United States Air and Radiation EPA420-R-03-002
Environmental Protection February 2003
Agency
&EPA The Effect of Cetane Number
Increase Due to Additives on
NOx Emissions from
Heavy-Duty Highway Engines
Final Technical Report
> Printed on Recycled Paper
-------
EPA420-R-03-002
February 2003
The Effect of Cetane Number Increase Due to
Additives on NOx Emissions from
Heavy-Duty Highway Engines
Final Technical Report
Assessment and Standards Division
Office of Transportation and Air Quality
U.S. Environmental Protection Agency
NOTICE
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 which
may form the basis for a final EPA decision, position, or regulatory action.
-------
Abstract
This report presents a technical analysis of the NOx emissions impacts of increases in
cetane number brought about through the use of diesel fuel additives. The purpose of this
technical report is to provide information to parties who may be evaluating the value,
effectiveness, and appropriateness of the use of cetane improver additives. It was originally
provided to the public in draft form so that interested parties could comment on the methodology,
assumptions, and conclusions. Where deemed appropriate, modifications were made to the
report on the basis of those comments. This technical report is now being released in final form.
The analysis described in this report uses statistical regression analysis to correlate diesel
cetane number increases brought about through additives with NOx emissions from heavy-duty
highway engines. It relies upon pre-existing data from publicly available sources. The result is
that the effect of cetane improver additives on NOx emissions can be predicted for the in-use
fleet using a correlation which is a function of the natural cetane number of the unadditized base
fuel and the increase in cetane number brought about through additives. These NOx emission
effects are also a function of the calendar year in order to account for lower cetane sensitivity of
advanced technology engines. For instance, if cetane improvers are added to a national average
base fuel so that the total cetane number is increased by 5 numbers, the percent reduction in NOx
for calendar year 2003 is predicted to be 2.0 for highway engines. The correlations predict lower
NOx benefits for future calendar years. Nonroad engines may exhibit slightly higher NOx
benefits due to the slower introduction of cetane-insensitive technologies into the nonroad fleet.
-------
Table of Contents
Abstract i
I. Context 1
A. Nature and purpose of this technical report 1
B. Public participation 2
II. Introduction 3
A. Background on cetane analysis 3
B. Cetane number as an emission control strategy 3
IE. Analytical approach 9
A. Database preparation 9
1. Representativeness of fuels 10
2. Representativeness of engines 14
B. Summary of analysis 17
IV. Conclusions 20
A. SAS modeling results 20
B. Confirmation of explanatory power 23
C. Application to the in-use fleet 28
D. Some predicted NOx impacts of additized cetane 31
Appendix 33
References 38
-------
I. Context
A. Nature and purpose of this technical report
This report presents a technical analysis of the NOx emissions impacts of increases in
cetane number brought about through the use of certain diesel fuel additives. It analyzes pre-
existing data from various emissions test programs to investigate these effects. The conclusions
drawn in this technical report represent the current understanding of this specific technical issue,
and are subject to re-evaluation at any time.
The purpose of this technical report is to provide information to interested parties who
may be evaluating the value, effectiveness, and appropriateness of the use of cetane improver
additives. This report informs any interested party as to the potential air emission impacts of
increases in cetane number brought about through use of these additives. It was originally
provided to the public in draft form so that interested parties would have an opportunity to
review the methodology, assumptions, and conclusions. The Agency also requested independent
peer reviews on the draft technical report from experts outside the Agency.
This technical report is not a rulemaking, and does not establish any legal rights or
obligations for any party. It is not intended to act as a model rule for any State or other party.
This report is by its nature limited to the technical analysis included, and is not designed to
address the wide variety of additional factors that could be considered by a State when initiating
a fuel control rulemaking. For example, this report does not consider isues such as air quality
need, cost, cost effectiveness, technical feasibility, fuel distribution and supply impacts, regional
fleet composition, and other potentially relevant factors.
State or local controls on motor vehicle fuels are limited under the Clean Air Act (CAA) -
certain state fuel controls are prohibited under the Clean Air Act, for example where the state
control applies to a fuel characteristic or component that EPA has regulated (see CAA Section
21 l(c)(4)). This prohibition is waived if EPA approves the State fuel control into the State
Implementation Plan (SIP). EPA has issued guidance describing the criteria for SIP approval of
an otherwise preempted fuel control. See "Guidance on the Use of Opt-in to RFG and Low RVP
Requirements in Ozone SIPs," (August, 1997) at: http://www.epa.gov/otaq/volatility.htm.
The SIP approval process, a notice and comment rulemaking, would also consider a
variety of technical and other issues in determining whether to approve the State fuel control and
what emissions credits to allow. An EPA technical report like this one can be of value in such a
rulemaking, but the SIP rulemaking would need to consider a variety of factors specific to the
area, such as fleet make-up, refueling patterns, program enforcement and any other relevant
factors. Additional evidence on emissions effects that might be available could also be
considered. The determination of emissions credits would be made when the SIP rulemaking is
concluded, after considering all relevant information. While a technical report such as this may
-------
be a factor in such a rulemaking, the technical report is not intended to be a determination of SIP
credits for a State fuel program.
B. Public participation
This technical report was made available to the general public in draft form on June 17,
2002. A comment period was established during which reviewers could submit written
comments on the analyses contained in the draft technical report. This comment period ended on
July 15, 2002. We received comments from the following companies/organizations:
Alliance of Automobile Manufacturers
American Petroleum Institute
American Trucking Association
California Truckers Association
Cummins, Inc.
Defense Energy Support Center
Ethyl Corporation
Flint Hills Resources, LP
Kern Oil & Refining Company
National Petrochemical & Refiners Association
We also requested and received comments from EPA's Office of Research and Development.
We reviewed the comments received from stakeholders and updated this technical report
accordingly. We also submitted a draft of this technical report to two outside experts to obtain
independent peer review. Following the guidelines in EPA's Science Policy Council Handbook
on Peer Review, we generated responses to all the recommendations provided by the peer
reviewers and modified this technical report and the analyses contain herein where it was deemed
appropriate.
Parties interested in the draft technical report, background materials, and support
documents associated with this final technical report can find them at the following Web site:
http://www.epa.gov/otaq/models/analysis.htm
This Web site also contains copies of all comments received from our stakeholders, the reports
generated by the independent peer reviewers, and our responses to the peer reviewers'
recommendations.
-------
II. Introduction
A. Background on cetane analysis
The control of diesel fuel properties as a means for reducing emissions of regulated
pollutants continues to be of interest to parties interested in or charged with reaching various air
quality goals. Previous test programs have shown various levels of benefits for changes to such
fuel properties as cetane number, aromatics, density, sulfur, and distillation properties. For areas
that are out of attainment for ozone or particulate matter, State air quality managers may consider
changes to diesel fuel as one element of their overall strategy for meeting the National Ambient
Air Quality Standards.
One example of a State that has implemented a diesel fuel approach as part of its overall
strategy to reach attainment is Texas. When its Low Emission Diesel (LED) program was first
proposed for Houston and Dallas in the fall of 2000, EPA determined that the NOx emission
reduction benefits claimed by Texas were based on a small amount of outdated data. As a result
we initiated an effort to evaluate the emission benefits of varying diesel fuel parameters. In July
of 2001, we issued a Staff Discussion Document1 with the preliminary results of this analysis.
Our process in conducting this evaluation involved reviewing existing engine emissions
data rather than conducting new emissions tests. Where data was available, we used a regression
model approach to analyze results and to develop a quantitative set of relationships between fuel
parameters and emissions changes (in the remainder of this technical report, this work will be
referred to as the Staff Discussion Document model). As part of our process, we met with
numerous stakeholders to review our preliminary conclusions, beginning in May of 2001, and in
response to requests from stakeholders, held a public workshop on August 28, 2001 to hear
comments on our Staff Discussion Document and our analysis.
