EPA-AA-TSS-PA-88-1
Technical Report
Derivation of Technology Specific Effects
of the Use of Oxygenated Fuel Blends on
Motor Vehicle Exhaust Emissions
October, 1988
NOTICE
Technical Reports do not necessarily represent final EPA
decisions or positions. They are intended to present
technical analysis of issues using data which are
currently available. The purpose in the release of such
reports is to facilitate the exchange of technical
information and to inform the public of technical
developments which may form the basis for a final EPA
decision, position or regulatory action.
Technical Support Staff
Emission Control Technology Division
Office of Mobile Sources
Office of Air and Radiation
U. S. Environmental Protection Agency
-------
1.0 INTRODUCTION
In the past decade there has been a large amount of
research into the effects of fuel composition on the emissions
of motor vehicles. Blends of gasoline with oxygenates such as
alcohols and ethers have received particular attention, and it
has been widely demonstrated that use of these blends can
reduce emissions of carbon monoxide (CO). Some organizations
have proposed that this beneficial effect could help alleviate
CO air quality problems which have been experienced in some
areas. It has also been shown that such blends can affect
exhaust and evaporative emissions of volatile organic compounds
(VOC) and exhaust emissions of oxides of nitrogen (NOx). The
blends generally cause small increases in NOx emissions, and
they can cause increases or reductions in the net emissions of
VOC, depending on the blending components and volatility of the
final product. State and local governments have proposed
including programs of increased use of these blends in their
plans in order to help local areas meet the National Ambient
Air Quality Standard for CO.
To assist local and state planners in considering this
issue, EPA has developed guidance for estimating the fleetwide
effects that these fuels can have on vehicular emissions of CO,
VOC, and NOx. The technical report, "Guidance on Estimating
Motor Vehicle Emission Reductions from the Use of Alternative
Fuels and Fuel Blends"1, discusses EPA's procedure- for
estimating these fleetwide effects. A draft of the report was
released for public comment in July, 1987, and the final
version was released in January, 1988.
The procedure is based on estimates of the effects on
emissions that gasoline/oxygenate blends can have on groups of
vehicles within the fleet. This report discusses the methods
and data which EPA used in calculating these estimates. Only
the data used in the development of the guidance report are
discussed here; no additional data have been included. Some
errors in calculation (Data were duplicated on more than one
reference) were discovered and corrected after the guidance
report was released. Therefore, the effects which are
presented here are different from those listed in the guidance
report. These differences are small.
It should be noted that all of the calculation steps
presented in this report are intermediate steps in a larger
calculation described in the Guidance Report. To prevent the
proliferation of truncation errors, numerical figures are not
truncated to their known number of significant digits.
-------
-2-
1.1 Effects of Fuel Oxygen Content; a General Discussion
Most vehicles meter fuel by volume with the goal of
controlling the ratio of air to fuel. A useful reference point
in discussing air/fuel ratios is the stoichiometric point, at
which all of the fuel can theoretically react with the oxygen
in the air. When the fuel control system of a vehicle puts too
much fuel into the engine at a time (a condition known as
"running rich"), there will not be enough oxygen to allow
complete combustion of the fuel. The result will be increased
amounts of the products of incomplete combustion, CO and VOC,
in the exhaust. Most gasoline fueled engines run rich some of
the time, and it is during this time that much of the emissions
of these pollutants occur.
The feature of gasoline/alcohol and gasoline/ether blends
which has the largest effect on emissions is oxygen content.
When these fuels are used, the oxygen in the fuel reduces the
amount of air needed to burn a given amount of fuel in two
ways. It increases the amount of oxygen which is available for
combustion, and it displaces some of the carbon which would
otherwise consume oxygen. This effect is called "enleanment."
A vehicle which uses a gasoline/oxygenate blend fuel will spend
less time running rich, and will have less severe richness
during that time than when it uses a non-oxygenated fuel. A
vehicle which spends a large portion of its operating time in a
rich mode will have higher base emissions than another vehicle
which is seldom running rich. The use of oxygenated blends
will cause a larger reduction in the emissions of the former
vehicle than the latter.
1.2 Quantifying These Effects
It is difficult to quantify the effects of fuel oxygen on
exhaust emissions both in terms of predicting the effects on
the emissions of a vehicle and in terms of predicting- the
effects on the emissions of a fleet of vehicles. This
difficulty stems from the wide variety of other factors which
affect emissions and can interact with fuel oxygen content to
further affect emissions. Such factors include, but are not
limited to the age of the vehicle, the emission control
technology used in the vehicle design, the maintenance history
of the vehicle, the ambient operating temperature, the altitude
of operation, the initial calibration of the fuel metering
system, and the fuel volatility. A further complicating factor
is the wide range of emission levels from different vehicles in
a given fleet and the non-normal distribution of these levels.
These issues are discussed in more detail in Section 3.
-------
-3-
EPA has examined many studies in an attempt to predict the
effects that gasoline/oxygenate blend fuels will have on
vehicle emissions. This document presents in a consistent form
all available and relevant data which EPA has used in its
examination of this issue. Also discussed are the methods
which EPA used to evaluate these data, and the conclusions
which EPA has drawn from this effort.
2.0 THE DATA BASE
EPA applied a number of criteria to the data from each
known vehicle test program before including it in the data
base. To be included, each program had to report emissions
data from Federal Test Procedures performed on vehicles using
at least two different fuels. Preferably, one of these fuels
would be similar to non-oxygenated gasolines currently
available at retail outlets, while the other fuel(s) would
represent a gasoline/oxygenate blend which could be sold at a
retail outlet. EPA excluded data on vehicles using certain
fuel blends, such as blends of ethanol in greater than 10
percent concentrations, blends of methanol in greater than 5
percent concentrations, and blends of methanol without
cosolvents. In one case a low level methanol/cosolvent blend
was excluded because it had an unrepresentatively low RVP of
8.0 psi. The data include test results from high and. low
altitude facilities, results of tests on light duty and heavy
duty gasoline fueled vehicles and trucks, results of tests
using blends of gasoline with ethanol, MTBE, and
methanol/cosolvent. No data from low temperature emission
tests are included. The resulting data base includes tests of
about 350 vehicles in 21 studies.
2.1 Sources
EPA has considered data which were found in its own
extensive literature searches, as well as all of the data which
were brought to light by commenters to the July 1987 draft, of
the report, "Guidance on Estimating Motor Vehicle Emission
Reductions from the Use of Alternative Fuels and Fuel Blends".
The data were gathered by State and Federal governmental
testing programs, programs conducted by the petroleum- and
automobile industries, and independent testing laboratories.
Two organizations offered large lists of additional sources of
data.
Energy and Environmental Analysis, Inc. (EEA) submitted a
report which it prepared for the Maricopa Association of
Governments, "Feasibility of Using Alternative Fuels as an Air
Pollution Control Strategy,"2 which contains a list and
grouped analysis of thirty-two studies. Some of the data
-------
-4-
contained in these studies were excluded from EPA's final
analysis because of the characteristics of the fuels which they
used, as was discussed previously, but other studies referenced
in this list provided EPA with additional information about the
effects of blends at high altitudes and emissions of vehicles
using fuels with intermediate levels of oxygen content. (An
oxygen content of 3.7% by weight is the maximum allowable under
existing regulations. Some blends, especially those which
contain MTBE, have oxygen contents close to 2.0%.) Nine of the
studies cited by EEA are directly included in EPA's data base.
Several others are referenced in a summary study which EPA also
included.
The Ad Hoc Ethanol Committee submitted extensive comments
including a list of thirteen studies. There was considerable
overlap between this list and the references of the EEA
report. The EPA data base was expanded to include six
additional studies from this list. Most of these compare a
blend of 10% ethanol with gasoline. A large proportion of the
vehicles covered have open loop fuel control.
EPA received additional data from the Colorado Department
of Health (CDH)5 during the comment period, some of which
were also referenced in the EEA report. The testing programs
of CDH have provided the majority of EPA's information on the
effects of blends at high altitude. They cover a wide range of
vehicle technologies and fuel types.
The CDH laboratory and more recently EPA's Office of
Mobile Sources have conducted testing programs on the effects
of long term use of gasoline/oxygenate blends in late model
vehicles with "adaptive learning systems." These systems
increase the control of vehicle designers over the air/fuel
ratio during the operation of the vehicle. Some vehicle
manufacturers have claimed that these systems can almost
completely negate the effects of fuel oxygen content on
emissions after the vehicle has used a blend for a certain
amount of time. These two laboratories have been testing
vehicles equipped with such systems, and the results of these
tests are also included in this analysis.
