EPA420-P-98-006
-Draft-
Fuel Oxygen Effects on Exhaust CO Emissions
Recommendations for MOBILE6
Report Number M6.FUL.002
March 16,1998
Venkatesh Rao
U.S. Environmental Protection Agency
Office of Air and Radiation
Office of Mobile Sources
Assessment and Modeling Division
-1-
-------
Objective
This document describes EPA's effort to estimate simple relationships between fuel oxygen
content and exhaust carbon monoxide (CO) emissions as a function of vehicle technology, fuel
blending scenario, vehicle emitter classification, ambient temperature, and oxygenate type for
gasoline-powered vehicles. These relationships will be used in MOBILE6 to supplant the current
estimates of the effects of oxygen on exhaust CO emissions. This report is divided into two major
parts: the first part describes the methodology used to determine effects for Tier 0 and older vehicles
and the second part outlines the methodology for Tier 1 and Low Emitting Vehicles (LEVs). An
attached report provides more details of the analysis and the actual numbers to be used in MOBILE6
for Tier 0 and older vehicles.
Background
The Clean Air Act Amendments of 1990 mandated use of oxygenated gasolines in areas that
did not meet the Federal ambient air standard for CO. Motor vehicle emissions are the primary
source of ambient CO levels in most areas and CO is generally at its highest levels during the cold
weather months. Oxygenated gasoline is designed to increase the combustion efficiency of gasoline,
thereby reducing exhaust CO emissions.
As more and more areas of the country come into attainment with the current ambient CO
standard, less emphasis has been placed on CO control in the on-highway fleet. However, accurate
modeling of the impacts of oxygenated fuels is important for those areas that remain out of
attainment for CO. The MOBILE5/MOBILE5B model is the standard tool currently available to
make fleet-wide estimates of the effects of oxygenated fuels for area-wide inventories.
MOBILES/MOBILE5B estimates emission levels in grams of pollutant per vehicle mile (g/mi) under
a wide variety of conditions.
MOBILES/MOBILE5B estimates exhaust emission benefits of oxygenated fuels as a function
of fuel oxygen content, baseline emission rate, and model year. Embedded in the model-year-
specific corrections are assumptions regarding the interaction of specific vehicle technologies and
specific oxygenates. However, MOBILE5/MOBILE5B predictions of oxygenated fuel CO benefits
are not supported by current ambient CO monitoring data. For example, under the following
conditions:
An area without an Inspection/Maintenance (I/M) program that requires
oxygenated fuel with 3.5 weight percent oxygen content,
Calendar year 2000
MOBILE5B predicts a light-duty gasoline vehicle (LDGV) fleet-average CO reduction of
approximately 33% due to: 1) relatively large benefits ascribed to oxygenated fuels in MOBILE5B
for vehicles with high CO emissions, and 2) high deterioration rates predicted under a non-I/M case.
Such large reductions are not supported by ambient CO monitoring data. For this reason, EPA is
-2-
-------
currently revisiting the oxygenated fuel adjustment factors for MOBILE6. This document presents
EPA's final proposal for determining revised oxygenated fuel adjustment factors in MOBILE6.
EPA Preliminary Proposal for MOBLE6 on 10/1/97
At the October 1, 1997 MOBILE6 workshop held in Ann Arbor, ML, EPA presented its first-
cut proposal for modeling the effects of fuel oxygenates on exhaust CO emissions. To keep the
analysis as simple as possible while at the same time matching MOBILE6 predictions with ambient
CO data, the following proposal was made:
The recently developed CO exhaust emissions model would be used for 1988 and
later model year Tier 0 vehicles1.
For 1981-1987 model year vehicles, a recent analysis performed for ARCO Chemical
Company would be used. This analysis was performed by Air Improvement
Resources using the California Air Resources Board predictive database2.
For all pre-1981 model year vehicles, EPA proposed using the current (MOBILESa)
estimates for older technology vehicles3.
Table 1 summarizes the emission reductions associated with the approach outlined above.
Table 1
CO Emission Reductions Proposed for MOBILE6
Assumes Splash-Blended Fuels
Based on October 1, 1997 MOBILE6 Workshop
Model Year
Group
1988+
1981-1987
Pre-1981
Open-Loop
Pre-1981
Non-Catalyst
2.7 wt% MTBE
Normal
13.8%
High
9.33 %
18.2%
26.9 %
18.9%
3.5 wt%Ethanol
Normal High
15.4% 10.1%
19.1 %
34.9%
24.5 %
Some comments received since this proposal was made questioned whether use of the CO
model and the methodology outlined above would provide the most accurate estimates of the effects
of fuel oxygen on CO emissions. In response to these comments, EPA decided to use existing data
to attempt an alternative, less-complicated "paired" data analysis (i.e., data pairs in which a vehicle
was tested on both a baseline, non-oxygenated fuel and a corresponding oxygenated fuel with all
-3-
-------
other fuel parameters as matched as much as possible) to develop estimates of the impacts of fuel
oxygen on CO emissions. Sierra Research conducted the bulk of the analysis and their final report
is attached. The next section briefly describes the conclusions reached by Sierra Research. These
conclusions have been reviewed by EPA. The last part of the report provides final estimates for
scenarios that the Sierra report did not address, namely oxygen's effect on CO emissions from
vehicles certified to Tier 1 and cleaner emission standards.
EPA's Proposed Final Methodology for MOBILE6
Pre-1981 Vehicles
Because new data on older technology vehicles (i.e., pre-1981 oxidation catalyst and non-
catalyst vehicles) have not been collected in recent years and since the last version of MOBILE was
developed, the CO emissions impacts for those vehicles will be based on EPA's 1988 Guidance
Document on the emissions impacts of oxygenated fuels3. These impacts are already embedded in
MOBILE5/MOBILE5B.
Tier 0 Vehicles (Model Years 1981-1994)
The attached report entitled "Effects of Fuel Oxygen Content on CO Emissions" (February
13, 1998) outlines the data sources, final methodology, and emission effects for Tier 0 and older
vehicles that EPA proposes to insert into MOBILE6 for the effect of oxygen/oxygenates on exhaust
CO emissions. A summary of the major conclusions reached in the report are as follows:
For model years 1981-1994 (Tier 0 vehicles), the results described in the report apply
for estimating effects of oxygen/oxygenates on exhaust CO emissions. Table 17
represents the effects that will be input into MOBILE6.
The effect of oxygen content on exhaust CO emissions is linear for all model years.
The percentage CO emissions impacts developed from FTP-composite data can be
applied to both starting and running exhaust emission estimates.
No clear, significant trends were found for the CO effects of oxygenates at different
ambient temperatures. Thus, in MOBILE6, the effect of oxygen on exhaust CO
emissions will be the same at all ambient temperatures.
For fuels with the same oxygen content, the effect of the type of oxygenate (MTBE,
ethanol, etc.) on exhaust CO emissions was found to be statistically insignificant.
The exception to this finding is splash blended ethanol fuels qualifying for an RVP
waiver. In this case, the emissions resulting from the RVP boost must be calculated
and offset against oxygen-based exhaust CO benefits using the existent RVP flag in
MOBILE.
-4-
-------
Sierra's analysis recommends that results in Table 18 of the attached report (which
accounts for the RVP changes when splash blending) be used directly for ethanol
splash blends at 75°F, a linear combination of the results shown in Tables 17 and 18
should be used for temperatures between 75°F and 45 T (where RVP will have a
smaller effect on emissions than at 75°F), and the results in Table 17 should be used
for temperatures less than 45°F since RVP will not have an effect on emissions at low
ambient temperatures. EPA's proposal differs from what Sierra recommends. EPA's
final proposal for ethanol splash blends is to use the results in Table 17 along with
the RVP flag in MOBILE6 to offset any emission disbenefits accrued by elevated
RVP levels in ethanol splash-blended fuels. This procedure will be explained in
detail in the MOBILE6 User's Guide.
The report does not address how to calculate effects for Tierl and advanced
technology vehicles due to the severe paucity of data for these types of vehicles.
Estimating effects of oxygen on exhaust CO emissions for these class of vehicles is
discussed below.
Tier 1 and Advanced Technology Vehicles
One other question that must be addressed relates to the oxygenated fuel CO emissions
impact for vehicles certified to Tier 1 and lower emission standards. It is expected that the effects
of oxygen on exhaust emissions will diminish with advanced technology vehicles since the oxygen
effect is dependent on the inability of the engine to maintain the proper air-to-fuel ratio under all
conditions. Due to improved engine technology, this inability has been decreasing over time and will
continue to decrease. However, the question is to what degree. The only way to address this is to
have oxygen-effect data on Tier 1 and advanced technology vehicles. There is only a minimal
amount of data that allows for examination of these effects and those data will be used to bracket the
possible oxygen-related CO emission impacts of these types of vehicles.
Available Data and Analysis
Six certified Tier 1 vehicles (five passenger cars and one Class 2 Light Duty Truck (LOT))
and six advanced technology vehicles (five prototype passenger car models and one prototype Class
1 LDT), not necessarily representing eventual production technology, were tested as part of the
Auto-Oil program (technical bulletin 17) in which a baseline non-oxygenated fuel (Industry Average
reference gasoline) was tested along with two reformulated fuels on a series of four vehicle fleets
designed for progressively lower emission standards4. The two reformulated gasolines were a
gasoline meeting 1996 California Phase 2 regulatory requirements (fuel C2) and a gasoline blended
to the same specifications, but without an oxygenated component (fuel Cl). The Industry Average
gasoline represented a 1988 national average composition. It was found that CO emission
differences between reformulated test gasoline C2 with oxygenate (MTBE at a 2 weight percent
level) and a very similar gasoline Cl without oxygenate were generally not statistically significant.
-5-
-------
The percent differences in exhaust CO emissions for the Tier 1 and advanced fleets is summarized
in Table 2 below:
Table 2
Auto/Oil Technical Bulletin 19 Results for Effects
of Fuel Oxygen (at a 2 weight percent level) on Exhaust CO
Emissions for Tier 1 and Advanced Technology Fleet of Vehicles
Vehicle Fleet
Average CO
Emissions for fuel
Cl* (grams/mile)
Average CO
Emissions for fuel
C2** (grams/mile)
Percent Difference in
Emissions, C1->C2
Certified Tier 1
Vehicles (n=6)
1.392
1.376
-1.20
Prototype Advanced
Technology Vehicles
(n=6)
0.893
0.902
+1.00
* Fuel Cl contains no oxygen
** Fuel C2 contains 11.2 volume percent (2 weight percent oxygen) MTBE oxygenate
It should also be noted that while oxygen was the primary fuel parameter varied between fuels Cl
and C2, other parameters (sulfur, olefins, RVP) also varied slightly which could have confounded
the effects of oxygen alone between these fuels on exhaust emissions. Thus, due to the small effects
seen and the uncertainty in these effects, it will be assumed that the effect of oxygen on exhaust CO
emissions from normal emitting Tier 1 vehicles (please see discussion below on high emitters for
clearer definitions of emitter classes) is zero. These effects can be revisited and revised in the future,
if necessary, as more data become available on the effects of oxygen on CO emissions from Tier 1
vehicles.
Recently, the Coordinating Research Council (CRC) performed a study5 to examine the
effects of fuel properties (mainly sulfur) on LEV certified vehicles. Specifically, the CRC study
involved 6 Light Duty Vehicle (LDV) models which were certified for sale in California in 1997.
Two vehicles from each model type were tested on seven fuels. One fuel was a California RFG with
40 ppmW sulfur, while another was the same fuel doped to 150 ppmW sulfur. The other five fuels
were national average conventional gasolines, except that their sulfur levels were 40, 100, 150, 330,
and 600 ppmW. The same base gasoline was used for all five of these fuels and the sulfur levels
were varied by adding representative sulfur-containing hydrocarbons. The vehicles were leased from
rental companies and averaged 10,200 miles of use. The vehicles were tested in an as-received
condition and with their catalyst aged to 100,000 miles by the manufacturer of each vehicle. All
testing was conducted at a single laboratory.