After reviewing the comments made at the workshop, we estimated the NOx emission
factors for the Texas diesel fuel program based on this analysis. Our conclusions were
summarized in a memorandum to EPA's Region VI2. In this memorandum, we limited the use of
the draft NOx model presented in the Staff Discussion Document to the evaluation of the benefits
of the Texas diesel fuel program. As a result, there currently exists no widely-applicable, EPA-
approved model for estimating the emission impacts of more general changes in diesel fuel
properties. At this time, EPA has no plans to pursue such a comprehensive model.
B. Cetane number as an emission control strategy
Of the various diesel fuel properties that could be controlled in order to produce emission
benefits, cetane number holds the greatest interest, particularly with regard to NOx. Even in the
absence of a more comprehensive model correlating all diesel fuel properties with emissions,
some States are still considering implementing cetane control programs. In such cases, the
-------
emission benefits might be based on results from various individual test programs that attempted
to control cetane number. Since any NOx benefits claimed in a SIP as a result of cetane control
must eventually be approved by EPA, we have determined that it is now appropriate to
investigate the NOx benefits of cetane control in a comprehensive fashion to preclude the use of
less robust estimates that may misrepresent benefits.
The cetane number of diesel fuel can be increased in two different ways: naturally, and
via the use of additives. The "natural" approach involves the modification of various physical
properties of diesel fuel, via changes to the concentration of various diesel fuel components. The
result is that multiple properties/components may change at the same time. This "colinearity" is
best illustrated for cetane, aromatics, and specific gravity, as shown in Figure II.B-1. The data in
these graphs is drawn from a survey of in-use diesel fuels3.
Figure II.B-1
Colinearities for natural cetane number
0.81 0.82 0.83 0.84 0.85 0.86 0.87 0.88
Specific gravity
0.84 0.85 0.86
Specific gravity
10 15 20 25 30 35 40 45 50
Tota aromatics, vol% by F A
These graphs provide strong evidence that changes in one fuel property will generally be
accompanied by concurrent changes in other fuel properties, confounding attempts to correlate
any one fuel property with emissions.
This colinearity can also be seen in variance inflation factors. Variance inflation factors
-------
are based on the r2 value (coefficient of determination) resulting from a least-squares regression
in which one fuel property is made a function of all other fuel properties. The formula for
variance inflation factors (VIF) is:
VIF = 1 / (1 - r2)
A value close to 1 indicates that no correlation exists. A value higher than 5 indicates a
moderately strong correlation, while values approaching 10 indicate very strong correlations. A
listing of variance inflation factors is shown in Table II.B-1 for the same in-use survey data
shown in Figure HB-1.
Table II.B-1
Variance inflation factors for in-use survey data
Natural cetane
Additized cetane"
Sulfur
Aromatics
T10
T50
T90
Specific gravity
7.1
1.1
1.3
10.0
4.8
11.1
4.2
16.7
"" "Additized cetane" is the increase in total
cetane number brought about through the
addition of cetane improver additives.
From this table it is clear that natural cetane is highly correlated with many other fuel properties.
In addition, the high VIFs for aromatics and specific gravity indicate that these three properties
are highly correlated with one another (the distillation properties are primarily correlated with
each other). However, since fuels in test programs are sometimes the product of more carefully
designed blending that might avoid some of these correlations between fuel properties, the values
in Table n.B-1, based on a survey of in-use fuels, may be misleading. We therefore repeated the
calculation of VIFs for the Staff Discussion Document model database. The result was that the
VIFs were indeed lower than the values in Table II.B-1, but not significantly.
Finally, a correlation matrix is another way to investigate colinearities between fuel
properties. Using our emissions database, we generated a correlation matrix by standardizing all
of the fuel property measurements (by subtracting the mean from every observation and dividing
by the standard deviation), multiplying the fuels matrix by its transpose, and normalizing the
results. The results for the primary fuel properties of interest are shown in Table II.B-2. The
matrix is necessarily identical on either side of the diagonal.
-------
Table II.B-2
Correlation matrix for diesel emissions database
Natural
Cetane
Additized
cetane
Sulfur
Aromatics
T10
T50
T90
Specific
gravity
Natural
Cetane
1
-0.35
-0.04
-0.57
0.16
0.26
0.32
-0.61
Additized
cetane
1
-0.17
0.20
-0.08
-0.07
-0.10
0.25
Sulfur
1
0.30
0.01
0.11
0.10
0.21
Aro-
matics
1
0.13
0.30
0.23
0.75
T10
1
0.69
0.30
0.30
T50
1
0.70
0.41
T90
1
0.23
Specific
gravity
1
"" "Additized cetane" is the increase in total cetane number brought about through the addition of cetane
improver additives.
As the values in the correlation matrix approach 1 (or -1), the correlation between the two fuel
properties in question approaches a perfect linear correlation. Thus the Table n.B-2 values can
be viewed as correlation coefficients for one fuel property as a function of another fuel property.
The highest values are for T50 as a function of the adjacent distillation properties T10 and T90,
and the intercorrelation between natural cetane, aromatics and specific gravity. The correlation
matrix also highlights the fact that colinearity between diesel fuel properties is more pronounced
than for gasoline, likely due to the fact that gasoline is composed of 7-8 blending streams while
diesel fuel is composed of 2-3 streams. For instance, the average of the values in Table II.B-2
(ignoring the diagonal and using the absolute values) is 0.28. By way of comparison, the average
correlation coefficient for the fuels used in developing the Complex Model4 for the reformulated
gasoline program was 0.15. Although the selection of fuel properties of interest is somewhat
arbitrary in both cases, we can conclude that the degree of colinearity between fuel properties in
our diesel emissions database is significantly greater than that in our Complex Model database.
Given the strong colinearities between natural cetane and other fuel properties, it is not
surprising that many stakeholders questioned the draft emissions model described in Section HA.
In particular, the NOx model contained no natural cetane term, despite the fact that many test
programs had shown a strong correlation between natural cetane and NOx emissions. On closer
examination, the absence of a natural cetane term appears to be a result of "aliasing". The NOx
model contained both an aromatics and a specific gravity term, and it is likely that these two
-------
terms were sufficient to describe the combined effect of aromatics, specific gravity, and natural
cetane on NOx emissions. The natural cetane effect on NOx was, therefore, included in the
effects exhibited by aromatics and specific gravity. This concept is supported by the fact that our
estimate of NOx benefits for the Texas diesel fuel program was the same regardless of whether
we used the NOx model proposed in our Staff Discussion Document (the model which did not
contain a natural cetane term) or an alternative model that contained a total cetane term
representing the sum of natural and additized cetane.
The model proposed in our Staff Discussion Document could still be used to predict the
NOx impacts of changes in natural cetane so long as there existed a means for translating natural
cetane changes into the aromatics and specific gravity changes that would likely occur
collinearly. Although currently there is no commonly accepted way to do this, we were fortunate
in the case of the Texas LED program to have survey data for fuels sold in California. Since
Californian diesel fuel was deemed a reasonable representation of fuels that would be produced
under the Texas LED program, we could use them to estimate the benefits of the Texas program.
The result was that the colinearities between natural cetane and other fuel properties were
inherent in the California survey data, and we could place confidence in the resulting predictions
from the Staff Discussion Document model.
There exists an alternative way to estimate the impact of changes in cetane number on
NOx emissions, one that avoids the complication of colinearity between fuel properties. This
approach uses additized cetane as the independent variable instead of natural cetane. Additized
cetane is here defined as increases in the total cetane number brought about through the use of
small quantities (~1 vol% or less) of compounds designed to specifically and solely bring about
this result. Examples of such cetane improver additives include 2-ethylhexylnitrate and di-
tertiary butyl peroxide. Other bulk blending components that could increase the natural cetane
number of conventional diesel fuel but which would generally be added at significantly larger
concentrations, such as biodiesel or Fischer-Tropsch diesel, would not be considered to be cetane
improver additives nor as sources of "additized cetane" for the purposes of this analysis, since
they always affect other fuel properties in addition to cetane number.