3.0 ISSUES IN ANALYSIS OF DATA
EPA considered several issues in the process of analyzing
the data. Since the database includes information from so many
studies of different fuels and vehicles, care must be taken to
combine the results in a manner which will give the best
approximation of the fleetwide effects of a particular fuel.
It is also important to separate groups of data in which
vehicles could be expected to respond in different ways to the
presence of oxygenates in fuel. Factors which were considered
when forming these groups are discussed later in this report;
-------
-5-
3.1 Choosing a Model for Effects on Emissions
A model must be developed which predicts the emission
effects of oxygenated fuel blends within each appropriately
segregated vehicle group. These effects will be weighted and
combined as described in Appendix D of the Guidance Document.
There are several ways in which these effects could be
modeled. The simplest of these is to assume that a given
oxygenated blend causes an offset, in grams per mile, above or
below the base emission level. This model is easy to develop
and apply, but, as has been discussed before, vehicles with
already low emissions should not, theoretically, see as much.of
an absolute reduction as a group of high emitters.
Another simple model could predict emissions changes in
each group as a constant percentage of the base emissions. In
this model, high emitters would achieve a larger absolute
change than low emitters. This model is consistent with other
models which have been developed4, but it does not allow for
the possibility that vehicles with already very low emissions
would see little or no benefit from the enleanment effect of
blends.
It has been proposed2 that quadratic or other non-linear
models could be used to describe the effect of oxygen content
on emissions. There are two problems with this approach. The
first is in developing the model; there is so much scatter in
the data that deciding which model to use and determining >the
parameters of such a model can be a difficult and arbitrary
procedure. The other problem with non-linear models (and-even
linear but non-proportional models) is that their application
to a fleet is a complicated process. To apply a non-linear
model properly it is necessary to know the distribution of
emissions within a given group and integrate the results of the
model over the whole group. It would be very difficult to
predict and to use the emissions distributions of the many
groups that would be required to obtain a fleetwide effect.
EPA's emission factor model, MOBILE3, does not provide such
distributions.
EPA has elected to use the model of percentage changes.
The average emission reduction for the portion of the vehicle
fleet of a given type is taken to be a constant percentage of
those vehicles' base emissions on oxygen free gasoline,
regardless of average base emissions, age, odometer, speed, and
ambient temperature. The constant percentage is calculated as
described below from the available test data on vehicles of the
given type. This model fits observed patterns in the data
better than the constant mass reduction model, and it is easier
to apply than a non-linear model would be.
-------
-6-
Most analyses of emission data have used the technique of
comparing means to report emission changes. At least one
report (EEA) has used regression techniques to make predictions
of emissions changes which depend on the base emissions
themselves. EPA recognizes the advantages that regression
techniques can offer in such analyses, such as the separation
of the effects of several independent variables. However, the
results present among the studies which EPA considered were
widely scattered and not normally distributed. Curve fitting
techniques are less convincing when applied to data such as
emissions results which are not normally distributed and have
large amounts of scatter. Therefore, EPA has elected to use
the simpler technique of finding changes in mean results.
The primary drawback to this method is that there is no
way to perform statistical tests of significance. Since the
data are lumped together in groups of different sizes (the
number of vehicles tested in each study varies widely), and
since many of the study wide means are reported without
information about their respective variances, there is no way
to examine the distribution of individual points in the data
base. Therefore, the interpreted significance of many of the
effects which will be discussed in this report must be based on
engineering judgement. It is likely that such judgement would
be equally important even if all of the individual data points
were known, because of the non-normality of emissions
distributions and the wide ranges of emissions from different
vehicles, which would make statistical tests less valid than
they would otherwise be.
3.2 Combining Many Studies in a Single Database
EPA has attempted in this analysis to combine the results
of as many studies as possible in a consistent manner so that
the aggregate could be extrapolated to in-use vehicles. In
this process, EPA was careful to consider the differences among
the studies and between the studies and in-use conditions.
The emission levels of the vehicles used in many of the
individual studies are not representative of emissions which
occur with in-use vehicles. Table 1 shows the emission levels
which EPA's mobile source emission factor model (MOBILES)
predicts for vehicles operating at low altitude in 1990. Three
vehicle classes are shown: light duty gasoline fueled vehicles,
and two classes of light duty gasoline fueled trucks. (Trucks
in the LDGT1 class weigh 6000 Ibs or less, and LDGT2 includes
trucks which weigh between 6000 and 8500 Ibs.) Within each
vehicle class, the vehicles are split into groups by model
year, to represent the different emission control technologies
which have been most common in different model years.
-------
-7-
Table 1
Emissions of In-Use Vehicles in 1990*
LDGV
Model yrs
HC (g/mi)
CO (g/mi)
NOx (g/mi)
LDGT1
Model yrs
HC (g/mi)
CO (g/mi)
NOx (g/mi)
LDGT2
Model yrs
HC (g/mi)
CO (g/mi)
NOx (g/mi)
NO
Catalyst
1968-74
6.15
80.44
3.98
1968-74
6.94
89.58
4.13
1968-78
9.02
99.94
5.61
Oxidation
Catalyst
1975-80
4.16
47.90
3.36
1975-83
5.41
58.05
3.88
1979-83
5.01
51.21
3.83
Closed
Loop/3way
1981-83
1.57
21.97
1.71
1984-87
1.82
20.12
2.54
1984-87
1.86
20.52
2.58
Recent Cl
Loop/3way
1984 +
0.66
8.97
1.04
1988 +
1.00
12.08
1.38
1988+
1.01
12.19
1.39
As predicted by MOBILES for vehicles operating at low
altitude with no I/M or anti-tampering program in effect.
High altitude operation causes increased emissions of. HC
and CO. I/M and anti-tampering programs cause the levels
of HC and CO emissions to be lower.
Many of the studies which EPA examined had vehicle groups
with average CO emissions close to 2 grams/mile. The emissions
measured in these studies clearly do not represent in-use
distributions. Table 1 shows that even among relatively- new
vehicles the average emissions are close to or higher than 9
grams/mile. However, under the percentage change model
selected by EPA, the base emission level should not, strictly
speaking, affect the percentage change. Given the extreme
differences between the emissions of the test samples and the
in-use vehicles they are taken to represent for purposes of
calculating percent changes, EPA kept alert for possible
contradictions as it proceeded through the analysis.
EPA considered whether to use individual test results or
to find some way to group the data before performing any
statistical observations. For example, the EEA summary
report2 used as the basic unit of observation the percentage
reduction in the average emissions of a given vehicle when
using a given fuel blend. While many of the reports included
individual test results, some included only averages. Using
-------
8
the results of individual tests would require that data from
some testing programs be eliminated from the data base.
Entering all of the available individual test results into a
data base would have required a large effort while adding
little additional information. As will be discussed in the
next section, EPA divided the total data base into subgroups,
based on technology features of the vehicles and the test
programs in which the data were gathered. EPA elected to use
study-wide reductions in the average emissions of groups of
vehicles operating on a given fuel. In other words, these
reductions were recorded, and statistical analyses were
performed on the groups' reductions. To give extensive studies
a larger effect on the final results than more limited ones,
each study is considered to have a weight within a technology
group equal to the number of vehicles of that group which were
used in the program.
To summarize the procedure, the data are divided into
groups, as will be explained in Section 3.3. The emissions of
vehicles which are in the same study and group are averaged for
all cases in which they use the same fuel. The averages of
vehicles using an oxygenated fuel blend are compared to the
averages of the same vehicles when operating on an HC-only fuel
and expressed as a percentage change. Each of these percentage
changes are normalized to represent fuels with matched
volatilities and constant oxygen content, as will be shown in
Section 3.3.4. When vehicles in the same study used more than
one oxygenated fuel blend, the percentage reductions are
averaged over all of the fuel blends for that study. The
resulting changes are averaged over an entire group, with the
results of each study weighted by the number of vehicles
represented in the group.
3.3 Subdividing the Data Base
There are factors which change the way the presence of
oxygenates in vehicle fuel affects emissions. These factors
fall into three categories; features of the vehicle (such as
emission control technology and fuel delivery system), fuel
attributes (volatility and oxygen content), and ambient
conditions (altitude and temperature). EPA decided to examine
the interactions between these effects and to subdivide the
data base when appropriate so that the effects could be
calculated separately. EPA has been careful keep the data
groups as large as possible; conclusions drawn on small samples
of emissions data may not be very accurate because of the wide
variation in emission levels from different vehicles.
-------
-9-
3.3.1 Technology
Perhaps the most important subdivision in this database is
based on the vehicle technology. Since 1970 there have been
several developments in emission control technology which may
interact with the way oxygenated fuel blends affect emissions.
