Tier 3 Certification Fuel Impacts
Test Program
United States
Environmental Protection
tl Agency
-------
Tier 3 Certification Fuel Impacts
Test Program
This technical report does not necessarily represent final EPA decisions or
positions. It is intended to present technical analysis of issues using data
that are currently available. The purpose in the release of such reports is to
facilitate the exchange of technical information and to inform the public of
technical developments.
Assessment and Standards Division
Office of Transportation and Air Quality
U.S. Environmental Protection Agency
NOTICE
&EPA
United States
Environmental Protection
Agency
EPA-420-R-18-004
January 2018
-------
Table of Contents
1. Executive Summary 2
2. Introduction / Purpose 2
3. Study Design 3
3.1. Test Fuels 3
3.2. Test Vehicles 5
3.3. Emission Test Design 7
3.3.1. Test Cycles 7
3.3.2. Test Site and Emission Measurements 9
3.4. Test Procedures 10
3.5. Prospective Power Analysis 14
4. Results 17
4.1. CO2 Emissions 19
4.2. Fuel Economy 20
4.3. Paired Tests 21
4.4. Drive Quality Statistics 26
4.5. Fuel Order Effect 29
5. Additional Study of Acura and Fuel Octane 31
6. Discussion 32
Appendices
A. Supplemental Information on Test Fuels 34
B. Tests Used in Analysis 40
C. Tests Excluded from Analysis 44
D. Supplemental Emissions Data 47
1
-------
1. Executive Summary
The U.S. Environmental Protection Agency (EPA) has the responsibility to determine the test
procedures to be used when performing emission and fuel economy testing for vehicles and
engines. Part of this responsibility involves defining the properties of the test fuels that are
required for the testing performed by manufacturers and laboratories, and also providing the
proper analytical equations to calculate both emission and fuel economy for the test fuels. When
test fuel properties change, as they have recently for the Tier 3 vehicle emissions program, EPA
must make the proper test procedure adjustments to maintain the intended level of stringency for
existing or new emission and fuel economy (FE) standards. The test procedure adjustments
include changes to the method of calculating the emission and FE results that are subject to
applicable standards. To determine the appropriate test procedure adjustments from the changes
to the test fuel properties included in the Tier 3 program, the agency performed a study on eleven
vehicles operating over the two required test cycles using the two test fuels, the Tier 2 and the
new Tier 3 test fuel. The overall results across the test fleet showed a reduction in CO2 of 1.78%
for the FTP and 1.02% for the HFET tests for Tier 3 compared to Tier 2 test fuel. For fuel
economy the overall reduction was 2.29% for the FTP and 2.98% for the HFET tests for Tier 3
compared to Tier 2 test fuel. Throughout, the high levels of statistical significance observed, both
for CO2 and fuel economy, suggest that the measured differences in these parameters are actual
and in reasonable agreement with the difference projected during the planning of the study.
2. Introduction / Purpose
EPA adopted a new set of "Tier 3" fuel and motor vehicle emission standards in 2014 to
reduce air pollution.1 The Tier 3 emission standards include changes to several properties of
emission test fuel to make it more representative of in-use fuel, and some of these changes are
expected to affect emissions and fuel economy. Among the property changes as specified in
Section 3.1 below, the property changes of interest for greenhouse gas (GHG) emissions and fuel
economy included total aromatics, aromatics distribution and ethanol content. This test program
was initiated to compare Tier 2 certification fuel, the fuel on which the Phase 1 and Phase 2
GHG and Corporate Average Fuel Economy (CAFE) standards were established for light-duty
and heavy duty-vehicles, with the new Tier 3 certification fuel from the Tier 3 program. The
program results will be used as a basis for test procedure adjustments to ensure consistent
stringency of GHG and fuel economy standards as vehicle certification makes the transition to
Tier 3 test fuel.
1 For additional information on the light-duty Tier 3 program, see https://www.epa.gov/regulations~emissions-
vehicles-and-engi nes/final-rule-a me ndments-related-tier-3-motor-vehicle
2
-------
3. Study Design
To determine the impact of the new Tier 3 fuel on GHG emissions and fuel economy, we
designed a study that would test vehicles with the most recent technologies deployed in vehicles
presently available to consumers. These technologies have largely been implemented in gasoline
vehicles to reduce GHG emissions and improve fuel economy in response to more stringent
GHG and fuel economy standards. The study was designed to test these vehicles while operating
on the Tier 2 gasoline certification fuel, which was the fuel used to determine the stringency
levels when the 2017 to 2025 GHG and CAFE standards were finalized, and then test the same
exact vehicles while operating on the Tier 3 certification fuel finalized in the Tier 3 rule. The
GHG and fuel economy results of each vehicle in the test program on the two different test fuels
would then have a statistical analysis performed to determine the impact of the fuel change on
CO2 and fuel economy that is to be expected on these advanced technology vehicles when the
required GHG and CAFE test fuel becomes Tier 3 fuel in model year 2020 and later. Unless
otherwise specified, the term fuel economy in this analysis refers to the carbon balance result,
not the adjusted CAFE value.
3.1. Test Fuels
The two test fuels used to generate the emission and fuel economy results in this study were
Tier 2 EEE and Tier 3 regular grade. Both fuels were dispensed from underground tanks through
the conditioning and metering system at the EPA National Vehicle and Fuel Emission
Laboratory (NVFEL). The Tier 2 fuel was the same fuel used daily for ongoing certification and
in-use surveillance programs at NVFEL. The Tier 3 fuel was procured for use in validating test
methods and emission calculations in anticipation of its phase-in as the required test fuel. Table
3.1 summarizes the test fuel properties, with more detailed data available in Appendix A. Both
fuels met their respective regulatory specifications, also available in Appendix A.
Regulatory fuel economy calculations include measured values for three fuel properties:
specific gravity, carbon weight fraction, and net heat of combustion. ASTM publishes
reproducibility values for the test methods, which are determined by performing statistical
analysis on tests of the same sample made over a period of time by different operators in
different locations. This value represents the variability inherent in the test method and can be
used as a guide as to whether two different test results indicate an actual difference in the sample.
Specific gravity (SG) by ASTM D4052 has a small ASTM reproducibility value relative to the
other two fuel property methods. At least two SG test results were obtained for each fuel to
produce the average used in the analysis. For the other two properties, test results from at least
four laboratories were combined into an average value. After collecting all test results, relative
standard deviation (RSD), defined as the standard deviation divided by the mean, among the labs
was found to be 0.5% or less for each fuel property.
Carbon weight fraction (CWF) values used in this study were derived from percent mass
results by ASTMD5291. This method combusts a small sample of fuel and measures the
gaseous products. It was developed for diesel and other low-volatility fuels but can be run for
3
-------
gasoline by a skilled operator, and several labs produce acceptable results. This method also
reports hydrogen content, but not oxygen. The Tier 2 test fuel contains negligible oxygen so
interpreting D5291 results is straightforward and all labs reported C+H sums falling with the
range 99-101%. For oxygenated fuels, such as Tier 3 test fuel, D5291 results are typically
normalized to 100% using oxygen content determined by D5599 or D4815. Thus, since the Tier
3 CWF relied on oxygen data, we obtained three oxygen content measurements by D5599.
These were averaged, and that average was used to normalize each lab's Tier 3 CWF value, with
those results then being averaged to produce the CWF result used in the fuel economy analysis.
Net heat of combustion (or net heating value, NHV) for both test fuels was determined
according to ASTM D4809. This method combusts a small sample inside an oxygen-purged
calorimeter and reports gross heat of combustion (Cv) as the direct results, with an equation to
convert the result to NHV (Cp) using mass percent hydrogen in the sample. This step was carried
out individually for each lab using its D5291 result for hydrogen, after which the NHVs were
averaged to produce the final value used in the fuel economy analysis. Comparing the two test
fuels, we see Tier 3 fuel has 3.46% less energy on a mass basis, or 2.77% less on a volumetric
basis, than Tier 2 fuel. In terms of carbon intensity, Tier 3 fuel has 1.33% less carbon per unit
energy.
4
-------
Table 3.1: Test fuel properties.
ASIM
Psirsimeler Tier 2 Tier 3 I nils Method
Initial Boiling Point
89
100
°F
D86
T10
126
129
T50
223
210
T90
317
322
Final Boiling Point
406
387
Recovery
97.5
97.3
vol %
Residue
1.1
0.9
Loss
1.4
1.8
Specific Gravity, 60°F
0.7437a
0.7490 a
-
D4052
Density, 60 °F
0.7430 a
0.7482 a
g/cm3
DVPE (EPA equation)
9.0
8.8
psi
D5191
Ethanol
0.0
10.15
vol %
D5599
Oxygenates other than EtOH
0.0
0.0
Oxygen
0.0
3.7 a
mass %
Carbon
86.8 a
82.7 a
D5291
Hydrogen
13.2 a
13.6 a
Carbon
86.6
82.6
mass %
D3343
Hydrogen
13.4
13.7
Sulfur
39.6
8.3
mg/kg
D2622
Aromatics
30.6
22.9
vol %
D1319
Olefins
0.6
5.4
Saturates
68.8
71.7
Aromatics
32.3
23.8
D5769
Water Content
70
930
mg/kg
E1064
Research Octane Number
96.5
91.0
-
D2699
Motor Octane Number
88.7
83.5
-
D2700
AKI (R+M)/2
92.6
87.3
-
D2699/2700
Net Heat of Combustion
18,446
_b
Btu/lb
D3338
Net Heat of Combustion, 25 °C
18,529 a
17,889 a
Btu/lb
D4809
Net Heat of Combustion, 25 °C
114,870
111,689
Btu/gal
Calculated
Carbon Intensity
21,252
20,968
gC/MMBtu
Calculated
a Value is an average of measurements from multiple laboratories.
b Method is not valid for oxygenated fuels.
3.2. Vehicles
EPA selected a set of test vehicles that represent a variety of technologies likely to be used to
meet the GHG emission and fuel economy standards in the future. Eleven vehicles, described in
Table 3.2, are included in the analysis of this test program. Most were acquired by EPA's
National Center for Advanced Technology (NCAT) to help validate EPA's Advanced Light-duty
Powertrain and Hybrid Analysis (ALPHA) model, a full-vehicle simulation tool for predicting
5
-------
fuel economy and CO2 emissions. The ALPHA model was central to the 2017-2025 Light-Duty
CAFE and GHG Emission Standards rulemaking. Just over half are vehicles with Gasoline
Direct Injection (GDI) engines which enables higher compression ratio for improved fuel
efficiency and combustion control. This technology approach to improving FE and GHG
emissions has been adopted by several manufacturers and is prevalent in a large portion of the
fleet today. Some of the vehicles are equipped with downsized turbocharged engines. This
technology is most prevalent in light-duty cars and trucks produced by Ford and is also used in
the Volvo S60 T5, the Honda Civic as well as the Ford F150 in this program. The Mazda 3 uses
a naturally aspirated high compression engine with a high degree of valve timing authority in
order to operate as an Atkinson Cycle engine when required. The use of this technology is
starting to increase as it can be found in several models currently entering the market and in the
future plans of other vehicles. The Silverado 1500 pickup truck uses cylinder deactivation
technology which is also popular in several larger engine displacement models from various
manufacturers. When cruising, the six cylinders deactivate down to using only five, four and
even three cylinders to propel the vehicle down the road. This technology has been popular in
certain larger vehicles such as the GM Silverado pick-up which uses the GM Active Fuel
Management (formerly known as Displacement on Demand or DoD) to deactivate cylinders
when they are not needed resulting in reductions in emissions and fuel usage. The Ram has stop-
start technology and an 8-speed automatic transmission. Transmissions are moving to a higher
number of gears for greater efficiencies in the future, even going to 9 and 10 speeds. The stop-
start feature reduces time the engine spends idling such as at traffic lights and in traffic jams,
which can reduce fuel consumption and emissions. The Altima and the Civic in this program
both have Continuously Variable Transmissions (CVT), which varies the ratio while the engine
stays at the most efficient RPM window, allowing for greater fuel efficiency. A heavy-duty Class
2b truck, the Chevrolet Silverado 2500, was also tested in the program to determine whether
heavy-duty gasoline engines are likely to show an effect similar to light-duty vehicles.
6
-------
Table 3.2: Test vehicle information.
Model
Yesir
\ ehicle Msikc/Modd
K n«ine
Odometer
Miles
Technologies
2014
Ram 1500
3 .6L V6PFI
5,300
8 speed automatic
transmission, start-stop
disabled
2016
Acura ILX
2.4L 14 GDI
4,100
8 speed DCT with a torque
converter
2013
Nissan Altima
2.5L 14 PFI
8,700
CVT
2016
Honda Civic
1.5L 14 GDI
9,000
CVT, downsized
turbocharged engine
2015
Ford F150 Eco-Boost
2.7L V6 GDI
7,600
Downsized turbocharged
engine, start-stop disabled
2013
Chevrolet Malibu
2.4L 14 GDI
8,900
2016
Chevrolet Malibu
1.5L 14 GDI
5,400
Downsized turbocharged
engine
2014
Mazda 3
2.0L 14 GDI
16,300
High compression ratio
engine
2014
Chevrolet Silverado
1500
4.3L V6 GDI
8,700
Cylinder deactivation
2015
Volvo S60 T5
2.0L 14 GDI
8,000
Downsized turbocharged
engine
2016
Chevrolet Silverado
2500
6.0L V8 PFI
14,600
Class 2b truck
3.3. Emission Test Design
3.3.1. Test Cycles
The emission tests were conducted on a chassis dynamometer using the drive cycles required
for certification of the light-duty GHG emission and corporate average fuel economy standards.
The vehicles were tested on both fuels over the Federal Test Procedure (FTP) and Highway Fuel
Economy (HFET) certification test cycles. The speed versus time schedule for these cycles are
shown in Figures 3.3.1 and 3.3.2.
7
-------
70
60
-50
CL
E
x> 40
OJ
^30
at
> 20
EPA Federal Test Procedure
Duration = 1879 s; Distance = 11.04 mi; Avg Speed = 21.1 mph; Max Speed = 56.7 mph
Cold Stabilized (Phase 2)
Duration = s; Distance = 3 .86 mi
10
o J
Cold Transient (Phase 1)
Duration = SOS s; Dislarn e = mi
I
Hot Transient (Phase 3)
Duration = SOS s; Distance = 3.59 mi
A
250 500 750 1000 1250 1500
TestTime(s)
1750
Figure 3.3.1. EPA FTP test cycle speed vs. time profile.
EPA Highway Fuel Economy Test
Duration = 765 s; Distance - 10.26 mi; Avg Speed = 48.3 mph; Max Speed = 59.9 mph
60
o- 40
0 100 200 300 400 500 600 700
Test Time (s)
Figure 3.3.2. EPA FIFET test cycle speed vs. time profile.
8
-------
3.3.2. Test Site and Emission Measurements
The test site used for this study is compliant with 40 CFR part 1066 requirements for
regulated gaseous and particulate measurements. Table 3.3 gives information on the test site
equipment. The focus of the study was primarily on fuel economy and carbon dioxide (CO2), a
GHG which is also the most important emission input for the fuel economy calculation.
Methane (CH4) was measured for all tests, but due to an analyzer malfunction some tests on the
Civic and Silverado 1500 are missing measured values. The median emission value from similar
tests on the same vehicle was substituted for the missing values when fuel economy was
calculated. These values are shown in Appendix D. Overall this is a tiny adjustment, as methane
emission rates fall below 0.005% of CO2 across the dataset. Measurement of N2O was also
initiated but ongoing equipment problems resulted in unreliable results for most tests, therefore
an analysis of the N2O emissions impact from the fuel could not be performed. Other pollutants
including total hydrocarbons (THC), methane (CH4), nonmethane organic gases (NMOG),
carbon monoxide (CO), oxides of nitrogen (NOx), and particulate matter (PM as PM2.5) were
also measured or calculated.