-------
While sulfur was the main parameter varied and tested for its effect on emissions in the CRC
program, emissions from the base fuel (containing no oxygenate-fuels Cl and C3 with sulfur levels
of 30 and 150 ppmW, respectively) can be compared to the two CA RFG fuels with matching sulfur
levels (fuels SI and S2 containing 28 and 147 ppmW sulfur, respectively), to get a rough estimate
of non-sulfur effects on emissions. The fuel properties for these fuels are listed in Table 3 below:
Table 3
Summary of Non- Sulfur Fuel Properties for
Fuels Cl, C3, SI and S2 from the CRC Testing Program
Fuel Property
RVP, psi
Aromatics, vol%
Olefms, vol%
MTBE, vol%
T50, F
T90, F
FuelCl
(sulfur = 30
ppmW)
7.6-8.6
27-35
6-14
0.1 maximum
(Taken as 0)
210-220
330-340
Fuel SI
(sulfur = 28
ppmW)
6.7-7.0
22-25
4-6
10.8-11.2
(Taken as 2
weight percent
oxygen)
200-210
290-300
Fuel C3
(sulfur = 150
ppmW)
7.6-8.6
27-35
6-14
0.1 maximum
(Taken as 0)
210-220
330-340
Fuel S2
(sulfur = 147
ppmW)
6.7-7.0
22-25
4-6
10.8-11.2
(Taken as 2
weight percent
oxygen)
200-210
290-300
Since this is the only data available at this time on LEVs for the effects of oxygen addition on
exhaust emissions, it will be used to gauge this effect. The discussion following the numbers in
Table 5 outlines the difficulty in isolating an oxygen effect from these data. Emissions data are
compared for fuels Cl and SI and for fuels C3 and S2 in Table 4 for the 10K catalyst as well as the
aged 100K catalyst. Table 5 shows the average effects of changing non-sulfur fuel parameters on
exhaust CO emissions based on the average g/mile emissions shown in Table 4.
-7-
-------
Table 4
Average CO Emissions in grams/mile for Fuels Cl, C3, SI and S3
from Vehicles with 10K and 100K Catalysts
Vehicle Id
Escort
Taurus
Civic
Sentra
Camry
Metro
FuelCl
10K/100K
0.586/1.005
0.349/0.439
1.055/1.983
0.433/0.553
0.662/0.987
0.316/0.616
Fuel SI
10K/100K
0.653/0.990
0.457/0.588
1.007/1.912
0.534/0.653
0.692/1.015
0.356/0.545
Fuel C3
10K/100K
0.742/2.008
0.395/0.723
1.374/2.355
0.733/0.851
0.875/1.098
0.446/0.654
Fuel S2
10K/100K
0.788/1.602
0.599/0.885
1.433/2.283
0.870/0.860
0.927/1.268
0.425/0.611
Table 5
Average Percent Change in CO Emissions for the Fuel Pairs
C1/S1 (with Sulfur ~ 50 ppmW) and C3/S2 (with Sulfur -150 ppmW)
with 10K and 100K Catalysts
Vehicle Type
Escort
Taurus
Civic
Sentra
Camry
Metro
Fleet Averages:
Fuel Cl >S1
1 OK Catalyst
11.4
31.0
-4.54
23.3
4.53
12.7
13.1
100K Catalyst
-1.49
34.0
-3.60
18.1
2.84
-11.5
6.4
Fuel C3 >S2
1 OK Catalyst
6.20
51.6
4.29
18.7
5.94
-4.71
13.6
100K Catalyst
-20.2
22.4
-3.06
1.06
15.4
-6.57
1.5
Table 5 indicates an overall increase in CO emissions when oxygen is added to base fuel
regardless of the age of the catalyst. The Auto/Oil data in Table 2 also indicates a small insignificant
increase in CO emissions when oxygen is added to base fuel. It is likely that the increase for LEVs
in CO emissions in this test program is due to other-than-oxygen parameters changing in fuel
combinations Cl, SI and C3, S2. Though LEV vehicles are expected to have a smaller response for
-------
the effects of oxygen on emissions, it is unlikely that oxygen will cause an increase in tailpipe CO
emissions.
Thus, with this limited amount of data on these types of vehicles for the effects of oxygen
on exhaust CO emissions, it will be assumed that the effects of oxygen addition on CO emissions
from normal emitting LEV vehicles (see discussion below on high emitters for more details on
emitter class definitions) is zero. This effect can be revisited and revised, if necessary, as more
testing data becomes available on the effects of oxygen alone on advanced technology vehicles.
High Emitters
Tier 0 CO high emitters (defined as any vehicle emitting more than 7 grams/mile on base,
non-oxygenated fuel) were seen in Table 17 of Sierra's report to have slightly different emission
effects than Tier 0 normal emitters. However, there is no test data available on the effects of Tier
1 and LEV "higher" emitters. There are several ways to define higher emitters for these advanced
technology vehicles. However, since we have data only on the effects of Tier 0 higher emitters ( in
the attached report the nomenclature for this is "1981 + TWC/CL systems"), it will be assumed that
the higher emitter effects of oxygen on exhaust CO emissions can be applied only to those Tier 1 and
LEV vehicles that are higher emitters as defined for Tier 0 (CO emissions on base fuel must be 7
grams/mile or higher). All other Tier 1 and LEV vehicles will be considered to be normal emitters
for which the effects of oxygen on exhaust CO emissions will be zero.
Summary
Table 6 (next page) summarizes the effects from the addition of oxygen on exhaust emissions
for all vehicle technologies, oxygenate type, oxygen content of fuel, and vehicle emitter
classifications.
-9-
-------
Table 6
Recommended CO Effects (CO Impact (in percent) Per Weight Percent Oxygen)
From the Use of Oxygenated Fuels on Light-Duty Gasoline Powered Vehicles
Emitter Classification
Normal Emitting
Vehicles
(CO Emissions on
base, non-oxygenated
fuel of < 7
grams/mile)
Higher Emitting
Vehicles
(CO Emissions on
base, non-oxygenated
fuel of > 7
grams/mile)
Vehicle Technology
LEV and Advanced
Technology (Year
1999+)
Tier 1 (1994-1999)
1988 + TWC/ADL
1986-1987
TWC/ADL
1986+TWC/NoADL
1981-1985 TWC/CL
OX/OL
Non-Catalyst
LEV and Advanced
Technology
Tier 1
1981 +
OX/OL
Non-Catalyst
Start Emissions
0.00
0.00
-3.10
-4.80
-5.70
-4.0
-9.40
-6.60
-5.30
-5.30
-5.30
-9.40
-6.60
Running Emissions
0.00
0.00
-3.10
-4.80
-5.70
-4.0
-9.40
-6.60
-5.30
-5.30
-5.30
-9.40
-6.60
Some notes on the entries in Table 6:
LEV stands for "Low Emitting Vehicles"
TWC stands for "three-way catalysts"
ADL stands for "adaptive learning"
OX stands for "oxygen sensors"
OL stands for "open loop" operation
The entries under the "OX/OL" and "Non-Catalyst" are already programmed into MOBILES.
The same numbers will be used in MOBILE6.
The numbers from Table 6 must be used in conjunction with the RVP flag (which will be
unchanged from MOBILES) for fuels in which the RVP increases significantly upon addition
-10-
-------
of oxygenate, specifically for splash-blended ethanol fuels. Details will be provided in the
guidance document.
The numbers in Table 6 will be used at all ambient temperatures (no temperature correction
factors for the effect of oxygen/oxygenates on CO emissions)
With the exception of splash-blended ethanol fuels, all oxygenates will be treated the same
and only the oxygen content of a fuel will determine exhaust CO emission effects as
identified by the emission factors in Table 6.
The CO effects of oxygenates on Light-duty trucks (LDTs) and heavy-duty gasoline vehicles
(HDGVs) will be assumed to be the same as the effects on LDVs
References
1. Rao, V., "Development of an Exhaust Carbon Monoxide Emissions Model," SAE Paper No.
961214, 1996.
2. Rykowski, R., Air Improvement Resources Draft Analysis: "Effect of Oxygenated Gasoline
on CO Emissions Based on the ARB Predictive Model Database for Older Vehicles," for
ARCO Chemical Company, September 1997.
3. "Derivation of Technology Specific Effects of the Use of Oxygenated Fuel Blends on Motor
Vehicle Exhaust Emissions," U.S. Environmental Protection Agency, October 1988.
4. "Auto/Oil Air Quality Improvement Research Program, Technical Bulletin No. 17: Gasoline
Reformulation and Vehicle Technology Effects on Exhaust Emissions," Coordinating
Research Council, August 1995.
5. "CRC Sulfur/LEV Program" Presentation Materials from the Coordinating Research
Council, December 1997.
-11-
-------
February 13, 1998
Memo To: David H. Lax
Senior Environmental Scientist
American Petroleum Institute
1220 L Street, NW
Washington, D.C. 20005-4070
From: Philip Heirigs
Subject: Effects of Fuel Oxygen Content on CO Emissions
This memorandum transmits the findings of Sierra's analysis of the effects of fuel oxygen
content on exhaust emissions. The review has been focused on FTP composite CO
emissions under "winter" fuel scenarios (i.e., simple addition of oxygenate with minimal
changes to fuel parameters other than those related to dilution or distillation
characteristics). If you have any questions regarding this material, please call me or
Bob Dulla at (916) 444-6666.
Background
The use of oxygenated gasoline is mandated in many urban areas as a wintertime CO
control strategy. As such, EPA has included the emissions impacts of oxygenated fuel in
its MOBILE series of on-road motor vehicle emission factors models since the
MOBILE4.1 version of the model. As more and more areas of the country come into
attainment with the ambient CO standard, less emphasis has been placed on CO control in
the on-highway fleet. However, proper modeling of the impacts of oxygenated fuels is
important for those areas that remain out of attainment so that air quality planners can
make intelligent decisions about the mix of control measures that are available to them.
(This is also an issue for attainment areas developing maintenance plans.) Thus, EPA has
indicated that it will update the oxygenated fuel correction factors in the next release of
the MOBILE model, MOBILE6.
Modeling the Effects of Fuel Oxygen Content With MOBILES a - The exhaust emission
benefits of oxygenated fuels modeled by MOBILESa are a function of fuel oxygen
content (wt%), baseline (non-oxygenate) emission rate, and model year. Embedded in the
model-year-specific corrections are assumptions regarding the impact of vehicle
technology on the effect of oxygenates on emissions. (The model-year-specific
corrections are developed from a subroutine contained in the TECHS model.) Although
-------
David H. Lax
-2-
February 13, 1998
the model was structured to account for the differences in HC benefits between alcohol
and ether blends, MOBILESa assumes the same benefit (as a function of oxygen content)
for both.
In the TECHS model, the impact of fuel oxygen is modeled separately for the following
technology groups:
Multipoint fuel injection/closed-loop;
Throttle-body injection/closed-loop;
Carbureted/closed-loop; and
Open-loop (carbureted and fuel-injected).
To generate the oxygenate effect for a specific model year, the reductions estimated for
each of the above technology groups are weighted by the fraction of those vehicles in the
fleet. To illustrate the magnitude of the projected impact of oxygenated fuels determined
by MOBILESa, Figure 1 shows the calculated CO benefits (as a percent reduction from
the base emission rate) for 1990 model year light-duty gasoline vehicles (LDGVs). The
benefits increase with increasing wt% oxygen content and with increasing base emission
rate, resulting in a maximum benefit of approximately 37.5% at an oxygen content of
3.5%. For alcohol blends (e.g., ethanol), the maximum oxygenate content modeled is
3.5%, while the maximum content modeled for ether blends (e.g., MTBE) is 2.7%.
40
35 --
r so --
3! 25 +
co
m
E 20 f
c 15 --
o
3 10 --
MOBILESa CO Ex h a u st Em ission Benefits
from Oxygenated Fuels
(1990 Model Year LDGV)
3.5% Oxygen
- - - - 2.7% Oxygen
2.0% Oxygen
10 20 30 40 50
CO Em ission Level (g/m i)
60
70
Figure 1
-------
David H. Lax -3- February 13, 1998
The relatively large benefits ascribed to oxygenated fuels in MOBILESa for vehicles with
high CO emissions, coupled with high deterioration rates under a non-I/M case, result in a
LDGV fleet-average CO reduction of approximately 33% in calendar year 2000 for an
area without an I/M program that requires oxygenated fuel with a 3.5% oxygen content.
Because reductions of this magnitude are not supported by ambient CO monitoring, EPA
intends to revisit the oxygenated fuel adjustment factors for MOBILE6.
EPA Preliminary Proposal for MOBILE6 - At the October 1, 1997 MOBILE6 workshop
held in Ann Arbor, EPA presented its preliminary proposal for modeling the effects on
CO emissions resulting from the addition of fuel oxygen. Due to a desire to keep the
analysis as simple as possible (primarily because EPA feels that the CO nonattainment
problem is diminishing), EPA proposed the following for MOBILE6:
The recently developed CO exhaust emissions model1* that was based on the data
and methodologies used for the reformulated gasoline (RFG) Complex model
would be used for 1988 and later model year vehicles.
For 1981 to 1987 model year vehicles, a recent analysis performed for ARCO
Chemical Co. would be used. In that study, a model to estimate the CO impact of
fuel parameter changes was constructed for 1981 to 1987 model year vehicles
based on the CARB Predictive model database. The methodologies used to
develop that model were based on the Complex model methodologies.
For pre-1981 model year vehicles, EPA proposes to use the current (MOBILESa)
estimates for older technology vehicles.
A summary of the emission reductions associated with the approach outlined above, as
presented at the MOBILE6 workshop, is presented in Table 1.
Concerns with the EPA MOBILE6 Proposal - After reviewing the CO model developed
by V. Rao, the Western States Petroleum Association (WSPA) expressed concern about
its use.2 The more important concerns with respect to modeling the impacts of
oxygenated fuels include the following:
Weighting factors used to combine normal and high emitters were based on the
RFG Complex model VOC weighting factors - CO-specific factors were not
developed.