The introduction of the oxidation catalytic converter in 1975
and then the widespread use of the three-way catalytic
converter, starting in 1981, are examples. Since oxygenates
have different products of combustion at a given air/fuel ratio
than gasoline and since they tend to burn at different
temperatures, it is reasonable to assume that a catalytic
converter might behave differently in the exhaust stream of. a
blend fueled vehicle than in the exhaust of the same vehicle
fueled with gasoline. EPA has chosen to examine the emissions
of vehicles equipped with catalytic converters separately from
those not so equipped.
A further development in technology has been the
introduction of closed loop fuel control systems. Such systems
readjust the air/fuel ratio during some phases of vehicle
operation based on measurements of the oxygen concentration in
the exhaust. Too much oxygen in the exhaust indicates that the
mixture is too lean, while too little oxygen in the exhaust
indicates that the mixture is rich. Closed loop fuel control,
by maintaining a mixture that is close to stoichiometric, has
been very effective at controlling emissions. Most vehicles
which have three-way catalysts also have closed loop- fuel
control, and vice-versa. Such systems can, in theory,
counteract the enleanment effect of the fuel oxygen content by
adding more fuel. EPA has chosen to examine the effects of
blends on the emissions of these systems separately from those
of open loop (fuel controlled without measurements of exhaust)
systems.
Some studies have separated vehicles with three-way
catalytic converters and open loop fuel control from other
groups. EPA considered this course of action and noted that
three-way catalysts are similar to oxidation catalysts in their
effect on VOC and CO. Oxidation catalysts often use some
formulation of platinum and palladium as the catalytic
material. Three-way catalysts usually have platinum and
rhodium, and perhaps some palladium. Based on this similarity
and the small amount of data on vehicles with open loop fuel
control and three-way catalytic converters, EPA decided to
group such vehicles together with open loop vehicles with
oxidation catalysts.
-------
-10-
Closed loop systems operate in an open loop mode under
certain circumstances, such as cold starting, idling, or
extreme acceleration, depending on the design. In such a mode,
the air/fuel mixture is controlled based on some predetermined
value, rather than a measurement of oxygen in the exhaust.
Most closed loop vehicles use only fixed calibrations to set
the air/fuel ratio when operating in an open loop mode.
However, some new systems (available since about 1984) adjust
the mixture during open loop mode based on electronically
stored values from previous closed loop modes. These systems
are said to have "adaptive learning" algorithms. Manufacturers
do not routinely identify individual vehicles or even vehicle
models as being equipped with an adaptive learning feature.
Not all vehicles are, even in the latest model year. EPA has
assumed that the mix of 1984 and later vehicles that have been
tested in the various studies adequately represent the mix of
adaptive and non-adaptive models in recent and future model
years.
In theory, vehicles with adaptive learning systems should
be less affected by oxygenates in the fuel than older style
closed loop systems. In actuality, the proof must be sought in
data, since vehicle designers have implemented the adaptive
learning concept in different ways.
The last issue with respect to technology division was
whether all closed loop vehicles that have been tested should
be grouped together to represent the entire population of
closed loop vehicles, including those from the most recent and
future model years. The alternative would be to divide the
closed loop test sample into "old technology" and "adaptive
learning" (or "new technology" more generally) and use only the
latter to represent the latest and future production vehicles.
To test the emissions of such vehicles when operating on a
gasoline/oxygenate blend fuel, it is important that the
vehicles be operated with the fuel for a certain amount of time
(which may vary for different vehicle designs) prior to the
test.
EPA has examined separately the tests of vehicles of the
1984 and later model years which were tested with this
preconditioning. The results of this comparison are shown in
Table 2. EPA could not determine in all cases which vehicles
actually did have adaptive learning systems and to what extent
the systems operate during test conditions.
-------
-11-
Table 2
Comparison of Effects;
Preconditioned* 1984 and Newer Vehicles Versus
Unpreconditioned and/or Pre-1984 Closed Loop Vehicles
Vehicle Group/ Emissions Changes on Oxygenated Fuels**
References N HC CO NOx
Preconditioned 50 -11.1% -24.6% +10.6%
1984 and Newer
B, G
Other Closed Loop 88 -2.3% -19.5% +8.0%
C, E, J, S, U, V,
W, X, Y, Z, AA, AB
* Preconditioning consists of use of a particular fuel in a
vehicle for a certain amount of time (perhaps several LA4
test cycles) prior to the actual emissions test.
Preconditioning procedures are those recommended by the
vehicle manufacturers to each study's project manager.
** Results include high and low altitude data. Results
normalized to matched RVP and 3.7% oxygen content.
Though the variance of the emissions was too large to
perform statistical tests comparing the two groups in Table 2,
it is possible to draw a meaningful conclusion from: it.
Theoretically, a vehicle with adaptive learning should be
affected less by fuel oxygen content than a vehicle without.
Thus, the null hypothesis would be that the vehicles without
adaptive learning are not affected more than the vehicles-which
have such algorithms, and the data confirm this hypothesis.
EPA concludes that to whatever extent adaptive learning, is
being used in production, it does not cause a significant
change in the response to oxygenated fuels.
Though not a new technology, the use of fuel injection has
become more common in new vehicles in recent years. Since it
is difficult to separate data regarding vehicles with fuel
injection from vehicles with carburetion (many of the studies
do not record that information) and since engine control logic
is probably more important than the mechanical approach, EPA
elects to consider vehicles with both types of fuel systems as
part of the same group in its analysis of exhaust emissions.
It should be noted, however, that the fuel delivery system is
important in the analysis of evaporative emissions.
-------
-12-
3.3.2 Altitude
The data may also be subdivided by the altitude (high or
low) at which they were collected. There is a large amount of
information available on both altitudes, so sample sizes remain
sufficiently large. At high altitude, a given volume of air
contains less oxygen. Vehicles operating at high altitudes
must compensate for the lack of oxygen, or they will run rich
for a larger portion of their operating time than when they
operate at low altitudes. The introduction of extra oxygen in
the fuel can mitigate this effect. Fuel oxygen content
theoretically should have an absolute effect on high altitude
emissions that is greater than or equal to the effects at low
emissions, but the percentage effect may be about the same.
EPA has analyzed separately the data taken at high altitude.
However, no studies are yet available which compare the
emissions of the same vehicles at both high and low altitudes
with oxygenated fuel blends.
Table 3
Emission Changes with Oxygenated
Blend Fuels* at High and Low Altitude
Vehicle
Group
No **
Catalyst
Oxidation
Catalyst
Closed
Loop
Altitude
High
Low
High
Low
High
Low
Number of
Vehicles
24
24
63
76
46
76
Effect of
HC
-11.3%
+0.4%
-15.8%
-15.3%
-1.9%
-4.2%
Blend on
CO
-23.9%
-25.0%
-31.4%
-37.3%
-18.4%
-20.6%
Emissions
NOx
+8.1%
+3.8%
+6.5%
+2.2%
+5.8%
+11.6%
**
Results normalized to represent matched RVP fuels with
3.7% oxygen.
Appendix A contains complete listings of the references
used in generating this table.
As with the results listed in Table 2, it would be
difficult to disprove a null hypothesis that the percentage
effect at high altitude is larger than that at low altitude,
because of the large variance in the data. However, the fact
that the average effects on CO are larger for the low altitude
tests of each vehicle type tends to confirm the null hypothesis
for CO. Similarly, the hypothesis can be confirmed for all
three pollutants from closed loop vehicles.
-------
-13-
EPA believes that the data do not show a significant
difference between the effects of fuel oxygen content at high
and low altitudes, especially for CO which is of most interest
to planners considering oxygenated fuels. EPA has elected to
use data gathered at high and low altitudes together.
3.3.3 Oxygen Content
A potentially important issue is the difference between
the effects of fuels with 3.. 7% oxygen content and those with 2%
oxygen content. Blends of ethanol in gasoline which are
currently sold at the retail level contain 3.7% oxygen.
Gasoline/methanol blends are not currently marketed in the
U.S., but they would generally be expected to contain about
3.7% oxygen. Blends of MTBE in gasoline may contain up to 2%
oxygen, limited by EPA's decision that MTBE is "substantially
similar" to gasoline and as such may be added in limited
concentrations without a Clean Air Act waiver. It is generally
accepted that the effects of fuel oxygen increase with
increasing oxygen content, but the increase is not necessarily
a linear relationship. Different studies have shown different
trends in this respect. One may show that the benefit of the
lower oxygen content fuel is larger than what would be
predicted by a proportional model, while another shows that the
benefit is smaller. After examining the data, EPA elected to
assume that the effect on emissions is linearly (in* fact,
proportionately) related to oxygen content at least up to the
3.7% oxygen level, and the data are normalized to represent a
constant oxygen content over the entire data base before
averages are computed.