Fuel economy was calculated as shown in Equation 3.1, which is the carbon-balance form of
the equation being proposed by EPA for use with Tier 3 test fuel.2 While this equation is
specified for low-level ethanol blends, it was applied to both test fuels for the results reported
here in order to provide as precise a comparison as possible.
FE =
CWFfuel -SpecificGravityfuel -3781.7
CWFexh • NMOG +0.749-CH4 + 0.429- CO + 0.273 C02
Eq. 3.1
NMOG values were determined using the calculation method described 40 CFR 1066.635(c)
for each emission phase (bag), then the FTP composite value was determined using the typical
emission weighting factors.3 The carbon weight fraction (CWF) of exhaust was assumed to be
the same as that of the fuel, consistent with current certification practices.
2 The carbon-balance form does not attempt to adjust results back to a baseline fuel using NHV. This Tier 3 version
also uses NMOG+CH4 in place of THC emissions to better account for combustion products of oxygenated fuels.
3 The weighting factors are 0.43 for the cold transient phase (bag 1), 1.0 for the cold stabilized phase (bag 2), and
0.57 for the hot transient phase (bag 3).
9
-------
Table 3.3: Test site equipment.
Test site
40 CFR part 1066 compliant.
Dynamometer
MAHA medium duty 48" roll 4WD dynamometer capable of
testing up to 20,000 lbs GVWR single and dual drive axle
vehicles
CVS
8" diameter dilution tunnel with tailpipe pressure control,
heated dilution air capability, and HEPA filtration; Horiba
CVS-7200T flow control utilizing four critical flow Venturis
capable of 15 flow rates up to 1200 CFM
Test cell host
Horiba CDTCS 5000
Gaseous
analytical
equipment
Horiba MEXA-ONE CI and D1 dilute and raw benches that
measure CO, C02, THC, CH4, NOx, and N20 (dilute only)
PM equipment
Horiba MEXA-ONE PM consisting of three sample trains to
allow measurement of 3 samples simultaneously
All analyzer checks were performed according to 40 CFR part 1066 specifications. There
were several calibration and maintenance activities conducted in the test site including, but not
limited to the following:
• Daily: bag leak checks4, air handling system tests, and zero spans.
• Weekly: repeatable car checks5, coastdowns for MAHA 2WD and 4WD dynamometer6,
Dynamometer Parasitic Losses Verification7, Gravimetric Propane Injection for THC8,
Vehicle Sampling Analysis Correlations for bag checks on CO, CO2, CH4, NOx.9
• Every 35 days: CH4 Gas Chromatography column efficiency check, NOx converter
check, chemiluminescent detector CO2 + H2O Quench Check, MEXA-ONE analyzer
linearizations per Horiba instructions and linearity checks per 40 CFR part 1066.
• Typically, annually: FID O2 inference check, FID response factor check, and NDIR
interference checks.
3.4. Test Procedures
The test program was designed to develop datasets that could distinguish small differences in
CO2 emissions and fuel economy between the two fuels. Special efforts were taken to reduce
4 EPA National Vehicle Fuel and Emissions Laboratory Work Instructions WI-1029
5 EPA National Vehicle Fuel and Emissions Laboratory Work Instructions WI-1262
6 EPA National Vehicle Fuel and Emissions Laboratory Work Instructions WI-1109
7 EPA National Vehicle Fuel and Emissions Laboratory Work Instructions WI-1078
8 EPA National Vehicle Fuel and Emissions Laboratory Work Instructions WI-1269
9 EPA National Vehicle Fuel and Emissions Laboratory Work Instructions WI-1199
10
-------
test-to-test variability with a minimum number of tests. The procedures were designed with the
goal of completing testing of one vehicle on one fuel in a single work week. However, there
were instances when additional testing was required so that testing extended to an additional
week. We attempted to utilize the same test site and driver throughout the program across all
fuels and vehicles to reduce site-to-site and driver variability but a subset of the tests was
conducted with a different driver. In addition, the fuel order was reversed for the second half of
the vehicles to avoid any potential fuel-order bias. As shown in Tables 3.5.4 and 3.5.5,
approximately half of the vehicles were tested first with the Tier 2 fuel and then with the Tier 3
fuel.
The fuel preparation procedures, as shown in Table 3.4.1, were outlined prior to testing to
ensure that the fuel was completely flushed and the vehicle control systems appropriately
adapted to the new fuel. For each fuel change, a triple drain and fill procedure was used to
completely flush out the previous fuel. Prior to testing, each vehicle ran three LA4 cycles,
defined as the first two phases of the FTP, to allow for the engine to adapt to the new fuel
properties, including ethanol.
The emissions testing followed the procedures laid out in Table 3.4.2. Two changes were
implemented during the program to further reduce the testing variability. A plug-in trickle
charger was used to keep the battery on the vehicle completely charged during the soak times
over the weekend and each overnight soak. In addition, we made an adjustment to run the same
test sequence and number of tests each day. At the beginning of the program we maximized the
number of tests run in a day, but this led to a different number of tests conducted on a daily basis.
We observed a higher variability with this approach. Therefore, we then required the same
sequence on each day and the test data became more repeatable.
In addition to the procedures in Table 3.4.2, weekly background and dilution air checks were
also run throughout the program to maintain integrity of the data and to understand the variables
in case of fluctuating measurements. These were within normal range and did not appear to
affect the data. The test cell conditions such as temperature, pressure and relative humidity were
monitored and remained stable throughout the program per 40 CFR parts 1065 and 1066
requirements.
The complete valid dataset can be found in Appendix B in chronological order for each
vehicle. We did experience some additional issues which resulted in excluded tests, which are
shown in Appendix C. The issues included the following:
• PM filer holder issue during HFET testing: The first two PM filter holders were
being left open while sample collection occurred on the fourth filter. Since the system
was pulling airflow through all filters this resulted in an unknown error on all gaseous
emissions. Therefore, it was necessary to repeat all HFET tests conducted in the first
portion of the program until the situation was fixed.
• Modal bench malfunction: A modal bench malfunction caused an error in the
calculated results for the gaseous bag emissions which resulted in invalid tests for the
Ram and the first Malibu (Malibu 1).
11
-------
• Inconsistent Acura results: The initial HFET data for the Acura showed no clear
effect of the test fuel, which was inconsistent with the FTP results. We added a
controlled experiment with additional replicates comparing the fuels and the result
suggested an opposite fuel effect. This latter experiment used a different driver, and
we did not want to mix data from different drivers so we used only the newer data in
the overall analysis. Later tests with Tier 3 premium fuel indicated an effect of octane
on CO2 emissions, which could explain the opposite fuel effect observed with Tier 3
regular fuel. However, the Tier 3 premium data was not included in the overall
analysis or this report. This is further discussed in Section 5.
• Inconsistent Altima results: The Altima initial HFET tests were repeated at a later
time because of the filter holder issue. In this later experiment, the initial fuel effect
was inconsistent with the FTP results so a decision was made to conduct additional
confirmatory tests. The first data of testing with the first fuel showed a large spread
with descending CO2 values. With further investigation we realized the trickle
charger had not been on the vehicle during several weeks of vehicle non-use prior to
these additional tests. After ensuring the vehicle battery was charged, another test set
was performed. This final set was retained in the dataset.
• Inconsistent Malibu 1 results: The initial HFET tests for the first Malibu were also
repeated at a later time because of the filter holder issue. Again, we saw an
inconsistent trend as compared to the FTP data. These tests also had some
repeatability issues. This vehicle had been used by another testing group and had been
returned to us with a fault code. Necessary actions were taken to resolve any issues
before an additional test set was performed. This final data set was retained in the
dataset.
• Inconsistent Silverado results: The initial Silverado HFET data showed no clear
effect of test fuel, which was inconsistent with the FTP results. An additional set of
data with both fuels was collected and confirmed the original results from the first
dataset. Since a different driver was used and we wished to keep to our resolve of not
mixing drivers, only the newer data was used in the analysis.
12
-------
Table 3.4.1: Fuel change procedure.
Stop
Description
With the ignition key in OFF position, drain vehicle fuel completely via
installed fuel drain or the fuel rail.
1
Note: Contact oroi ect engineer for additional instructions if the vehicle has
a saddle tank and does not have the fuel drain installed.
2
Turn vehicle ignition to RUN position for 30 seconds to allow the fuel
level reading to stabilize. Confirm the return of fuel gauge reading to
zero.
3
Turn ignition off. Fill fuel tank to 50% with next test fuel in sequence.
Fill-up fuel temperature must be less than 60°F.
4
Repeat Steps 1-3. (If repeated steps 1-3, move to Step 5)
5
Repeat Steps 1-3, but fill the fuel tank to 100%.
6
Fully warm up the vehicle and run three sampled LA4 prep cycles
(gaseous emissions only), with key-off after every test.
7
Run vehicle coastdowns following the 3 LA4 prep cycles.
8
Allow the vehicle to idle in neutral for two minutes, then shut the
engine down in preparation for the soak. Report results of those tests to
project engineer.
9
Move vehicle to soak area without starting the engine (or can leave on
dyno for soak).
Park vehicle in soak area at proper temperature (75 °F) for no more
than 100 hours.
10
Note: During the soak period, maintain the nominal charge of the vehicle's
battery using an appropriate charging device.
Note: If 100-hour soak time is exceeded before Step 3 of the Vehicle Test
Protocol is executed, drain the fuel tank and fill it to 100% with the same
fuel the following Friday, then repeat Steps 7 and 8.
13
-------
Table 3.4.2: Vehicle test procedure to be performed during one work week.
Stop
Description
1
Move the vehicle to an appropriate area and without starting the engine
take an 8 oz. sample of fuel from the fuel rail. Submit the sample to the
Chemistry Lab for the measurement of density at 60°F by ASTM D4052
and ethanol content by ASTM D5599. Request that results be available
within 24 hours, if possible.
Note: Make sure the label on the fuel container includes vehicle designation,
fuel FTAG number and date sample was taken.
2
Move vehicle to test area without starting engine.
3
Perform a sampled FTP test followed by a sampled HFET test and a
sampled US06 test.
4
Move vehicle to soak area without starting the engine (or can leave on
dyno for soak).
5
Park vehicle in soak area at proper temperature for 12-52 hours.
Note: During the soak period, maintain the nominal charge of the vehicle's
battery using an appropriate charging device.
Note: If the 52-hour soak time limit is exceeded or if there is no chance to
complete Steps 6 and 7 in the course of the same week, drain the fuel tank
and fill it to 100% with the same fuel the following Friday, then repeat Steps
7 and 8 of the Fuel Change and Vehicle Preparation Procedure.
6
Move vehicle to test area without starting the engine.
7
Repeat Steps 3-5 until the required number of replicates is met.
3.5 Prospective Power Analyses
In the implementation of the project an important consideration concerned the number of
replicate tests required for each vehicle for each cycle. Rather than performing a pre-determined
number of replicates, decisions were made dynamically, based on estimates of statistical power
to detect a difference in CO2 emissions between the two fuels on each vehicle. The precision
goal for FTP cycle data from each vehicle was to detect an effect of 1.5%, with 80% power at a
95% confidence level. However, in some cases the results achieved differed from the target due
to logistical issues involving test-site and vehicle availability. The estimated effect size of 1.5%
14
-------
was based on the CO2 model from the EPAct/V2/E-89 study and typical certification fuel
properties.10
For each vehicle, we performed three initial replicates on the first test fuel selected. We then
calculated power to detect the specified effect size, based on the observed variance and the mean
of the first three replicates, assuming that the variance of three replicates on the second test fuel
would equal that of the first fuel. If the power was below the target, an additional replicate test
was performed, and this process repeated until the target was met. At this point the fuel was
changed and the process was repeated on the second fuel until the target power was achieved
using the actual variance values.
For the majority of vehicles, the numbers of HFET replicates matched those on the FTP
cycle, as these tests were conducted in pairs. A subset of vehicles required additional HFET
replicates to confirm or replace tests where procedural or equipment problems occurred.
The calculation was performed as for a one-tailed two-sample test, under an assumption of
equal variances. We followed a one-tailed assumption because we thought it reasonable to expect
a reduction in CO2 emissions on the Tier 3 fuel, due primarily to its lower aromatics level.
To estimate power, we assigned the acceptance region under the null hypothesis, as defined
by its upper and lower confidence limits, calculated as shown in Eq. 3.5.1
acceptance region = 0 + 2\i,licence Eq. 3.5.1
where the standard error of the difference is defined as
1 1
^difference ~~ Spooled J + Eq. 3.5.2
In estimating the standard error, the pooled variance for the difference in means is defined as
(nT2 - l).vf2 + (nT 3 - 1 ).s
_ _ 2
2 _ \'lT 2 lPt 2 1PT3 r ~ c ~
pooled Eq. 3.5.3
3 _
where:
nj2 = number of replicates on the Tier 2 fuel,
«T3 = number of replicates on the Tier 3 fuel,
.vt2 = standard deviation of replicate measurements on the Tier 2 fuel, and
,s't3 = standard deviation of replicate measurements on the Tier 3 fuel.
The test statistics for the acceptance region, representing lower and upper confidence limits
under the null hypothesis, were then calculated with reference to the assumed difference under
the alternative hypothesis, as shown in Eq. 3.5.4,
10 Memo to Tier 3 final rulemaking docket by James Warila, February 28. 2013. Docket entry number EPA-HQ-
OAR-2011-0135-0605.
15
-------
, CL,o„,r-(-0 015r„,CLupl„-(-0.015x,
lower ? upper ^4*
^difference ^difference
with the standard error of the difference defined as above.
The probability of accepting the null hypothesis when it is false (Type-II error) was then
estimated as the probability that the actual test result would fall in the acceptance region. This
probability is defined as shown in Eq. 3.5.5.
/? = PrRc,uai> Viewer and t3CtU3l< /uppJ Eq. 3.5.5
In cases where both ft0Wer and fupPer were of the same sign, either positive or negative, fi was
calculated as the difference in probabilities for the lower and upper ^-statistics.
If Slower and ^pper were both negative, fi was estimated as
p = abs( Pr {/< /uppcr} - Pr {/ < /lnwcr}) Eq. 3.5.6
However, if t\0Wer and /upper were both positive, fi was estimated as in Eq. 3.5.7.
P = abs( Vr{t> /iower }-Pr{/>/upper}) Eq. 3.5.7
In both cases, power for the test was calculated simply as 1-/?.
However, if fewer and /upper were of opposite sign, power was calculated directly as the sum of the
probabilities for the lower and upper ^-statistics.
Specifically, if fewer < 0 and /upper > 0, power was estimated as
\-P = power = Pr{//upper} Eq. 3.5.8
But if Aower > 0 and /upper < 0, power was estimated as
1 - fi = power = Pr{//lower} Eq. 3.5.9
In all cases, the degrees of freedom for test statistics were given as «T2+«T3-2.
Power results for tests on the FTP and HFET test cycles are shown in Tables 3.5.4 and 3.5.5,
respectively. For the FTP cycle, calculated power exceeded the target level for all vehicles. For
the HFET cycle, power was below the target for two vehicles, but exceeded the target for the
remaining nine vehicles. For most vehicles, three replicates on each fuel proved adequate to
achieve the target power level. However, in some cases, up to six replicates were performed on
one of the fuels.
16
-------
Table 3.5.4: FTP Cycle: Prospective power analyses for CO2 emissions, by vehicle.