Additized cetane is largely uncorrelated with other diesel fuel properties, as shown by the
low variance inflation factor in Table II.B-1. This result is expected since the additives used to
increase cetane generally are used in concentrations of 1 volume percent or less. These low
additive concentrations mean that the other components of diesel fuel are not diluted in any
measurable way. Also, the non-cetane properties of the additives themselves are not so extreme
that physical properties of diesel fuel such as distillation properties or specific gravity are
affected. Table II.B-2 indicates that the correlation coefficient between natural and additized
cetane is -0.35, a value that may appear to be non-negligible. However, this degree of colinearity
is due to human intervention and the need to enhance fuels with poor natural cetane by using
additives to meet minimum cetane requirements. It does not reflect properties inherent in the
fuel makeup as is the case for other fuel property pairs. As described in more detail in Section
in.A below, the colinearity between natural and additized cetane is reduced substantially when
-------
the database is limited to those observations for which both an additized fuel and its associated
unadditized base fuel were tested.
There is good reason to believe that additized cetane and natural cetane describe identical,
or at least similar, combustion mechanisms, since both additized and natural cetane are measures
of a fuel's propensity to auto-ignite. Any differences in NOx impacts between additized and
natural cetane may be related to the aromatics and specific gravity effects that are inherent with
natural cetane. That is, natural cetane increases accompanied by typical reductions in aromatics
and specific gravity might be expected to produce somewhat larger NOx benefits than additized
cetane alone. However, we have insufficient information at this time to quantify any potential
differences between natural and additized cetane effects on NOx, excluding the inherent effects
of aromatics and specific gravity.
As a result of our review of colinearities between diesel fuel properties and our current
understanding of the impacts of increased cetane number on combustion activity, we have
determined that correlating additized cetane with NOx emissions is an appropriate means for
providing inventory impact information to anyone considering the use of higher cetane diesel
fuel. The remainder of this technical report describes the analyses we conducted to investigate
additized cetane effects on NOx emissions.
-------
III. Analytical approach
A. Database preparation
The data that we used for our analysis was a subset of the database used to develop the
Staff Discussion Document models (discussed in Section HA). That database was composed of
three portions covering engine characteristics, emission measurements, and fuel properties. The
database included values for a fuel property titled "CETANE_DIF" which represented the
increase in cetane number resulting from the addition of a cetane improver additive to
conventional diesel fuel. CETANE_DIF is thus the same as "additized cetane" referred to in
Section II. In order to use this database for the analysis of additized cetane effects on NOx
emissions, we simply deleted (blanked out) any CETANE_DIF values in the fuel properties
database for any test fuels that either did not contain a cetane improver additive, or were not the
base fuel to which a cetane improver was added. Base fuels with CETANE_DIF values of zero
were left intact. In the course of curve-fitting using SAS, any fuels with a missing value for
CETANE_DIF were automatically dropped from the analysis. The result was a collection of
fuels whose additized cetane values were essentially uncorrelated with other fuel properties.
Although the correlation coefficients for additized cetane in Table HB-2 were low, the data
subset used for our analysis produced correlation coefficients that were even lower, as shown in
Table in.A-1. A listing of the fuels that were retained for this analysis is given in the Appendix.
Table IE. A-1
Correlation coefficients for additized cetane
Natural Cetane
Additized cetane
Sulfur
Aromatics
T10
T50
T90
Specific gravity
Full database
-0.35
1
-0.17
0.20
-0.08
-0.07
-0.10
0.25
Subset of database used for
additized cetane analysis
-0.17
1
-0.07
0.01
-0.10
-0.12
-0.10
0.04
The final database used in our analysis provided a wide range of measurements for
additized cetane and natural cetane number. Figure ni.A-1 shows the distribution of cetane
values in the database and the lack of correlation between natural and additized cetane.
-------
Figure IE. A-1
Additized cetane versus natural cetane measurements
15
0
m
Additized cet
Ol O
0
3
n n
1 n
D n *
* g
n, "
9 D fc n n _.
w ŁT n
1 1 '
1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
5 40 45 50 55 60
Natural cetane
Although we eliminated all irrelevant fuels from the fuels dataset, doing so did not
guarantee that every engine in the database was tested on both an additized fuel and its
corresponding unadditized base fuel. Therefore, we took additional steps to ensure that every
engine used in the analysis was tested on both an unadditized fuel and an additized fuel. The
result was that all engines from the EPEFE study which received modifications to their injection
timings were excluded. No other engines were excluded for this reason. Note that complete
descriptions of the full database can be found in the July 2001 Staff Discussion Document.
As in the development of the Staff Discussion Document model, we excluded all data
collected on the Japanese 13-mode cycle as being unrepresentative of the federal FTP, since the
overall loads in this steady-state cycle are too low in comparison to the FTP. We also included
all repeat emission measurements (i.e. the same fuel tested on the same engine and cycle multiple
times) in the database without averaging those repeats or limiting their inclusion in the database
to some maximum number of observations.
1.
Representativeness of fuels
The goal of our analysis was to determine the impact on NOx emissions of increases in
cetane number brought about through additives. Generally the amounts of the additive required
is small, one percent or less, so that the impacts on fuel properties other than cetane number is
negligible. Since only one fuel property changes when cetane improver additives are used, we
10
-------
have assumed that the other fuel properties play no role in the effect that increased cetane number
has on emissions. The one exception to this assumption is the base (natural) cetane number,
which was taken into account explicitly in our analysis as described in Section ni.B.
However, to ensure that fuel properties other than cetane would not bias the emission
effects we estimated, we reviewed the composition and properties of the base fuels in the
database to determine the degree to which they could be said to represent in-use fuels. In this
comparison we made use of survey data collected by the Alliance of Automobile Manufacturers
in 2000. We created distributions for each fuel property and placed the results from the base
fuels in our database side-by-side with distributions created from the in-use survey data. The
results are shown in Figures in.A. 1-1 through in.A. 1-6.
Figure IE. A. 1-1
Comparison of base fuels from database to in-use fuels for sulfur content
30.0%
25.0%
20.0%
0)
o 15.0%
o>
°~
10.0%
5.0%
0.0%
90% of database
90% of in-use surveys
IL
Database
AAM survey
0 50 100 150 200 250 300 350 400 450 500 550 600
25 75 125 175 225 275 325 375 425 475 525 575
Sulfur, ppm
11
-------
Figure IE. A. 1-2
Comparison of base fuels from database to in-use fuels for natural cetane
40.0%
g
U-*
ro
30.0%
o 20.0%
10.0%
0.0%
90% of database
90% of in-use surveys
J
Database
AAM survey
36 38 40 42 44 46 48 50 52 54 56 58 60
Natural cetane
Figure IE. A. 1-3
Comparison of base fuels from database to in-use fuels for aromatics content
30.0%
25.0%
o
| 20.0%
1 15.0%
M
O
Ł 10.0%
°~ 5.0%
OfW
.
k.
^ 90% of database ^
"^ 90% of in-use surveys w'
n n n
LtiuUra
r
12 16 20 24 28 32 36
14 18 22 26 30 34
r
~~
n
40
38
42
I
44
^b^
PI Database
D AAM survey
48 52
46 50
Aromatics, vol%
12
-------
Figure IE. A. 1-4
Comparison of base fuels from database to in-use fuels for specific gravity
40.0%
| 30.0%
ro
1 20.0%
M
O
c
fc 10.0%
CL
0.0%
^ ^
^ 90% of database **
90% of in-use surveys
, JJ*
0.8 0.81 0.82 0.83 0.84
I
1
1
1
0.85
-,
\
rn
\ 1
0.86
0.805 0.815 0.825 0.835 0.845 0.855
-,
I
PI Database
D AAM survey
0.87 0.88
0.865
0.875 0.885
Specific gravity
Figure IE. A. 1-5
Comparison of base fuels from database to in-use fuels for T50
40.0%
)
c
g
^4-*
(0
20.0%
10.0%
0.0%
90% of database
90% of in-use surveys
[ffL^
Database
AAM survey
410 430 450 470 490 510 530 550 570 590
420 440 460 480 500 520 540 560 580
T50, OF
13
-------
Figure IE. A. 1-6
Comparison of base fuels from database to in-use fuels for T90
20.0%
15.0%
"ro
-g 10.0%
M
O
fc 5.0%
0.0%
90% of database
90% of in-use surveys
Database
AAM survey
550 560 570 580 590 600 610 620 630 640 650
555 565 575 585 595 605 615 625 635 645
T90, OF
With the possible exception of aromatics, the range of property values for base fuels in the
database encompases the range of values for in-use fuels. This result ensures that NOx emission
effects that we estimate for additized cetane will not be unduly influenced by other fuel
properties, and thus can be used to represent the NOx emission effects that would be expected in
the field. For aromatics, the database range substantially overlaps the in-use survey range, and
extends to lower aromatics levels which may be more representative of clean diesel fuel such as
that required in California. Thus we believe that our database is fully representative of in-use
fuels.