Planners will be considering as a pollution control
strategy the mandated use of oxygenated blends. By specifying
the amount of oxygen which will be required, they can
effectively limit or allow certain blend types. By some
measures, the cost (in terms of loss of consumer choice,
vehicle effects, reduced price competition, loss of highway
revenue through tax credits, etc.) of restricting the retail
market to higher oxygen level fuels is high. Planners will
want to be sure that sufficient additional emissions benefits
are needed before incurring these additional costs.
More data exist on emissions effects with 3.7% oxygen
fuels than with 2% oxygen fuels. There are several studies in
which vehicles were ,tested with both levels of blends. There
is enough variability in the results of these studies that the
shape of a curve which relates oxygen content to emissions
effects could only be arbitrarily determined. Such a curve
would be based on the theoretical enleanment effects of the two
types of fuels. The 2% oxygen fuel will provide a certain
-------
-14-
amount of enleanment during rich modes of operation. The
enleanment might reduce the richness of some modes beyond the
point of stoichiometry (after which additional enleanment has
little additional effect). The additional enleanment of a 3.7%
oxygen fuel will only further reduce emissions during operation
modes that are still rich when using the 2% fuel. Thus, there
would be some levelling off of the benefits at high oxygen
contents. This argument would indicate that the 2% fuel has a
larger effect than simply 2/3.7 of the effect of a 3.7% oxygen
fuel (as would be predicted by an oxygen proportional model).
However, given that vehicles often emit pollutants at higher
levels than their certification standards even when using 3.7%
oxygen fuel, it appears that there are still significant
periods of rich operation. It seems reasonable to assume that
most of the significant levelling off occurs at even higher
levels of oxygen content. The correct answer to this question
could vary with vehicle technology. EPA has chosen to assess
these questions by examining the emissions of vehicles which
have been tested with both fuels, accounting for other factors,
such as RVP, which could otherwise affect the results.
Table 4
Comparison of Effects on RVP-Adjusted CO per
Percent Oxygen; Two Levels of Oxygen Content
Vehicle Group/ N Change from Base Emissions/%0x*
References vehs** 2.0% oxygen 3.7% oxygen
No Catalyst/ 9 -7.8% -4.9%
E, H, AB
Open Loop/ 28 -10.5% -8.8%
E, R, S, U, V, AB
Closed Loop/ 25 -3.4% -4.2%
E, S, V, AB
* *
Effects listed are the actual effects after adjustment to
to a constant RVP divided by the oxygen content of the
fuels, which demonstrates the idea that the effect of the
blend fuel is proportional to the oxygen content.
Includes vehicles tested on fuels of both oxygen levels.
-------
-15-
Table 4 shows the ratios of average CO effects to fuel
oxygen content for three different technology types and two
different oxygen levels. It shows that for vehicles with
catalysts (both open and closed loop)/ the ratio stays roughly
the same, regardless of the actual oxygen level. The constant
ratio indicates that the effect of oxygen on these vehicles can
be estimated as directly proportional to oxygen content. The
sample size of the vehicles without catalytic converters which
were tested on both fuels is small, so the comparison is less
conclusive in their case. EPA has elected to extend the
assumption of proportionality to these older vehicles. The
null hypothesis that the ratio stays the same regardless. of
total oxygen content could not be disproved based on these data
for any of the groups.
Since the time that the Guidance Document was released,
EPA has tested more than sixty vehicles using fuels of two
different oxygen levels. Data have become available from other
sources as well, and this additional information will be
considered in the next revision of the Guidance Document.
3.3.4 Other Fuel Related Factors; RVP and Oxygenate Species
Fuel related factors other than oxygen content may affect
emissions. For example, it is well documented that the Reid
Vapor Pressure (RVP) of the fuel can affect FTP exhaust
emissions. Since the RVP of any large group of fuels forms a
continuous spectrum, EPA has found no appropriate way to
subdivide the data base to separate the effects of RVP from
fuel oxygen content. Instead of subdividing the data base, EPA
has elected to adjust emissions test data using known relations
to volatility when RVP of the fuels are known (or can be
reasonably assumed, since some reports do not include
volatility information). The adjustment takes place before the
percent changes between group means on two fuels are
calculated. The adjustments are based on data from variable
RVP tests in EPA's emission factor program. They were
determined using only tests with gasoline, but EPA assumes that
they also apply to gasoline/oxygenate blends. Few data are
available to test this assumption, however. The following
table shows the functions which EPA applied to make these
adjustments.
-------
-16-
Table 5
RVP adjustments from MOBILE3.9
Model Years 1971 - 1980 HC adjustment:
HC = (HC @11.5 RVP) * (0.79622 + 0.01772(RVP))
Model Years 1981 and later HC adjustment:
HC - (HC (§11.5 RVP) * (0.57112 + 0.03729(RVP))
Model Years 1971 - 1980 CO adjustment:
CO = (CO @11.5 RVP) * (0.65094 + 0.03035(RVP))
Model Years 1981 and later CO adjustment:
CO = (CO eil.5 RVP) * (0.18753 + 0.07065(RVP))
MOBILE3.9 is the last version of the MOBILE3 emission
factor model. It was intended to incorporate many of the ideas
which will be included in the MOBILE4 model which is still
under development as of this writing.
The model year groups above separate those years in which
open loop and closed loop fuel control systems predominate. In
the analysis of gasoline/oxygenate blends, the pre-1981 model
year expression is applied to open loop vehicles, and., the
expression for 1981 and later model years is applied to closed
loop vehicles.
The type of oxygenate used in a fuel blend might also have
an effect on emissions. There are significant physical
differences between the various oxygenates, such as polarity,
miscibility with water, latent heat of vaporization, etc. EPA
has attempted to determine the effect of the species of
oxygenate (ethanol, methanol, MTBE, etc.) in the fuel blend on
emissions independent of total fuel oxygen content. This is
directly possible only for comparisons of ethanol blends with
methanol blends, since blends with MTBE have lower oxygen
contents than most blends with methanol or ethanol. This
comparison can be made in paired tests with the relatively very
few vehicles which received emissions tests using both types of
fuels, as shown in Table 6.
-------
-17-
Table 6
Paired Oxinol, TBA, and Gasohol Tests*
Vehicle
Group
Oxidation
Catalyst
Closed Loop
N
Vehs
8
Fuel
Blend
10% EtOH
16% TBA
Oxinol
10% EtOH
16% TBA
Oxinol
Changes in Emissions **
HC CO NOx
-35.4%
-30.2%
-33.9%
-28.8%
29.7%
-19.7%
-47.3% -4.5%
-47.5% -10.6%
-49.6% -13.1%
-49.0% -3.8%
-51.0% -7.0%
-29.2% -10.8%
* Data taken from ARCO waiver request for Oxinol.
** Emissions changes are relative to emissions
non-oxygenated base gasoline.
on
The theoretical effect of oxygenate species (independent
of oxygen content and volatility) is not well defined. EPA
could find no empirical reason to separate the data by
fuel-related factors other than volatility and oxygen content.
3.3.5 Ambient Temperature
Low ambient temperatures can extend the time that it takes
for an engine to reach operating temperature. During this
ti«e, catalytic converters operate at less than peak
efficiency, closed loop systems operate in open loop mode, and
fuel control systems use a rich mixture to keep the engine from
stalling. These effects may interact with the effects of fuel
oxygen. EPA does not presently have enough data to
characterize this interaction.
3.3.6 Base Emission Levels
While a percentage reduction model has a theoretical
attraction (since higher base emissions imply that there are
more modes with opportunity for enleanment), it may not
strictly hold for extreme emission levels. Because the
emission levels represented in the data (as shown in Appendix
A) are so low compared to in-use levels (as shown in Table 1),
this effect, if present, could reduce the accuracy of the
extrapolations from the data to a fleet.
-------
-18-
In EPA's analysis, the results of individual studies are
kept distinct up to a point, so it is possible to look for a
relationship between base emission levels and the percentage
reductions. EPA could find no such relationship, but the
difference between the emission levels of the different studies
is not as great as the difference between the levels in that
data base and in-use emissions. The reader is encouraged to
examine the results in Appendix A to confirm this conclusion.
4.0 RESULTS
The following table shows the technology-specific effects
on emissions that would be caused by switching from a
non-oxygenated gasoline to a gasoline/oxygenate blend with the
same volatility and a 3.7% oxygen content.