Vehicle
first
fuel
Acura
T2
Altima
T2
Malibu 1
T2
Mazda
T2
Ram
T2
Volvo
T2
Civic
T3
F150
T3
Malibu 2
T3
Silverado
T3
Silverado (2b)
T3
Mean1
Difference
Replicates
Standard
Deviations
(g/mi)
(g/mi)
nT 2
nT 3
T2
T3
275.66
-4.135
4
3
1.7579
1.4539
276.19
-4.143
3
5
0.9739
1.0967
314.53
-4.718
3
3
0.5403
0.7474
242.12
-3.632
3
3
1.2054
0.4800
423.94
-6.359
6
3
1.8872
3.7524
305.98
-4.590
6
3
1.7998
0.4699
213.37
3.201
3
3
0.8475
0.5022
376.87
5.653
3
3
1.4747
1.5390
268.64
4.030
3
5
0.4382
0.8567
419.88
6.298
3
6
1.5009
2.5931
706.83
10.60
3
3
0.5384
0.8865
Margin-
f-values:
p -values:
of-error3
acceptance region
acceptance region
(g/mi)
lower
upper
lower
upper
2.5287
1.280
5.310
0.128
0.002
1.5004
3.422
7.308
0.007
0.0002
1.1351
6.729
10.993
0.001
0.0002
1.5969
2.717
6.980
0.027
0.001
3.4330
1.615
5.404
0.075
0.001
2.0653
2.316
6.105
0.027
0.0002
1.2124
-7.759
-3.496
0.001
0.012
2.6235
-6.725
-2.462
0.001
0.035
1.0555
-9.362
-5.475
0.00004
0.001
3.1265
-5.711
-1.922
0.0004
0.048
1.2766
-19.84
-15.57
0.00002
0.00005
P
Power
0.12679
0.8732
0.00689
0.9931
0.00108
0.9989
0.02548
0.9745
0.07469
0.9253
0.02662
0.9734
0.01175
0.9882
0.03351
0.9665
0.00073
0.9993
0.04766
0.9523
0.00003
1.0000
1 Mean FTP cycle aggregate result on the first fuel tested.
2 Expected difference under the alternative hypothesis, calculated as 1.5% of the mean on the first fuel.
3 Calculated as the product of the critical f-statistic, for 95% confidence and n T2+w t3_2 degrees of freedom, and the standard error for the difference in means.
Table 3.5.5: HFET Cycle: Prospective power analyses for CO2 emissions, by vehicle.
Vehicle
first
fuel
Acura
T2
Altima
T2
Malibu 1
T2
Mazda
T2
Ram
T2
Volvo
T2
Civic
T3
F150
T3
Malibu 2
T3
Silverado
T3
Silverado (2b)
T3
Mean1
Difference
Replicates
Standard
(ffj2
Deviations
(g/mi)
(g/mi)
n T2
« 13
T2
T3
171.31
-2.570
3
3
0.3619
0.3842
165.49
-2.482
6
3
0.4796
1.1564
189.15
-2.837
3
4
2.3365
1.0493
161.87
-2.428
3
3
0.5045
0.5376
262.76
-3.941
3
3
0.5913
1.2250
175.61
-2.634
3
3
0.8586
1.6970
144.75
2.147
3
3
0.3345
0.6830
244.79
3.629
3
3
1.5878
0.8544
166.02
2.454
3
5
0.5799
0.8100
281.37
4.216
3
3
1.4616
0.6754
447.66
6.647
3
3
1.7182
0.9522
Margin-
f-values:
p -values:
of-error3
acceptance region
acceptance region
(g/mi)
lower
upper
lower
upper
0.6496
6.301
10.565
0.002
0.0002
0.9903
2.855
6.644
0.012
0.0001
2.5956
0.188
4.218
0.429
0.004
0.9074
3.573
7.836
0.012
0.001
1.6743
2.887
7.150
0.022
0.001
2.3408
0.267
4.531
0.401
0.005
0.9361
-7.022
-2.758
0.001
0.025
2.2193
-5.618
-1.354
0.002
0.124
1.0519
-6.476
-2.589
0.0003
0.021
1.9817
-6.667
-2.403
0.001
0.037
2.4178
-7.992
-3.729
0.001
0.010
P
Power
0.00139
0.99861
0.01211
0.98789
0.42509
0.57491
0.01094
0.98906
0.02134
0.97866
0.39601
0.60399
0.02438
0.97562
0.12112
0.87888
0.02030
0.97970
0.03573
0.96427
0.00950
0.99050
1 Mean HWFET result on the first fuel tested.
2 Expected difference under the alternative hypothesis, calculated as 1.5% of the mean on the first fuel
3 Calculated as the product of the critical /"-statistic, for 95% confidence and «T2+«T3-2 degrees of freedom, and the standard error of the difference in means.
4. Results
This section describes and presents statistical analyses of the results for CO2 and carbon-
balance fuel economy. In addition to CO2, the calculation of fuel economy involves the use of
test results for several other species, including CO, THC, NMOG, and CH4. Results for these
species are not summarized in this section. However, we have included test results for each in
Appendix D. For completeness, we have also included results for NOx and PM.
As a counterpart to the prospective power calculations performed during testing to determine
sample sizes for each vehicle, we performed tests of significance for each vehicle after the
completion of testing.
As with the power calculations, the retrospective tests were performed as two-sample /-tests
for differences in means. However, it is important to note that the retrospective analysis of
results incorporated several important differences from the prospective power analyses:
17
-------
• The retrospective tests were performed as two-tailed, rather than one-tailed tests. The
prospective one-tailed assumption was revised due to the realization that increases in CO2
emissions on the Tier 3 fuel were possible for some vehicles, although unlikely.
• The retrospective tests were performed assuming unequal, rather than equal variances.
The unequal-variances assumption was adopted following observation of two- to three-
fold differences in standard deviations for replicates on the two fuels for several vehicles.
Consistent with this assumption, the degrees of freedom for each test were estimated
using a Satterthwaite approximation, as described below.
• The difference in means is consistently calculated as the mean on Tier 3 fuel minus that
on Tier 2 fuel ( x7 3 -xT2), without respect to testing order.
• Tests were performed for carbon-balance fuel economy, as well as for CO2 emissions.
Accordingly, the tests were formulated as follows:
Null hypothesis: Ho: xT3-xT2=0
Alternative hypothesis: Ha: xn-xT2^0
Accordingly, the test statistic was calculated as shown in Eq. 4.1,
nj2 = number of replicates on the Tier 2 fuel,
«T3 = number of replicates on the Tier 3 fuel,
.s'T2 = standard deviation of replicate measurements on the Tier 2 fuel, and
.s'T3 = standard deviation of replicate measurements on the Tier 3 fuel.
Note that, consistent with the two-tailed assumption, the critical ^-statistic for each test was
the value corresponding to the 97.5% confidence level. The corresponding degrees of freedom
for each test is based on the variances of the two means and was calculated using the
Satterthwaite approximation, shown in Eq. 4.2.
_ (XT3 ~xr2)~0
Eq. 4.1
where:
Eq. 4.2
nT2 J \nT3
flj2 ^ ^ 1
18
-------
4.1 CO2 Emissions
The CO2 emission results of the two-sample individual vehicle tests for the FTP and HFET
cycles shown in Tables 4.1.1 and 4.1.2, respectively.
For the FTP cycle (Table 4.1.1), all differences are negative, and are significant for 10 of the
11 vehicles at the 95% confidence level. The difference for the remaining vehicle, the Acura, is
marginally significant. Absolute differences are generally proportional to emission levels, and
range from -15 to -3 grams/mile, with an average of -6.37 grams/mile. Percent differences range
from -2.3% to -1.0%, relative to the Tier 2 level, with an average of -1.8%.
For the HFET cycle (Table 4.1.2), differences are negative for 10 of the 11 vehicles, except
for the Acura, which shows a positive difference. Overall, the degree of statistical significance is
not as pronounced as for the FTP. Of the 11 vehicles, six show significant differences at the 95%
confidence level, including the Acura, which shows a significant positive difference. Of the
remaining five vehicles, three show marginal significance (0.05 0.10). Of the 11 vehicles, the Silverado shows the smallest and
least significant difference (-0.32 g/mi, -0.1 %,p=0.75). Overall, absolute differences range from
-5.14 to 1.27 g/mi, averaging -2.16 g/mi, with percent differences ranging from -2.12% to 0.74%,
averaging -1.02%.
Table 4.1.1: FTP Cycle: Two-sample t-tests for differences in CO2 emissions, by vehicle.
Vehicle
Means (g/mi)
Difference1
Replicates
Standard Deviations
d.£
standard
error
^actual
p -value
T2
T3
(g/mi)
(%)
«T2
»T3
T2
T3
Acura
275.66
272.74
-2.92
-1.06
4
3
1.7579
1.4539
4.88
1.2154
-2.406
0.06240
Altima
276.19
270.60
-5.59
-2.02
3
5
0.9739
1.0967
4.81
0.7461
-7.486
0.00080
Malibu 1
314.53
307.37
-7.16
-2.28
3
3
0.5403
0.7474
3.64
0.5324
-13.450
0.00031
Mazda
242.12
238.57
-3.55
-1.47
3
3
1.2054
0.4800
2.62
0.7491
-4.742
0.02401
Ram
423.94
414.49
-9.46
-2.23
6
3
1.8872
3.7524
2.52
2.2993
-4.114
0.03605
Volvo
305.98
299.83
-6.15
-2.01
6
3
1.7998
0.4699
6.17
0.7832
-7.849
0.00020
Civic
216.98
213.37
-3.61
-1.66
3
3
0.8475
0.5022
3.25
0.5687
-6.340
0.00621
F150
380.61
376.87
-3.74
-0.98
3
3
1.4747
1.5390
3.99
1.2306
-3.041
0.03845
Malibu 2
274.00
268.64
-5.36
-1.96
3
5
0.4382
0.8567
5.98
0.4591
-11.676
0.00002
Silverado
427.69
419.88
-7.81
-1.83
3
6
1.5009
2.5931
6.57
1.3681
-5.707
0.00091
Silverado (2b)
721.57
706.83
-14.7
-2.04
3
3
0.5384
0.8865
3.30
0.5988
-24.616
0.00007
[Means 1 | 350.84 | 344.47 | -6.37 | -1.78
1 Calculated as T3 - T2. and as % relative to the T2 fuel.
2 Degrees of freedom for the difference in means, based on the Satterthwaite approximation.
3 Two-tailed value at the 95% confidence level.
19
-------
Table 4.1.2: HFET Cycle: Two-Sample ^-tests for differences in CO2 emissions, by vehicle.
Vehicle
Means (g/mi)
Difference1
Replicates
Standard Deviations
d.f.
standard
error
^actual
p -value
T2
T3
(g/mi)
(%)
«T2
n i3
T2
T3
Acura
171.31
172.58
1.27
0.74
3
3
0.3619
0.3842
3.99
0.3047
4.173
0.014
Altima
165.49
163.37
-2.13
-1.29
6
3
0.4796
1.1564
2.35
0.6958
-3.060
0.075
Malibu 1
189.15
184.01
-5.14
-2.72
3
4
2.3365
1.0493
2.61
1.4474
-3.554
0.047
Mazda
161.87
160.32
-1.54
-0.95
3
3
0.5045
0.5376
3.98
0.4256
-3.625
0.022
Ram
262.76
260.67
-2.09
-0.79
3
3
0.5913
1.2250
2.88
0.7854
-2.658
0.080
Volvo
175.61
173.22
-2.39
-1.36
3
3
0.8586
1.6970
2.96
1.0980
-2.179
0.119
Civic
144.75
143.16
-1.59
-1.10
3
3
0.3345
0.6830
2.91
0.4391
-3.627
0.038
F150
244.79
241.92
-2.87
-1.17
3
3
1.5878
0.8544
3.07
1.0410
-2.758
0.069
Malibu 2
166.02
163.58
-2.44
-1.47
3
5
0.5799
0.8100
5.59
0.4933
-4.953
0.003
Silverado
281.37
281.05
-0.32
-0.11
3
3
1.4616
0.6754
2.82
0.9296
-0.344
0.755
Silverado (2b)
447.66
443.11
-4.54
-1.02
3
3
1.7182
0.9522
3.12
1.1341
-4.007
0.026
Means
219.16
217.00
-2.16
-1.02
Calculated as T3 - T2. and as % relative to the T2 fuel.
Degrees of freedom for the difference in means, based on the Satterthwaite aoDroximation.
Two-tailed value at the 95% confidence level.
4.2 Fuel Economy Results
The fuel economy results of the individual two-sample tests for fuel economy are shown in
Tables 4.2.1 and 4.2.2.
For the FTP cycle (Table 4.2.1), all vehicles showed reductions in fuel economy on the Tier
3 fuel, as they did with CO2. Absolute differences range from -1.01 to -0.24 mpg, with an
average of -0.66 mpg. In relative terms, these values correspond to differences of -3.1 % to -
1.73%, with an average of -2.29%. All reductions are significant at the 95% confidence level for
all vehicles except the Ram, which is marginally significant (p=0.054). In contrast to its
behavior for CO2, the Acura has the second largest reduction in fuel economy both in absolute
and percentage terms.
For the HFET cycle (Table 4.2.2), results are similar. All vehicles show fuel economy
reductions on Tier 3 fuel, ranging from -2.49 to -0.36 mpg, and averaging -1.34 mpg. Percentage
differences are larger than those for the FTP on the whole, ranging from -4.8% to -0.76%, and
averaging -3.0%. The Acura is distinguished in having the largest reduction, both in absolute
and relative terms. For this cycle, all reductions are significant at the 95% level, except for the
Malibu 1, which has the minimum absolute and relative differences (-0.36 mpg, -0.76%,/>=0.47).
This Malibu is also conspicuous for the size of its standard deviations, particularly on the Tier 2
fuel.
20
-------
Table 4.2.1: FTP Cycle: Two-sample ^-tests for differences in fuel economy, by vehicle.
Vehicle
Means (mpg)
Difference1
Replicates
Standard Deviations
d.£
standard
error
^actual
p -value
T2
T3
(mpg)
(%)
nT 2
KT3
T2
T3
Acura
32.43
31.45
-0.98
-3.02
4
3
0.2051
0.1666
4.91
0.1406
-6.972
0.00101
Altima
32.29
31.63
-0.66
-2.03
3
5
0.1017
0.1313
5.33
0.0830
-7.913
0.00038
Malibu 1
28.32
27.83
-0.49
-1.73
3
3
0.0575
0.0643
3.95
0.0498
-9.862
0.00063
Mazda
36.90
35.93
-0.97
-2.63
3
3
0.1950
0.0604
2.38
0.1179
-8.235
0.00839
Ram
21.06
20.68
-0.39
-1.83
6
3
0.0947
0.1846
2.54
0.1134
-3.399
0.05437
Volvo
29.15
28.54
-0.62
-2.12
6
3
0.1590
0.0513
6.59
0.0713
-8.665
0.00008
Civic
41.18
40.18
-1.01
-2.44
3
3
0.1635
0.0895
3.10
0.1076
-9.345
0.00227
F150
23.47
22.74
-0.73
-3.10
3
3
0.0899
0.0945
3.99
0.0753
-9.668
0.00065
Malibu 2
32.58
31.89
-0.68
-2.10
3
5
0.0434
0.1080
5.63
0.0544
-12.585
0.00002
Silverado
20.85
20.38
-0.47
-2.25
3
6
0.0740
0.1181
6.27
0.0644
-7.297
0.00028
Silverado (2b)
12.34
12.11
-0.24
-1.91
3
3
0.0109
0.0121
3.96
0.0094
-25.092
0.00002
[Means 1 | 28.23 | 27.58 | -0.66 | -2.29
1 Calculated as T3 - T2. and as % relative to the T2 fuel.
2 Degrees of freedom for the difference in means, based on the Satterthwaite approximation.
3 Two-tailed value at the 95% confidence level.
Table 4.2.2: HFET Cycle: Two-sample ^-tests for differences in fuel economy, by vehicle.