2. Representativeness of engines
The degree to which we can apply the results of our analysis to the in-use fleet of heavy-
duty engines also depends on the degree to which engines in our database are representative of
in-use engines. To evaluate this issue, we focused on the distribution of model years among
engines in our database. Since we excluded all 2-stroke and EGR-equipped engines as described
in Section in.B below, all engines used in our analysis were built in the 1990's. Figure in.A.2-1
shows the distribution of model years for these engines.
14
-------
Figure IH.A.2-1
Distribution of engine model years in the database
35%
30%
) 25%
0
c
'o>
5 20%
"o
Ł 15%
0)
a
>
O>O>O>O>O>O>O>O>O>O>O>O>O>O>O>O>
Model year
fleet
, i
o Ť- cvj
C
3
o
/Vl
C
3
O
rvi
C
)
O
/Vl
I
CO
O
o
/Vl
(numbers above bars represent the actual number of engines in the database)
The distribution of model years is more broad for the in-use 2003 fleet than for the
database. However, the current in-use fleet is comprised primarily of engines built in the 1990's,
consistent with the database. Also, the average model year for the database is 1994, compared to
an average of 1996 for the in-use 2003 fleet. We do not consider this difference to be significant
in the context of estimating the impact of cetane on NOx emissions. Finally, engines built after
1999 will not constitute a majority of the highway engine NOx inventory for several more years.
Thus we have determined that the distribution of model years in our database is sufficient to
represent the current fleet for the purposes of our analysis.
As time goes on and the fleet turns over, however, it is possible that the emission effects
estimated in our analysis will become less representative of effects exhibited by the in-use fleet.
One step we have taken to account for this possibility is to assume that all engines built after
2002 will be equipped with EGR and will, as a result, exhibit no benefits from higher cetane (see
Section IV.C for details). This is a conservative assumption, since there are indications that
some manufacturers will not use EGR in their 2002+ engines. For those manufacturers that do
use EGR, early indications are that EGR will continue to be used on engines built to meet the
2007 standards.
We also investigated the degree to which engines built in the later 1990's were less
responsive to higher cetane than engines built in the early 1990's. There has been some limited
15
-------
information suggesting that the more frequent use of injection rate-shaping and diffusion burning
could mute the impact that cetane has on NOx emissions by reducing the amount of pre-mix
burning and thus raising the adiabatic flame temperature. To do this analysis, we calculated
engine-specific NOx effects for every unique pair of base fuel and its additized counterpart. For
each engine in the database, the average % change in NOx was calculated from all repeat
emission measurements on a given base fuel and its associated additized counterpart. The
average % change in NOx was then divided by the change in cetane brought about through
additives to produce an estimate of the % reduction in NOx for a unit increase in cetane number.
We then plotted these effects by model year. To determine if engines were becoming less
responsive to higher cetane over the course of the last decade, we conducted a least-squares
regression on these values, with model year as the independent variable. We also conducted a
least-squares regression that included dummy variables for the engines. The results are shown in
Figure m.A.2-2.
Figure IH.A.2-2
Effects of increased cetane on NOx by model year
0 2.50%
5
0 2.00%
O>
C 1.50%
(0
.E
o
^ 1.00%
0
.E 0.50%
g
o 0.00%
Ł
go -0.50%
-1 00%
1Ł
90
H
H
3
1992
,
!
=
|
|
19!
I
14 1
Model year
j
1
D
9
Average
1
1
F
i i i i
36 1998 20
00
The graph shows that the least-squares regressions result in lines that are very similar to the flat
average. Although the slope was negative for both regression lines, suggesting that more recent
engines may indeed exhibit less NOx sensitivity to cetane than older engines, in both cases the
curve was not statistically significant at the p = 0.10 level. The regression coefficients and p-
values are shown in Table ni.A.2-1
16
-------
Table IH.A.2-1
Correlations between model year and % reduction in NOx/change in cetane
Overall regression
Regression including dummy
variables for engines
Coefficient
-0.000209
-0.000062
p-value
0.11
0.35
Based on this investigation, it is reasonable to assume that the effects of cetane on NOx are the
same for all engines built in the 1990's. In addition, since we do not have conclusive data to the
contrary, it appears reasonable to assume that these effects are also representative of model years
1999-2002.
B. Summary of analysi s
In correlating additized cetane with NOx emissions, we generally followed the approach
described in our July 2001 Staff Discussion Document. This included using the procedure "proc
mix" in the statistical analysis package SAS to permit the simultaneous treatment of cetane
number as a continuous independent variable and engines as random effects. We included all
repeat emission measurements in the database without averaging those repeats or limiting their
inclusion in the database to some maximum number of observations. The SAS curve-fitting
procedure "proc mix" treats these repeat observations in a manner that precludes them from
overweighting the results.
We chose to use the natural logarithm of NOx emissions to mitigate the heteroscedastic
nature of the NOx measurements. We also standardized the independent variables by subtracting
the mean and dividing by the standard deviation. The means and standard deviations are shown
in Table UI.B-1. Standardization removes the scale differences between fuel terms, and also
reduces some of the colinearity between first and second-order terms.
Table HI.B-1
Means and standard deviations used for standardizing independent variables
Additized cetane
Natural cetane
Mean
5.03963
45.13889
Standard deviation
4.94910
4.27954
In our earlier work, we found that technology groups B and L produced different
cetane/NOx relationships than other engine technology groups (see Table UI.B.3-2 in the Staff
Discussion Document). Technology group B represents 2-stroke engines, while technology
group L represents engines equipped with exhaust gas recirculation (EGR). For our analysis of
17
-------
additized cetane effects, we chose to exclude all technology group B data. Two-stroke engines
currently represent approximately 1 percent of the heavy-duty highway fleet5, and are expected to
become an even smaller part of the fleet in the future. As a result we do not believe that
excluding the group B data from our analysis materially affected the applicability of the results to
the in-use fleet.
We also chose to exclude all group L data from our analysis. There are essentially no
EGR-equipped engines in the current fleet, but they are expected to become a significant portion
of the fleet over the next decade. The Heavy-Duty Engines Workgroup6 is our primary source for
data on the effects of additized cetane on an EGR-equipped engine, and that data suggests that
these engines exhibit no discernable NOx response to cetane. The July 2001 Staff Discussion
Document did conclude that a group L model adjustment term was statistically significant, but its
magnitude was very similar to the overall term and of opposite sign. The result was that the
predicted effect of cetane on NOx for group L engines was nearly zero (the correlation predicted
a very small increase in NOx for EGR-equipped engines). However, the adjustment term was
significant only because EGR-equipped engines could not be assigned the same NOx effect as
non-EGR engines. When we attempted to generate an independent model for EGR-equipped
engines, the effect on NOx was not significant, consistent with the findings of the Heavy-Duty
Engines Workgroup. Therefore, the NOx impacts of additized cetane that resulted from the
analyses described below are expected to apply to the entire fleet except EGR-equipped engines,
as described more fully in Section IV.C.
Engines were identified in proc_mix as categorical random variables in our modeling.
This is equivalent to specifying dummy variables for each engine in a fixed model, except that
tested engines are treated as a random sample of engines drawn from the full population of
engines in the fleet. We also recognized and accounted for two other types of random effects in
our modeling. The first is the cetane/NOx relationship that is specific to every test engine, and
the second is the effect of the unadditized base fuel on NOx emissions for each engine. By
controlling for these types of random effects, we believe that the overall estimated effect of
additized cetane on NOx can be confidently applied to the in-use fleet.