Table 7
Technology Specific Effects
on Emissions of a Fuel with 3.7%
Oxygen and Volatility Matched to Base Fuel
Vehicle Number of Effect of Blend on Emissions
Group Vehicles HC CO
No Catalyst 48 -5.5% -24.5%
Oxidation
catalyst 160 -15.5% -34.7% +4.1%
Closed
Loop 138 -2.3% -19.5% +8.0%
Note that these values are slightly different from those
listed in Table 3-1 of the Guidance Document. The difference
stems from an error which was found in the original data base
after the final version of the document was published. The
correction of this error yielded values which are in all cases
but one within 0.3 percentage points of the values originally
published. The estimate of the effect on NOx emissions from
closed loop vehicles increased by 1.1 percentage points. EPA
does not consider these corrections to be large enough to
require re-issuing the Guidance Document. A complete listing
of the data and sources used in developing these values is
presented in Appendix A.
If the fuel has an oxygen content other than 3.7%, then
these effects must be adjusted in proportion to the actual
oxygen content. If the blend is of a different volatility than
the fuel it replaces, then Table 5 should be used to adjust the
resulting effect.
-------
-19-
References
1. "Guidance on Estimating Motor Vehicle Emission Reductions
From the Use of Alternative Fuels and Fuel Blends," EPA
Technical Report, EPA-AA-TSS-PA-87-4, January 1988.
2. "Feasibility of Using Alternative Fuels as an Air
Pollution Control Strategy," Energy and Environmental
Analysis, October 1987, prepared for The Maricopa
Association of Governments.
3. "Comments of the Ad Hoc Ethanol Committee Regarding the
July 1987 Draft Technical Report, 'Guidance on Estimating
Motor Vehicle Emission Reductions From the Use of
Alternative Fuels and Fuel Blends'," Rivkin, Radler,
Dunne, & Bayh, December 1987.
4. "Ethanol-Blended Fuel as a CO Reduction Strategy at High
Altitude," Section C, Vol. II, Colorado Department of
Health, August 1985.
5. "Fleet Demonstration," Colorado Department of Health,
November 1987.
6. "Effects of Ethanol-Blended Fuel on Motor Vehicles at High
Altitude," Colorado Department of Health, September 1983.
7. "Exhaust Emissions and Fuel Economy from Automobiles using
Alcohol/Gasoline Blends Under High Altitude Conditions,"
David Richardson, EPA, October 1978.
8. "The Effects of. Two Different Oxygenated Fuels on Exhaust
Emissions at High Altitude," Colorado Department of
Health, January 1987.
9. EPA Ann Arbor 1987-88 In-House Gasohol Test Data as of
December 1987.
10. "Exhaust and Evaporative Emissions from a Brazilian
Chevrolet Fueled with Ethanol-Gasoline Blends," R. Furey
and M. Jackson, GM Research Publication GMR-2403, June
1977.
11. "Analysis of Gasohol Fleet Data to Characterize the Impact
of Gasohol on Tailpipe and Evaporative Emissions," EPA
Mobile Source Enforcement Division Report, December 1978.
12. "Characterization and Research Investigation of Alcohol
Fuels in Automobile Engines," Santa Clara University,
prepared for DOE as DOE/CS/51737-1, February 1982.
-------
-20-
13. "Exhaust Emissions, Fuel Economy, and Driveability of
Vehicles Fueled with Alcohol-Gasoline Blends," Brinkman,
Gallopoulos, Jackson, SAE Paper 750120 1975.
14. "Evaporative and Exhaust Emissions of Two Automobiles
Fueled with Volatility Adjusted Gasohol," David Lawrence
and Daniel Niemczak, EPA Report EPA-AA-TEB-81-12, December
1980.
15. "Exhaust and Evaporative Emissions from Alcohol and Ether
Fuel Blends," T. M. Naman and J. R. Allsup, SAE Paper
800858, 1980.
16. "Gasohol; Laboratory and Fleet Test Evaluation," M. D.
Gurney, et al, SAE Paper 800892, 1980.
17. "Gasohol, TBA, MTBE Effects on Light-Duty Emissions," B.
Bykowski, Southwest Research Institute, EPA Report
460/3-79-012, NTIS PB 80224082, October 1979.
18. "Evaporative and Exhaust Emissions from Cars Fueled With
Gasoline Containing Ethanol or MTBE," R. Furey and J.
King, SAE Paper 800261.
19. "Performance Evaluation of Alcohol-Gasoline Blends in 1980
Model Automobiles: Phase 1 - Gasoline-Ethanol Blends," CRC
Report No. 527, July 1982.
20. Clean Air Act Waiver Application for Oxinol, ARCO, May
1981.
21. Clean Air Act Waiver Application for the DuPont blend,
July 11 1984.
22. "In-House 23 Car In-House Oxinol Blend Test Program," EPA
memo from Craig Harvey to Charles Gray, ECTD, November 19,
1984.
23. "Characterization of Emissions from Vehicles Using
Methanol and Methanol-Gasoline Blended Fuels," P. Gabele
et al, JAPCA Vol. 35, no. 11, 1168-1175, 1985.
24. "A Generic Report of Toxics from Oxygenated Fuels,"
Colorado Department of Health, Spring 1987; Note: CDH has
stated these preliminary data are released for use in our
report.
-------
Appendix A
Data and Sources
-------
Dunary of Fuel Data, by Study
Page A-l
I. No Catalyst
$ Alt
Altitude Ref # #Vehs Fuels Notes
high A
high 3
high D
high AB
high E
High Total
Low if
low ?
lOH T
Low Total
10
4
*?
4
4
24
l
i
22
24
i all HDU's
1 CDH fleet deao
i big cars
: CDH Toxics
: Trucks
3 very approx
'^
I sostly LDT's
'/are
9.0
10.7
10.8
Emissions on Base Fuel
in (g/ii)
SUP 'JOC CO
4.66 80.50
2.72 72.97
4.00 89.80
5.29 51.94
NOx
3.23
3.53
2.59
1.30
5.91 109.30 3.42
3.00 120.00 0.50
1.50 25.40 1.93
4.58 32.30 4.60
Splash
10% Eton
RVP <;oc% GO; nox%
N.D. -14.61 -23.7'. 5.0%
-15.0V -16.6V 60.0V
-2.0'. -30.0-°. -23.0'.
2.8V -22.3V -2.81
TOTAL Son-Cat
II. Ox cat. Open loop
i Alt
Eaissions on Base Fuel
in (g/ii).
Altitude Ref #
high B
high
high
high
high
high
High
low
low
low
low
low
low
low
low
low
low
low
low
low
low
low
low
Low
b
C
D
AB
E
Total
J
j
S
5
s
T
a
u
\!
H
U
K
X
Y
y
7
w
Total
JKUehs Fuels
4 1
7
33
8
q
q
70
24
19
i
b
c
C
u
31
2
">
L
1
7
7
(
8
3
3
i
7
90
1
1
1
2
V
1
^
3
i
I
1
X
1
i
3
2
1
2
1
1
1
1
Notes
CDH fleet deao
other vehs
large engines
CDH Toxics
CDH Oxinl/MTBE
tests not aatched
conn base fuel
other blend
fleet tests
other blend
other blend
ARCO waiver
other blend
DuPont
other blend
EPA 23 car
RUP
uars
vars
9
9
10
10
9
10
9
3
3
9
9
9
9
9
12
12
9
9
11
.0
.0
.7
.8
.0
.0
.3
.6
.6
.1
.1
.0
.6
.6
.5
.5
.2
.2
.6
VOC
2.45
1
1
2
1
3
0
0
0
0
0
2
0
0
0
0
0
0
0
0
0
0
..42
.42
.06
.91
.06
.80
.83
.86
.85
.85
.08
.63
.63
.65
.27
.27
.96
.96
.53
.16
.54
CO
63.
38.
24.
31.
27
55.
8.
n.
10.
3.
8.
14.
7.
7.
8.
4.
4.
38
80
73
76
.4
70
77
95
44
56
56
80
58
58
52
67
67
20.40
20.40
6.67
5
u
20
5.43
NOx
1.77
1.62
1.32
1.49
2.38
1.97
1.53
1.59
2.41
1.63
1.53
3.36
0.97
0.97
1.79
1.43
1.43
1.98
1.98
1.80
1.02
0.99
Splash
10% EtOH
RVP von cm NOX*
uars
uars
9.8 -15.4% -34.8% 8.3%
-23.4% -24.8% 2.4%
9.6 -16.4% -39.0% 11.6%
10.9 -20.9V -32.6V 7.4%
10.2 -23.3V -39.5% -2.5%
-13.0V -34.5V -8.6%
9.7 -13.5V -28.1V -22.3%
10.5 0.0% -25.3% 16.8%
13.0 -35.4% -47.3V -4.5%
TOTAL Open Loop 160
-------
Suiiary of Fuel Data, by Study
Page A-2
!!I. Closed loop (pre '84 model year or not preconditioned)
* Alt
Altitude Ref # #Uehs Fuels Notes
high C 19 1
high SB 7 I CDH Toxics
high £ 13 2 CDH Oxinl/MTBE
High Total 39
TOTflL
Emissions on Base Fuel
in (g/mi)
RUP UK CO NOx
9.0 0.47 8.21 5.68
Splash
10% EtOH
SUP UOC% C0% NOX%
9.S -1.8% -13.61 7.9%
low
low
low
low
lew
low
iOH
low
low
low
low
low
low
low
low
low
Low
J
j
X
s
3
u
u
'J
»
H
;{
X
Y
Y
z
AA
Total
11
9
1
-\
0
3
1
1
>
7
7
3
3
5
2
14
2
49
j.