The CO model is based primarily on summer fuels, with the majority of testing
performed with fuel RVP between 6.5 and 10 psi.
Superscripts denote references provided at the end of this memorandum.
-------
David H. Lax
-4-
February 13, 1998
Table 1
CO Emission Reductions Proposed for MOBILE6
Assumes Splash-Blended Fuels
(From the October 1, 1997 MOBILE6 Workshop Handouts)
Model Year
Group
1988+
1981-1987
Pre-1981
Open-Loop
Pre-1981
Non-Catalyst
2.7 wt% MTBE
Normal
13.8%
High
9.33%
18.2%
26.9%
18.9%
3.5 wt%Ethanol
Normal High
15.4% 10.1%
19.1%
34.9%
24.5%
The CO response to RVP reduction shows a minima at approximately 8 psi. It is
unclear that such a trend would exist at the cold temperatures under which winter
oxygenated fuels are normally used. (And it is clear that the RVP used under
winter conditions is nowhere near 8 psi.)
Normal emitters show a stronger response to oxygen than high emitters. Based on
the physical mechanisms involved, the opposite would have been expected.
Given the above, it is unclear that use of the CO Complex model will provide the most
accurate estimates of the effects of fuel oxygen on CO emissions. Thus, Sierra has
analyzed available "paired" data (i.e., data pairs in which a vehicle was tested on both a
baseline, non-oxygenated fuel and a corresponding oxygenated fuel) to develop
alternative estimates of the impacts of fuel oxygen on emissions.
Data Sources
Since the development of the fuel oxygen correction factors for MOBILESa, a number of
test programs aimed at investigating the impacts of gasoline modifications on emissions
have been conducted. Primary among those is the Auto/Oil Air Quality Improvement
Research Program (AQIRP). In addition, smaller programs have been sponsored by
industry and government agencies, and EPA has generated additional data that were not
used in the development of the MOBILESa factors. Much of this testing was generated
with summertime reformulated gasoline in mind; however, some of the test results can be
used to predict the emissions impact of adding oxygenates to wintertime fuels. A brief
summary of the data sources investigated in this effort is presented below.
-------
David H. Lax -5- February 13, 1998
Auto/Oil - The Auto/Oil AQIRP represents perhaps the best-designed, most extensive test
program ever conducted to study the effects of fuel parameters on emissions from light-
duty vehicles. The primary drawback from the use of those data to predict CO changes
from the addition of fuel oxygen is that the fuels investigated in that effort were generally
reflective of summertime fuels. This is particularly true of the Phase 2 testing conducted
in the program in which many of the fuels had RVP levels below 7 psi, and there were
few paired tests conducted with and without oxygenate in the fuel. However, two data
sets are available from Phase 1 that can be used to estimate the impact of fuel oxygen on
CO emissions, and one data set from Phase 2 can be used to assess the impact of fuel
oxygen on advanced technology vehicles. These data sets are described below.
AMOTdata set - The AMOT (AromaticsMTBE/Olefms/T90) data set consists of
tests conducted with 16 fuels with combinations of high and low levels of those
four fuel parameters.3 The RVP of the fuels in this series of tests was held around
8.7 psi for all of the fuel blends. Comparing the high (2.7 wt%) and low (0 wt%)
MTBE-paired tests (keeping all other fuel parameters constant) gives an estimate
of the impact of fuel oxygen on emissions.
RVP/Oxygenate data set - Following the AMOT testing, a test program was
conducted in which the impact of both RVP and fuel oxygen was investigated.4
Three oxygenates were assessed in this program - MTBE (15 vol%, 2.7 wt%
oxygen) , ethanol (10 vol%, 3.5 wt% oxygen), and ETBE (17 vol%, 2.5 wt%
oxygen). The ethanol blends were "splash" blends (with and without an RVP
adjustment to maintain constant RVP) and reflected dilution effects, while the
MTBE and ETBE were "match" blends that did not reflect dilution of other fuel
parameters. Fuel blends were prepared with both a 1990 commercial gasoline
(RF-A) and a "reformulated" base gasoline that had low aromatics, low olefins,
and low T90.
Technology effects data set - As part of Phase 2 of the program, emissions tests
were conducted with a reformulated gasoline meeting the 1996 California
specifications and a gasoline blended to the California specifications but without
an oxygenated component.5 Tests on these two fuels (and RF-A) were conducted
on the AQIRP "current" fleet, a fleet of vehicles meeting federal Tier I standards,
and a fleet of advanced technology vehicles. Although the testing was not
extensive, it provides some insight regarding the effects of fuel oxygen on
emissions from advanced technology vehicles.
EPA Emission Factors Database - The historical EPA emission factors database has a
considerable number of tests in which vehicles were tested with a baseline gasoline
(typically Indolene) and an oxygenate blend (primarily ethanol and MTBE). However, in
many cases the RVP was not controlled on the oxygenate blend, which causes difficulty
when attempting to evaluate only the effect of fuel oxygen on CO emissions. In addition
to testing at 75 °F, the EPA emission factors database contains results from vehicles tested
at50°Fand20°F.
-------
David H. Lax -6- February 13, 1998
EPA RFG Study - This study was conducted to support the development of EPA's
Complex model for RFG and consisted of three phases:
Phase I was an initial evaluation of the impact of oxygenate, volatility, distillation
parameters and sulfur on emissions;6
Phase //was a continuance of Phase 1 investigating the effects of oxygenate
content, oxygenate type, volatility, sulfur, olefms, and distillation parameters;7 and
Phase III investigated sulfur, olefms, volatility, aromatics, and the interactions
between olefms and volatility or sulfur.8
Although this was an extensive test program, the focus of the testing was on summertime,
reformulated fuels. For the most part, oxygen levels were kept near 2.0 wt% in most of
the fuels, and the non-oxygenated fuels in the program were typically industry average
(RF-A) or Indolene. There are few paired data points to make a direct comparison
between a non-oxygenated gasoline and an oxygenated gasoline. Although there were
two levels of oxygen (2.0 wt% versus 3.7 wt%) investigated in the Phase I program in
which a fuel pair was blended with similar base gasoline, those fuels utilized different
oxygenates (MTBE and ethanol) and the results reveal only the difference between one
level of oxygen and another. (Unfortunately, the non-oxygenated base gasoline used to
derive these fuels was not included in the test matrix for this program.)
API RVP/Oxygenates Test Program - In 1988, API initiated a two-year study of the
effects of changes in RVP and oxygenates on emissions from gasoline-fueled vehicles at
varying ambient temperature.9 Eleven vehicles were tested in that effort, ranging in
model year from 1981 to 1989. The RVP of the fuels in the program ranged from 7 to 13
psi, and three temperatures were investigated: 80°, 55°, and 35°F. Although the number
of vehicles tested in this program was not extensive, the wide range of temperatures and
fuels investigated allows for an evaluation of the impact of ambient temperature on the
benefits attributed to oxygenated fuel. In this program, the magnitude of the response of
CO to fuel oxygen content was greatest at 55 °F and lowest at 35 °F.
CARS Low Oxygenate Gasoline Blends Test Program - This study, sponsored by the
California Air Resources Board (CARB), investigated the impacts of oxygenated gasoline
on emissions of light-duty vehicles under both summer and winter conditions.10 The
winter fuel study included gasolines blended with ethanol, MTBE, and ETBE to a
2.7 wt% oxygen level. Thirteen vehicles were tested in this program, ranging in model
year from 1973 to 1991. Vehicles were tested at 75 °F and 50°F. As with the API test
program, the small number of vehicles tested in the program (which covered a broad
range of control technologies) makes it difficult to generate accurate estimates of the
emissions impacts of fuel oxygen on CO emissions; however, the varying test
temperatures may provide insight into how the oxygen effect changes with temperature.
CRC Study of Winter Gasoline/Oxygenate Blends11 - In the late 1980s, the Coordinating
Research Council (CRC) sponsored a test program investigating the effects of fuel
-------
David H. Lax -7- February 13, 1998
oxygen content on exhaust emissions for three technology types (six vehicles equipped
with three-way catalysts and adaptive learning, six vehicles equipped with three-way
catalysts without adaptive learning, and four oxidation catalyst vehicles) at three
temperatures (75°, 50°, and 35°F). Tests were conducted at sea level and at high
altitude. The primary fuel set used in this testing included a 13 psi RVP non-oxygenate
base fuel, all vol% MTBE blend, and a 10 vol% ethanol splash blend. The sea level
tests also included a 10 vol% ethanol matched volatility blend. Again, the small number
of vehicles tested in the program makes it difficult to generate accurate estimates of the
emissions impacts of fuel oxygen on CO emissions, but the varying test temperatures may
provide insight into how the oxygen effect changes with temperature.
Analysis of Paired Data at 75 °F
This section of the memorandum presents the mean emission levels of paired non-
oxygenate/oxygenate tests run at 75 °F from the programs summarized above. In general,
the results are presented by emitter category using a 7.0 g/mi CO emission rate (on
Indolene at 75 °F) as the cut-off between normal- and high-emitters. In addition, the data
were analyzed according to the following model year/technology groups, as it is expected
that the CO response to fuel oxygen content will be a function of the emission control
system employed on the vehicle:
1981 and later model year oxidation catalyst (OX) or open-loop three-way catalyst
(TWC);
1981-85 model year closed-loop (CL) TWC;
1986 and later model year closed-loop TWC without adaptive learning (ADL);
1986-87 model year closed-loop TWC with ADL; and
1988 and later model year closed-loop TWC with ADL.
In some cases, sample sizes were not sufficient to analyze the data as outlined above and
broader model year/technology groups were utilized. In addition, it was not always
possible to determine if a vehicle was equipped with ADL; in those cases, all TWC/CL
were combined. A vehicle was considered to have ADL if it was so specified in the
database being analyzed or if it was included in the Complex model database. (EPA had
carefully screened all of those vehicles to ensure that they were "1990 technology"
vehicles with adaptive learning capability.)
In the tables that follow, test pairs were generally selected in which the oxygenated blend
was as close as possible to the base gasoline, with only dilution causing the other fuel
parameters to differ. Particular attention was paid to differences in RVP between the
non-oxygenate and oxygenate blends. Much of the available oxygenated fuel data have
been collected using simple splash blends of ethanol and gasoline. Although this reflects
-------
David H. Lax -8- February 13, 1998
what occurs in practice, there is concern that the extra 1 psi RVP of the oxygenate blend
will mask the impact of the oxygen, particularly with a 75 °F test. (In fact, some of the
test results from the EPA emission factors database show an increase in CO emissions
from late-model cars when tested on a 10.2 psi RVP Indolene/EtOH versus a 9.0 psi RVP
Indolene base fuel.) Since MOBILE has a separate exhaust correction for RVP, the
analysis of fuel oxygen effects should eliminate the RVP effect as much as possible.
Alternatively, the splash-blend ethanol data could be used to develop a combined
RVP/oxygenate adjustment for cases in which an RVP waiver is granted. Such an
approch would be valid for higher temperature regions, but for colder temperatures (i.e.,
below 35 ° to 45 °F), the impact of the RVP difference is expected to diminish.
Auto/Oil AMOT Data set - Mean CO emissions from the Auto/Oil AMOT testing
(aromatics - MTBE - olefins - T90) are shown in Table 2. In each case, the difference
between the non-oxygenated fuel and the oxygenated fuel is the MTBE content (either 0
or 15 vol%). As in the A/O reports and SAE papers presenting the results of this testing,
lower case refers to the lower levels of these fuel parameters; upper case refers to the
higher levels of these fuel parameters. This fuel set was tested in both the current fleet
(noted by the 1988+ model year group in the table) and the older fleet (noted by the 83-85
model year group in the table). One of the vehicles in the current fleet was eliminated
from the Complex model database because it was not considered representative of 1990
technology (a carbureted 1989 Honda Accord). Hence, it is labeled as not having
adaptive learning, and results are presented separately for that vehicle (actually, that pair
of vehicles). As seen in the table, this vehicle had a substantial impact on the overall
results. It is also interesting to note that the addition of MTBE had a larger impact on the
older fleet than the current fleet, even for the normal emitters. Finally, the high emitters
in the table had a strong response to MTBE, but it should be noted that four of the five
were open-loop/oxidation catalyst vehicles.
Auto/Oil RVP/Oxygenate Data set - Table 3 summarizes the paired data from the
Auto/Oil RVP/Oxygenate test program. Comparisons are made for the same RVP fuels
as well as for the splash-blend ethanol fuels. One item of particular interest in this table
is the small effect observed for the MTBE blends relative to the EtOH and ETBE blends.
This was surprising because it is inconsistent with other available data. It is interesting to
note, however, that the SAE paper presenting the results of this test program reports a
benefit of 9.3 ± 6.7% in going from 0% to 15% MTBE (based on a regression analysis of
the data rather than a comparison of paired tests). Also, the EtOH benefit of 14.9%
shown in Table 3 for the ADL fleet on the F-T fuel pair (matched RVP) is very close to
the results from the EPA database for 10% EtOH in Indolene (also matched RVP)
presented below.