The primary independent variable included as a fixed effect in our model was additized
cetane, defined as the increase in cetane number brought about through the addition of a cetane
improver to conventional diesel fuel. We made no distinction between different types of cetane
improver additives since we were not concerned with the effectiveness of a given additive in
terms of cetane increase per unit concentration of additive. Instead we treated a given increase in
cetane number as having the same effect on NOx regardless of the specific additive used to bring
about that cetane number increase. This approach is consistent with conclusions reached in
several previous studies7'8'9. To account for potential nonlinear effects we also included a
squared additized cetane term.
Based on past studies of additized cetane effects on emissions, we had some evidence that
increases in cetane brought about through the use of additives produced diminishing NOx
18
-------
impacts as the base (natural) cetane number of the diesel fuel increased. Thus, for instance, an
increase in cetane number from 45 to 50 might be expected to produce larger NOx impacts than
an increase in cetane number from 50 to 55. To account for this possibility, we introduced a term
into the model that represented the interaction of additized cetane and natural cetane number.
Figure IHA-1 shows that there is good separation between additized cetane values and natural
cetane values, i.e. no correlation exists between the two, which is an important prerequisite for
investigating interactive terms. Since the inclusion of an interactive term meant that the natural
cetane number was now represented in the model, we decided to also investigate the need for
linear and squared natural cetane number terms. The complete list of five terms investigated in
this analysis are given in Table ni.B-2.
Table HI.B-2
Independent variables investigated
in correlation between cetane number and NOx
Additized cetane
Additized cetane2
Natural cetane
Natural cetane2
Additized cetane * natural cetane
Only those terms that were statistically significant at the p = 0.05 level were retained
using a backwards stepwise approach. Once all the remaining terms were statistically significant,
we identified outliers as any data points whose residual exceeded four standard deviations from
the predicted effect, removed them, and regenerated the model.
19
-------
IV. Conclusions
A. SAS modeling results
The initial modeling run indicated that the squared natural cetane number term was not
statistically significant. After it was dropped and the model regenerated, the remaining terms
were all significant. Four outliers were then identified out of 540 total observations. These
outliers are listed in Table IV. A-1 by the labels used in the database.
Table IV. A-1
Outliers excluded from final model
Study
SAE902173
SAE902173
SAE902173
SAE902173
Engine
902173-1
902173-1
902173-1
902173-1
Fuel
A3
A3
B2
Dl
NOx, g/bhp-hr
3.66
3.67
4.38
4.63
Once the outliers were excluded, the model was regenerated a final time. The coefficients for the
standardized variables and the associated P-values are given in Table IV. A-2.
Table IV.A-2
SAS proc mix output for final model
Variable
Intercept
Additized cetane
Additized cetane 2
Natural cetane
Additized cetane x natural cetane
Coefficient
1.5060
-0.01677
0.004139
-0.02093
0.004720
P-value
< 0.0001
< 0.0001
< 0.0001
0.0057
< 0.0001
Using the mean and standard deviation for the standardized independent variables (Table ni.B-
1), we converted the coefficients back into unstandardized form. The fixed effects portion of the
resulting model is shown below:
20
-------
ln(NOx, g/bhp-hr) = 1.79883 (1)
- 0.015151 x (additized cetane)
+ 0.000169 x (additized cetane)2
- 0.006014 x (natural cetane)
+ 0.000223 x (additized cetane) x (natural cetane)
We can convert this equation into one that provides a percent change in NOx emissions as a
function of additized cetane and natural cetane. During this conversion, the natural cetane term
drops out since natural cetane is the same for the base fuel and the additized fuel. The constant
also drops out for the same reason. If the base fuel is assumed to contain no cetane improver
additives, the resulting equation is:
% change in NOx = (exp[ - 0.015151 x (additized cetane) (2)
+ 0.000169 x (additized cetane)2
+ 0.000223 x (additized cetane) x (natural cetane) ] - 1 } x 100%
The predicted NOx impacts are shown graphically in Figure IV. A-1.
21
-------
Figure IV. A-1
Predicted effect of additized cetane on NOx for all heavy-duty highway engines
except 2-strokes and those equipped with EGR
6%
x 5%
O
Ł 4%
o
(U
cu
Q_
3%
o3 2%
1%
0%
5 10 15
Increase in cetane number due to additives
The predicted NOx impact of a given increase in cetane number brought about through
the use of additives diminishes as the natural cetane increases, consistent with expectations. If
the average natural cetane number in a particular area happens to be close to 40, the use of cetane
improver additives will have a substantially larger benefit than would be produced in areas with
an average natural cetane number close to 50. Thus regional differences in base fuel properties
can and should be taken into account when using equation (2). There is also a possibility that
natural cetane numbers nationwide will increase with the introduction of ultra-low sulfur
highway diesel fuel in 2006, since the hydrogenation typically used to remove sulfur tends also to
increase natural cetane. Although we do not have the means at present to quantify this effect,
early indications are that any increase in natural cetane number will be quite small.
We note that there are certain conditions under which a "turnover" appears in the
predicted effects in Figure IV. A-1. For instance, when the natural cetane is 50, the slope of the
additized cetane curve changes from positive to negative at a value of approximately 11.8 . We
do not believe that these turnovers represent real impacts of additized cetane on NOx emissions,
but rather are artifacts of the squared additized cetane term that we used to represent nonlinear
22
-------
effects. We believe it would be appropriate to insert a flat-line extrapolation at the point of any
turnover, so that additional increases in cetane number brought about through additives would
cause no additional changes in NOx emissions. To do this, we derived a formula that identifies
the location of all turnovers as a function of the natural cetane number. This formula is shown
below:
Turnover in additized cetane = 44.83 - 0.6598 x (natural cetane) (3)
Thus for any values of the additized cetane that are larger than the value calculated from equation
(3), the predicted NOx impact should be the value calculated from equation (2) using the
additized cetane value calculated using equation (3). In practice, however, we do not believe that
this flat-line extrapolation will be necessary. Cetane improver additives are rarely used to
increase the cetane number of fuels that already have natural cetane numbers above 55, and even
in these cases the additive would have to add more than 8 cetane numbers (for a total cetane
number of more than 63) before the turnover would be reached and the extrapolation would be
necessary. Far more common is for cetane improver additives to be added to fuels having a
natural cetane number in the range of 40 - 50, and then only to raise the total cetane number 5 -
10 numbers. Within these ranges, no turnovers are encountered.
B. Confirmation of explanatory power
Although we did not have an independent set of emissions data with which to validate the
model due to the fact that we used all the available data to construct the model, we were able to
confirm the explanatory power of the correlations described in Section IV. A using two
alternative approaches. First, we compared equation (2) with regression equations generated by
other researchers. A review of the literature produced seven different regression models from
four different studies, each of which included either an additized cetane term or a total cetane
(natural + additized) term. For natural cetane and other fuel properties present as independent
variables in some of the models, we used the nationwide average values given in Table ni.D-1 of
the July 2001 Staff Discussion Document. Figure IV.B-1 presents this comparison.
23
-------
Figure IV.B-1
Comparison of EPA correlation to those from other researchers
X
O
g
's
CD
CD
Q_
8.0%
7.0%
6.0%
5.0%
4.0%
3.0%
2.0%
1.0%
0.0%
0
*
5 10 15
Increase in cetane number due to additives
SAE 950251
SAE 950250, 5g
SAE 950250, 4g
A VE-10, 5gDDC60
T VE-10, 4g DDC60
-O- SAE 922267, 1994
* SAE 922267, 1991
~ EPA analysis
Our correlation falls in the middle of the range of effects estimated via regression analysis from
other researchers. In addition, our correlation predicts that cetane improver additives proOduce
less NOx benefit as the cetane number increases. Most other models are strictly linear,
suggesting that our correlation provides a more conservative prediction at higher values of
additized cetane.