Vehicle
Means (mpg)
Difference1
Replicates
Standard Deviations
d.f.
standard
error
^actual
p -value
T2
T3
(mpg)
(%)
n T2
nT3
T2
T3
Acura
52.20
49.71
-2.49
-4.78
3
3
0.1109
0.1098
4.00
0.0901
-27.679
0.00001
Altima
53.88
52.42
-1.46
-2.71
6
3
0.1574
0.3655
2.38
0.2206
-6.609
0.014
Malibu 1
46.97
46.61
-0.36
-0.76
3
4
0.7011
0.2645
2.43
0.4259
-0.843
0.474
Mazda
55.22
53.49
-1.73
-3.13
3
3
0.1826
0.1743
3.99
0.1458
-11.847
0.0003
Ram
34.01
32.90
-1.11
-3.26
3
3
0.0726
0.1557
2.83
0.0992
-11.193
0.0020
Volvo
50.82
49.42
-1.41
-2.77
3
3
0.2423
0.5049
2.87
0.3233
-4.346
0.025
Civic
61.70
59.86
-1.84
-2.98
3
3
0.1493
0.2783
3.06
0.1823
-10.089
0.0019
F150
36.51
35.44
-1.07
-2.94
3
3
0.2387
0.1258
3.03
0.1558
-6.882
0.0061
Malibu 2
53.80
52.41
-1.39
-2.59
3
5
0.1982
0.2608
5.40
0.1634
-8.522
0.0002
Silverado
31.75
30.50
-1.25
-3.94
3
3
0.1659
0.0767
2.82
0.1055
-11.841
0.0017
Silverado (2b)
19.95
19.36
-0.59
-2.97
3
3
0.0757
0.0413
3.09
0.0498
-11.903
0.0011
Means
45.17
43.83
-1.34
-2.98
1 Calculated as T3 - T2, and as % relative to the T2 fuel.
2 Degrees of freedom for the difference in means, based on the Satterthwaite approximation.
3 Two-tailed value at the 95% confidence level.
4.3 Paired Tests
The absolute differences in CO2 and fuel economy, by vehicle, are summarized graphically
below. Figure 4.3.1 shows absolute changes in CO2 emissions, by cycle. As described
previously, the chart makes it clear that CO2 reductions were larger and more significant on the
FTP than on the HFET.
21
-------
The reverse is true for the differences in fuel economy, shown in Figure 4.3.2. Reductions on
the HFET are definitely larger for most vehicles although the degree of significance is not
markedly higher.
n
D
Acura
Altima
Malibu 1
Mazda
Ram
Vo Ivo
Civic
F150
Malibu 2
Silverado
Silverado (2b)
on -
i _ .x
T
1 11
cud c n
if
\m\
L ll l I
CD
U
c
I1 ' II I
CD
it
Q
-1 n
¦ FTP
-20.0 —
~ HWFET
Figure 4.3.1: CO2 Emissions: Absolute differences (g/mi), on the FTP and HFET cycles, with
95% confidence intervals (calculated as the result on T3 fuel minus that on T2 fuel).
-Q
r\i
~ HWFET
-2.5
-3.0
Figure 4.3.2: Fuel Economy: Absolute differences (mpg), on the FTP and HFET cycles
(calculated as the result on T3 fuel minus that on T2 fuel).
22
-------
The culminating step in analysis is to assess the differences in CO2 and fuel economy for the
vehicle sample as a whole. The step was achieved through the application of paired t tests to
absolute emissions and fuel economy. The paired test is based on the mean and variance of the
difference in means, with the difference for each vehicle calculated as shown in Eq. 4.3.1.
d1=xTXl-xT2tl Eq. 4.3.1
The mean difference for the vehicle sample is calculated as in Eq. 4.3.2.
_veh * v _veh
_ V*T3,i — XT2,i ) ^'
d =
Eq. 4.3.2
To account for the fact that the each of the vehicle means incorporated multiple replicates
with their associated variability, we calculated between-vehicle and within-vehicle variance
components in estimating the variance of d.
The between-vehicle variance component was calculated very simply as the sum of squared
errors for the dt, with the degrees of freedom reflecting the number of vehicles in the sample (Eq.
4.3.3).
"1^-df
Eq. 4.3.3
Su =¦
"veh"1
The within-vehicle variance component, analogous to an error sum of squares, was calculated
by summing the pooled-sums of squares for each of the di, as previously shown in Eq. 4.3.4. The
degrees of freedom for the within-vehicle variance component was also the sum of the degrees of
freedom for individual vehicle differences.
[(^T17 i ^)ST2,i + (Pf3,i l)^r3,i]
^veh , v
+rlT3,i ~2)
•*w =— 7- ; Eq. 4.3.4
Z=1
Incorporating the between-vehicle and within-vehicle variance components, the standard
error of d is then calculated as shown in Eq. 4.3.5,
2 2 Eq. 4.3.5
with the total of all measurements on all vehicles and fuels, denoted as Htotai, calculated as
23
-------
"veil ,
Z(»
Eq 4.3.6
After calculating these parameters, the test statistic for the paired test, under a null hypothesis
of no difference in means, is
t
d-0
actual
Eq. 4.3.7
The critical value of t for this two-tailed test was taken at the 97.5% confidence level, with nveh-1
degrees of freedom.
The results of the paired tests for CO2 and fuel economy on both cycles are shown in Table
4.3.1. As shown above, the mean CO2 differences for this vehicle sample are -6.4 and -2.2
grams/mile on the FTP and HFET cycles, respectively. Corresponding mean differences for fuel
economy are -0.66 and -1.34 mpg. All results are highly significant. The tests for fuel economy
are apparently more significant than those for CO2 emissions. However, it is not clear that this
result is of interpretive significance, given the importance of the carbon balance in the fuel
economy calculation, i.e., that fuel economy is itself dependent on CO2 emissions.
The importance of the variability of replicate measurements on the estimation of the standard
error of the difference is shown in the relationships between the two variance components, each
divided by their respective degrees of freedom. For CO2, examination of the two terms in Eq.
4.3.5, prior to taking the square root, shows that the within-vehicle component accounts for 2.8%
of the variance of the difference for the FTP, and 5.0% for the HFET. While the contribution of
the within-vehicle component is small in both cases, it's contribution is twice as large for the
HFET, reflecting the somewhat greater relative variability of replicate measurements on this
cycle.
Table 4.3.1: Paired ^-tests for CO2 emissions and fuel economy, on the FTP and HFET cycles.
CO2
Cycle
Mean Difference
Wveh
Wtotal
d*f* within
2
sb
2
sd
factual
/>-value
(g/mi)
(%)
FTP
-6.37
-1.78
11
80
58
11.83
2.457
1.052
-6.06
0.00012
HFET
-2.16
-1.02
11
72
50
3.11
1.090
0.546
-3.96
0.00267
Carbon-balance Fuel Economy
Cycle
Mean Difference
Wveh
Wtotal
d*f* within
si
Aw
sd
factual
/>-value
(mpg)
(%)
FTP
-0.66
-2.29
11
80
58
0.0645
0.0146
0.0778
-8.45
0.000019
HFET
-1.34
-2.98
11
72
50
0.340
0.0645
0.1784
-7.49
0.000010
In the interpretation and application of these results, the ostensible transportability of the
relative differences will be the most relevant factor.
24
-------
Figures 4.3.3 and 4.3.4 show relative reductions in CO2 and fuel economy on the Tier 3 fuel,
relative to the Tier 2 fuel, by vehicle. Mean relative differences in CO2 are -1.78 and -1.02% on
the FTP and HFET cycles, respectively. Corresponding values for fuel economy are -2.29 and -
2.98%. The charts illustrate the patterns described above, namely, that the relative differences in
CO2 are generally larger on the FTP than on the HFET (9 of 11 vehicles). The reverse holds true
for fuel economy (10 of 11 vehicles).
One conclusion from this work is that the mean relative differences, at least for the FTP
cycle, are in the neighborhood of the prediction from the model used to define the expected
effect on CO2 emissions in the design and conduction of the study (-1.5%). The agreement on the
FTP may be due to the fact that the model was itself based on cycle aggregate emissions from the
LA92 cycle, thus incorporating start and running emissions.
~ HWFET
Figure 4.3.3: Relative differences in CO2 emissions on the FTP and
HFET cycles, by vehicle (calculated as 100x(T3 - T2) /T2).
25
-------
~ HWFET
Figure 4.3.4: Relative differences in fuel economy on the FTP and
HFET cycles, by vehicle (calculated as 100x(T3 - T2) ITT).
4.4 Drive Quality Statistics
After data collection the drive quality statistics were reviewed to ensure the tests were driven
in a repeatable manner. All metrics were calculated according to SAE J2951, and are included in
the reports generated by the dynamometer control software. Tables 4.4.1 and 4.4.2 summarize
these statistics as average values by vehicle and fuel for the FTP and HFET cycles, respectively.
The percentages are calculated for each metric as the driver value minus the target divided by the
target, times 100 percent. These tables only include vehicles that were considered "good" tests
in this analysis (see Appendix B).
Overall the results indicate the vehicles were driven consistently between the test fuels and
that driver performance did not bias the CO2 or fuel economy results. Figures 4.4.1 and 4.4.2
show the drive cycle energy metrics, where in all cases there was less than 0.5% difference in
energy between test fuels on any vehicle, and roughly an equal number of vehicles where slightly
more energy was used on Tier 2 fuel versus on Tier 3 fuel. There was no attempt to normalize
results based on drive quality statistics within a test group or between vehicles.
The Absolute Speed Change % for HFET shows larger differences than other driver metrics
especially for the Malibu 1. Since these HFETs were rerun with a different driver we would
expect some differences in drive characteristics. The same driver also did the revised tests for the
Altima, Acura and Silverado which also have slightly higher Absolute Speed % changes than the
other vehicles. All tests fell within the required limits of the trace and for most vehicles no
26
-------
warnings or alarms were triggered. The Malibu 1 did have an accelerator fault code occasionally
occur on the Tier 2 HFET tests with a message indicating reduced engine power, but since there
was not a violation of speed trace there was not sufficient grounds to exclude the test.
Table 4.4.1: Drive quality statistics for the FTP.
Vehicle
Fuel type
Energy
Distance
Energy
Economy
Absolute
Speed
Change
Inertial
Work
Malibu 1
Tier 2
Tier 3
0.07%
-0.11%
0.00%
0.08%
-0.07%
0.34%
0.30%
0.13%
0.30%
0.32%
Civic
Tier 2
Tier 3
-0,61%
-0.47%
-0.41%
-0.31%
0.21%
0.16%
-0.06%
0.09%
-0.28%
0.34%
Ram
Tier 2
Tier 3
-0.06%
-0.06%
-0.09%
-0.11%
-0.03%
-0.06%
0.77%
0.53%
1.19%
0.83%
Altima
Tier 2
Tier 3
1,35%
0.94%
-0.34%
-0.23%
-1.66%
-1.16%
1.63%
1.20%
2.15%
1.52%
Silverado
Tier 2
Tier 3
-0.78%
-0.85%
-0.33%
-0.34%
0.45%
0.51%
1.08%
0.91%
1.72%
1.44%
Silverado (2b)
Tier 2
Tier 3
-0.30%
-0.60%
-0.07%
-0.20%
0.23%
0.41%
1.20%
1.44%
1.73%
2.19%
Volvo
Tier 2
Tier 3
-0.64%
-0.72%
0.00%
0.06%
0.65%
0.79%
0.23%
0.37%
0.29%
0.52%
Acura
Tier 2
Tier 3
-0,15%
-0.48%
-0.33%
-0.35%
-0.18%
0.13%
0.01%
-0.31%
-0.10%
-0.54%
Malibu 2
Tier 2
Tier 3
-0.80%
-0.89%
-0.01%
-0,09%
0.79%
0.80%
0.99%
0.84%
2.38%
1.97%
Mazda
Tier 2
Tier 3
-1.09%
-1.13%
-0.02%
-0.14%
1.09%
1.00%
-0.40%
-0.37%
-0.69%
-0.64%
F150
Tier 2
Tier 3
-1,36%
-1.22%
-0.09%
-0.05%
1.29%
1.18%
-1.11%
-1.05%
-1.70%
-1.60%
27
-------
Table 4.4.2: Drive quality statistics for the HFET.
Vehicle
Fuel type
Energy
Distance
Energy
Economy
Absolute
Speed
Change
Inertial
Work
Malibu 1
Tier 2
Tier 3
0,03%
0.31%
-0,02%
0.09%
-0.06%
-0.22%
9.82%
10.12%
12.17%
12.75%
Civic
Tier 2
Tier 3
-0.70%
-0.59%
-0.14%
-0.12%
0.56%
0.48%
0.69%
0.85%
1.00%
1.00%
Ram
Tier 2
Tier 3
0.73%
0.52%
0.00%
-0.03%
^ -0.73%
-0.54%
4.37%
2.56%
5.24%
3.13%
Altima
Tier 2
Tier 3
0.62%
0.70%
-0.04%
0.01%
-0.65%
-0.69%
6.26%
5.38%
8.06%
7.23%
Silverado
Tier 2
Tier 3
-0.65%
-0.15%
-0.20%
0.45%
0.45%
4.88%
4.19%
5.87%
5.00%
Silverado (2b)
Tier 2
Tier 3
-0.69%
-0.28%
-0.19%
-0.14%
0.50%
0.13%
3.86%
3.19%
4.49%
3.95%
Volvo
Tier 2
Tier 3
-0.80%
-0.76%
-0.18%
-0.14%
0.62%
0.63%
2.61%
2.15%
3.43%
2.60%
Acura
Tier 2
Tier 3
0.15%
0.27%
0.03%
-0.01%
-0.11%
-0.27%
2.99%
5.70%
3.97%
7.44%
Malibu 2
Tier 2
Tier 3
-0.76%
-0.86%
-0.17%
-0.14%
0.60%
0.73%
6.13%
6.20%
7.46%
7.69%
Mazda
Tier 2
Tier 3
-0.63%
-0.67%
-0.12%
-0.19%
0.52%
0.49%
2.45%
2.51%
3.23%
3,33%
F150
Tier 2
Tier 3
-0.91%
-0.76%
-0.17%
-0.09%
0.74%
0.66%
-0.03%
0.94%
0.13%
1.54%
28
-------
Percent Change in Energy for FTP
1.50%
£? 1.00%
QJ
C
c 0-50%
c 0.00%
ro
_c \
U O
^ -0.50% nJP
-
WD
is
c
LU
c
Ml
C
ro
_c
u
X-i
c
-------
This analysis was performed by conducting paired Wests for the two groups of vehicles, as
described in 4.1 above. We then tested the two mean differences against each other, as shown in
Eq. 4.1, as for a two-sample test, in which the standard error for the difference in the mean
differences was calculated from the standard errors for each.
^T3-lst ^T2-lst
V2 _i_ 2
^,T3-lst +5-d,T2-lst
Eq. 4.1
The critical value for the test was taken as a two-tailed t statistic at the 95% confidence level
with 9 degrees of freedom, estimated as 6+5-2. Results for the tests are shown in Tables 4.5.1
and 4.5.2 for results on the FTP and HFET cycles, respectively.