-------
David H. Lax
-9-
February 13, 1998
Table 2
Mean CO Emissions from Paired Tests In the Auto/Oil AMOT Database
("Current" and "Older" Fleets)
Emitter
Class
Normal
Normal
All
Normal
High
Model
Year
1988+
1988+
83-85
83-85
83-85
ADL
All
Vehicles
with and
without
ADL
Yes
No
No
No
Non-Oxy
Fuel
D- amOT
F - amot
G - AmOt
I - AmoT
K - Amot
L- AmOT
P - amOt
Q - amoT
D- amOT
F - amot
G - AmOt
I - AmoT
K - Amot
L- AmOT
P - amOt
Q - amoT
D- amOT
F - amot
G - AmOt
I - AmoT
K - Amot
L- AmOT
P - amOt
Q - amoT
D- amOT
F - amot
G - AmOt
I - AmoT
K - Amot
L- AmOT
P - amOt
Q - amoT
D- amOT
F - amot
G - AmOt
I - AmoT
K - Amot
L- AmOT
P - amOt
O - amnT
Oxy
Fuel
M- aMOT
N - aMot
O - AMOt
R- AMoT
C - AMot
E- AMOT
H - aMOt
J - aMoT
M- aMOT
N - aMot
O - AMOt
R- AMoT
C - AMot
E- AMOT
H - aMOt
J - aMoT
M- aMOT
N - aMot
O - AMOt
R- AMoT
C - AMot
E- AMOT
H - aMOt
J - aMoT
M- aMOT
N - aMot
O - AMOt
R- AMoT
C - AMot
E- AMOT
H - aMOt
J - aMoT
M- aMOT
N - aMot
O - AMOt
R- AMoT
C - AMot
E- AMOT
H - aMOt
.1 -aMnT
wt%
Oxygen
2.6
2.5
2.6
2.7
2.8
2.7
2.6
2.7
2.6
2.5
2.6
2.7
2.8
2.7
2.6
2.7
2.6
2.5
2.6
2.7
2.8
2.7
2.6
2.7
2.6
2.5
2.6
2.7
2.8
2.7
2.6
2.7
2.6
2.7
2.6
2.7
2.8
2.7
2.6
n
20
20
20
20
20
20
20
20
18
18
18
18
18
18
18
18
14
14
14
14
14
14
14
14
9
9
9
9
9
9
9
9
5
5
5
5
5
5
5
5
Mean CO (g/mi)
Non-Oxy
2.55
2.73
2.96
3.14
2.96
3.11
2.69
2.70
2.59
2.74
2.99
3.31
2.95
3.23
2.69
2.77
5.33
6.08
6.54
6.49
6.19
6.52
7.41
6.26
3.85
4.48
4.55
4.13
4.27
4.37
5.05
4.14
7.99
8.95
10.12
10.75
9.65
10.38
11.67
10 07
Oxy
2.33
2.46
2.74
3.16
2.68
2.92
2.45
2.32
2.44
2.56
2.84
3.37
2.78
3.12
2.53
2.44
4.89
5.41
5.74
5.36
5.29
4.79
5.58
4.69
3.38
4.33
4.20
4.14
3.97
3.75
4.19
3.43
7.60
7.36
8.50
7.55
7.66
6.68
8.09
fi Qfi
CO
Impact
-8.6%
-9.5%
-7.4%
+0.6%
-9.5%
-5.8%
-8.9%
-13.7%
-6.2%
-6.6%
-5.0%
+ 1.8%
-6.1%
-3.4%
-5.9%
-11.9%
-8.3%
-10.9%
-12.2%
-17.6%
-14.5%
-26.5%
-24.7%
-25.1%
-12.2%
-3.3%
-7.7%
+0.2%
-7.0%
-14.4%
-16.8%
-17.1%
-4.9%
-17.8%
-16.0%
-29.8%
-20.6%
-35.6%
-30.8%
-30 9%
CO Impact
Per Wt% O
-3.3%
-3.8%
-2.9%
+0.2%
-3.4%
-2.1%
-3.4%
-5.1%
-2.4%
-2.6%
-1.9%
+0.7%
-2.2%
-1.3%
-2.3%
-4.4%
-3.2%
-4.3%
-4.7%
-6.5%
-5.2%
-9.8%
-9.5%
-9.3%
-4.7%
-1.3%
-3.0%
+0.1%
-2.5%
-5.3%
-6.5%
-6.4%
-1.9%
-7.1%
-6.2%
-11.0%
-7.4%
-13.2%
-11.8%
-11 4°/r
-------
David H. Lax
-10-
February 13, 1998
Table 3
Mean CO Emissions from Paired Tests In the Auto/Oil RVP/Oxygenate Database
Constant RVP Between Fuel Pairs
Emtter
Class
Normal
Normal
Model
Year
1988+
1988+
ADL
All
Vehicles
with and
without
ADL
Yes
Non-Oxy
Fuel
A- Ind Ave/9 psi
F - Reform/9 psi
F - Reform/9 psi
S - Reform/8 psi
F - Reform/9 psi
A - Ind Ave/9 psi
F - Reform/9 psi
F - Reform/9 psi
S - Reform/8 psi
F - Reform/9 psi
Oxy
Fuel
W- 10%EtOH
T-10% EtOH
N2-15%MTBE
MM-15%MTBE
NN- 17%ETBE
W- 10% EtOH
T-10% EtOH
N2- 15%MTBE
MM- 15%MTBE
NN-17%ETBE
Wt%
Oxygen
3.5
3.5
2.6
2.7
2.5
3.5
3.5
2.6
2.7
2.5
n
20
20
20
20
20
18
18
18
18
18
Mean CO
(g/mi)
Non-
Oxy
2.85
2.74
2.74
2.50
2.74
2.99
2.76
2.76
2.55
2.76
Oxy
2.32
2.27
2.53
2.43
2.40
2.45
2.35
2.64
2.51
2.46
CO
Impact
-18.6%
-17.2%
-7.7%
-2.8%
-12.4%
-18.1%
-14.9%
-4.3%
-1 .6%
-10.9%
CO
Impact
Per
Wt% O
-5.3%
-4.9%
-2.9%
-1 .0%
-5.0%
-5.2%
-4.2%
-1.7%
-0.6%
-4.3%
Splash-Blended Ethanol Pairs with an Increase of 1 psi RVP for the Ethanol Blend
Emitter
Class
Normal
Normal
Model
Year
1988+
1988+
ADL
All
Vehicles
with and
w/o ADL
Yes
Non-Oxy
Fuel
A- Ind Ave/9 psi
V- Ind Ave/8 psi
F - Reform/9 psi
S - Reform/8 psi
A - Ind Ave/9 psi
V- Ind Ave/8 psi
F - Reform/9 psi
S - Reform/8 psi
Oxy
Fuel
X- 10% EtOH
W- 10% EtOH
U- 10% EtOH
T-10% EtOH
X- 10% EtOH
W- 10% EtOH
U- 10% EtOH
T - 1 0% EtOH
Wt%
Oxygen
3.5
3.5
2.6
2.7
3.5
3.5
2.6
2 5
n
20
20
20
20
18
18
18
18
Mean CO
(g/mi)
Non-
Oxy
2.85
2.61
2.74
2.50
2.99
2.76
2.76
2 55
Oxy
2.42
2.32
2.59
2.27
2.54
2.45
2.72
235
CO
Impact
-15.1%
-11.1%
-5.5%
-9.2%
-15.1%
-1 1 .2%
-1 .4%
-7 8%
CO
Impact
Per
Wt% 0
-4.3%
-3.2%
-2.1%
-3.4%
-4.3%
-3.2%
-0.6%
-3 1 %
EPA Emission Factors Database - Matched RVP Data Pairs - Table 4 shows the mean CO
emission rates for paired data from the EPA Emission Factors database for the fuel pairs
in which the RVP between the non-oxygenate and oxygenated blend matched. Results of
three different fuels are included in this table:
The baseline, non-oxygenated tests were conducted on Indolene (nominally 9.0
psi RVP) and are recorded as "RECV" in the EPA database;
-------
David H. Lax
-11-
February 13, 1998
The MTBE blend is a 15 vol% volatility matched blend of MTBE and gasoline
with an RVP of 9.1 psi (these tests are logged as "TST93" in the EPA database); and
The ethanol blend is a 10 vol% volatility matched blend of ethanol and gasoline
with an RVP of 9.0 psi (these tests are logged as "TST92" in the EPA database).
Because the results from this series of tests were included in the development of the
Complex model, it was a simple matter of merging the EPA vehicles that were included
in the Complex model database with the EPA emission factors database to determine
which vehicles in the EPA emission factors database were equipped with adaptive
learning (i.e., vehicles in the EPA database that were also included in the Complex model
database were assumed to have ADL, those not in the Complex model database were
assumed not to have ADL).
Table 4
Mean CO Emissions From Paired Tests in the EPA Emission Factors Database
(Based on Matched RVP Non-Oxy/Oxy Pairs)
Emitter
Group
Normal
High
Tech Group
81+OX/OL
81-853W/CL
86+ 3W/NoADL
86-87 3W/ADL
81+OX/OL
81-853W/CL
86+ SW/NoADL
86-87 3W/ADL
1988+3W/ADL
81+OX/OL
81-853W/CL
86+ 3W/NoADL
86-87 3W/ADL
1988+3W/ADL
81+OX/OL
81-853W/CL
86+ SW/NoADL
86-87 3W/ADL
1 988+ 3W/ADL
Oxy Fuel
10%EtOH
10%EtOH
10%EtOH
10%EtOH
10%EtOH
15% MTBE
15% MTBE
15% MTBE
15% MTBE
15% MTBE
10%EtOH
10%EtOH
10%EtOH
10%EtOH
10%EtOH
15% MTBE
15% MTBE
15% MTBE
15% MTBE
1 5% MTBE
Wt%
Oxygen
3.5
3.5
3.5
3.5
3.5
2.7
2.7
2.7
2.7
2.7
3.5
3.5
3.5
3.5
3.5
2.7
2.7
2.7
2.7
27
n
1
8
141
89
110
6
12
62
65
58
0
7
25
22
4
8
18
20
23
2
Mean CO
(a/mi)
Non-
Oxy
5.99
4.65
3.22
3.19
2.99
3.70
4.42
3.06
3.38
2.77
na
13.67
17.16
19.03
8.33
32.16
22.84
18.61
18.37
762
Oxy
5.09
3.89
2.55
2.59
2.54
2.35
3.77
2.56
2.89
2.49
na
10.11
12.29
16.36
4.84
20.11
18.33
14.71
17.93
485
CO Impact3
-15.0% ± na
-16.3% ± 10.8%
-20.8% ± 3.4%
-18.8% ± 4.3%
-15.1% ± 3.9%
-36.8% ± 27.8%
-14.7% ± 9.7%
-16.3% ± 6.4%
-14.5% ± 4.5%
-10.1% ± 5.5%
na ± na
-26.0% ± 37.1%
-28.4% ± 1 1 .0%
-14.1% ± 8.3%
-41.9% ± 29.2%
-37.5% ± 10.7%
-19.7% ± 8.1%
-21.0% ± 12.3%
-2.3% ± 20.3%
-36 5% + 53 3%
CO
Impact
Per Wt%
0
-4.3%
-4.7%
-5.9%
-5.4%
-4.3%
-13.5%
-5.4%
-6.1%
-5.4%
-3.7%
na
-7.4%
-8.1%
-4.0%
-12.0%
-13.9%
-7.3%
-7.8%
-0.9%
-13 5%
'Confidence interval reflects a 90% confidence level.
The results presented in Table 4 show that for normal-emitting vehicles, newer
technology vehicles are less impacted by the presence of oxygen in the fuel than older
technology vehicles. This is true for both the ethanol and the MTBE blends. (Because of
-------
David H. Lax
-12-
February 13, 1998
the small sample size of the 1981+ OX/OL and 1981-85 TWC/CL vehicles, it is difficult
to draw conclusions about those technologies from this test program.) For high-emitting
vehicles, the effect of fuel oxygen on CO emissions is more variable, likely because of
the inherent variability of high-emitting vehicles.
Another series of tests included in the EPA emission factors database consisted of testing
with a 17.5 vol% MTBE blend with an RVP of 9 psi. (This test is logged as "TST15" in
the database.) The results of that testing are given in Table 5, which compares non-
oxygenated fuel results (based on Indolene) to the 17.5 vol% MTBE blend. For normal-
emitting vehicles, those results are similar to the 15 vol% MTBE results given in Table 4.
As with the previous results, the impact of fuel oxygen content on high-emitter CO levels
is more variable.