We also compared predictions from our model to the individual observations in our
database. To do this, we first had to convert the emission measurements in the database into %
change values. Since repeat emission measurements sometimes made it difficult to match
specific base fuel measurements with specific additized fuel measurements, we first averaged any
repeat emission measurements of a given fuel on a given engine. We then calculated a % change
in NOx value for every unique pair of base fuel and its additized counterpart. Equation (2) was
then used to predict a % change value for each fuel pair. Finally, the predicted and observed
values were plotted against one another. The results are shown in Figure IV.B-2.
24
-------
Figure IV.B-2
Comparison of predicted and observed impacts of additized cetane
15.00%
-5.00%
-5.00% 0.00% 5.00% 10.00%
Predicted % reduction in NOx
15.00%
If the model perfectly predicted all the data, all the observations would fall exactly on the
diagonal. The fact that the observations are distributed around the diagonal can be explained by
measurement variability. For instance, 9% of the observations in the database indicated that
NOx emissions actually increased when a cetane improver additive was added to a base fuel (the
observations below the horizontal line in Figure IV.B-2). But this same result could be predicted
from a knowledge of NOx measurement variability and an expected cetane improver NOx
benefit of 2%l. These NOx increases do not change the fact that the fleet-wide effect of cetane
improver additives is to lower NOx in a statistically significant way.
In addition, a least-squares regression on the data in Figure IV.B-2 does produce a slope
that is close to 1.0 (actual value was 0.86, with an associated p-value of <0.0001). This
regression line, along with the 90% confidence interval around the regression line, is shown in
Figure IV.B-3. Note that the 1:1 diagonal falls entirely within the confidence interval, providing
another indication that equation (2) provides a good explanation of the observed values.
We assumed a NOx measurement standard deviation of 0.037 g/bhp-hr based on repeat measurements
presented in the July 2001 Staff Discussion Document, and a mean NOx emission level of 3.7 g/bhp-hr, representing a
1997 engine. Under these conditions, a large sample of emission measurements on additized and unadditized fuels would
result in approximately 91% of paired observations exhibiting a reduction in NOx and 9% of observations exhibiting an
increase in NOx.
25
-------
Figure IV.B-3
Least-squares regression of predicted versus observed impacts
of additized cetane, including 90% confidence limits
15.00%
-5.00%
-5.00%
0.00% 5.00% 10.00%
Predicted % reduction in NOx
15.00%
For multi-parameter regression equations of the sort generated by our maximum
likelihood curve-fitting approach, there is no straightforward mechanism for combining term-
specific standard error estimates into confidence limits around equation (2). The SAS procedure
"proc mix" employed in our analysis does have the capability of generating confidence intervals
around every observation in the database, but this falls short of confidence limits around the
regression equation itself. In order to generate a quantitative measure of uncertainty for equation
(2), we used the predicted and observed % reduction values to calculate the standard error
associated with the residuals, where the residuals were simply the difference between predicted
and observed values for every observation. These values were calculated for several different
ranges of values for natural cetane and additized cetane. Although this approach provides only
an estimate of uncertainty in the regression equation at several discrete points on the curve, it is
useful to illustrate the potential uncertainty in the predicted effects. The results are shown in
Table IV.B-1.
26
-------
Table IV.B-1
Standard errors and number of observations for % change residuals
Natural cetane range
37.5 to 42.5
42.5 to 47.5
47.5 to 52.5
Median = 40
Median = 45
Median = 50
Additized cetane range
2.5 to 7.5
Median = 5
0.357 (35)a
0.553 (21)
0.269 (44)
7.5 to 12.5
Median= 10
0.271 (43)
0.310(72)
0.285 (54)
12.5 to 17.5
Median =15
0.468 (14)
0.267 (24)
1.722(2)
a Values in parentheses are the number of observations
The standard errors for the residuals were then converted into confidence intervals by multiplying
the standard errors by the appropriate value oft (from Student's t-distribution):
Confidence interval = mean predicted effect ą t * standard error
Table IV.B-2 presents the t * standard error values assuming a 90% confidence band, while
Figure IV.B-4 shows how the confidence intervals would appear for the case of a base fuel with
natural cetane of 45.
Table IV.B-2
90% t x standard error values about predicted % reduction values
Natural cetane range
37.5 to 42.5
42.5 to 47.5
47.5 to 52.5
Median = 40
Median = 45
Median = 50
Additized cetane range
2.5 to 7.5
Median = 5
0.60
0.95
0.45
7.5 to 12.5
Median= 10
0.46
0.52
0.48
12.5 to 17.5
Median =15
0.83
0.46
10.87
27
-------
Figure IV.B-4
Predicted effect of additized cetane on NOx with 90% confidence intervals
for base fuel with natural cetane of 45
0 5 10 15
Increase in cetane number due to additives
C. Application to the in-use fleet
As described in Section HIE, we excluded 2-stroke and EGR-equipped engines from our
analysis. As a result, equation (2) does not predict NOx effects for these types of engines. For 2-
stroke engines we do not believe that this result presents a hindrance to the application of
equation (2) to the in-use fleet. There were no heavy-duty 2-stroke diesel engines certified for
highway use for model years 1998 and 1999, and we expect that this trend will continue in the
future. As described in Section in.B, 2-stroke diesel engines currently account for approximately
1% of the heavy-duty highway fleet, so the application of equation (2) to the entire (non-EGR)
highway fleet for future years should introduce only negligible error.
Engines equipped with EGR, however, are expected to become an increasingly important
part of the highway fleet beginning this year. EGR-equipped engines are expected to exhibit no
discernable NOx response to cetane, based on testing done by the Heavy-Duty Engines
Workgroup under the auspices of the Mobile Source Technical Review Subcommittee. This test
program did not include an evaluation of the impacts of injection rate-shaping or other
combustion management techniques that some contend are the primary, or possibly additional,
reasons for the cetane-insensitivity of future engines. However, for our purposes it is not
necessary to identify the specific reasons for cetane-insensitivity of future engines if we make the
simplifying assumption that all future engines will employ cetane-insensitive technologies.
28
-------
Thus our approach to estimating fleet-wide NOx effects of cetane improver additives was to use
a weighted combination of equation (2), representing engines without cetane insensitive
technologies, and the zero effect attributable to cetane-insensitive engines. Because the relative
NOx inventories between these two categories of engines change over time, these weighting
factors would be dependent on calendar year.
To estimate these weighting factors, one might use NOx inventories that represent the
specific areas where the cetane improver additives are intended to be used. For the purposes of
generating example weighting factors, we used the nationwide inventory modeling done in the
context of our rulemaking setting new standards for heavy-duty engines beginning in 2007 [66
FR 5002]. Using this modeling, we determined how the NOx inventory will be distributed
among the various model years in the nationwide fleet10. We assumed that all 2003 and later
heavy-duty diesel engines will employ EGR or some other cetane-insensitive technology despite
the fact that this may be not true for all manufacturers. Lacking a robust means for estimating the
fraction of new engine sales that will employ cetane-insensitive technologies in the future, this
assumption assures that we are not overestimating NOx benefits of cetane control via additives
for future years. From this information we were able to estimate the fraction of the NOx
inventory that derived from cetane-sensitive engines for any calendar year. These fractions are
given in Table IV.C-1.