For the FTP cycle, the difference in the mean differences between the fuel-order groups is -
1.246 g/mi. This value is roughly half of its standard error, resulting in a low t statistic and a
highly insignificantp-walue. The level of significance would be even lower, but for the inclusion
of the 2b Silverado in the T3-first group. Its relatively large absolute difference increases the
mean difference and standard error for this group, as shown graphically in Figure 4.5.1.
For the HFET cycle, the difference in the mean differences is -0.35 g/mi. This value is less
than one third of its standard error, giving an insignificant p-value. This result would be even less
significant if the Acura's result did not differ in sign from those for the other vehicles, thus
increasing the variance for the Tier-2-first group. This is shown in Figure 4.5.2.
On the whole, the conclusion is that this analysis shows no evidence of a bias or artifact
related to test fuel-order for measurements on either cycle.
Table 4.5.1. FTP Cycle: Comparison of paired /-tests for groups of vehicles, by fuel order.
Group
Differences
/7veh
Sd
/-statistic
p-value
(g/mi)
%
T2 first
-5.805
-1.85
6
1.008476
-5.756
0.00222
T3 first
-7.051
-1.69
5
2.184972
-3.388
0.01378
Difference
-1.246
11
2.312548
-0.538
0.6031
Table 4.5.2. HFET Cycle: Comparison of paired Wests for groups of vehicles, by fuel order.
Group
Differences
/7veh
Sd
/-statistic
p-value
(g/mi)
%
T2 first
-2.004
-1.06
6
0.852321
-2.351
0.0654
T3 first
-2.354
-0.97
5
0.723467
-3.254
0.0313
Difference
-0.350
11
1.117969
-0.3131
0.7614
30
-------
Tier 2 Fuel first
Tier 3 Fuel first
-2
1
-2
¦ ¦ |
7/. Z
1 1
1
¦| -4
3 -6
! 4
-6
' 1
OJ
O -8
C
a
i'10
° 0
0 -8
C
a
10
a
Li
-14
-14
i
Figure 4.5.1: Absolute differences in CO2 emissions on the FTP cycle, by test-fuel order.
Figure 4.5.2: Absolute differences in CO2 emissions on the HFET cycle, by test-fuel order.
5. Additional Study of Acura and Fuel Octane
During the course of testing the Acura, we encountered vehicle behavior in the HFET that
was unexpected and counter to what we had observed in other vehicles, as well as being
inconsistent with its own behavior over the FTP portion of the testing (see Figure 4.3.3). In an
effort to better understand these observations, we performed additional testing on this vehicle
using a Tier 3 premium grade test fuel to examine whether octane level could be a contributing
factor.
Figure 5.1 shows CO2 results for replicate HFET tests on the Tier 2 and Tier 3 test fuels, as
well as Tier 3 premium (T3P). Note that Tier 3 regular grade and Tier 3 premium grade fuels are
very closely matched in property specifications other than octane rating.11 These results,
showing equivalent performance on Tier 2 and Tier 3 premium fuels, which have very similar
11 Tier 3 premium test fuel properties are available in Table A-5 of Appendix A.
31
-------
octane but significantly different aromatics and ethanol levels, are suggestive that the vehicle is
deriving a performance benefit from the additional octane value. The Tier 3 program does allow
the use of higher octane fuel for any vehicle labeled as requiring premium grade fuel, however,
the Acura does not have a premium fuel manufacturer recommendation and was specifically
included in this study because it was not expected to be sensitive to fuel octane level.
174 -
172 -
E
3
CO
c
o
"co
c/>
E
LU
170
168 -
166 -
T2
T3
Fuel
T3P
Figure 5.1: Acura CO2 emissions over replicate HFEr
cycles
using Tier 2, Tier 3 regular, and Tier 3 premium grade test fuels.
6. Discussion
This test program and analysis examined effects of a test fuel change on both CO2 emission
rates and fuel economy. Fuel economy is calculated from CO2 but when comparing two fuels,
these values can move in the same or opposite directions depending on fuel properties. Relative
to Tier 2 test fuel, Tier 3 test fuel has less carbon per unit energy primarily due to its lower
aromatic content. This difference is the primary driver for the observed lower emissions of CO2
in grams per mile and is consistent with results of the EPAct/V2/E-89 study. At the same time,
the Tier 3 fuel's lower aromatic content, as well as the presence of ethanol, lowers its volumetric
energy density (Btu/gal), resulting in lower fuel economy. In other words, when using Tier 3 test
fuel these results show that vehicles emit less CO2 per mile while consuming more fuel volume
than when using Tier 2 test fuel.
The test fleet consisted of eleven vehicles covering a range of advanced technologies that
reasonably represent a future fleet meeting the greenhouse gas and fuel-economy standards for
32
-------
light- and heavy-duty applications. We observed a consistent response to the fuel change across
the vehicle sample regardless of specific engine technology, lending confidence to application of
these results to other engine technologies.
Fundamental to determining the small effect of this test fuel change with statistical
confidence was careful study design and consistent execution of procedures from day to day.
Also important was the analysis of key fuel properties at several laboratories to establish accurate
inputs to the calculation of fuel economy. Additionally, driver metrics were recorded and
reviewed to confirm driver behavior did not bias the fuel comparison.
The overall results across the test fleet showed a reduction in CO2 of 1.78% for the FTP and
1.02% for the HFET tests for Tier 3 compared to Tier 2 test fuel. For fuel economy the overall
reduction was 2.29% for the FTP and 2.98% for the HFET tests for Tier 3 compared to Tier 2
test fuel. Throughout, the high levels of statistical significance observed, both for CO2 and fuel
economy, suggest that the measured differences in these parameters are actual and in reasonable
agreement with the difference projected during the planning of the study.
33
-------
Appendix A
Supplemental Information on Test Fuels
34
-------
Table A-l. Detailed data for test fuels used in this study.
Propi-m
Test Method
I nils
Tier 2 Tesi I'uel
Tier 3 Re» Tesi I'uel
Miiiii USD1, ii--
Meiin RSI)1, n'
API Gravity, 60°F
ASTMD4052
°API
58.8
0.36%
2
57.5
0.18%
2
Density, 60°F
g/cm3
0.7430
0.10%
2
0.7482
0.06%
4
Specific Gravity, 60°F
-
0.7437
0.10%
2
0.7490
0.06%
4
DVPE (EPA equation)
ASTMD5191
psi
8.95
-
1
8.75
-
1
Ethanol
ASTMD5599
vol %
-
-
0
10.15
3.92%
3
Oxygenates other than EtOH
-
-
0
0
0.00%
3
Oxygen
mass %
-
-
0
3.74
3.81%
3
Carbon
ASTMD5291
mass %
87.01
0.40%
5
82.88
0.59%
5
Hydrogen
13.21
1.27%
5
13.59
0.89%
5
Mass% sum with oxygen
calc from D5291
mass %
100.22
0.30%
5
100.20
0.41%
5
C, normalized
calc from D5291
mass %
86.82
0.20%
5
82.70
0.19%
5
H, normalized
calc from D5291
mass %
13.18
1.32%
5
13.56
1.17%
5
Mass% sum, normalized
calc from D5291
mass %
100.00
0.00%
5
100.00
0.00%
5
Carbon Weight Fraction
calc from D5291
-
0.8682
0.20%
5
0.8270
0.19%
5
Elydrogen
ASEM D3 343®
mass %
13.35
-
-
13.68
-
-
Carbon
calc from D3343
mass %
86.65
-
1
82.58
-
1
Sulfur
ASTMD2622
mg/kg
41.8
7.44%
2
9.2
13.14%
2
Aromatics
ASTMD1319
vol %
30.6
-
1
22.9
-
1
Olefins
0.6
-
1
5.4
-
1
Saturates
68.8
-
1
71.7
-
1
Olefins
ASTMD6550
mass %
-
-
-
6.4
-
1
Water Content
ASEM El 064
mg/kg
70
-
1
930
-
1
Existent Gum, washed
ASEMD381
mg/lOOml
-
-
-
0.5
-
1
Research Octane Number
ASEM D2699
-
96.5
0.15%
2
91
-
1
Motor Octane Number
ASEM D2700
-
88.65
0.40%
2
83.5
-
1
AKI (R+M)/2
D2699/D2700
-
92.6
0.31%
2
87.25
-
1
Octane sensitivity
D2699/D2700
-
7.85
2.70%
2
7.50
-
1
Net Eleat of Combustion
ASEMD3338(3)
Btu/lb
18,446
-
-
18,527
-
-
Gross Eleat of Combustion, 25°C
ASEM D4 809
Btu/lb
19,734
0.35%
4
19,124
0.17%
4
Gross Eleat of Combustion, 25°C
ASEM D4 809
MJ/kg
45.900
0.35%
4
44.482
0.17%
4
Net Eleat of Combustion, 25°C
ASEM D4 809
MJ/kg
43.100
0.43%
4
41.610
0.25%
4
Net Eleat of Combustion, 25°C
ASEM D4 809
Btu/lb
18,529
0.43%
4
17,889
0.25%
4
(1) Relative standard deviation, calculated as standard deviation divided by mean.
(2) Number of replicate fuel property measurements from different labs included in the mean.
(3) This method is a calculated result which used the mean of other property measurements, thus n and RSD were omitted.
35
-------
Table A-2. Detailed data for test fuels used in this study.
Propi-m
Test Mi-ihixl
I nils
Tier
2 l isl l iiil
Tier3 Km: Tesi I'm-I
Mi-;in
USD 1
n':'
RSI)1
n'21
Distillation - IBP
89.4
-
1
93.7
-
1
5%
111.9
-
1
120.9
-
1
10%
125.6
-
1
129.2
-
1
20%
147.0
-
1
140.5
-
1
30%
171.7
-
1
148.8
-
1
40%
201.9
-
1
154.4
-
1
50%
°F
222.6
-
1
210.0
-
1
60%
ASTMD86
232.4
-
1
239.9
-
1
70%
241.9
-
1
255.6
-
1
80%
259.8
-
1
286.3
-
1
90%
317.3
-
1
322.0
-
1
95%
340.7
-
1
340.2
-
1
Distillation - EP
405.9
-
1
387.0
-
1
Recovery
97.5
-
1
97.3
-
1
Residue
vol %
1.1
-
1
0.9
-
1
Loss
1.4
-
1
1.8
-
1
Benzene
0.03
-
1
0.55
-
1
Toluene
25.2
-
1
6.8
-
1
C 8 Aromatic s
0.9
-
1
6.4
-
1
C 9 Aromatic s
9.1
-
1
6.2
-
1
C10+Aromatic s
3.1
-
1
5.6
-
1
Total Aromatics
38.3
-
1
25.4
-
1
C4 Paraffins
0.7
-
1
3.5
-
1
C5 Paraffins
24.7
-
1
OO
OO
-
1
C6 Paraffins
ASTM D6729
vol %
6.5
-
1
OO
OO
-
1
C7 Paraffins
3.7
-
1
4.8
-
1
C8 Paraffins
17.7
-
1
13.4
-
1
C9 Paraffins
2.4
-
1
4.3
-
1
C10+ Paraffins
2.2
-
1
4.1
-
1
Total Paraffins
57.9
-
1
47.7
-
1
Cycloparaffms
1.0
-
1
9.7
-
1
Olefins
0.1
-
1
6.4
-
1
Ethanol
0.0
-
1
9.7
-
1
Unidentified
2.7
-
1
1.0
-
1
PM Index
See note (3)
-
1.86
-
1
1.52
-
1
Benzene
0.05
-
1
0.56
-
1
Toluene
20.0
-
1
6.2
-
1
C 8 Aromatics
ASTM D5769
vol %
0.9
-
1
6.2
-
1
C 9 Aromatics
9.9
-
1
5.5
-
1
C10+Aromatics
1.5
-
1
5.4
-
1
Total Aromatics
32.3
-
1
23.8
-
1
(1) Relative standard deviation, calculated as standard deviation divided by mean.
(2) Number of replicate fuel property measurements from different labs included in the mean.
(3) Calculated as described in SAE technical paper 2010-01-2115.
36
-------
Table A-3. Tier 3 regular grade emission test fuel specifications for a low-level ethanol-
gasoline blend (also Table 1 of 40 CFR 1065.710).
Propi-m I nil
Ki-I'i'iviuv pnnvilinv1
(li-iu-nil l.
-------
Table A-4. Tier 2 emission test fuel specifications for gasoline without ethanol (also Table 1
of 40 CFR 86.113-04).
1(0111
Uoiiiihir
Reference pmccdu iv1
Research octane, Minimum
93
ASTM D2699; ASTM D2700
Octane sensitivity2
7.5
ASTM D2699; ASTM D2700
Distillation Range (°F):
Evaporated initial boiling point3
75-95
ASTM D86
10% evaporated
120-135
50% evaporated
200-230
90% evaporated
300-325
Evaporated final boiling point
415 Max.
Hydrocarbon composition (vol %):
Olefins
10% Max.
ASTM D1319
Aromatics
35% Max.
Saturates
Remainder
Lead, g/gallon (g/liter), Maximum
0.050 (0.013)
ASTM D3237
Phosphorous, g/gallon (g/liter), Maximum
0.005 (0.0013)
ASTM D3231
Total sulfur, wt. %4
0.0015-0.008
ASTM D2622
Dry Vapor Pressure Equivalent (DVPE), psi (kPa)5
8.7-9.2 (60.0-63.4)
ASTM D5191
!ASTM procedures are incorporated by reference in §86.1.
2Octane specifications are optional for manufacturer testing.
3Fortesting at altitudes above 1,219 m (4000 feet), the specified range is 75-105 °F.
4Sulfur concentration will not exceed 0.0045 weight percent for EPA testing.
5For testing unrelated to evaporative emission control, the specified range is 8.0-9.2 psi (55.2-63.4 kPa). For testing at altitudes
above 1,219 m (4000 feet), the specified range is 7.6-8.0 psi (52.4-55.2 kPa). Calculate dry vapor pressure
equivalent, DVPE, based on the measured total vapor pressure, pT, using the following equation: D VPE (psi) = 0.956 ¦ />T~0.347
(or DVPE (kPa) = 0.956 ¦ pT-2.39). DVPE is intended to be equivalent to Reid Vapor Pressure using a different test method.
38
-------
Table A-5. Properties of Tier 3 premium grade emission test fuel used in this study.
Pnipi-iiv
IVsi Mi-ihixl
I nils
Tier 3 Pmiiium IVsi I'liil
API Gravity, 60°F
ASTM D4052
°API
58.23
Density, 60°F
g/cm3
0.7451
Specific Gravity, 60°F
-
0.7458
DVPE (EPA equation)
ASTM D5191
psi
8.77
Ethanol
ASTM D5599
vol %
9.80
Oxygenates other than EtOH
<0.01
Oxygen
mass %
3.62
Carbon
ASTM D5291
mass %
82.67
Hydrogen
13.70
Mass% sum with oxygen
Calc.
mass %
99.99
Aromatics
ASTM D5769
voP/o
23.94
Sulfur
ASTM D5453
mg/kg
8.9
Distillation IBP
ASTMD86
°F
103.3
10%
132.6
50%
208.4
90%
325.0
Distillation EP
385.0
Olefins
ASTM D6550
vol %
5.2
Existent Gum, washed
ASTMD381
mg/lOOml
0.5
Research Octane Number
ASTM D2699
-
97.8
Motor Octane Number
ASTM D2700
-
88.4
AKI (R+M)/2
ASTM D2699/D2700
-
93.1
Octane sensitivity
ASTM D2699/D2700
-
9.4
Net Heat of Combustion, 25°C
D240
Btu/lb
17,967
39
-------
Appendix B
Tests Used in Analysis
40
-------
Table B-l. List of tests used in CO2 and FE analysis (page 1 of 3).