.1
1
i
i
1
1
]_
J
O
1
i
1
1
1
1
1
Indol base
coaserc base
very approx
other blend
other blend
prototype vehs
other blend
ARCO waiver
other blend
DuPont
other blend
EPft 23 car
RTP
9.0
10.0
3.6
3.6
9.1
.1
.0
.6
.6
12.5
12.5
9.2
9.2
11.6
11.7
0.34
0.37
0.27
0.43
0.43
0.24
0.24
0.23
0.30
0.30
0.37
0.37
0.22
0.21
0.55
0.44
4.86
5.70
2.60
6.23
5.23
3.41
3.41
1.38
5.08
5.08
5.78
5.78
3.44
2.65
5.59
4.60
0.66
0.65
0.38
0.48
0.48
0.79
0.79
0.70
0.62
0.52
1.57
1.57
0.72
1.04
0.91
1.08
9.6 -6.2'.
10.9 11.61
-22.2'.
9.7 13.0%
10.5 16.7V
13.0 -28.8%
-19.4%
-7.5%
-23.1%
3.7%
-26.0%
-49.0%
3.7%
S.5%
13.0%
1.4%
11.3%
-3.8%
IU. Closed loop (84+ lodel year, preconditioned)
# Alt
Altitude Ref # Wehs Fuels Notes
high 3 7 i CDH fleet de»o
high b 10 1 different uehs
low G 33 1 EPA In-house
TOTAL 50
Eiissions on Base Fuel
_in (g/ui).
RUP
uars
uars
11.7
roc
0.63
0.37
0.55
i:0
16.41
6.02
9.41
NOx
1.33
1.13
0.79
RUP
Splash.
10% EtOH
'JOC% C0%
NOK*
-------
Suiiary of Fuel Data, by Study Page A-3
Emission Changes caused by blend use
Other Fuel
'3.7'. Ox Other Fuel, - 21. Ox
type *0xy RVP VOC* CO*. NOx* type *0xy RVP VOC* CO'.' tIOx%
splash EtOH 3.2 uars 3,1'. -16.2°. 15.7V
£10 adj 3.7 10.2 -14.6* -20.6'.- -1.0-*, 11* MTBE 2 10.5 -1.1'. -11.6'. -4.5%
OxinolSO 3.5 11 -12.0* -21.4* -0.9% in MTBE 2 9.8 -14.4* -23.8* 15.4%
S10 adj 3.7 adjus -25.0'. -20.8% 80.0'. 5*EtOH splash 1.9 -7.5'. -3.3'. 40.0%
Emission Changes caused by blend use
Other Fuel
3.7'. Ox
type
splash EtOH
E10 adj
OxinolSO
E10 adj
E10 adj
E10 adj
E10 adj 2
E10 adj 3
TBA16
Oxinol
Fuel 82
Fuel93(E10)
Oxinol
'.Oxy
2.92
3.7
3.5
3.7
3.7
3.7
3.7
3.7
3.5
3.7
3.7
3.7
3.5
CTP
uars
10.2
11
10
9.4
9
9.4
9.4
12.4
14.9
9.1
9.1
11.2
VOC*
10.0*
-7
-2
-.
-3
-21
f
- 1
-7
-30
-33
-18
-12
7
.91
.6%
.2'.
.1'.
.5V
.4*
.4*
.2*
.9".
.9%
.5%
.4*
CO*.
-19.0*
-32.5*
-22.4%
-26.5*
-34.7*
-32.7*
-45.6*
-29.1*
-47.5*
-49.6*
-41.5*
-28.1*
-14.5*
N0x%
21.3*
5
1
7
-21
-26
23
22
-10
-13
-4
-1
13
.5*
.0%
.7*
.6*
.3*
.3*
.4*
.6*
.1*
.4%
.0*
.2*
Other
type *0xy RUP
MTBE 1 - 2* 1.54 vars
11% MTBE 2 10.5
11% HTBE 2 9.8
5* EtOH 1.9 10.4
7* MTBE 1.3 8.7
7* TEA 1.5 9.1
7* MTBE 1.3
7* TBA 1.5
15* MTBE 2.7 9.1
Fuel, "
uoc%
-13.2*
0.7*
-14.1*
2.2*
-20.0*
-3.2*
-7.7*
-10.3*
-20.0*
2* Ox
CO*
-12.0*
-11.9*
-21.5%
-4.2*
-32.5*
-7.9*
-39.7*
-44.8*
-25.9*
H0x%
-3.1%
4.0V
8.1%
2.5%
-0.6*
7.4%
10.0*
10.0%
-21.8%
-------
Suiiary of Fuel Data, by Study Page A-4
Saission Changes caused by blend use
Other Fuel
3.71 UH
type
E10 adj
Gxinol 50
E10 adj
£10 adj
E10 adj 2
£10 adj 3
TBftl6
Oxinol
Fuel 82
Fuel93{E10)
Oxinol
",0xy
j .
3.
3,
;
j
3
3
3
^
3
3
.1
.5
'T
. /
.7
.7
.7
.5
.7
.7
.7
.5
RVP
10.2
11
10
9.4
9.4
12.4
14.9
9.1
9.1
11.2
vocv
i.
s
13
15
-13
0
-29
-19
-4
9
7
.3'.
.2V
.9;
.0".
.3',
.0;
.7'.
.7%
.5'.
.5;
.3'.
CO'.
-11.5*
-16,
-0
33
-43
-19
-51
-29
-24
-35
-10
,4V
.34
.3".
.9V
.7'.
.01
.2%
.4V
.8V
.6V
NOxV
-2,
f\
11
Q
19
16
^
- i
-10
-5
-1
0
L,
.4V
.7V
.0V
.0V
.4V
.IV
.0V
.8V
.6V
.9-V
.2V
Other Fuel. ~ 2V Ox
type VOxy RVP VOC% COV MOxV
11V HTBE 2 10.5 -2.8V -3.8V -0
11V HTBE 2 9.8 3.3V -9.6V 10
7', HTBE 1.3 8.7 -27.9V -27.5V 35
7V TEA 1.5 9.1 -25.5V -21.5V 20
7VTBB 1.5 -26.7V -39.6V 0
7V MTBE 1.3 -26.7V -34.4V 2
15V HTBE 2.7 9.1 -4.3V 2.7V 6
.6V
.7V
.0%
,8V
.0%
.01
.4V
Oxinol(hiRVP) 3.5 12.9 -20.5V 1.1% 0.9%
Eaission Changes caused by blend use
Other Fuel
/3.7V Ox other Fuel,
type ",Cxy RVP VOCV COV MOxV type VOxy RVP UOCV
splash EtOH 3.27 vars -22.4V -30.3V -3.8%
MTBE 1 - 2V 1.34 vars -3.1V -13.1V 0.9%
£10 adj 3.7 11.8 -6.3V -21.5V 15.3V
-------
Siuiary of Fuel Data, by Study Page A-5
.Calculations: Account for Different Oxygen Content and Volatility
Forcing Matched RVP Forcing Matched RVP & Guy Forcing rlatched RVP L7orcir.g 3.7*. Ox
Splash 10% Ethanol froa other "3.7 Ox fuels _ * 2*. Ox froa "2'.Ox frca 2% fuels
Ref # ID psi UOC% CO: D psi i/OC». C0% N0x% D psi
-------
Suiiary of Fuel Data, by Study
Page fl-6
.Calculations: Account for Different Oxygen Content and Volatility
Sef I
A3
U
a
'}
'A
w
ii
X
Y
y
z
AA
Forcing Matched RVP Forcing Matched RVP & Oxy
Splash 101 Ethanoi froa other "3.7 Ox fuels
3 psi von coi o psi yon coi NOXI
0.80 -4.71 -23.21
Q.60 -3.3* -22.81
0.90 7.9% -13.«
0.76 -24.4* -27.2'.
0.70 10.11 -1.4°,
0.90 12.8'. -30.7".
0.50 -30.11 -50.8-1
-0.5
0.2
6.21
.... -5.41 -2.41
n.S1 -13.61 -2.91
0.0 13.91 -6.31 11.01
0
0
.-%
L
0
0
I]
1
.0
f\
.1
.1
.4
.1
.1
.4
.2
15.