29
-------
Table IV.C-1
Potential yearly weighting factors 'k' for additized cetane model
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
Fraction of diesel highway NOx inventory
which comes from cetane-sensitive engines
0.93
0.84
0.77
0.70
0.65
0.61
0.57
0.55
0.54
0.53
0.51
0.50
0.48
0.46
0.44
0.41
0.39
0.36
Using the values in Table IV.C-1, equation (2) could be modified to represent the entire in-use
fleet of heavy-duty diesel highway engines. The result is shown below:
% change in NOx =
k x 100% x (exp[
- 0.015151 x (additized cetane)
+ 0.000169 x (additized cetane)2
+ 0.000223 x (additized cetane) >
(4)
(natural cetane) ] - 1 }
There were no nonroad engines in the database we used to evaluate the effects of
additized cetane on NOx emissions. However, most nonroad engines (excluding locomotive
engines) use technologies similar to those found in highway engines, although in a given year the
highway vehicle technology is generally more advanced. Since our previous modeling showed
few technology-specific effects of cetane on NOx, and many of those distinctions have been
accounted for in our current analysis by excluding technology groups B and L, any differences
between highway and nonroad technology may not be important for additized cetane effects on
NOx. As a result, it might be appropriate to apply equation (2) to heavy-duty nonroad engines,
excluding locomotives. This approach is consistent with the conclusion drawn by the American
Petroleum Institute in a letter to the Ozone Transport Commission11, in which API concluded that
nonroad engines would exhibit NOx responses to cetane that were "similar" to highway engine
30
-------
effects. However, we caution that there exists no robust set of data to validate the use of
equation (2) for nonroad, and the concerns we raised in Section VHB.6 of our Staff Discussion
Document regarding this type of extrapolation have not yet been fully addressed. Thus the
decision to apply equation (2) to nonroad in any particular context may be dependent on the
availability of supporting data or other relevant factors.
Finally, we discussed in Section II.B how natural and additized cetane are related, and the
fact that additized cetane is more easily analyzed since it can be disassociated from changes in
other fuel properties. Based on our current understanding of diesel ignition properties, natural
and additized cetane likely represent overlapping effects on combustion and thus on NOx.
Although we do not at this time have sufficient information to quantify the differences in NOx
effects between natural and additized cetane, preliminary analyses suggest that changes in natural
cetane, if accompanied by the collinear changes in aromatics and specific gravity shown in
Figure II.B-1, would produce larger NOx benefits than equivalent changes in additized cetane. If
so, then equation (4) would provide environmentally conservative predictions of changes in NOx
due to changes in natural cetane. Equation (4) can be modified for application to changes in
natural cetane to produce the following:
% change in NOx = (5)
k x 100% x (exp[ - 0.015151 x (NATCETf - NATCET;)
+ 0.000169 x (NATCETf-NATCET;)2
+ 0.000223 x (NATCETf - NATCET;) x (NATCET;) ] - 1 }
where
k = Factor from Table IV.C-1 representing engines without EGR
NATCET; = Initial value of natural cetane number
NATCETf = Final value of natural cetane number
Equation (5) would be relevant for changes in natural cetane brought about through conventional
means, namely lowering aromatics content or other refinery-based changes to the composition of
the fuel. It would not be appropriate to use equation (5) to represent changes in natural cetane
brought about through the additional of high cetane bulk blending components, such as biodiesel
or Fischer-Tropsch fuels, to conventional diesel fuel.
The only alternatives to equation (5) that we considered for correlating natural cetane
with NOx emissions were those in the Staff Discussion Document. Since that analysis showed
that the effects of natural cetane on NOx could actually be better represented by aromatics and
density alone, equation (5) permits one to directly predict the NOx impact of changes in natural
cetane instead of inferring the impacts through changes in aromatics and density.
D. Some predicted NOx impacts of additized cetane
31
-------
The predicted NOx impact of a given change in cetane number is a function of both the
calendar year (Table IV.C-1) and the natural (or initial) cetane number of the base fuel. We can
choose some representative years and base fuel cetane values to predict specific NOx impacts
using equations (2) and (4). For instance, one of the primary years in which many current non-
attainment areas must show attainment with the ozone standard is 2007. Thus we have made
NOx predictions both for the next full calendar year 2003 and for 2007. In this example we also
used the current national average cetane number to represent the base fuel for areas that have not
implemented a clean diesel fuel program. According to survey data collected by the Alliance of
Automobile Manufacturers, the current average cetane number is approximately 45. If we
wanted to raise the cetane number of such base fuels to 50, equations (2) and (4) would predict
the NOx impacts shown in Table IV.D-1.
Table IV.D-1
Examples of predicted NOx effects (% reduction in NOx)
Cetane number increased 5 numbers to 50
National average base fuel assumed
Highway engines
Nonroad engines
2003
2.0
2.1
2007
1.4
2.1
The modest NOx reductions predicted by equation (4) suggest that cetane improver
additives are, by themselves, an unlikely candidate for producing NOx emission reductions that
are equivalent to those produced for clean diesel fuel in California or other areas with California-
like fuel. For instance, the July 2001 Staff Discussion Document concluded that California
diesel fuel produces NOx reductions of approximately 6.2 percent (see Table UI.F-2 in the Staff
Discussion Document). Assuming a natural cetane number of 45 as the base, Figure IV. A-1
shows that the NOx benefits of cetane improver additives could not exceed 4 percent even for
large concentrations of additive. In fact, cetane improver additives are already used in California
by many refiners under the equivalent formulation provision of that state's clean fuel regulations.