Vehicle
Test Date Time
Fuel
TestCycle
Driver
C02 (g/mi)
FE (mpg)
Notes
Ram
2/24/2016 11:46
Tier 2
FTP
Primary
425.07
21.04
Ram
2/25/2016 13:02
Tier 2
FTP
Primary
423.81
21.09
Ram
2/26/2016 8:48
Tier 2
FTP
Primary
424.77
21.07
Ram
3/1/2016 12:30
Tier 2
FTP
Primary
420.25
21.28
Ram
3/2/2016 10:22
Tier 2
FTP
Primary
425.35
21.04
Ram
3/3/2016 8:22
Tier 2
FTP
Primary
424.43
21.07
Ram
3/29/2016 10:22
Tier 3
FTP
Primary
410.44
21.03
Ram
3/30/2016 10:04
Tier 3
FTP
Primary
417.86
20.65
Ram
3/31/2016 7:30
Tier 3
FTP
Primary
415.15
20.79
Ram
4/7/2016 15:02
Tier 3
HFET
Alt
261.77
32.82
Ram
4/7/2016 15:50
Tier 3
HFET
Alt
260.90
32.93
Ram
4/7/2016 16:38
Tier 3
HFET
Alt
259.35
33.12
Ram
4/8/2016 14:00
Tier 2
HFET
Alt
263.13
33.83
Ram
4/8/2016 15:02
Tier 2
HFET
Alt
263.08
33.84
Ram
4/8/2016 15:52
Tier 2
HFET
Alt
262.08
33.96
Silverado (2b)
6/22/2016 9:09
Tier 3
FTP
Primary
706.28
12.19
Silverado (2b)
6/22/2016 10:21
Tier 3
HFET
Primary
442.23
19.42
Silverado (2b)
6/23/2016 13:39
Tier 3
FTP
Primary
706.36
12.18
Silverado (2b)
6/23/2016 14:44
Tier 3
HFET
Primary
442.99
19.39
Silverado (2b)
6/24/2016 7:08
Tier 3
FTP
Primary
707.86
12.16
Silverado (2b)
6/24/2016 8:24
Tier 3
HFET
Primary
444.12
19.34
Silverado (2b)
6/28/2016 13:15
Tier 2
FTP
Primary
721.16
12.34
Silverado (2b)
6/28/2016 14:33
Tier 2
HFET
Primary
449.34
19.79
Silverado (2b)
6/29/2016 10:10
Tier 2
FTP
Primary
722.18
12.33
Silverado (2b)
6/29/2016 11:21
Tier 2
HFET
Primary
447.73
19.86
Silverado (2b)
6/30/2016 9:20
Tier 2
FTP
Primary
721.38
12.36
Silverado (2b)
6/30/2016 10:49
Tier 2
HFET
Primary
445.90
19.94
Acura
5/3/2016 12:22
Tier 2
FTP
Primary
276.45
32.40
Acura
5/4/2016 7:30
Tier 2
FTP
Primary
273.92
32.68
Acura
5/5/2016 10:45
Tier 2
FTP
Primary
277.74
32.25
Acura
5/6/2016 7:20
Tier 2
FTP
Primary
274.54
32.61
Acura
5/10/2016 7:19
Tier 3
FTP
Primary
271.51
31.78
Acura
5/11/2016 13:05
Tier 3
FTP
Primary
274.34
31.48
Acura
5/12/2016 9:07
Tier 3
FTP
Primary
272.37
31.71
Acura
6/14/2016 14:46
Tier 3
HFET
Alt
172.16
49.91
Acura
6/14/2016 16:03
Tier 3
HFET
Alt
172.91
49.69
Acura
6/14/2016 16:57
Tier 3
HFET
Alt
172.68
49.76
Acura
6/16/2016 12:53
Tier 2
HFET
Alt
171.68
51.87
Acura
6/16/2016 14:12
Tier 2
HFET
Alt
170.95
52.09
Acura
6/16/2016 14:57
Tier 2
HFET
Alt
171.31
51.98
Altima
3/22/2016 9:11
Tier 2
FTP
Primary
275.07
32.40
Altima
3/23/2016 14:39
Tier 2
FTP
Primary
276.61
32.26
Altima
3/24/2016 8:05
Tier 2
FTP
Primary
276.88
32.27
Altima
4/6/2016 13:06
Tier 3
FTP
Primary
270.26
31.92
Altima
4/7/2016 7:57
Tier 3
FTP
Primary
271.23
31.78
Altima
4/19/2016 9:10
Tier 3
FTP
Primary
269.12
32.08
Altima
4/20/2016 12:03
Tier 3
FTP
Primary
270.37
31.94
Altima
4/21/2016 7:28
Tier 3
FTP
Primary
272.04
31.71
Altima
5/5/2016 15:01
Tier 2
HFET
Primary
164.62
53.95
Altima
5/5/2016 16:34
Tier 2
HFET
Alt
165.70
53.60
Altima
5/5/2016 17:27
Tier 2
HFET
Alt
165.53
53.67
Altima
6/9/2016 14:55
Tier 2
HFET
Alt
165.72
53.55
Altima
6/9/2016 16:00
Tier 2
HFET
Alt
165.38
53.65
Altima
6/9/2016 16:48
Tier 2
HFET
Alt
166.02
53.51
Altima
6/10/2016 15:49
Tier 3
HFET
Alt
162.05
52.92
Altima
6/10/2016 17:09
Tier 3
HFET
Alt
163.83
52.34
Altima
6/10/2016 17:58
Tier 3
HFET
Alt
164.22
52.24
41
-------
Table B-l. List of tests used in the CO2 and FE analysis (page 2 of 3).
Vehicle
Test DateTime
Fuel
TestCycle
Driver
C02 (g/mi)
FE (mpg)
Notes
Civic
7/19/2016 7:31
Tier 3
FTP
Primary
213.20
40.40
Methane analyzer malfunction (1)
Civic
7/19/2016 10:03
Tier 3
HFET
Primary
143.90
59.65
Methane analyzer malfunction (1)
Civic
7/20/2016 12:31
Tier 3
FTP
Primary
213.94
40.28
Methane analyzer malfunction (1)
Civic
7/20/2016 13:42
Tier 3
HFET
Primary
142.56
60.19
Methane analyzer malfunction (1)
Civic
7/21/2016 9:11
Tier 3
FTP
Primary
212.97
40.45
Methane analyzer malfunction (1)
Civic
7/21/2016 10:29
Tier 3
HFET
Primary
143.00
60.00
Methane analyzer malfunction (1)
Civic
7/26/2016 13:00
Tier 2
FTP
Primary
216.86
41.14
Civic
7/26/2016 14:13
Tier 2
HFET
Primary
145.10
61.27
Civic
7/27/2016 7:44
Tier 2
FTP
Primary
217.88
40.96
Civic
7/27/2016 9:00
Tier 2
HFET
Primary
144.70
61.46
Civic
7/28/2016 9:11
Tier 2
FTP
Primary
216.19
41.28
Civic
7/28/2016 10:25
Tier 2
HFET
Primary
144.44
61.57
F150
7/6/2016 12:42
Tier 3
FTP
Primary
375.14
22.95
F150
7/6/2016 13:59
Tier 3
HFET
Primary
241.00
35.63
F150
7/7/2016 7:42
Tier 3
FTP
Primary
378.07
22.76
F150
7/7/2016 8:57
Tier 3
HFET
Primary
242.07
35.47
F150
7/8/2016 10:14
Tier 3
FTP
Primary
377.40
22.83
F150
7/8/2016 11:20
Tier 3
HFET
Primary
242.69
35.38
F150
7/12/2016 12:22
Tier 2
FTP
Primary
382.28
23.33
F150
7/12/2016 13:39
Tier 2
HFET
Primary
246.08
36.16
F150
7/13/2016 9:27
Tier 2
HFET
Primary
245.27
36.29
F150
7/14/2016 13:17
Tier 2
FTP
Primary
379.49
23.52
F150
7/14/2016 14:33
Tier 2
HFET
Primary
243.02
36.62
F150
7/15/2016 7:39
Tier 2
FTP
Primary
380.06
23.49
Malibu 1
3/15/2016 11:00
Tier 2
FTP
Primary
314.27
28.44
Malibu 1
3/16/2016 10:48
Tier 2
FTP
Primary
315.15
28.39
Malibu 1
3/17/2016 14:20
Tier 2
FTP
Primary
314.18
28.49
Malibu 1
3/22/2016 14:57
Tier 3
FTP
Primary
308.16
28.03
Malibu 1
3/24/2016 14:11
Tier 3
FTP
Primary
306.68
28.16
Malibu 1
3/25/2016 8:03
Tier 3
FTP
Primary
307.28
28.08
Malibu 1
6/6/2016 15:19
Tier 2
HFET
Alt
186.68
47.47
Accelerator fault code (2)
Malibu 1
6/6/2016 16:17
Tier 2
HFET
Alt
191.32
46.08
Accelerator fault code (2)
Malibu 1
6/6/2016 17:04
Tier 2
HFET
Alt
189.45
46.77
Accelerator fault code (2)
Malibu 1
6/15/2016 14:23
Tier 3
HFET
Alt
182.99
46.94
Malibu 1
6/15/2016 15:35
Tier 3
HFET
Alt
184.31
46.61
Malibu 1
6/15/2016 16:33
Tier 3
HFET
Alt
185.34
46.35
Malibu 1
6/15/2016 17:20
Tier 3
HFET
Alt
183.39
46.84
Malibu 2
5/17/2016 12:33
Tier 3
FTP
Primary
269.68
31.92
Malibu 2
5/17/2016 13:43
Tier 3
HFET
Primary
163.83
52.40
Malibu 2
5/18/2016 7:22
Tier 3
FTP
Primary
267.65
32.18
Malibu 2
5/18/2016 8:36
Tier 3
HFET
Primary
163.38
52.53
Malibu 2
5/19/2016 12:52
Tier 3
FTP
Primary
269.38
31.98
Malibu 2
5/19/2016 13:57
Tier 3
HFET
Primary
163.30
52.59
Malibu 2
5/24/2016 7:24
Tier 2
FTP
Primary
273.56
32.60
Malibu 2
5/24/2016 9:28
Tier 2
HFET
Primary
166.60
53.39
Malibu 2
5/25/2016 11:41
Tier 2
FTP
Primary
274.01
32.58
Malibu 2
5/25/2016 12:55
Tier 2
HFET
Primary
166.03
53.55
Malibu 2
5/26/2016 12:29
Tier 2
FTP
Primary
274.43
32.53
Malibu 2
5/26/2016 14:06
Tier 2
HFET
Primary
165.44
53.78
Malibu 2
6/14/2016 7:53
Tier 3
FTP
Primary
268.18
32.10
Malibu 2
6/14/2016 9:03
Tier 3
HFET
Primary
164.79
52.10
Malibu 2
6/15/2016 10:31
Tier 3
FTP
Primary
268.30
32.09
Malibu 2
6/15/2016 11:48
Tier 3
HFET
Primary
162.59
52.81
42
-------
Table B-l. List of tests used in the CO2 and FE analysis (page 3 of 3).
Vehicle
Test DateTime
Fuel
TestCycle
Driver
C02 (g/mi)
FE (mpg)
Notes
Mazda
5/17/2016 7:28
Tier 2
FTP
Primary
241.64
36.96
Mazda
5/17/2016 9:22
Tier 2
HFET
Primary
161.62
55.07
Mazda
5/18/2016 12:35
Tier 2
FTP
Primary
243.50
36.68
Mazda
5/18/2016 13:53
Tier 2
HFET
Primary
161.53
55.11
Mazda
5/19/2016 7:17
Tier 2
FTP
Primary
241.24
37.03
Mazda
5/19/2016 10:18
Tier 2
HFET
Primary
162.45
54.78
Mazda
5/24/2016 11:43
Tier 3
FTP
Primary
238.07
36.17
Mazda
5/24/2016 13:09
Tier 3
HFET
Alt
160.94
53.37
Mazda
5/25/2016 7:39
Tier 3
FTP
Primary
238.61
36.10
Mazda
5/25/2016 8:50
Tier 3
HFET
Primary
160.06
53.66
Mazda
5/26/2016 9:01
Tier 3
FTP
Primary
239.03
36.04
Mazda
5/26/2016 10:31
Tier 3
HFET
Primary
159.97
53.69
Silverado
7/6/2016 7:44
Tier 3
FTP
Primary
421.72
20.41
Silverado
7/7/2016 13:55
Tier 3
FTP
Primary
416.27
20.62
Silverado
7/8/2016 7:06
Tier 3
FTP
Primary
417.00
20.61
Silverado
7/12/2016 7:16
Tier 2
FTP
Primary
427.19
20.87
Silverado
7/13/2016 12:34
Tier 2
FTP
Primary
429.37
20.78
Silverado
7/14/2016 9:25
Tier 2
FTP
Primary
426.49
20.90
Silverado
7/19/2016 13:06
Tier 3
FTP
Primary
420.36
20.45
Methane analyzer malfunction (1)
Silverado
7/20/2016 7:47
Tier 3
FTP
Primary
422.12
20.38
Methane analyzer malfunction (1)
Silverado
7/21/2016 13:04
Tier 3
FTP
Primary
421.79
20.41
Methane analyzer malfunction (1)
Silverado
7/26/2016 9:18
Tier 3
HFET
Primary
281.83
30.45
Silverado
7/26/2016 10:08
Tier 3
HFET
Primary
280.61
30.59
Silverado
7/26/2016 11:17
Tier 3
HFET
Primary
280.71
30.58
Silverado
7/28/2016 13:54
Tier 2
HFET
Primary
282.62
31.47
Silverado
7/28/2016 14:45
Tier 2
HFET
Primary
279.76
31.79
Silverado
7/28/2016 15:39
Tier 2
HFET
Primary
281.72
31.57
Volvo
4/19/2016 13:22
Tier 2
FTP
Primary
309.08
28.90
Volvo
4/20/2016 7:36
Tier 2
FTP
Primary
306.57
29.11
Volvo
4/22/2016 7:30
Tier 2
FTP
Primary
304.16
29.35
Volvo
5/3/2016 7:09
Tier 2
FTP
Primary
306.18
29.21
Volvo
5/3/2016 9:10
Tier 2
HFET
Primary
176.36
50.39
Volvo
5/4/2016 12:29
Tier 2
FTP
Primary
305.57
29.25
Volvo
5/4/2016 13:54
Tier 2
HFET
Primary
175.80
50.56
Volvo
5/5/2016 7:29
Tier 2
FTP
Primary
304.33
29.34
Volvo
5/5/2016 8:44
Tier 2
HFET
Primary
174.67
50.87
Volvo
5/10/2016 10:53
Tier 3
FTP
Primary
300.35
28.64
Volvo
5/10/2016 13:37
Tier 3
HFET
Primary
173.75
49.33
Volvo
5/11/2016 7:25
Tier 3
FTP
Primary
299.73
28.71
Volvo
5/11/2016 8:39
Tier 3
HFET
Primary
174.59
49.08
Volvo
5/12/2016 13:02
Tier 3
FTP
Primary
299.42
28.74
Volvo
5/12/2016 14:08
Tier 3
HFET
Primary
171.32
50.06
Footnotes
(1)
Malfunctioning methane analyzer could have produced a very small effect on calculated fuel economy but no effect on
C02. Not considered sufficient grounds for excluding tests.
(2)
Accelerator fault code occasionally occurred on this vehicle, with message indicating reduced engine power. However, no
violation of speed trace recorded during test. Not considered sufficient grounds for excluding tests.
43
-------
Appendix C
Tests Excluded from Analysis
44
-------
Table C-l. List of tests not used in the data analysis, with explanatory notes (page 1 of 2).