-1 9
0.
-31.
-26.
-4.
0.
9.
-25.
01
M
; o
71
11
91
11
91
41
41
33
-43
-13
-53
-41
-23
-35
-3
-7
.31
.11
.61
.51
.21
.91
.31
.51
.9%
')
19
16
-10
-5
-1
2
1
.01
.4-1
.11
.41
.81
.61
.91
.31
.01
Forcing Matched RVP
. & 2-1 Ox froa "ZIGx
psi
0.2
1.0
0.1
0.5
0.0
0.0
0.1
von
_2
7
-43
-35
-35
-41
-3
.11
i'
.31
.91
.61
.11
.41
C01 N0x%
-7
*
-43
-32
-52
-52
i
.51
7"
.21
.41
.81
.91
.51
-0.61
107"
C7 Qo
Jv* . 0 t
27.71
0.01
3.1".
4.71
Forcing 3.71 Ox
froa 21 fuels
VOX
-3.
13
-30.
-66.
-65.
-76.
-5.
CO* NOx*
81
>
21
31
91
01
41
-4
^
-80
-59
-97
-97
2
.5*
9°,
.01
.9%
.71
.9-1
.71
-1
19
99
51
0
5
8
.11
.81
.61
.31
.01
.71
.81
Ref #
6
b
G
.Calculations; Account for Different Oxygen Content and Volatility
Forcing Matched RVP Forcing Matched RVP & Oxy
Splash 101 Ethanoi
0 psi VOC1 C01
t'roa other
0 psi von
3.7 Ox fuels
CGI H0x%
0.8 -27.81 -33.51 -4.31
0.1 -6.61 -22.11 16.31
Forcing Matched RVP
_ i 21 Ox froi "2 ".Ox
D psi VOn C01 NOxl
Forcing 3.71 Ox
froa 21 fuels
;;on cov HOX%
0.0 -3.81 -14.21 1.01 -16.31 -26.31 1.81
-------
Susiary of Fuel Data, by Study Page A-7
Average reductions by study
aatched 3.71 Ox and RVP.
weighted together by ueh Base g/ai.
Hi alt
Lo alt
otal
u
rr
A
3
U
AB
E
tot
H
p
T
tot
VOC%
-15.81
2.01
-9.4%
-7.6%
-13.41
-11.4%
-19.31
-3.31
1.41
9.4%
-:.51
CO*
-25.51
-21.01
-21.71
-19.9%
-31.41
-24.51
-19.81
-51.21
-24.11
-25.01
-24.81
N0x%
5.01
13.21
20.11
-4.61
1-1.71
a. 51
72.61
-23.01
-2.81
-0.51
4.0-1
!/OC CO HOx
4.60 81.73 3.10
4.59 35.67 4.32
4.59 58.70 3.71
Average reductions by study
latched 3.71 Ox and RVP,
weighted together by veh Base g/ii
VOC CO NOx
Hi alt
io alt
total
#
B
b
C
0
AB
E
tot
.]
j
S
5
3
T
U
u
V
W
w
X
X
Y
y
T
tot
uoc-;
10.8%
-31.71
-17.6%
-24.4%
-2.5%
-13.21
-15.7%
-17.31
-12.2".
-10.71
-57.31
""7 *>«
-LL.t* d
-14.21
-21.91
-25.4%
-22.91
-4.3%
-7.1%
-33.9%
-36.71
-18.8%
-12.31
8.6%
-15.31
-15.4%
C0%
-26.4%
-23.8%
-36.4%
-26.5%
-26.2%
-29.8%
-31.8%
-40.11
-30.51
-30.11
-93.1%
-22.91
-36.01
-100.01
-100.0%
-32.71
-36.3%
-28.71
-49.1%
-53.31
-41.31
-27.91
-14.21
-37.51
-35.0%
NOX%
0.0%
-7.41
8.3%
2.4%
5.4%
3.0%
5. Si-
ll. 61
7.51
-6.4%
-1.71
'18.31
-3.6%
28.51
24.71
-25.21
20.31
22.41
-7.9-1
-13.11
-4.4%
-1.0%
19.2%
2.21
3.6%
1.33 33.47 1.62
1.18 11.34 2.13
i.46 21.02 1.93
-------
Suasary of Fuel Data, by Study Page A-8
Average reductions by study
matched 3.7"* Ox and RVP.
weighted together by veh Base g/ai
CO HOs
8.30 3.80
if
\_,
A3
P
Hi ait tot
;
i
j
i\
.^
;3
u
u
;
;<
W
FT
.1
X
Y
?
Z
M
lo alt tot
voc%
-4.7'i
1.2;
10.5'.
i.4".
-8.3'.
13.4V
-24. 41
-SO. 2".
-66.3*
-65. 9".
-76.0'.
6.2*.
3. It
0.7*.
-30.6'.
-26.9%
-4.1*
9.9*
9.4%
-25.4%
-5.1%
cot
-23.2%
-6.51.
-12.3%
-16.6%
-22.8V
-6.8V
-27.2V
-30. 0V
-59.9V
-97.7V
-U7.9-V
11.5V
-36.9V
-15.6V
-52.2V
-41.2V
-23.9%
-35.3%
-3.5%
-7.9%
-22.2%
H0x%
7.9V
-1.7%
3.5V
5.4V
3.7V
9.3V
13.0V
99.6V
51.3V
0.0V
» 7'
3.4V
15.4V
16.1V
-5.6%
-10.8V
-5.6%
-1.9%
2.3%
1.0%
9.3%
HOC
G.53
0.3!
4.95 0.80
Total -2.2% -19.7% 3.0% 0.45 6.43 2.13
Average reductions by study
siatched 3.7V Ox and RVP.
weighted together by veh Base g/ui
70C CO HOX
u
:T
3
b
G
Both alt
voc%
-27.8%
-16.3V
-6.6%
-11.5%
COV
-33.5V
-25.3V
-22.1V
-25.2V
N0x%
-4.3%
1.8%
16.3V
10.5V
0.52 9.71 0.93
all CL -5.6% -21.7% 3.9% 0.5 7.6 1.7
-------
Page A-9
Pages A-l through A-8 list all of the data which were used
in this report. These pages also show results at intermediate
steps in the calculations which lead to the report's
conclusions. The information is presented as a spreadsheet, so
the easiest way to look at the entire process is to separate
the pages from the report and tape them together side by side
in the following arrangement (although it is possible to follow
the process without doing so):
Page A-l Page A-3 Page A-5 Page A-7
Page A-2 Page A-4 Page A- 6 Page A-8
The data on vehicles without catalytic converters and
vehicles with oxidation catalysts and open loop fuel control
are on odd numbered pages (top row), while the even numbered
pages have the data concerning all of the vehicles with closed
loop fuel control. The following is an explanation of the
columns in the spreadsheet:
Altitude: Indicates the altitude at which each study is
conducted. Most of the high altitude studies were conducted in
the Denver area by the Colorado Department of Health.
Ref tt: The letter shown in this column indicates the study frcvn
which the data in each row was taken. The letter/study key is
listed on the last pages of this Appendix. Lower case letters
indicate that the row is a second entry for the study in the
previous row. Studies are listed in more than one entry
whenever they include more than one fuel of the same general
type, such as two different gasoline/methanol blends. Also, if
a study includes vehicles of different emission control groups,
the study is indicated (with a capital letter) in each emission
control group in this table. Some letters are skipped because
the studies which would have been indicated by those letters
are already represented in a summary study. Though some
vehicles are represented twice in this and the next column, all
operations which add the results of different studies together
are designed to count the results from each vehicle only once.
KVehs: This column indicates the number of vehicles which are
represented in the row.
tt Alt Fuels: Indicates the number of gasoline/oxygenate blends
which are represented in the row.
Notes: Notes were included to help keep track of important
points regarding each study.
-------
Page A-10
Emissions on Base Fuel: The four columns under this heading
list the RVP of the base fuel, when known, and the average
exhaust VOC, CO, and NOx emissions of all the vehicles in the
group when tested on the non-oxygenated base fuel.
Splash 10% EtOH: The four columns under this heading are
filled only when the vehicles represented on the row were
tested on a splash blend of gasoline with 10% ethanol. The
columns list the RVP of the blend and the percent reductions in
the exhaust emissions of VOC, CO, and NOx relative to the
levels in the previous columns (base fuel).