Table 5
Mean CO Emissions From Paired Tests in the EPA Emission Factors
Database
(Based on Matched RVP Non-Oxy/Oxy Pairs with 17.5% MTBE Fuel)
Emitter
Group
Normal
High
Tech Group
81+ OX/OL
81-853W/CL
86+ 3W/NoADL
86-87 3W/ADL
1988+3W/ADL
81+ OX/OL
81-853W/CL
86+ 3W/NoADL
86-87 3W/ADL
1988+3W/ADL
Oxy
Fuel
17.5%
MTBE
17.5%
MTBE
17.5%
MTBE
17.5%
MTBE
17.5%
MTBE
17.5%
MTBE
17.5%
MTBE
17.5%
MTBE
17.5%
MTBE
17.5%
MTRF
Wt%
Oxyge
n
3.1
3.1
3.1
3.1
3.1
3.1
3.1
3.1
3.1
3.1
n
6
4
24
24
38
8
11
9
7
2
Mean CO (g/mi)
Non-Oxy
3.70
3.95
3.30
3.06
2.68
32.16
28.68
27.03
28.13
7.62
Oxy
2.32
4.45
2.42
2.68
2.44
21.20
21.23
24.20
37.94
5.16
CO Impact3
-37.3% ± 17.1%
+ 12.7% ± 89.7%
-26.7% ± 8.3%
-12.4% ± 8.9%
-9.0% ± 6.1%
-34.1% ± 11.5%
-26.0% ± 13.5%
-10.5% ± 24.1%
+34.9% ± 57.3%
-32.3% ± 111.3%
CO Impact
Per Wt% O
-12.0%
+4.1%
-8.6%
-4.0%
-2.9%
-11.0%
-8.4%
-3.4%
+ 11.2%
-10.4%
a Confidence interval reflects a 90% confidence level.
A final comparison of paired tests performed with matching RVP fuel sets in the EPA
database is given in Table 6. That table compares a non-oxygenated commercial fuel
(termed "CMFUEL" in the database) to an 11 vol% MTBE blend ("TST68" in the
database) and a 10 vol% ethanol blend ("TST65" in the database). All fuels in this series
of tests had a nominal RVP of 11.7 psi. Because the results of these tests were not
-------
David H. Lax
-13-
February 13, 1998
included in the Complex model database, it was not possible to determine if vehicles
were equipped with ADL. Thus, the following broader technology groups were used for
the analysis:
1981 and later OX catalyst or open-loop TWC;
1981 to 1985 TWC/CL; and
1986 and later TWC/CL.
The results from normal emitters in this testing are reasonably consistent with previous
results for the ethanol blend; however, the MTBE blend results indicate an increase in CO
emissions for the 1986+ TWC/CL group when using the oxygenated fuel. It should be
noted, however, that this increase is not significant at the 90% confidence level. (In fact,
the confidence interval for that technology group is very broad.) As observed in the
previous results, the impact of fuel oxygen on high-emitter CO levels is quite variable.
Table 6
Mean CO Emissions From Paired Tests in the EPA Emission Factors
Database
Based on Nominal 11.7 psi Matched RVP Non-Oxy/Oxy Pairs
Emitter
Group
Normal
High
Tech Group
81+OX/OL
81-853W/CL
86+ 3W/CL
81+OX/OL
81-853W/CL
86+ 3W/CL
81+OX/OL
81-853W/CL
86+ 3W/CL
81+OX/OL
81-853W/CL
86+ 3W/CL
Oxy Fuel
11% MTBE
11% MTBE
11% MTBE
10%EtOH
10%EtOH
10%EtOH
11% MTBE
11% MTBE
11% MTBE
10%EtOH
10%EtOH
1 0% EtOH
Wt%
O
2.0
2.0
2.0
3.6
3.6
3.6
2.0
2.0
2.0
3.6
3.6
36
n
2
42
47
2
54
83
1
31
3
5
36
7
Mean CO (g/mi)
Non-
Oxy
7.09
4.81
3.23
7.09
5.10
3.44
53.10
23.02
91.14
34.72
23.46
5861
Oxy
5.28
4.41
3.41
3.37
4.26
2.76
41.50
21.81
86.54
17.20
20.71
4847
CO Impact3
-25.5% ± 188.9%
-8.3% ± 6.8%
+5.6% ± 14.3%
-52.5% ± 297.2%
-16.5% ± 5.1%
-19.8% ± 5.6%
-21.8% ± na
-5.3% ± 9.4%
-5.0% ± 19.4%
-50.5% ± 23.8%
-11.7% ± 32.0%
-173% + 125%
CO
Impact
Per
Wt% 0
-12.8%
-4.2%
+2.8%
-14.6%
-4.6%
-5.5%
-10.9%
-2.6%
-2.5%
-14.0%
-3.3%
-4 8%
a Confidence interval reflects a 90% confidence level.
EPA Emission Factors Database - Non-Matching RVP Data Pairs - In addition to the
matched RVP oxygenated fuel test results in the EPA emission factors database, a series
of tests was performed with non-matching RVP fuel sets. In this testing, the following
fuels were used:
-------
David H. Lax -14- February 13, 1998
The baseline, non-oxygenated tests were conducted on Indolene (nominally 9.0
psi RVP) and are recorded as "RECV" in the EPA database;
The MTBE blend is a 15 vol% "splash" blend of MTBE and Indolene with an
RVP of 9.6 psi (these tests are logged as "TST69" in the EPA database); and
The ethanol blend is a 10 vol% "splash" blend of ethanol and Indolene with an
RVP of 10.2 psi (these tests are logged as "TST34" in the EPA database).
Mean CO emission results from this testing are summarized in Table 7, which shows a
generally smaller reduction as a result of adding oxygen to the fuel for normal-emitting
vehicles than that observed for the matched RVP fuel set (i.e., Table 4). In fact, the
1988+ TWC/ADL group shows an increase in CO emissions for both the MTBE and
ethanol blend relative to non-oxygenated gasoline (although those increases are not
significant at the 90% confidence level). It appears that these vehicles are more sensitive
to the increased RVP level than they are to the presence of oxygen in the fuel. The results
for the high-emitting vehicles are more consistent with previous findings, indicating less
sensitivity to changes in fuel volatility.
-------
David H. Lax
-15-
February 13, 1998
Table 7
Mean CO Emissions From Paired Tests in the EPA Emission Factors
Database
Based on Splash-Blended, Non-Matching RVP Non-Oxy/Oxy Pairs
Emitter
Group
Normal
High
Tech Group
81+OX/OL
81-853W/CL
86+ 3W/NoADL
86-87 3W/ADL
1988+3W/ADL
81+OX/OL
81-853W/CL
86+ SW/NoADL
86-87 3W/ADL
1988+3W/ADL
81+OX/OL
81-853W/CL
86+ 3W/NoADL
86-87 3W/ADL
1988+3W/ADL
81+OX/OL
81-853W/CL
86+ SW/NoADL
86-87 3W/ADL
1 988+ 3W/ADL
Oxy Fuel
10%EtOH
10%EtOH
10%EtOH
10%EtOH
10%EtOH
15%MTBE
15%MTBE
15%MTBE
15%MTBE
15%MTBE
10%EtOH
10%EtOH
10%EtOH
10%EtOH
10%EtOH
15% MTBE
15%MTBE
15% MTBE
15% MTBE
1 5% MTBE
Wt%
0
3.5
3.5
3.5
3.5
3.5
2.6
2.6
2.6
2.6
2.6
3.5
3.5
3.5
3.5
3.5
2.6
2.6
2.6
2.6
26
n
1
30
24
24
9
2
30
26
36
11
2
12
4
3
0
2
12
5
3
0
Mean CO (g/mi)
Non-
Oxy
5.92
3.81
3.33
2.50
2.13
5.96
3.81
3.31
2.38
2.02
7.45
37.42
26.55
13.76
0.00
7.45
37.42
23.78
13.76
000
Oxy
3.16
3.13
3.20
2.17
2.39
3.57
3.32
2.94
2.06
2.11
4.90
26.13
17.39
10.11
0.00
5.13
28.52
13.48
11.65
000
CO Impact3
-46.6% ± na
-17.8% ± 7.9%
-3.9% ± 10.3%
-13.2% ± 7.1%
+12.2% ± 23.9%
-40.1% ± 41.2%
-12.9% ± 7.0%
-11.2% ± 9.8%
-13.4% ± 7.2%
+4.5% ± 12.7%
-34.2% ± 106.1%
-30.2% ± 23.2%
-34.5% ± 82.2%
-26.5% ± 42.3%
na ± na
-31.1% ± 10.2%
-23.8% ± 17.2%
-43.3% ± 96.3%
-15.3% ± 20.2%
na + na
CO
Impact
Per
Wt% 0
-13.3%
-5.1%
-1.1%
-3.8%
+3.5%
-15.4%
-4.9%
-4.3%
-5.2%
+1 .7%
-9.8%
-8.6%
-9.9%
-7.6%
na
-12.0%
-9.1%
-16.7%
-5.9%
na
' Confidence interval reflects a 90% confidence level.
-------
David H. Lax
-16-
February 13, 1998
Linearity of Response to Oxygen Content
In previous versions of MOBILE, the emissions response to oxygenated fuel has been
assumed to be proportional to the wt% oxygen in the fuel. For a 15 vol% MTBE blend,
this typically corresponds to 2.7 wt% oxygen, whereas a 10 vol% ethanol blend generally
contains from 3.5 to 3.7 wt% oxygen. Thus, based on an assumption of linearity, a
10 vol% ethanol blend would show a 33% increase in "CO effect" relative to a 15 vol%
MTBE fuel. For high-emitting vehicles that are running rich a significant portion of the
time, the linearity assumption is probably valid. However, for normal-emitting vehicles,
it has been suggested that a point may be reached where additional oxygen does little to
further reduce CO emissions. (This issue is discussed in detail in a 1989 SAE paper.12)
To investigate this issue, results from vehicles in the EPA emission factors database that
were tested on both a 10 vol% ethanol blend and a 15 vol% MTBE blend were analyzed.
(The fuel set used in this analysis is the same as that used to construct Table 4.) The
mean CO emission levels, by emitter category, technology group, and fuel blend, are
summarized in Table 8. The oxygen content in the ethanol blend was 3.5 wt% and the
Table 8
Mean CO Emissions From Paired Tests in the EPA Emission Factors Database
(Vehicles Tested on Both EtOH and MTBE Matched RVP Oxygenated Blends)
Emitter
Group
Normal
High
Tech Group
81-853W/CL
86+ SW/NoADL
86-87 3W/ADL
1988+3W/ADL
81-853W/CL
86+ SW/NoADL
86-87 3W/ADL
1988+3W/ADL
81-853W/CL
86+ SW/NoADL
86-87 3W/ADL
1988+3W/ADL
81-853W/CL
86+ SW/NoADL
86-87 3W/ADL
1 988+ 3W/ADL
Oxy Fuel
10% EtOH
10% EtOH
10% EtOH
10% EtOH
15% MTBE
15% MTBE
15% MTBE
15% MTBE
10% EtOH
10% EtOH
10% EtOH
10% EtOH
15% MTBE
15% MTBE
15% MTBE
1 5% MTBE
Wt%
O
3.5
3.5
3.5
3.5
2.7
2.7
2.7
2.7
3.5
3.5
3.5
3.5
2.7
2.7
2.7
27
n
8
57
58
58
8
57
58
58
7
11
19
2
7
11
19
2
Mean CO
(a/mi)
Non-
Oxy
4.65
3.00
3.34
2.77
4.65
3.00
3.34
2.77
13.67
11.72
19.47
7.62
13.67
11.72
19.47
762
Oxy
3.89
2.39
2.83
2.40
3.87
2.50
2.83
2.49
10.11
7.18
17.03
5.38
10.99
7.49
19.34
4 85
CO Impact3
-16.3% ± 10.8%
-20.3% ± 6.3%
-15.3% ± 5.7%
-13.4% ± 4.8%
-16.8% ± 13.3%
-16.7% ± 6.9%
-15.3% ± 4.9%
-10.1% ± 5.5%
-26.0% ± 37.1%
-38.7% ± 12.0%
-12.5% ± 9.2%
-29.5% ± 76.8%
-19.6% ± 25.2%
-36.2% ± 11.1%
-0.7% ± 23.4%
-36 5% + 53 3%
CO Impact
Per Wt% 0
-4.7%
-5.8%
-4.4%
-3.8%
-6.2%
-6.2%
-5.7%
-3.7%
-7.4%
-11.1%
-3.6%
-8.4%
-7.3%
-13.4%
-0.2%
-135%
' Confidence interval reflects a 90% confidence level.