Vehicle
Test DateTime
Fuel
TestCycle
Driver
C02 (g/mi)
FE (mpg)
Notes
Ram
2/24/2016 13:51
Tier 2
HFET
Primary
266.83
33.37
PM procedural issue (1)
Ram
2/25/2016 14:21
Tier 2
HFET
Primary
267.46
33.29
PM procedural issue (1)
Ram
2/26/2016 10:47
Tier 2
HFET
Primary
268.10
33.21
PM procedural issue (1)
Ram
3/1/2016 15:20
Tier 2
HFET
Primary
268.04
33.22
PM procedural issue (1)
Ram
3/2/2016 11:48
Tier 2
HFET
Primary
267.10
33.34
PM procedural issue (1)
Ram
3/3/2016 9:48
Tier 2
HFET
Primary
266.98
33.35
PM procedural issue (1)
Ram
3/8/2016 15:36
Tier 3
FTP
Primary
411.66
20.96
Modal bench malfunction (2)
Ram
3/9/2016 11:53
Tier 3
HFET
Primary
267.35
32.13
PM procedural issue (1)
Ram
3/9/2016 13:39
Tier 3
HFET
Primary
266.32
32.26
PM procedural issue (1)
Ram
3/10/2016 14:19
Tier 3
FTP
Primary
411.28
20.98
Modal bench malfunction (2)
Ram
3/10/2016 15:41
Tier 3
HFET
Primary
262.19
32.77
PM procedural issue (1)
Ram
3/11/2016 7:26
Tier 3
FTP
Primary
414.17
20.83
Modal bench malfunction (2)
Ram
3/29/2016 11:26
Tier 3
HFET
Primary
272.27
31.55
PM procedural issue (1)
Ram
3/30/2016 11:49
Tier 3
HFET
Primary
272.07
31.58
PM procedural issue (1)
Acura
5/3/2016 13:34
Tier 2
HFET
Primary
169.66
52.49
Repeatability concerns (3)
Acura
5/4/2016 8:51
Tier 2
HFET
Primary
169.98
52.39
Repeatability concerns (3)
Acura
5/5/2016 11:55
Tier 2
HFET
Primary
172.50
51.62
Repeatability concerns (3)
Acura
5/6/2016 8:45
Tier 2
HFET
Primary
171.97
51.78
Repeatability concerns (3)
Acura
5/10/2016 8:48
Tier 3
HFET
Primary
171.86
50.00
Repeatability concerns (3)
Acura
5/11/2016 14:25
Tier 3
HFET
Primary
171.95
49.97
Repeatability concerns (3)
Acura
5/12/2016 10:30
Tier 3
HFET
Primary
171.33
50.15
Repeatability concerns (3)
AHima
3/22/2016 11:22
Tier 2
HFET
Primary
170.85
51.94
PM procedural issue (1)
AHima
3/24/2016 9:24
Tier 2
HFET
Primary
172.05
51.64
PM procedural issue (1)
AHima
3/24/2016 10:17
Tier 2
HFET
Primary
170.65
52.06
PM procedural issue (1)
AHima
4/5/2016 8:45
Tier 3
FTP
Primary
278.87
30.91
Irregular test/prep sequence (4)
AHima
4/5/2016 10:50
Tier 3
HFET
Primary
174.27
49.19
PM procedural issue (1)
AHima
4/6/2016 14:14
Tier 3
HFET
Primary
166.23
51.60
PM procedural issue (1)
AHima
4/7/2016 9:10
Tier 3
HFET
Primary
170.60
50.26
PM procedural issue (1)
Ahima
4/19/2016 10:27
Tier 3
HFET
Primary
168.92
50.77
PM procedural issue (1)
Ahima
4/20/2016 13:53
Tier 3
HFET
Primary
171.54
50.00
PM procedural issue (1)
Ahima
4/21/2016 10:53
Tier 3
HFET
Primary
167.46
51.19
PM procedural issue (1)
AHima
4/29/2016 14:52
Tier 3
HFET
Alt
169.28
50.62
Repeatability concerns (5)
AHima
4/29/2016 15:49
Tier 3
HFET
Alt
166.88
51.35
Repeatability concerns (5)
AHima
4/29/2016 16:35
Tier 3
HFET
Alt
167.45
51.20
Repeatability concerns (5)
AHima
6/7/2016 15:35
Tier 2
HFET
Alt
169.75
52.24
Repeatability concerns (5)
AHima
6/7/2016 16:47
Tier 2
HFET
Alt
167.90
52.90
Repeatability concerns (5)
AHima
6/7/2016 17:38
Tier 2
HFET
Alt
165.97
53.48
Repeatability concerns (5)
Malibu 1
3/8/2016 10:17
Tier 2
FTP
Primary
314.12
28.46
Irregular test/prep sequence (4)
Malibu 1
3/8/2016 11:31
Tier 2
HFET
Primary
190.44
46.64
PM procedural issue (1)
Malibu 1
3/9/2016 16:14
Tier 2
FTP
Primary
319.21
27.95
Modal bench malfunction (2)
Malibu 1
3/10/2016 7:55
Tier 2
FTP
Primary
318.15
28.05
Modal bench malfunction (2)
Malibu 1
3/10/2016 9:20
Tier 2
HFET
Primary
191.49
46.40
PM procedural issue (1)
Malibu 1
3/10/2016 10:33
Tier 2
HFET
Primary
191.89
46.29
PM procedural issue (1)
Malibu 1
3/15/2016 13:35
Tier 2
HFET
Primary
190.50
46.64
PM procedural issue (1)
Malibu 1
3/16/2016 12:59
Tier 2
HFET
Primary
189.99
46.76
PM procedural issue (1)
Malibu 1
3/17/2016 15:31
Tier 2
HFET
Primary
189.60
46.86
PM procedural issue (1)
Malibu 1
3/23/2016 8:15
Tier 3
FTP
Primary
311.80
27.64
Irregular test/prep sequence (4)
Malibu 1
3/23/2016 10:14
Tier 3
HFET
Primary
187.43
45.78
PM procedural issue (1)
Malibu 1
3/23/2016 11:09
Tier 3
HFET
Primary
187.15
45.85
PM procedural issue (1)
Malibu 1
3/25/2016 9:57
Tier 3
HFET
Primary
186.67
45.97
PM procedural issue (1)
Malibu 1
5/20/2016 9:39
Tier 3
HFET
Alt
204.28
41.96
Repeatability concerns (6)
Malibu 1
5/20/2016 10:58
Tier 3
HFET
Alt
204.30
41.87
Repeatability concerns (6)
Malibu 1
5/20/2016 11:53
Tier 3
HFET
Alt
203.03
42.20
Repeatability concerns (6)
Malibu 1
6/2/2016 15:30
Tier 2
HFET
Alt
203.70
43.68
Repeatability concerns (6)
Malibu 1
6/2/2016 16:59
Tier 2
HFET
Alt
209.40
42.47
Repeatability concerns (6)
Malibu 1
6/2/2016 17:59
Tier 2
HFET
Alt
206.86
43.00
Repeatability concerns (6)
45
-------
Table C-l. List of tests not used in the data analysis, with explanatory notes (page 2 of 2).
Vehicle
Test DateTime
Fuel
TestCycle
Driver
C02 (g/mi)
FE (mpg)
Notes
Silverado
7/6/2016 9:40
Tier 3
HFET
Primary
280.42
30.59
Repeatability concerns (7)
Silverado
7/7/2016 14:57
Tier 3
HFET
Primary
275.07
31.22
Repeatability concerns (7)
Silverado
7/8/2016 8:18
Tier 3
HFET
Primary
281.84
30.44
Repeatability concerns (7)
Silverado
7/12/2016 8:54
Tier 2
HFET
Primary
279.16
31.88
Repeatability concerns (7)
Silverado
7/13/2016 13:36
Tier 2
HFET
Primary
280.46
31.72
Repeatability concerns (7)
Silverado
7/14/2016 10:44
Tier 2
HFET
Primary
279.21
31.86
Repeatability concerns (7)
Silverado
7/19/2016 14:27
Tier 3
HFET
Primary
282.72
30.33
Repeatability concerns (7)
Silverado
7/20/2016 9:00
Tier 3
HFET
Primary
279.82
30.67
Repeatability concerns (7)
Silverado
7/21/2016 14:06
Tier 3
HFET
Primary
278.28
30.84
Repeatability concerns (7)
Volvo
4/5/2016 15:46
Tier 2
HFET
Primary
184.30
48.19
PM procedural issue (1)
Volvo
4/6/2016 7:25
Tier 2
FTP
Primary
310.07
28.79
Irregular test/prep sequence (4)
Volvo
4/6/2016 8:36
Tier 2
HFET
Primary
183.60
48.40
PM procedural issue (1)
Volvo
4/7/2016 12:01
Tier 2
FTP
Primary
325.34
27.62
Irregular test/prep sequence (4)
Volvo
4/7/2016 13:04
Tier 2
HFET
Primary
181.72
48.89
PM procedural issue (1)
Volvo
4/19/2016 14:30
Tier 2
HFET
Primary
184.15
48.23
PM procedural issue (1)
Volvo
4/20/2016 8:49
Tier 2
HFET
Primary
183.00
48.55
PM procedural issue (1)
Volvo
4/22/2016 8:57
Tier 2
HFET
Primary
180.53
49.22
PM procedural issue (1)
Footnotes
' (i)
PM filter holders for unused test phases were configured inconsistently in this test, producing an unknown error in all gaseous
results and thus we excluded this test.
(2)
A malfunction in the modal bench caused an error in calculated results for gaseous bag emissions, which was sufficient ground
for excluding this test.
(3)
Initial HFET results for the Acura showed no clear effect of test fuel, which was inconsistent with FTP results. We added a
controlled experiment with additional replicates comparing the fuels, and the result suggested an opposite fuel effect. This
latter experiment used a different driver, and we did not want to mix data from different drivers so we used the newer data in
the overall analysis. Additional tests with Tier 3 premium fuel indicated an effect of octane on C02 emissions, which could
explain the opposite fuel effect observed with Tier 3 regular. The tests performed on T3 premium were not included in the
overall analysis.
(4)
We observed higher emission variability when prior day's test sequence was inconsistent with the test plan. We also generally
attempted to keep testing in 4-day work-week blocks.
(5)
Initial HFET tests for the Altima were repeated in a later experiment because of the filter holder issue (1). In this later
experiment, the intial fuel effect was inconsistent with FTP results so a decision was made to conduct additional confirmatory
tests. These results showed a large spread with descending C02 results, and it was determined that the trickle charger had not
been connected during several weeks of vehicle storage prior to these additional tests. After ensuring the vehicle battery was
charged, another test set was performed. This final set was retained in the dataset.
(6)
Initial HFET tests for the Malibu lwere repeated in a later experiment die to the filter holder issue (1). The initial fuel
comparison in this experiment showed a trend inconsistent with the FTP data. At this point we had concerns about
repeatability after vehicle had been used by another testing group and returned to us with a fault code. Actions were taken to
resolve any issues, and an additional test set was performed over the following two weeks. This final set was retained in the
dataset.
(V)
InitialHFET results for the Silverado showed no clear effect of test fuel, which was inconsistent with FTP results. We added
a controlled experiment with additional replicates comparing the fuels, which confirmed the original result of no effect. This
later experiment used a different driver, and we did not want to mix data from different drivers so we used the newer data in
the overall analysis.
46
-------
Appendix D
Supplemental Emissions Data
47
-------
Table Dl. Supplemental emissions data for tests included in the analyses (page 1 of 3).a
Vehicle
Fuel
TestCycle
Test Date/Time
THC (g/mi)
CH4 (g/mi)
CO (g/mi)
NMOG (g/mi)
NOx (g/mi)
PM (mg/mi)
Silverado (2b)
Tier 2
FTP
6/28/2016 13:15
0.1530
0.0286
1.8638
0.1295
0.0201
1.4843
Silverado (2b)
Tier 2
FTP
6/29/2016 10:10
0.0662
0.0248
1.7805
0.0450
0.0193
0.9988
Silverado (2b)
Tier 2
FTP
6/30/2016 9:20
0.0613
0.0225
1.4870
0.0417
0.0197
0.7846
Silverado (2b)
Tier 2
HFET
6/28/2016 14:33
0.0177
0.0133
0.3551
0.0054
0.0082
1.0188
Silverado (2b)
Tier 2
HFET
6/29/2016 11:21
0.0171
0.0132
0.3897
0.0049
0.0090
0.9497
Silverado (2b)
Tier 2
HFET
6/30/2016 10:49
0.0167
0.0128
0.3799
0.0048
0.0096
0.8718
Silverado (2b)
Tier 3
FTP
6/22/20169:09
0.0499
0.0165
1.0437
0.0386
0.0199
1.2380
Silverado (2b)
Tier 3
FTP
6/23/2016 13:39
0.0506
0.0174
1.2281
0.0391
0.0180
Silverado (2b)
Tier 3
FTP
6/24/2016 7:08
0.0486
0.0168
0.9187
0.0365
0.0213
Silverado (2b)
Tier 3
HFET
6/22/2016 10:21
0.0068
0.0067
0.0701
0.0005
0.0033
1.1232
Silverado (2b)
Tier 3
HFET
6/23/2016 14:44
0.0057
0.0062
0.0714
0.0000
0.0041
Silverado (2b)
Tier 3
HFET
6/24/2016 8:24
0.0062
0.0064
0.0632
0.0002
0.0037
Acura
Tier 2
FTP
5/3/2016 12:22
0.0147
0.0031
0.0465
0.0124
0.0180
Acura
Tier 2
FTP
5/4/2016 7:30
0.0228
0.0032
0.0558
0.0205
0.0200
0.1200
Acura
Tier 2
FTP
5/5/2016 10:45
0.0122
0.0031
0.0532
0.0099
0.0177
Acura
Tier 2
FTP
5/6/2016 7:20
0.0131
0.0028
0.0478
0.0108
0.0173
0.2067
Acura
Tier 2
HFET
6/16/2016 12:53
0.0004
0.0006
0.0026
0.0000
0.0052
Acura
Tier 2
HFET
6/16/2016 14:12
0.0004
0.0006
0.0002
0.0000
0.0034
Acura
Tier 2
HFET
6/16/2016 14:57
0.0003
0.0006
0.0004
0.0000
0.0042
Acura
Tier 3
FTP
5/10/2016 7:19
0.0162
0.0034
0.0256
0.0146
0.0237
Acura
Tier 3
FTP
5/11/2016 13:05
0.0121
0.0034
0.0249
0.0103
0.0215
0.2723
Acura
Tier 3
FTP
5/12/2016 9:07
0.0115
0.0030
0.0258
0.0099
0.0199
Acura
Tier 3
HFET
6/14/2016 14:46
0.0004
0.0005
0.0059
0.0000
0.0045
Acura
Tier 3
HFET
6/14/2016 16:03
0.0004
0.0005
0.0015
0.0000
0.0034
Acura
Tier 3
HFET
6/14/2016 16:57
0.0004
0.0005
0.0030
0.0000
0.0033
Alt
ma
Tier 2
FTP
3/22/2016 9:11
0.0239
0.0068
0.5167
0.0183
0.0124
0.5970
Alt
ma
Tier 2
FTP
3/23/2016 14:39
0.0112
0.0059
0.4765
0.0063
0.0104
0.6672
Alt
ma
Tier 2
FTP
3/24/2016 8:05
0.0158
0.0059
0.3861
0.0109
0.0108
0.5325
Alt
ma
Tier 2
HFET
5/5/2016 15:01
0.0030
0.0022
0.2792
0.0010
0.0020
Alt
ma
Tier 2
HFET
5/5/2016 16:34
0.0036
0.0039
0.2818
0.0000
0.0036
Alt
ma
Tier 2
HFET
5/5/2016 17:27
0.0048
0.0048
0.2411
0.0004
0.0051
Alt
ma
Tier 2
HFET
6/9/2016 14:55
0.0034
0.0035
0.3644
0.0001
0.0032
Alt
ma
Tier 2
HFET
6/9/2016 16:00
0.0042
0.0045
0.3814
0.0000
0.0036
Alt
ma
Tier 2
HFET
6/9/2016 16:48
0.0030
0.0035
0.2463
0.0000
0.0032
Alt
ma
Tier 3
FTP
4/6/2016 13:06
0.0100
0.0051
0.3258
0.0063
0.0088
0.3376
Alt
ma
Tier 3
FTP
4/7/2016 7:57
0.0145
0.0050
0.4100
0.0113
0.0095
0.2591
Alt
ma
Tier 3
FTP
4/19/2016 9:10
0.0127
0.0053
0.3849
0.0089
0.0097
0.4698
Alt
ma
Tier 3
FTP
4/20/2016 12:03
0.0096
0.0051
0.3294
0.0059
0.0088
0.2306
Alt
ma
Tier 3
FTP
4/21/2016 7:28
0.0122
0.0053
0.4017
0.0083
0.0128
0.3164
Alt
ma
Tier 3
HFET
6/10/201615:49
0.0011
0.0015
0.1977
0.0000
0.0017
Alt
ma
Tier 3
HFET
6/10/201617:09
0.0014
0.0018
0.2020
0.0000
0.0023
Alt
ma
Tier 3
HFET
6/10/201617:58
0.0015
0.0019
0.1601
0.0000
0.0033
Civ
c
Tier 2
FTP
7/26/2016 13:00
0.0162
0.0080
0.0899
0.0090
0.0121
0.7907
Civ
c
Tier 2
FTP
7/27/2016 7:44
0.0196
0.0078
0.0924
0.0127
0.0133
0.5581
Civ
c
Tier 2
FTP
7/28/20169:11
0.0148
0.0078
0.0822
0.0079
0.0116
0.6652
Civ
c
Tier 2
HFET
7/26/2016 14:13
0.0006
0.0011
0.1466
0.0000
0.0008
0.1530
Civ
c
Tier 2
HFET
7/27/20169:00
0.0006
0.0010
0.1200
0.0000
0.0008
0.1688
Civ
c
Tier 2
HFET
7/28/2016 10:25
0.0009
0.0012
0.1279
0.0000
0.0009
0.1988
Civ
c
Tier 3
FTP
7/19/2016 7:31
0.0195
0.0078
0.0684
0.0215
0.0116
0.4757
Civ
c
Tier 3
FTP
7/20/2016 12:31
0.0158
0.0078
0.0473
0.0174
0.0123
0.6495
Civ
c
Tier 3
FTP
7/21/2016 9:11
0.0159
0.0078
0.0770
0.0175
0.0130
0.5276
Civ
c
Tier 3
HFET
7/19/2016 10:03
0.0007
0.0011
0.0994
0.0007
0.0008
0.0977
Civ
c
Tier 3
HFET
7/20/2016 13:42
0.0008
0.0011
0.1159
0.0008
0.0012
0.0504
Civ
c
Tier 3
HFET
7/21/2016 10:29
0.0008
0.0011
0.1242
0.0009
0.0008
48
-------
i/mi)
.8749
.6695
.4875
.2706
.2200
.2579
.4866
.8411
.0318
.2800
.3505
,3465
8690
.3943
.2597
.0627
.8856
.0935
.5376
.7309
.7257
.5311
.5110
.3014
.6620
.8018
.5406
.5665
.5896
.8146
.7568
.6465
.5561
.4652
.2025
.3648
.1983
.0337
.0129
.0739
.1290
.8756
.0560
.0739
.0575
.0394
Supplemental emissions data for tests included in the analyses (page 2 of 3).