Other Fuel -3.7% Ox: The six columns under this heading are
filled only when the vehicles were tested on a blend of fuel
with an oxygen content of more than 3.0%, but which is not a
splash blend of ethanol. Such a blend might be a volatility
adjusted blend of gasoline with 10% ethanol or a blend of
gasoline with 5% methanol and a cosolvent. The columns name
the fuel, list its RVP and oxygen content, and show the average
emissions reductions relative to , the base fuel which were
reported when the fuel was used.
Other Fuel -2% Ox: The columns under this heading are
analogous to those listed under the previous heading, but art
included only when vehicles were tested with a fuel of about 2%
oxygen content, such as a blend of 11% MTBE or 5% ethanol.
The next four headings show the calculations which
normalize the listed reductions in emissions to reflect the
reductions that the vehicles would have achieved using a blend
with 3.7% oxygen and matched volatility to the base fuel. Each
of the headings indicates which group of fuels data (from among
the previous three headings) is being normalized. "D psi"
indicates the difference between the RVP of the blend and that
of the base fuel. The effect of fuel oxygen content was
calculated assuming a linear relationship between fuel oxygen
level and exhaust emission changes from the base fuel. The
effect of RVP on exhaust emissions was calculated using the
assumptions in MOBILES.9.
Forcing Matched RVP, Splash 10% Ethanol: These columns show
the results of the adjustments for RVP that are presented in.
section 3.3.4. Entries to these columns only occur when the
vehicles represented in a given row were tested on a splash
blend of 10% Ethanol. Since there is no adjustment for NOx due
to RVP, and a 10% ethanol blend should already have
approximately 3.7% oxygen content, there is no column listing
for NOx under this heading.
-------
Page A-11
Forcing Matched RVP and Oxy for Other -3.7 Oxy Fuels: Since
the blends represented in this column do not necessarily have
matched RVP to the base fuel, the VOC and CO reductions are
adjusted to reflect the results of matched RVP fuels. Since
the blends may have oxygen contents of other than 3.7%, the
VOC, CO, and NOx results are further adjusted in proportion to
the oxygen content to reflect a 3.7% oxygen content.
Forcing Matched RVP & 2% Oxygen from -2% Ox: This heading
contains entries when the group of vehicles represented in the
row was tested using a fuel of about 2% oxygen content. The
exhaust emissions are adjusted to reflect matched RVP and 2%
oxygen content.
Forcing 3.7% Oxygen from ~2% Oxygen: Under this heading, the
results of the previous heading are scaled to reflect a larger
oxygen content.
Average Reductions by Study: In each entry under this heading
is the average of the percentage changes in emissions as
adjusted to reflect a matched RVP and 3.7% oxygen content over
all of the fuels tested in the vehicles represented by this
line of the table. At the bottom of each technology group is
the average of these percentage figures, weighted by the number
of vehicles in each study. Each gasoline/oxygenate blend
within a given study receives weight equal to other fuels in
the same study, even when they are represented on more than one
line in the table.
Base g/i> i: Under this heading are the average base fuel
emission levels of the studies, weighted by the number of.
vehicles in each study. Averages are listed for different
altitudes within each technology group, and for the combined
high and low altitude results of each group. This information
is useful for comparing the emission levels of vehicles in the
studies to those which would be predicted of in-use vehicles.
Based on this comparison, a qualitative judgement can be made
regarding how well the data set represents an in-use fleet.
The following is a detailed example of the calculations
involved in one of these rows for one pollutant. Study V
included 2 closed loop vehicles tested at low altitude. The
base (non-oxygenated) fuel used in this study had an RVP of 9.0
psi, and the average exhaust CO emissions of these vehicles was
1.88 g/mi. These vehicles were also tested with a splash blend
of gasoline with 10% ethanol, which had an RVP of 9.7 psi.
When using this fuel, their average CO emissions increased by
3.7%. When the vehicles were tested using a volatility
adjusted blend of 10% ethanol in gasoline (RVP = 9.0 psi),
their average CO emissions changed by 33.3% from the baseline
levels. When they were tested on a blend with 15% MTBE, their
CO emissions changed by 2.7%.
-------
Page A-12
Before the results can be averaged, they must be adjusted
to offset the effects of different volatilities and oxygen
levels. The percentage reductions are changed to represent
those that would be achieved by a blend with 3.7% oxygen
content and matched volatility to the base fuel. When a splash
blend of 10% ethanol in gasoline is used, the oxygen content is
already approximately 3.7%, so the only adjustment that is made
is for volatility. This RVP adjustment is made according to
the equations presented in section 3.3.4 of this report.
These vehicles had closed loop fuel control systems, so
their emissions are adjusted using the relationship of RVP to
emissions listed for vehicles of the 1981 and later model
years. For this calculation, the following values are used.
The CO emissions of the vehicles when fueled with the blend
were 1.037 times the emissions when the base fuel was used (or
1 + 3.7%). The base fuel RVP was 0.7 psi lower than that of
the blend. To simplify the calculations at the cost of only a
small error, the effect of RVP is calculated using a base value
of 11.5 psi, so the difference between a 11.5 psi fuel and a
10.8 psi fuel is calculated first, and then applied to the
blend emission level.
CO @ -0.7 psi
= CO (blend) * (0.18753 + 0.07065 * (ll.Spsi - 0.7))
= CO (blend) * 0.95055
= CO (base) * 1.037 * 0.95055
= CO (base) * 0.98572 (a reduction of 1.4%)
This operation separates the effect of the 0.7 psi
difference in RVP from that of the oxygen content, now
calculated to be a 1.4% decrease. For the matched volatility
ethanol blend the RVP was already matched, so the matched RVP
column hc.s the same reductions as the actual test data.
In the next column test data from blends of roughly 2%
oxygen are adjusted to match the RVP of the base gasoline and
to have exactly 2.0% oxygen. In the case of this 15% MTBE
blend, the 0.1 psi RVP adjustment and the oxygen content
adjustment from 2.7 to 2.0% result in smaller effects on all
three emissions. This is followed by a column in which the
2.0% oxygen numbers are proportionally adjusted up to 3.7%
oxygen to make chem directly comparable to the other 3.7%
oxygen data. This increases the effects on VOC, CO, and NOx
to -6.4%, +2.7%, and +8.8%.
-------
Page A-13
The final combination of data is accomplished in the three
columns labelled, "Average Reductions by Study, matched 3.7% Ox
and RVP, weighted together by veh." In the row for the closed
loop vehicles in Study V the adjusted effects of each of the
three oxygenated fuels are combined by weighting the effects
for each fuel by the number of vehicles tested on that fuel in
that study, yielding a CO increase of 11.5%. The totals for
each column are similarly combined by weighting the results
from each study by the number of vehicles in that study. The
totals for all closed-loop, pre-'84 vehicles show a CO
reduction of 19.7%. The last three columns provide the base
gasoline gram/mile emissions, averaged in the same way for each
column. These values may be compared between different groups
of vehicles within this summary as well as with in-use emission
levels.
-------
Page A-14
Sources of Data
The reports from viich data were ta:
-------
Page A-15
R. "Evaporative and Exhaust Emissions of Two Automobiles
Fueled with Volatility Adjusted Gasohol," David Lawrence
and Daniel Niemczak, EPA Report EPA-AA-TEB-81-12, December
1. 80.
S. "Exhaust and Evaporative Emissions from Alcohol and Ether
Fuel Blends," T.~ M. Naman and J. R. Allsup, SAE Paper
800858, 1980.
T. "Gasohol; Laboratory and Fleet Test Evaluation," M. D.
Gurney, et al, SAE Paper 800892, 1980.
U. "Gasohol, TEA, MTBE Effects on Light-Duty Emissions", B.
Bykowski, Southwest Research Institute, EPA Rerort
460/3-79-012, NTIS PB 80224082, October 1S79.
V. "Evaporative and Exhaust Emissions from Cars Fueled With
Gasoline Containing Ethanol or MTBE," R. Furey and J.
King, SAE Paper 800261.
W. "Performance Evaluation of Alcohol-Gasoline Blends in 1980
Model Automobiles: Phase 1 - Gasoline-Ethanol Blends," CRC
Report No. 527, July 1982.
X. Clean Air Act Waiver Application for Oxinol, ARCO, May
1981.
Y. Clean Air Act Waiver Application for the DuPont blend,
July 11, 1984.
Z. "In-House 23 Car In-House Oxinol Blend Test Program," EPA
memo from Craig Harvey to Charles Gray, ECTD, November 19,
1984.
AA. "Characterization of Emissions from Vehicles Using
Methanol and Methanol-Gasoline Blended Fuels," P. Gabele
et al, JAPCA Vol. 35, no. 11, 1168-1175, 1985.
AB. "A Generic Report of Toxics from Oxygenated Fuels,"
Colorado Department of Health, Spring 1987; Note: CDH has
stated these preliminary data are released for use in our
report.
------- |