-------
David H. Lax
-17-
February 13, 1998
25
Figure 2
CO Reductions as a Function of Wt% Oxygen
for Normal- and High-Emitting Vehicles
20 -
15 -
u
D
i 10
o
o
5 -
-486+ TWC/No ADL - Normal
- - -86-87 TWC/ADL - Normal
-A - 1988+ TWC/ADL - Normal
O- 1981+TWC-High
0.5
1.5 2 2.5
Wt% Oxygen
3.5
oxygen content in the MTBE blend was 2.7 wt%. The results for normal- and high-
emitting vehicles (all data were combined for the high emitters) are illustrated in Figure 2,
which indicates a linear response to fuel oxygen content for the 1986+ TWC/No ADL
and 1988+ TWC/ADL technology groups (normal emitters). Although the mean CO
emission reductions from the normal-emitting 1986-87 TWC/ADL group and the high-
emitters show a non-linear trend, the confidence interval for those estimates is such that
the trend is linear within the error associated with the estimates. Given this, a linear
response should be assumed unless a more rigorous test program evaluating this effect is
conducted.
Temperature Effects
One issue that has received considerable attention in the past is whether the benefits
predicted for oxygenated blends at moderate temperatures also hold at lower
temperatures. A vast majority of the vehicle testing used to generate oxygenated fuel
adjustment factors has been conducted at the standard FTP temperature of 75 °F.
However, these fuels are typically used at much lower temperatures. Because ambient
CO studies generally support a lower benefit than that predicted by emission models, it
has been postulated that the benefits of oxygenated fuels are diminished at low
temperature.
In 1995, Hood and Farina13 published a review of studies on the effects of oxygenated
fuels on emissions at low ambient temperature. Although the available data are quite
-------
David H. Lax -18- February 13, 1998
scattered, the authors concluded that the directionality of the impact of oxygenates at low
temperature is consistent with testing at 75 °F, i.e., addition of 2% to 3% (by weight) of
oxygen generally decreases fleet emissions of HC and CO at lower temperatures.
However, they also concluded that the CO reductions are lower (on a percentage basis)
compared to what has been observed at 75 °F.
Because many of the test programs cited in the above paper tested only a limited number
of vehicles, and because the fuel/temperature combinations varied among programs, it is
difficult to quantitatively establish an adjustment to account for the impact of cold
temperature on the effects of oxygenated fuels. Nonetheless, data from several of the
papers cited by Hood and Farina were obtained and analyzed. In addition, the EPA
emission factors database contains a number of vehicles that were tested with and without
an oxygenated fuel at 75°, 50°, and 20°F. Those data were also evaluated in this effort.
CRC Study - A summary of paired tests from the CRC study is given in Tables 9 and 10.
To maintain consistency with other data analyzed in this effort, only the low-altitude data
were used in this analysis. The results are stratified by the technologies evaluated in that
study - three-way catalysts with adaptive learning and three-way catalysts without
adaptive learning are shown in Table 9. All of the vehicles included in Table 9 are
normal emitters (i.e., CO emissions less than 7 g/mi when tested at 75 °F on Indolene).
Results from the four oxidation catalyst vehicles in the program are shown in Table 10.
Two of those vehicles were high emitters, and the results are shown separately for normal
and high emitters.
-------
David H. Lax
-19-
February 13, 1998
Table 9
Effect of Oxygenated Gasoline and Temperature on Mean CO Emissions
of Paired Tests from TWC Vehicles in the CRC Study
Model Year/
Technology
1986-1988
TWC/ADL
1983-1986
TWC/No ADL
Sample
Size
5a
6
Base
Fuel
RVP (psi)
12.8
12.8
Oxygen
Blend
MTBE
2.0 wt% O
12. 5 psi RVP
EtOH-Splash
3.7 wt% O
13. 2 psi RVP
EtOH-
Matched
3.7 wt% O
12. 7 psi RVP
MTBE
2.0 wt% O
12. 5 psi RVP
EtOH-Splash
3.7 wt% O
13. 2 psi RVP
EtOH-
Matched
3.7 wt% O
12. 7 psi RVP
Tern
P
(°F)
75
50
35
75
50
35
75
50
35
75
50
35
75
50
35
75
50
35
Base CO
(g/mi)
2.44
5.43
8.20
2.44
5.43
8.20
2.44
5.43
8.20
3.42
10.39
16.79
3.42
10.39
16.79
3.42
10.39
16.79
Oxy CO
(g/mi)
2.23
5.03
7.70
2.12
4.74
7.98
2.43
5.13
7.04
3.14
10.12
16.43
2.87
9.16
14.24
3.18
9.06
14.49
Oxygen
Effect
-8.6%
-7.4%
-6.1%
-13.1%
-12.7%
-2.7%
-0.4%
-5.5%
-14.1%
-8.2%
-2.6%
-2.1%
-16.1%
-1 1 .8%
-15.2%
-7.0%
-12.8%
-13.7%
a Note that one of the six TWC/ADL vehicles in the original program had been removed from the database that
Sierra received from CRC because of its erratic behavior.
The results presented in Table 9 show a decrease in the magnitude of the oxygenated fuel
benefit as temperature is decreased for the MTBE fuel and for the ethanol splash blended
fuel. For the ethanol match blended fuel, the opposite occurs for both the TWC/ADL and
TWC/No ADL technologies. It is not clear why this occurs, since the RVP and
distillation characteristics of this fuel are nearly identical to the MTBE blend (as well as
the base gasoline). For the OX CAT vehicles, the results are more variable. This is likely
due to the small sample size for these vehicles.
-------
David H. Lax
-20-
February 13, 1998
Table 10
Effect of Oxygenated Gasoline and Temperature on Mean CO Emissions
of Paired Tests from OX CAT Vehicles in the CRC Study
Model Year/
Technology/
Emitter Cat.
1979-1980
OX CAT
Normal
Emitters
1979-1980
OX CAT
High
Emitters
Sample
Size
2
2
Base
Fuel
RVP (psi)
12.8
12.8
Oxygen
Blend
MTBE
2.0 wt% O
12. 5 psi RVP
EtOH-Splash
3.7 wt% O
13. 2 psi RVP
MTBE
2.0 wt% O
12.5 psi RVP
EtOH-Splash
3.7 wt% O
13. 2 psi RVP
Tern
P
(°F)
75
50
35
75
50
35
75
50
35
75
50
35
Base CO
(g/mi)
7.50
14.73
24.44
7.50
14.73
24.44
49.89
32.54
42.90
49.89
32.54
42.90
Oxy CO
(g/mi)
7.11
12.65
17.03
4.83
11.09
18.97
33.75
28.86
35.82
35.96
22.52
30.44
Oxygen
Effect
-5.2%
-13.9%
-30.3%
-35.6%
-24.7%
-22.4%
-32.4%
-1 1 .3%
-16.5%
-27.9%
-30.8%
-29.0%
API Study - A summary of paired data at different temperatures from the API study is
given in Tables 11 and 12 for two different oxygenated blends - a 15 vol% MTBE blend
and a 10 vol% ethanol splash blend. Table 11 summarizes the results for the nominal
13 psi RVP fuel set, while Table 12 contains the results for the nominal 9 psi RVP fuel
set. Three different technology types are shown in these tables: TWC/ADL, TWC/No
ADL, and open-loop. Although the results are somewhat scattered, the trends observed in
this test program are consistent with the CRC program. In general, the relative reductions
in exhaust CO observed with the use of oxygenated blends decrease with decreasing
temperature.
One item of note is that in some cases, particularly for the ethanol blends, the maximum
reduction is observed at the 55 °F test temperature (see Table 12). Because the ethanol
blend used in this program was a splash blend, its RVP was about 1 psi higher than the
other fuels in the fuel set. (Two fuel sets were tested in this work - one with a nominal
RVP of 13 psi and one with a nominal RVP of 9 psi.) It is likely that under the 80°F
testing the higher RVP of the ethanol blends is partially offsetting the benefit from the
addition of oxygenate (due to higher canister loading and subsequent purge). As found in
this test program, the effect of fuel volatility on CO emissions is much more pronounced
at 80°F than it is at 55° or35°F.
-------
David H. Lax
-21-
February 13, 1998
Table 11
Effect of Oxygenated Gasoline and Temperature on Mean CO Emissions
of Paired Tests from Vehicles in the API Study
13psiRVPFuelSet
Model Year/
Technology/
1986-1988
TWC/ADL
Normal
Emitters
1981 +
TWC/No ADL
1981 +
Open-Loop
Sample
Size
4a
3
3
Base
Fuel
RVP (psi)
13.1
13.1
13.1
Oxygen
Blend
MTBE
2.6 wt% O
12. 7 psi RVP
EtOH-Splash
3.9 wt% O
13. 9 psi RVP
MTBE
2.6 wt% O
12.7 psi RVP
EtOH-Splash
3.9 wt% O
13.9 psi RVP
MTBE
2.6 wt% O
12.7 psi RVP
EtOH-Splash
3.9 wt% O
13.9 psi RVP
Tern
P
(°F)
80
55
35
80
55
35
80
55
35
80
55
35
80
55
35
80
55
35
Base CO
(g/mi)
6.35
4.90
10.02
6.35
4.90
10.02
12.46
11.97
22.00
12.46
11.97
22.00
9.55
11.30
21.12
9.55
11.30
21.12
Oxy CO
(g/mi)
4.66
4.20
10.60
5.01
3.60
9.72
7.80
10.02
16.78
9.86
10.56
20.67
5.02
7.05
20.40
7.31
5.27
17.61
Oxygen
Effect
-26.6%
-14.3%
+5.8%
-21.1%
-26.5%
-3.0%
-37.4%
-16.3%
-23.7%
-20.9%
-1 1 .8%
-6.0%
-47.4%
-37.6%
-3.4%
-23.5%
-53.4%
-16.6%
aOne of the five TWC/ADL vehicles in the API study was a high emitter and
demonstrated erratic behavior.
-------
David H. Lax
-22-
February 13, 1998
Table 12
Effect of Oxygenated Gasoline and Temperature on Mean CO Emissions
of Paired Tests from Vehicles in the API Study
9 psi RVP Fuel Set
Model Year/
Technology/
1986-1988
TWC/ADL
1981 +
TWC/No ADL
1981 +
Open-Loop
Sample
Size
4a
3
3
Base
Fuel
RVP (psi)
8.9
8.9
8.9
Oxygen
Blend
MTBE
2.7 wt% O
9.1 psi RVP
EtOH-Splash
3.7 wt% O
9.9 psi RVP
MTBE
2.7 wt% O
9.1 psi RVP
EtOH-Splash
3.7 wt% O
9.9 psi RVP
MTBE
2.7 wt% O
9.1 psi RVP
EtOH-Splash
3.7 wt% O
9.9 psi RVP
Tern
P
(°F)
80
55
35
80
55
35
80
55
35
80
55
35
80
55
35
80
55
35
Base CO
(g/mi)
2.47
4.84
10.58
2.47
4.84
10.58
9.84
13.52
20.95
9.84
13.52
20.95
6.08
8.73
20.90
6.08
8.73
20.9
Oxy CO
(g/mi)
2.59
3.55
9.26
2.36
3.26
8.92
7.60
12.13
19.56
7.57
12.01
22.13
4.50
6.99
20.90
4.43
5.21
18.53
Oxygen
Effect
+4.9%
-26.7%
-12.5%
-4.5%
-32.6%
-15.7%
-22.8%
-10.3%
-6.6%
-23.1%
-1 1 .2%
+5.6%
-26.0%
-19.9%
0.0%
-27.1%
-40.3%
-1 1 .3%
a One of the five TWC/ADL vehicles in the API study was a high emitter and demonstrated erratic behavior.
It was excluded from the calculations presented in this table.
CARS Study - As with the test programs described above, the CARB Low Oxygenates
test program was limited in terms of the number of vehicles tested (13). In addition, the
winter fuel set used in this program was tested only at 75° and 50°F. Nonetheless, those
data were obtained and analyzed for this study. A summary of the results, by technology
type, is given in Table 13, which shows mixed results with respect to the impact of
oxygenate at the lower temperature. For TWC/ADL vehicles operating on an MTBE
blend, the oxygen effect diminishes at 50 °F relative to 75 °F. However, the results for
that
-------
David H. Lax
-23-
February 13, 1998
Table 13
Effect of Oxygenated Gasoline and Temperature on Mean CO Emissions
of Paired Tests from Vehicles in the CARB Study
Winter Fuel Set
Model Year/
Technology/
1987-1990
TWC/ADL
(All Normals)
1983-1985
TWC/No ADL
(All Highs)
1976-1978
Open-Loop
Ox Cat
(All Highs)
Sample
Size
5
3
2
Base
Fuel
RVP (psi)
9.6
9.6
9.6
Oxygen
Blend
MTBE
2.7 wt% O
9.6 psi RVP
EtOH-Match
2.7 wt% O
9.6 psi RVP
MTBE
2.7 wt% O
9.6 psi RVP
EtOH-Match
2.7 wt% O
9.6 psi RVP
MTBE
2.7 wt% O
9.6 psi RVP
EtOH-Match
2.7 wt% O
9.6 psi RVP
Tern
P
(°F)
75
50
75
50
75
50
75
50
75
50
75
50
Base CO
(g/mi)
3.65
5.08
3.65
5.08
13.36
15.19
13.36
15.19
27.70
24.31
27.70
24.31
Oxy CO
(g/mi)
3.25
5.15
3.89
4.79
11.37
11.90
12.01
11.72
20.13
36.46
19.21
38.35
Oxygen
Effect
-10.9%
+1 .3%
+6.7%
-5.8%
-14.9%
-21 .6%
-10.1%
-22.8%
-27.3%
+31 .6%
-30.7%
+38.5%
technology group tested on an ethanol matched RVP blend show an increase in the
oxygenate benefit at 50°F. (In fact, that fuel caused an increase in CO emissions for
these vehicles at 75 °F, and a slight decrease at 50 °F relative to the non-oxygenated base
fuel.) Again, the small number of vehicles in this test program makes it difficult to
definitively predict the effect of temperature on the impact of oxygenated fuels.