Fuel
TestCycle
Test Date/Time
THC (g/mi)
CH4 (g/mi)
CO (g/mi)
NMOG (g/mi)
Tier 2
FTP
7/12/2016 12:22
0.0322
0.0103
0.2214
0.0233
Tie
FTP
7/14/2016 13:17
0.0300
0.0000
0.2446
0.0308
Tie
FTP
7/15/2016 7:39
0.0385
0.0137
0.2104
0.0285
Tie
HFET
7/12/2016 13:39
0.0004
0.0009
0.1087
0.0000
Tie
HFET
7/13/20169:27
0.0008
0.0014
0.0934
0.0000
Tie
HFET
7/14/2016 14:33
0.0007
0.0000
0.0947
0.0007
Tie
FTP
7/6/2016 12:42
0.0401
0.0100
0.1730
0.0340
Tie
FTP
7/7/2016 7:42
0.0424
0.0123
0.2210
0.0344
Tie
FTP
7/8/2016 10:14
0.0322
0.0106
0.1635
0.0250
Tie
HFET
7/6/2016 13:59
0.0008
0.0011
0.1046
0.0000
Tie
HFET
7/7/2016 8:57
0.0008
0.0012
0.1052
0.0000
Tie
HFET
7/8/2016 11:20
0.0007
0.0011
0.1101
0.0000
Tie
FTP
3/15/2016 11:00
0.0234
0.0070
0.7375
0.0176
Tie
FTP
3/16/2016 10:48
0.0222
0.0084
0.8118
0.0151
Tie
FTP
3/17/2016 14:20
0.0177
0.0073
0.6824
0.0118
Tie
HFET
6/6/2016 15:19
0.0173
0.0106
0.5395
0.0077
Tie
HFET
6/6/2016 16:17
0.0299
0.0151
1.1752
0.0162
Tie
HFET
6/6/2016 17:04
0.0163
0.0104
0.5764
0.0068
Tie
FTP
3/22/2016 14:57
0.0149
0.0059
0.4838
0.0110
Tie
FTP
3/24/2016 14:11
0.0159
0.0070
0.5214
0.0110
Tie
FTP
3/25/2016 8:03
0.0200
0.0068
0.6101
0.0156
Tie
HFET
6/15/2016 14:23
0.0003
0.0008
0.0347
0.0000
Tie
HFET
6/15/2016 15:35
0.0007
0.0011
0.0272
0.0000
Tie
HFET
6/15/2016 16:33
0.0018
0.0021
0.0304
0.0000
Tie
HFET
6/15/2016 17:20
0.0026
0.0029
0.0328
0.0000
Tie
FTP
5/24/2016 7:24
0.0170
0.0047
0.3724
0.0130
Tie
FTP
5/25/2016 11:41
0.0077
0.0030
0.2475
0.0053
Tie
FTP
5/26/2016 12:29
0.0082
0.0032
0.2891
0.0057
Tie
HFET
5/24/20169:28
0.0001
0.0005
0.1313
0.0000
Tie
HFET
5/25/2016 12:55
0.0000
0.0002
0.1653
0.0000
Tie
HFET
5/26/2016 14:06
0.0000
0.0001
0.0945
0.0000
Tie
FTP
5/17/2016 12:33
0.0135
0.0044
0.2926
0.0103
Tie
FTP
5/18/2016 7:22
0.0146
0.0037
0.2114
0.0125
Tie
FTP
5/19/2016 12:52
0.0080
0.0030
0.2107
0.0058
Tie
FTP
6/14/2016 7:53
0.0078
0.0032
0.1726
0.0054
Tie
FTP
6/15/2016 10:31
0.0068
0.0025
0.1693
0.0051
Tie
HFET
5/17/2016 13:43
0.0004
0.0001
0.0864
0.0003
Tie
HFET
5/18/2016 8:36
0.0002
0.0002
0.1125
0.0001
Tie
HFET
5/19/2016 13:57
0.0000
0.0001
0.0595
0.0000
Tie
HFET
6/14/2016 9:03
0.0000
0.0001
0.0791
0.0000
Tie
HFET
6/15/2016 11:48
0.0000
0.0000
0.0689
0.0000
Tie
FTP
5/17/2016 7:28
0.0169
0.0045
0.0874
0.0131
Tie
FTP
5/18/2016 12:35
0.0159
0.0053
0.1752
0.0112
Tie
FTP
5/19/2016 7:17
0.0131
0.0039
0.0916
0.0097
Tie
HFET
5/17/2016 9:22
0.0034
0.0025
0.0387
0.0010
Tie
HFET
5/18/2016 13:53
0.0059
0.0032
0.0336
0.0030
Tie
HFET
5/19/2016 10:18
0.0078
0.0041
0.0650
0.0041
Tie
FTP
5/24/2016 11:43
0.0160
0.0054
0.1642
0.0121
Tie
FTP
5/25/2016 7:39
0.0115
0.0040
0.0770
0.0085
Tie
FTP
5/26/2016 9:01
0.0114
0.0042
0.0756
0.0084
Tie
HFET
5/24/2016 13:09
0.0008
0.0008
0.0272
0.0000
Tie
HFET
5/25/2016 8:50
0.0006
0.0007
0.0522
0.0000
Tier 3
HFET
5/26/2016 10:31
0.0005
0.0006
0.0368
0.0000
49
-------
Table Dl. Supplemental emissions data for tests included in the analyses (page 3 of 3).
Vehicle
Fuel
TestCycle
Test Date/time
THC (g/mi)
CH4 (g/mi)
CO (g/mi)
NMOG (g/mi)
NOx (g/mi)
PM (mg/mi)
Ram
Tier 2
FTP
2/24/2016 11:46
0.0178
0.0037
0.3175
0.0151
0.0081
0.1850
Ram
Tier 2
FTP
2/25/2016 13:02
0.0824
0.0060
0.4532
0.0787
0.0111
0.6961
Ram
Tier 2
FTP
2/26/2016 8:48
0.0188
0.0040
0.3852
0.0157
0.0073
0.1127
Ram
Tier 2
FTP
3/1/2016 12:30
0.0160
0.0033
0.3021
0.0135
0.0062
0.1441
Ram
Tier 2
FTP
3/2/2016 10:22
0.0183
0.0022
0.2619
0.0169
0.0107
0.1223
Ram
Tier 2
FTP
3/3/20168:22
0.0181
0.0031
0.2881
0.0157
0.0058
0.0734
Ram
Tier 2
HFET
4/8/2016 14:00
0.0000
0.0004
0.0639
0.0000
0.0059
Ram
Tier 2
HFET
4/8/2016 15:02
0.0000
0.0003
0.0639
0.0000
0.0072
Ram
Tier 2
HFET
4/8/2016 15:52
0.0000
0.0005
0.0909
0.0000
0.0044
Ram
Tier 3
FTP
3/29/2016 10:22
0.0143
0.0031
0.2769
0.0129
0.0073
0.0674
Ram
Tier 3
FTP
3/30/2016 10:04
0.0148
0.0032
0.2051
0.0132
0.0077
0.1006
Ram
Tier 3
FTP
3/31/2016 7:30
0.0178
0.0032
0.2153
0.0164
0.0091
0.0585
Ram
Tier 3
HFET
4/7/2016 15:02
0.0000
0.0001
0.0400
0.0000
0.0036
Ram
Tier 3
HFET
4/7/2016 15:50
0.0000
0.0002
0.0317
0.0000
0.0030
Ram
Tier 3
HFET
4/7/2016 16:38
0.0000
0.0002
0.0307
0.0000
0.0047
Silverado
Tier 2
FTP
7/12/2016 7:16
0.0436
0.0120
0.7421
0.0335
0.0042
1.3943
Silverado
Tier 2
FTP
7/13/2016 12:34
0.0382
0.0108
0.7539
0.0294
0.0028
1.6073
Silverado
Tier 2
FTP
7/14/2016 9:25
0.0325
0.0534
0.6472
0.0212
0.0043
1.4249
Silverado
Tier 2
HFET
7/28/2016 13:54
0.0026
0.0030
0.2048
0.0000
0.0004
Silverado
Tier 2
HFET
7/28/2016 14:45
0.0046
0.0040
0.2029
0.0009
0.0032
Silverado
Tier 2
HFET
7/28/2016 15:39
0.0070
0.0047
0.2417
0.0027
0.0077
Silverado
Tier 3
FTP
7/6/2016 7:44
0.0510
0.0113
0.5703
0.0450
0.0040
1.4320
Silverado
Tier 3
FTP
7/7/2016 13:55
0.0920
0.0144
0.7943
0.0869
0.0066
1.8988
Silverado
Tier 3
FTP
7/8/2016 7:06
0.0465
0.0125
0.6026
0.0386
0.0041
1.3235
Silverado
Tier 3
FTP
7/19/2016 13:06
0.0436
0.0115
0.6733
0.0479
0.0045
1.8905
Silverado
Tier 3
FTP
7/20/2016 7:47
0.0493
0.0115
0.4884
0.0544
0.0041
1.4426
Silverado
Tier 3
FTP
7/21/2016 13:04
0.0342
0.0115
0.5374
0.0377
0.0044
1.5833
Silverado
Tier 3
HFET
7/26/2016 9:18
0.0004
0.0012
0.2038
0.0000
0.0004
Silverado
Tier 3
HFET
7/26/2016 10:08
0.0002
0.0011
0.1521
0.0000
0.0005
Silverado
Tier 3
HFET
7/26/2016 11:17
0.0006
0.0014
0.1870
0.0000
0.0005
Volvo
Tier 2
FTP
4/19/2016 13:22
0.0123
0.0036
0.3190
0.0095
0.0056
0.8402
Volvo
Tier 2
FTP
4/20/2016 7:36
0.0205
0.0041
0.4900
0.0174
0.0075
0.9924
Volvo
Tier 2
FTP
4/22/2016 7:30
0.0178
0.0038
0.5293
0.0148
0.0067
0.8195
Volvo
Tier 2
FTP
5/3/2016 7:09
0.0199
0.0033
0.4080
0.0176
0.0082
0.8560
Volvo
Tier 2
FTP
5/4/2016 12:29
0.0117
0.0036
0.4879
0.0088
0.0070
0.8440
Volvo
Tier 2
FTP
5/5/2016 7:29
0.0205
0.0033
0.5200
0.0181
0.0084
0.7952
Volvo
Tier 2
HFET
5/3/2016 9:10
0.0000
0.0005
0.2284
0.0000
0.0005
0.0406
Volvo
Tier 2
HFET
5/4/2016 13:54
0.0000
0.0005
0.2158
0.0000
0.0006
0.0542
Volvo
Tier 2
HFET
5/5/2016 8:44
0.0000
0.0004
0.2505
0.0000
0.0005
0.0680
Volvo
Tier 3
FTP
5/10/2016 10:53
0.0113
0.0040
0.5327
0.0087
0.0057
0.9388
Volvo
Tier 3
FTP
5/11/2016 7:25
0.0162
0.0037
0.5130
0.0143
0.0069
0.8453
Volvo
Tier 3
FTP
5/12/2016 13:02
0.0123
0.0036
0.4327
0.0101
0.0062
Volvo
Tier 3
HFET
5/10/2016 13:37
0.0000
0.0003
0.2753
0.0000
0.0004
0.0352
Volvo
Tier 3
HFET
5/11/2016 8:39
0.0000
0.0002
0.2969
0.0000
0.0005
0.0203
Volvo
Tier 3
HFET
5/12/2016 14:08
0.0000
0.0002
0.2120
0.0000
0.0005
a Blank values indicate that data was not collected or there was a procedural problem that invalidated the
specific result but not overall test. Some missing CH4 values for the Silverado and Civic were replaced
with medians taken from similar tests on the same vehicle, as described in Section 3.3.2.
50
------- |