See 40 CFR 80.41 (e) and (f).
22
-------
of how much ethanol-oxygenated and non-oxygenated RFC would exist within the
Connecticut RFC area if an oxygen content waiver were granted.
EPA believes that the maximum adverse commingling impact occurs when
ethanol and non-ethanol gasolines are sold in significant quantities in the same
geographic area (e.g. the Connecticut RFC area). In its California waiver analysis, EPA
concluded that this condition could exist if California were granted an oxygen content
waiver, because the refinery modeling indicated that some refiners would blend ethanol
in RFC even if the waiver was granted, while other refiners would not. As noted
previously, DEP has not addressed the issue of possible use of ethanol in Connecticut
RFC if a waiver were granted. EPA believes that some refiners supplying RFC to
Connecticut would blend ethanol into RFC even if a waiver was granted, due to the
octane ethanol provides and to make up for the lost gasoline volume from MTBE
resulting from New York's MTBE ban. As discussed previously, there is no basis for
EPA to assume that under a waiver, all RFC will be free of ethanol. Consequently, the
maximum adverse commingling impact could occur in Connecticut if EPA grants an
oxygen content waiver, leading to corresponding increases in RVP and evaporative
VOC emissions. This maximum adverse commingling impact condition would not exist
in Connecticut without a waiver, since without a waiver only ethanol-oxygenated RFC
would be sold within the Connecticut RFC areas.
In summary, while some commingling could occur in the non-waiver scenario as
cars travel back and forth from Connecticut's RFC areas (where ethanol is used) to
other areas (where it may or may not be used), such commingling would likely be much
less prevalent than would be the case with an oxygen waiver. In the absence of
modeling showing how much ethanol will be used in Connecticut's RFC if the oxygen
content requirement were removed, EPA believes it most likely that commingling will be
more prevalent with a waiver than without a waiver. It follows that VOC evaporative
emissions resulting from commingling would likely be greater if a waiver were granted
than if it were not, however without further explanation it is not possible to estimate the
magnitude of this impact.
3. Evaporative VOC emissions due to permeation
EPA assumes that increased ethanol use would cause some increase in
permeation-related VOC emissions. However, DEP has not provided a quantitative
estimate of the increase in emissions that could occur. Ethanol may still be used in
some significant quantities even if a waiver were granted. EPA assumes that the "no
waiver" to "waiver" difference in permeation effect would be dependent on the amount of
ethanol used if a waiver were granted.
The change in permeation emissions between the non-waiver and waiver
conditions depends on the extent to which ethanol RFC would be used were a waiver
granted, and DEP has not provided estimates of ethanol use in RFC with a waiver.
Additionally, estimates of permeation emission changes would have to be combined
23
-------
with estimates of exhaust and other evaporative VOC emission changes in order to
estimate the total "no waiver" to "waiver" VOC change. DEP has not provided such
estimates or the underlying fuel property information needed to make such estimates.
Thus, even with the best available quantitative estimate of permeation emission
differences between ethanol-oxygenated and non-ethanol gasoline, EPA could still not
determine the net effect of the oxygen waiver on VOC emissions with the information
submitted by DEP. In summary, EPA believes that permeation-related VOC emissions
would likely be less with a waiver, because we assume that less ethanol will be used
with a waiver than without. However, without further information it is not possible to
estimate the magnitude of this impact.
C. Effect of a waiver on CO emissions
DEP did not address emissions of CO, although increases in CO can contribute
to increases in ozone Because of the lack of information on ethanol market share
should a waiver be granted, there is insufficient information for a quantitative estimate of
"waiver" to "no waiver" CO emission changes for Connecticut, since EPA cannot
determine the average oxygen content for the "no waiver" case or cases from the
available information. Directionally, EPA assumes that average oxygen content will
decrease with a waiver. CO emissions should increase for Tech 3 , Tech 4 and non-
road, and remain the same for Tech 5. While EPA can estimate this directional impact,
without further information it is not possible to estimate the magnitude of this impact.
D. Ethanol transport-related emissions
While there will be emissions resulting from the transport of ethanol to terminals,
the difference in emissions would depend, in part, on the amount of ethanol used in
Connecticut with and without a waiver. DEP has not provided that information.
Therefore, EPA concludes that without further information it is not possible to estimate
the magnitude of this impact.
24
-------
VI. CONCLUSIONS OF ERA'S ANALYSIS OF PEP'S SUBMISSION
The information that DEP has provided fails to demonstrate what effect a waiver
would have an on ozone or particulate matter levels in Connecticut. This is because: 1)
there are three pollutants whose emission rates could be altered by a waiver (NOx, CO
and VOC) and all three affect ozone formation to varying degrees; 2) the lack of
information on fuel qualities with and without a waiver and the lack of other relevant and
necessary information precludes even a directional estimate of the impact of a waiver
on NOx and VOC emissions; 3) the best estimate of the net impact of a waiver on CO
emissions is that CO emissions would be greater with a waiver than without, but the
difference cannot be quantified; 4) no analysis has been provided or performed, and the
information before the agency does not allow an analysis to be performed, on the
combined effect of these emissions changes on ozone.
In addition, (1) NOx emissions affect PM, (2) the lack of information on fuel
qualities with and without a waiver and the lack of other relevant and necessary
information precludes even a directional estimate of the impact of a waiver on NOx, (3)
no analysis has been provided or performed, and the information before the agency
does not allow an analysis to be performed, on the effect of any NOx emissions
changes on particulate matter.27
27 The table in Appendix A summarizes EPA's qualitative consideration of Connecticut's waiver
request.
25
-------
APPENDIX A: SUMMARY OF EPA'S ANALYSIS OF DEC'S SUBMISSION
The following table summarizes EPA's assessment of the directional change in "no
waiver" to "waiver" emissions for each on-road vehicle technology group, and for non-
road gasoline vehicles and engines. T indicates an increase, "D" a decrease, "0" little
or no change, and "?" means insufficient information to determine.
Tech 3
on-road
Tech 4
on-road
Tech 5
on-road
non-road
NOx
?
?
0
D
Exhaust
VOC
?
?
0
1
as blended
evap VOC
?
?
?
?
commingling
VOC28
lor?
lor?
lor?
lor?
Permeation
VOC
D
D
D
D
CO
1
1
0
1
28
If ethanol is used in Connecticut's RFC to a significant extent with a waiver, then EPA
expects commingling VOC emissions to increase. If ethanol is not used extensively with
a waiver, then the direction of this VOC change is uncertain. EPA believes that significant
ethanol use with a waiver is likely.
26
-------
27
------- |