EPA Emission Factors Database - The final database evaluated to assess the impact of
ambient temperature on the CO emissions effects of fuel oxygenates was the EPA
emission factors database. That database contains emissions data from vehicles tested on
a variety of fuels at a number of ambient temperatures (tests at 75°, 50°, and 20°F were
analyzed in this effort). For the most part, the oxygenated fuel low-temperature testing
has been conducted with 10 vol% ethanol blends, although there is a limited number of
low-temperature tests that were conducted with 15 vol% MTBE blends.
The most comprehensive (in terms of temperatures) and consistent (in terms of fuels)
series of tests in this database includes 29 vehicles that were tested with a base gasoline
-------
David H. Lax
-24-
February 13, 1998
with an RVP of 11.7 psi and a 10 vol% ethanol blend with an RVP of approximately
11.8 psi. The test sequence designations from the EPA database were as follows:
Temperature
75 °F
50°F
20°F
Non-Oxygenate Fuel
CMFUEL
TST23
TST25
Oxygenated Fuel
TST65
TST66
TST67
Although the EPA database does not have detailed fuel specification information, it has
been assumed here that the oxygenated blend used for this series of tests was generally
consistent with the 11.7 psi RVP commercial fuel used in the non-oxygenated tests.
The mean emissions and oxygen effect from the paired tests outlined above are
summarized in Table 14 by technology and emitter class. Only two of the 29 vehicles in
this series of tests were equipped with open-loop systems, so they were removed from the
analysis. The 21 normal-emitting three-way/closed-loop vehicles were segregated by
model year (1981 to 1985 and 1986+), while the high emitters were combined into a
single group. As observed in Table 14, the impact of oxygenated fuel is diminished at
50° and 20°F (relative to 75 °F) for the normal emitters. The results for the high-emitting
vehicles show considerable scatter.
Table 14
Effect of Oxygenated Gasoline and Temperature on Mean CO Emissions
of Paired Tests from TWC Vehicles in the EPA Database
Fuels Having a Nominal RVP of 11.7 psi - Tests Conducted at 75°, 50°,
and 20°F
Model Year/
Technology
1986+
TWC/CL
Normal
Emitters
1981-1985
TWC/CL
Normal
Emitters
1981 +
TWC/CL
High
Emitters
Sample
Size
13
8
6
Base
Fuel
RVP (psi)
11.7
11.7
11.7
Oxygen
Blend
EtOH
10vol%
1 1 .8 psi RVP
EtOH
10vol%
1 1 .8 psi RVP
EtOH
10vol%
1 1 .8 psi RVP
Tern
P
(°F)
75
50
20
75
50
20
75
50
20
Base CO
(g/mi)
3.14
7.04
13.78
5.21
6.89
11.11
55.53
34.65
39.76
Oxy CO
(g/mi)
2.48
5.87
11.32
3.88
5.80
9.64
35.31
28.53
22.38
Oxygen
Effect
-21 .0%
-16.6%
-17.9%
-23.4%
-16.2%
-16.3%
-36.4%
-17.7%
-43.7%
-------
David H. Lax
-25-
February 13, 1998
A similar analysis is shown in Table 15. However, in this case, only tests at 75° and
20 °F were considered, and all paired tests in which the oxygenated fuel was a 10 vol%
ethanol blend were included, regardless of whether the RVP between the non-oxygenated
fuel and the oxygenated fuel matched. (For the majority of those tests, the RVP for the
75 °F tests did match. For the 20 °F testing, the impact of differences in fuel volatility is
less of a concern.) As observed in the table, this approach increased the sample size
considerably. The results from this analysis show that the effect of fuel oxygen content
on CO emissions diminishes at lower temperatures. For this group of vehicles, both
normal and high emitters had lower CO reductions at the 20 °F test temperature relative to
75°F.
Table 15
Effect of Oxygenated Gasoline and Temperature on Mean CO Emissions
of Paired Tests from TWC Vehicles in the EPA Database
All Paired Ethanol Blend Tests Conducted at 75° and 20 F
Model Year/
Technology
1986+
TWC/CL
Normal
Emitters
1981-1985
TWC/CL
Normal
Emitters
1981 +
TWC/CL
High
Emitters
Sample
Size
45
15
16
Base
Fuel
RVP (psi)
9.0
or
11.7
9.0
or
11.7
9.0
or
11.7
Oxygen
Blend
EtOH
10vol%
Misc. RVP
EtOH
10vol%
Misc. RVP
EtOH
10vol%
Misc. RVP
Tern
P
(°F)
75
20
75
20
75
20
Base CO
(g/mi)
3.40
15.31
4.50
14.98
52.53
53.93
Oxy CO
(g/mi)
2.84
13.54
3.75
12.68
34.99
41.50
Oxygen
Effect
-16.5%
-1 1 .6%
-16.7%
-15.4%
-33.4%
-23.0%
In summary, many of the variable-temperature test programs support the contention that
as ambient temperature is decreased, the impact of fuel oxygen content on exhaust CO
emissions is diminished. However, the available data are highly variable, making a
precise quantitative correction for temperature difficult. At this time, there is no basis for
offering a specific recommendation to account for temperature effects on the CO benefits
attributable to oxygenated fuels, although EPA should revisit this issue if/when additional
cold-temperature oxygenated fuel data are collected.
-------
David H. Lax -26- February 13, 1998
Oxygen Impacts on FTP. Starting, and Running CO Emissions
One of the questions that came up during the course of this study was whether the fuel
oxygen effects on CO emissions calculated from FTP composite data could be applied
equally to all operating modes (i.e., cold start, hot start, and stabilized). Because EPA has
indicated that running and starting emissions will be modeled separately in MOBILE6,
this issue will take on greater importance with the release of that model.
As a check on the validity of using FTP-composite fuel oxygen impacts for all operating
modes, Sierra compared mean emissions from the EPA matched RVP data pairs for the
FTP-composite results (presented above in Table 4), hot-running emissions, and cold-
start emissions. Two approaches were used to estimate hot-running and cold-start
emissions from the individual bag results. First, Sierra has previously estimated cold-
start emissions (in terms of grams per engine start) from FTP bag data by subtracting bag
3 from bag 1 and multiplying the result by the length of bags 1 and 3 (3.59 miles).
Although bag 3 occurs after a 10-minute soak, it has been reported that the degree of fuel
enrichment is diminished and catalyst light-off is rapid in bag 3 for properly functioning
late-model vehicles. Thus, bag 3 is a reasonable estimate of stabilized emissions over the
same cycle as bag 1 and can be used to develop a cold-start emissions estimate. For
overall hot-running emissions, Sierra's methodology combines bag 2 with bag 3 of the
FTP using weighting factors of 0.521 for bag 2 and 0.473 for bag 3. These weighting
factors are consistent with the length (in miles) of bags 2 and 3. Shown elow are the
equations used by Sierra to calculate cold-start and hot-running emissions, where the bag
data are reported in g/mi.
CSsierra (grams/start) = (Bag 1 - Bag 3) * 3.59 miles
(grams/mile) = (Bag 2 * 0.521) + (Bag 3 * 0.473)
EPA is taking a similar approach to developing cold-start and hot running emission
estimates for MOBILE6. However, its approach is a little more sophisticated in that "Hot
Running 505" (HR505) emissions, defined as bag 3 without an engine start and in a
completely warmed-up state, are substituted for the bag 3 results in the above equations.
HR505 emissions can be estimated from individual bag data with regression equations
recently developed by EPA.14 For CO, that equation is:
HR505 = exp(-0.3452*ln(Bag 1) + 0.4304*ln(Bag 2) + 0.5375*ln(Bag 3) - 0.0674)
where exp denotes the exponential function and In is the natural logarithm. Using the
HR505 results obtained from the equation above, cold-start and running exhaust emission
rates are generated from individual bag data according to the following equations:
CSEPA (grams/start) = (Bag 1 - HR505) * 3.59 miles
HREPA (grams/mile) = (Bag 2 * 0.521) + (HR505 * 0.473)
-------
David H. Lax -27- February 13, 1998
The above methodology was applied to the EPA matched RVP data set (i.e., Table 4) to
obtain cold-start and hot-running emission estimates for each vehicle for the baseline fuel
and for the oxygenated blends. The CO emissions impacts of the oxygenated fuel blends
were then estimated for the composite FTP and for the hot-running and cold-start
emissions estimates using both the Sierra methodology and the EPA methodology
outlined above. The results of that analysis are summarized in Table 16. In general, for
the groups with larger sample sizes, the results show consistent impacts across the FTP
composite, hot-running, and cold-start modes. For cases in which the impacts show a
noticeable difference, e.g., normal-emitting 1988+ 3W/ADL vehicles operated on 10 vol
% ethanol, the FTP, hot-running, and cold-start impacts are within the uncertainty
intervals computed for these operating modes. Based on these results, the use of FTP -
composite CO emissions impacts for hot-running and cold-start emissions is valid.
Recommended Methodology for MOBILE6
As evidenced by the information presented above, there is a considerable body of data
with which to predict the emissions impacts of oxygenated fuels. One area of concern,
however, is whether the comparisons being made with the various data sets reflect only
the impact of the oxygenate, or whether other fuel parameter changes are contributing to
the differences observed (e.g., RVP differences). With this in mind, the following
recommendations are being made.
CO Impacts as a Function of Oxygen Content for Tier 0 Vehicles - To generate an overall
adjustment to account for the impact of oxygen content on CO emissions, paired test
results from the databases described above were combined into a single data set.* In all
cases, only testing conducted at a nominal temperature of 75 °F was included in the final
database used to develop the oxygen content fuel adjustments.
For each vehicle in the final database, the percentage CO impact from the use of
oxygenated gasoline was determined for each fuel pair (i.e., oxygenated versus non-
oxygenated gasoline) over which the vehicle was tested. These results were then put on a
consistent basis by dividing the impact by the oxygen content in the oxygenated test fuel
(i.e., the wt% oxygen), giving a CO impact per 1 wt% oxygen for each fuel pair. (This
approach is supported by the linearity demonstration presented above.) So that individual
vehicles receiving multiple tests (e.g., those in the Auto/Oil program) would not have an
undue influence on the final results, the average CO impact (per 1 wt% oxygen) was
determined for each vehicle by taking the mean of the impacts determined for each fuel
pair (i.e., an average CO impact per wt% oxygen was calculated for each vehicle).
Note that data from 1986 and later model year vehicles in the EPA 11.7 psi matched RVP testing (i.e.,
Table 6) were not used because it was not possible to determine which of these vehicles had ADL systems.
In addition, data from the Auto/Oil AMOT study (Table 2) were not used because it was felt that the results
from the RVP/Oxygenate testing (Table 3) for those vehicles were more applicable for determining a winter
CO impact of fuel oxygen content.
-------
oo
-8
PH
OO
CN
X
J
K
T3
to
C
o
E
HI
O)
C
r
(0
CO
D
(0
0> -^
C "*
^ '^3
C n
_ ^L
(£ >
n" O
^L S^
1 >
LJ_ X
s| 1
co 0) Z
^; in
2 j{{ f\
(0 * ^^
^^ DLJ ^^
(« *
C "D
O 0)
m^m ^
(/) ^
_(/) id
"c ^
HI ^
~* ^
o 5:
O UJ
c "~"
m
re
a)
§
2
O
CO
"o
re
Q.
c
a)
O)
^^
O
t
CO
cyp^
"o
O c:
<
CL
LJJ
,
O c "?j
^ 'E <\
C OO -
CO CD "^
C t- CO
+1 +1 +1
O ^t OO
LO CD CD
T- T- CN
1 1 1
T- OO 5
cc cc cc
E E E
o o o
Z Z Z
1
1+OX/OL
-85 3W/CL
3W/NoADl
oo <- +
CO CD
oo
III
o o o
LJJ LJJ LJJ
"5 "5 "5
> > >
o o o
j£ ^
CD CD
LO -
LO
CD o^
CO °°.
CN CD
+1 +1
0 0
>
LO LO
j£ ^
* T
CD CD
+1 +1
0 0
O LO
OO OO
i i
09- 09-
LO CD
c*~ CD
+i +i
-? o?
>
LO LO
CD ?-;
CO LO
CN T-
CO +| +|
f
CN CO
LO CD
1 CN
i
CD CO
CN - > >
o o o
CN ?-
CN T^
+1 +1
0 0
CO CO
oo -
CD ^
t- OO
+1 +1
LO OO
r^ CD
CO T-
1 '
00 °2
O) O)
I I
1+OX/OL
-85 3W/CL
oo <-
LJJ LJJ
00 00
| |
^ ^
"5 "o
> >
LO LO
O h- CN
CN - > >
LO LO LO
^
H
o
CT1
1>
a
O
p
bo
aa
i .
0
«-)
ffi
_ya
-^
1 * PH
g w
(D ^f~\
^ ^
g £
^ Qi
^Q -fj
*S 'o
O ^
|l
"S a
o o
O Q
-------
David H. Lax
-29-
February 13, 1998
Once the individual vehicle impacts were determined, the overall fleet impacts were
calculated by taking the mean of the per-vehicle impacts* according to the technology
groups described previously. This was done as a function of emitter category and
whether the fuel pairs had matched or mis-matched RVPs. For the matched RVP fuel
pairs, all oxygenate types were combined (i.e., ethanol, MTBE, and ETBE). For the mis-
matched RVP fuel pairs, only ethanol splash blends were considered since, in practice,
that is the fuel most likely to have a higher RVP than the non-oxygenated base fuel. The
results of this analysis are given in Table 17 for the matched RVP blends. Note that
because of the large variability observed in the high-emitting vehicles, all 1981 and later
three-way catalyst/closed-loop vehicles were combined. Also, because of the small
number of non-catalyst and oxidation catalyst/open-loop vehicles in the databases
evaluated in this effort, it is recommended that the CO impacts developed by EPA in the
1988 technical guidance on oxygenated fuels15 continue to be used for those technologies.
Table 17
Recommended CO Effects From the Use of Oxygenated Fuels
for Matched RVP Blends at 75 °F
Emitter
Category
Normal
High
Technology
1988+TWC/ADL
1 986-87 TWC/ADL
1986+TWC/NoADL
1981-85TWC/CL
OX/OLb
Non-Catalystb
1981+TWC/CL
OX/OLb
Non-Catalystb
CO Impact Per
Wt% Oxygen
-3.1%(n=133)a
-4.8%(n=104)
-5.7%(n=151)
-4.0% (n=73)
-9.4%
-6.6%
-5.3% (n=134)a
-9.4%
-6.6%
Typical
MTBE
Blend
(2.7 wt% O)
-8.4%
-13.0%
-15.4%
-10.8%
-25.4%
-17.8%
-14.3%
-25.4%
-17.8%
Typical
Ethanol Blend
(3.5 wt% O)
-10.9%
-16.8%
-20.0%
-14.0%
-32.9%
-23.1%
-18.6%
-32.9%
-23.1%
Sample size shown in parentheses.
' CO impacts for these technologies are based on reference 15.
One might argue that calculating the percent reduction in fleet-average CO is more reflective of the fleet-
wide impact of oxygenated fuels. However, because of the widely variable base fuels used in the test
programs evaluated in this study, and because the data were segregated by emitter category, it was felt that
the approach taken was less prone to being heavily influenced by a few vehicles that may or may not be
reflective of the in-use population.
-------
David H. Lax
-30-
February 13, 1998
Table 18 summarizes the results of the paired-data analysis for the ethanol splash blends
that have a nominal 1 psi RVP increase over the base fuel. Outlined below are several
items worth noting with respect to that table.
Table 18
Recommended CO Effects From the Use of Oxygenated Fuels
for Ethanol Splash Blends with Mis-Matched RVP at 75 F
Emitter
Category
Normal
High
Technology
1988+TWC/ADL
1 986-87 TWC/ADL
1986+TWC/NoADL
1 981-85 TWC/CL
1981+TWC/CL
CO Impact Per
Wt% Oxygen
+0.3% (n=34)a
-3.1%(n=44)
-3.6% (n=31)
-5.0% (n=35)
-4.5% (n=27)a
Typical
Ethanol Blend
(3.5 wt% O)
+1.1%
-10.9%
-12.6%
-17.5%
-15.8%
Sample size shown in parentheses.
In general, the CO emissions impacts listed in Table 18 are lower than those listed
in Table 17. This is the result of the higher RVP of the oxygenated fuel relative to
the base fuel (i.e., the increase in RVP leading to higher CO emissions partially
offsets the benefits of the fuel oxygen). This is particularly evident for the 1988+
TWC/ADL technology, which appears to show a stronger response to the RVP
increase than to the presence of fuel oxygen, resulting in a slight increase in CO
emissions with ethanol splash blends.
Recommendations for OX/OL and non-catalyst vehicles are not included in
Table 18 because EPA adjusted those data for RVP differences prior to
developing the CO impacts. MOBILE contains a separate RVP adjustment, and
that adjustment should be applied to the impacts listed in Table 17 to reflect fuels
receiving an RVP waiver for these technology types.
The results presented in Table 18 are valid at 75 °F and are valid only for ethanol
splash blends receiving an RVP waiver. As the test temperature decreases, it was
shown previously that the RVP effect will be less pronounced. For "warm-
weather" areas, the results in Table 18 should be used directly to determine the
CO impacts of ethanol splash blends receiving an RVP waiver. However, at
temperatures below 75 °F, some combination of Tables 17 and 18 should be used
-------
DavidH. Lax -31- February 13, 1998
because the effect of RVP on CO emissions decreases with decreasing
temperature. One means to account for this is to linearly interpolate between the
two tables from 75 °F (Table 18) down to 45 °F (Table 17), which is the
temperature at which the RVP adjustment in MOBILES is turned off. Below
45 °F, the results presented in Table 17 should be used for both matched RVP and
mis-matched RVP blends. This approach is supported by data from the API
RVP/Oxygenates program9 which show that the effect of RVP on CO emissions is
greatly reduced at 55°F and negligible at 35°F. In addition, data from recent
testing representative of the Las Vegas vehicle fleet showed that RVP has a
statistically significant effect on CO emissions collected at "afternoon" winter
temperatures (65 °F), while no statistically significant effect was observed for the
"morning" temperatures (40 °F).16
Tier 1 and Advanced Technology Vehicles - A significant question that remains relates to
the oxygenated fuel CO emissions impact assumed for vehicles certified to Tier 1 and
lower emission standards. Based on testing performed in the Auto/Oil program,5 the
response to fuel oxygen (at a 2.0 wt% level) in a Tier 1 fleet and in an advanced
technology fleet was very small (i.e., +1.2% and -1.0%, respectively). In addition, recent
testing conducted by CRC17 to support an investigation of fuel sulfur impacts on low-
emission vehicles indicates that the presence of oxygen has little or no effect on CO
emissions from advanced technology vehicles. Given the importance of these
technologies on future emission estimates prepared with MOBILE6, it is recommended
that testing be performed to better define the emissions impact of oxygenated fuels on
late-model vehicles.
Conclusions and Final Recommendations for MOBILES
Based on the analyses presented above, the following recommendations are being made
with respect to modeling the CO emissions impacts of oxygenated fuel blends in
MOBILE6:
For Tier 0 vehicles, the CO emissions impacts listed in Table 17 should be used
for matched RVP blends at all temperatures. Those impacts should also be used
for mis-matched RVP blends (i.e., ethanol splash blends receiving an RVP
waiver) at temperatures below 45 °F. The CO emissions impacts listed in Table
18 should be used for ethanol splash blends receiving an RVP waiver at
temperatures of 75 °F. Between 45° and 75 °F, the CO impacts of ethanol splash
blends receiving an RVP waiver should be modeled by interpolating between
Table 17 (= 45 °F) and Table 18 (= 75 °F).
Because new data on older technology vehicles (i.e., pre-1981 oxidation catalyst
and non-catalyst vehicles) have not been collected in recent years, the CO
emissions impacts for those vehicles should be based on EPA's 1988 Guidance
-------
David H. Lax -32- February 13, 1998
Document on the emissions impacts of oxygenated fuels.15 These effects are
currently used in MOBILES a.
With the exception of ethanol splash blends receiving an RVP waiver, the CO
emissions impacts are not a function of oxygenate type. Further, the CO
emissions impacts are a linear function of the wt% oxygen in the fuel.
The percentage CO emissions impacts developed from FTP-composite data can be
applied to both starting and running exhaust emission estimates.
Although available data appear to indicate that the CO impacts of oxygenated
fuels are diminished at low ambient temperatures, the data are highly variable,
making a precise quantitative estimate of this effect difficult. Thus, no specific
recommendation is being made at this time to adjust the oxygenated fuel CO
impacts for low ambient temperature. (Recall that the temperature adjustment of
the splash-blended ethanol CO impacts is an RVP adjustment, not an oxygenate
adjustment.)
Very few data are available with which to estimate the impact of oxygenated fuel
on CO emissions from Tier 1 and advanced technology vehicles. The data that are
available indicate only a small effect is observed for these technologies. Given
the importance of these vehicles on future CO emissions estimates prepared with
MOBILE6, testing needs to be conducted to better define the emissions impacts of
using oxygenated fuels with these vehicles. Until a more robust dataset is
available, estimates prepared with the existing data should be used in MOBILE6.
References
1. Rao, V., "Development of an Exhaust Carbon Monoxide Emissions Model," SAE
Paper No. 961214, 1996.
2. Western States Petroleum Association, "WSPA Comments on the CO Complex
Model," Letter from WSPA to Margo Oge, Director, Office of Mobile Sources,
U.S. Environmental Protection Agency, November 4, 1996.
3. Hochhauser, A.M., et al., "The Effect of Aromatics, MTBE, Olefins and T90 on
Mass Exhaust Emissions from Current and Older Vehicles - The Auto/Oil Air
Quality Improvement Research Program," SAE Paper No. 912322, 1991.
4. Reuter, R.M., et al., "Effects of Oxygenated Fuels and RVP on Automotive
Emissions - Auto/Oil Air Quality Improvement Program," SAE Paper No.
920326, 1992.
-------
David H. Lax -33- February 13, 1998
5. "Auto/Oil Air Quality Improvement Research Program, Technical Bulletin No.
17: Gasoline Reformulation and Vehicle Technology Effects on Exhaust
Emissions," Coordinating Research Council, August 1995.
6. Mayotte, S.C., et al., "Reformulated Gasoline Effects on Exhaust Emissions:
Phase I: Initial Investigation of Oxygenate, Volatility, Distillation and Sulfur
Effects," SAE Paper No. 941973, 1994.
7. Mayotte, S.C., et al., "Reformulated Gasoline Effects on Exhaust Emissions:
Phase II: Continued Investigation of the Effects of Fuel Oxygenate Content,
Oxygenate Type, Volatility, Sulfur, Olefins and Distillation Parameters," SAE
Paper No. 941974, 1994.
8. Korotney, D.J., et al., "Reformulated Gasoline Effects on Exhaust Emissions:
Phase HI: Investigation on the Effects of Sulfur, Olefins, Volatility, and Aromatics
and the Interactions Between Olefins and Volatility or Sulfur," SAE Paper
No. 950782, 1995.
9. "The Effects of Fuel RVP and Fuel Blends on Emissions at Non-FTP
Temperatures," Vol. I, Health and Environmental Sciences Department, API
Publication No. 4533, August 1994.
10. "Effect of Use of Low Oxygenate Gasoline Blends upon Emissions from
California Vehicles," Prepared by Automotive Testing Laboratories and Radian
Corporation for the California Air Resources Board and the South Coast Air
Quality Management District, February 1994.
11. Most, W. J., "Coordinating Research Council Study of Winter Exhaust Emissions
With Gasoline/Oxygenate Blends," SAE Paper No. 892091, 1989.
12. Gething, J.A., et al., "Are the Reductions in Vehicle Carbon Monoxide Exhaust
Emissions Proportional to the Fuel Oxygen Content?" SAE Paper No. 890216,
1989.
13. Hood, J. and R. Farina, "Emissions from Light Duty Vehicles Operating on
Oxygenated Fuels at Low Ambient Temperatures: A Review of Published
Studies," SAE Paper No. 952403, 1995.
14. Brzezinski, D.J., et al., "The Determination of Hot Running Emissions from FTP
Bag Emissions (Draft)," Report No. M6.STE.002, September 29, 1997.
15. "Derivation of Technology Specific Effects of the Use of Oxygenated Fuel Blends
on Motor Vehicle Exhaust Emissions," U.S. Environmental Protection Agency,
October 1988.
16. Rutherford, J.A., et. al., "Effects of RVP Reduction on Vehicle CO Emissions
During Las Vegas and Los Angeles Winter Conditions - Petroleum
-------
David H. Lax -34- February 13, 1998
Environmental Research Forum Project Number 95-06," SAE Paper No. 971726,
1997.
17. "CRC Sulfur/LEV Program," Presentation Graphics, Coordinating Research
Council, December 1997.
cc: James Uihlein, ARCO Products Company
------- |