On average, cetane improver additives are responsible for approximately one-third of the NOx
reductions generated by California diesel fuel, with the remaining NOx reductions being
generated by higher natural cetane, lower aromatics, and lower density.
32
-------
Appendix
Database used in additized cetane analysis
All studies are listed by their database STUDY_ID label. See Appendix A of Staff Discussion
Document for full citations for studies that comprise the full database.
Studies that contained no additized fuels are were therefore eliminated from
the additized
cetane analysis
ACEA SAE 852078 SAE 942053
CARB-LOCO SAE 88 11 73 SAE 96 1973
C ARB-TOXIC SAE 9222 1 4 SAE 96 1 974
SAE 1999-01-1 117 SAE 932685 SAE 971635
SAE 1999-01-3606 SAE 932731 SAE 972898
SAE 2000-0 1-2890 SAE 932734 VE-1 PHASE I
SAE 790490 SAE 932800
Studies/fuels included in the additized cetane analysis
FBATCH_ID
EPD1
EPD10
EPD11
EPD2
EPD3
EPD4
EPD5
EPD6
EPD7
EPD8
EPD9
HDE-
HDE-
HDE-
HDE-
HDE-
HDE-
HDE-
HDE-
HDE-
HDE-
HDE-
HDE-
HDE-
HDE-
HDE-
10N
11
12
14N
15
16
16N
18
IN
2
3
4N
5
6
7N
STUDY_
EPEFE
EPEFE
EPEFE
EPEFE
EPEFE
EPEFE
EPEFE
EPEFE
EPEFE
EPEFE
EPEFE
HDEWG
HDEWG
HDEWG
HDEWG
HDEWG
HDEWG
HDEWG
HDEWG
HDEWG
HDEWG
HDEWG
HDEWG
HDEWG
HDEWG
HDEWG
_ID CETANE_NUM CETANE_DIF BASE
51
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
58
57
50
50
50
50
50
49
54
59
42
48
52
42
47
52
53
47
42
48
53
42
47
53
42
1
2
3
6
2
5
8
1
3
1
7
1
9
2
4
9
8
2
2
7
8
7.6
0
0
4.5
8.8
0
5.8
10.4
0
5.8
10.1
0
5.2
10.4
0
5.3
10.6
0
EPD7
BASE
BASE
EPD4
EPD4
BASE
HDE-
HDE-
BASE
HDE-
HDE-
BASE
HDE-
HDE-
BASE
HDE-
HDE-
BASE
FUEL
ION
ION
14N
14N
IN
IN
4N
4N
33
-------
HDE-8
HDE-8N
HDE-9
HDE-R
FUEL1
FUEL1A
FUEL1B
FUEL1C
FUEL1D
FUEL1E
FUEL1F
FUEL1G
FUEL1H
FUEL1I
FUEL1J
FUEL2
FUEL2A
FUEL2B
FUEL2C
FUEL2D
FUEL2E
FUEL2F
FUEL2G
FUEL2H
FUEL2I
FUEL2J
0
1
2
2A
2B
2S
4
4B
5
5B
6
Al
A2
A3
A4
Bl
B2
B3
B4
B5
B6
HDEWG II
HDEWG II
HDEWG II
HDEWG II
SAE1999-01
SAE1999-01
SAE1999-01
SAE1999-01
SAE1999-01
SAE1999-01
SAE1999-01
SAE1999-01
SAE1999-01
SAE1999-01
SAE1999-01
SAE1999-01
SAE1999-01
SAE1999-01
SAE1999-01
SAE1999-01
SAE1999-01
SAE1999-01
SAE1999-01
SAE1999-01
SAE1999-01
SAE1999-01
SAE902172
SAE902172
SAE902172
SAE902172
SAE902172
SAE902172
SAE902172
SAE902172
SAE902172
SAE902172
SAE902172
SAE902173
SAE902173
SAE902173
SAE902173
SAE902173
SAE902173
SAE902173
SAE902173
SAE902173
SAE902173
-1478
-1478
-1478
-1478
-1478
-1478
-1478
-1478
-1478
-1478
-1478
-1478
-1478
-1478
-1478
-1478
-1478
-1478
-1478
-1478
-1478
-1478
48
48
52
46
39
42
43
45
47
51
41
42
45
47
51
46
48
51
54
59
60
48
51
54
58
61
42
39
39
47
55
39
46
61
48
67
51
45
49
51
55
39
46
48
48
50
56
1
6
9
7
1
2
8
9
1
6
5
7
9
1
3
9
6
6
1
5
3
1
9
4
2
5
9
6
1
5
6
4
9
6
8
3
8
9
6
2
3
7
5.3
9.8
0
2.4
3.5
6.1
8.2
11.4
1.9
2.8
6
8.2
11.4
0
2.6
5.3
8.3
12.8
14 .2
2
4.8
8.6
12.1
14 . 9
0
7.5
15.9
0
15.5
0
18 .4
0
4.5
5.7
10.6
0
6.4
8.4
8.6
10 .7
17.1
HDE-7N
HDE-7N
BASE
FUEL1
FUEL1
FUEL1
FUEL1
FUEL1
FUEL1
FUEL1
FUEL1
FUEL1
FUEL1
BASE
FUEL2
FUEL2
FUEL2
FUEL2
FUEL2
FUEL2
FUEL2
FUEL2
FUEL2
FUEL2
BASE
2
2
BASE
4
BASE
5
BASE
Al
Al
Al
BASE
Bl
Bl
Bl
Bl
Bl
34
-------
Cl
C2
C3
C4
Dl
D2
D3
D4
Cl
C2
C2I
C2S
CR
DD10
DD11
DD12
DD4
DD5
DD8
DD9
A
B
C
D
E
F
G
H
I
J
K
L
LS
LS-N
LS-P
A
A-DTBP
A-EHN
B
B-DTBP
B-EHN
C
C-DTBP
C-EHN
D
D-EHN
D-EHN/DTBP
SAE902173
SAE902173
SAE902173
SAE902173
SAE902173
SAE902173
SAE902173
SAE902173
SAE910735
SAE910735
SAE910735
SAE910735
SAE910735
SAE912425
SAE912425
SAE912425
SAE912425
SAE912425
SAE912425
SAE912425
SAE922267
SAE922267
SAE922267
SAE922267
SAE922267
SAE922267
SAE922267
SAE922267
SAE922267
SAE922267
SAE922267
SAE922267
SAE932767
SAE932767
SAE932767
SAE942019
SAE942019
SAE942019
SAE942019
SAE942019
SAE942019
SAE942019
SAE942019
SAE942019
SAE942019
SAE942019
SAE942019
47
50
55
55
49
55
55
59
42
44
52
43
50
62
53
47
50
60
50
50
51
55
54
42
47
49
38
47
47
52
39
50
43
52
53
46
57
56
41
58
57
38
49
49
43
53
53
7
5
1
4
8
1
3
7
8
6
8
8
7
1
7
7
2
7
6
1
6
3
9
4
6
3
6
4
0
2.8
7 .4
7 .7
0
5.3
5.5
9.9
0
7.8
0
10
0
3.5
0
5.6
6.7
0
9.2
0
9
10
0
11
10
0
17
16
0
11
11
0
10
10
BASE
Cl
Cl
Cl
BASE
Dl
Dl
Dl
BASE
C2
BASE
DD4
BASE
A
BASE
D
D
BASE
G
BASE
LS
LS
BASE
A
A
BASE
B
B
BASE
C
C
BASE
D
D
35
-------
A-l
A-10
A-2
A-3
A-4
A-5
A-6
A-7
A-8
A-9
MAN 18
MAN2
MAN2*
MAN 7
MAN 7*
A
B
C
D
E
F
VE
VE
VE
VE
VE
VE
VE
VE
VE
VE
VE
VE
VE
VE
VE
VE
VE
VE
VE
VE
VE
VE
VE
VE-
VE-
VE-
SAE970758
SAE970758
SAE970758
SAE970758
SAE970758
SAE970758
SAE970758
SAE970758
SAE970758
SAE970758
SAE972894
SAE972894
SAE972894
SAE972894
SAE972894
SAE972904
SAE972904
SAE972904
SAE972904
SAE972904
SAE972904
10
10
10
10
10
10
10
10
10
10
10
10
10
10
10
10
10
10
10
10
10
10
10
1A
IB
1C
A
AA
B
BB
C
CC
D
DD
E
EE
F
FF
G
GG
H
HH
I
II
J
JJ
K
KK
L
VE
VE
VE
VE
VE
VE
VE
VE
VE
VE
VE
VE
VE
VE
VE
VE
VE
VE
VE
VE
VE
VE
VE
VE
VE
VE
10
10
10
10
10
10
10
10
10
10
10
10
10
10
10
10
10
10
10
10
10
10
10
-1 PHASE II
-1 PHASE II
-1 PHASE II
56
53
65
43
60
58
56
51
58
43
58
50
56
56
60
44
50
54
55
55
55
44
41
52
47
58
50
50
51
51
55
51
44
50
52
54
53
59
57
46
58
54
51
54
42
51
55
7
9
5
9
8
5
7
4
6
1
3
1
8
5
8
9
5
2
4
7
5
7
2
6
4
5
3
9
3
7
3
6
4
1
10
0
7
0
0
5.1
0
4 .4
0
0
10.6
10.8
4.6
0
0
7 .7
5.9
14.5
9.4
10.8
14.1
0
8
8.9
12.8
14 .2
0
8.8
A-9
BASE
A-7
BASE
BASE
MAN2
BASE
MAN 7
BASE
BASE
A
A
B
BASE
BASE
VE
VE
VE
VE
VE
VE
10
10
10
10
10
10
A
AA
A
AA
AA
AA
BASE
VE
VE
VE
VE
10
10
10
10
FF
FF
FF
FF
BASE
VE-
1A
36
-------
VE-
VE-
VE-
VE-
VE-
VE-
VE-
VE-
VE-
VE-
1D
IE
IF
1G
1H
IK
1L
1M
IN
10
VE
VE
VE
VE
VE
VE
VE
VE
VE
VE
-1 PHASE
-1 PHASE
-1 PHASE
-1 PHASE
-1 PHASE
-1 PHASE
-1 PHASE
-1 PHASE
-1 PHASE
-1 PHASE
II
II
II
II
II
II
II
II
II
II
39
40
50
49
53
44
53
48
38
49
3
4
7
4
1
9
6
5
2
0
1 . 1
11.4
0
4 .4
0
9.8
BASE
VE-1D
VE-1D
BASE
VE-1G
BASE
VE-1K
37
-------
References
1. "Strategies and Issues in Correlating Diesel Fuel Properties with Emissions," Staff Discussion
Document, EPA report number EPA420-P-01-001, July 2001.
2. EPA Memorandum, "Texas Low Emission Diesel (LED) Fuel Benefits," from Robert Larson,
Transportation and Regional Programs Division, OAR, to Karl Edlund, Region VI. September
27,2001.
3. Alliance of Automobile Manufacturers International Diesel Fuel Survey, 1999. Summer and
winter #2 regular and premium diesel fuel blends for the United States only.
4. Defined at 40 CFR 80.45, and presented in the Federal Register at 59 FR 7725 (Feb. 16, 1994)
5. Personal communication with John Duerr of Detroit Diesel Corporation, June 5, 2001. Total
sales of heavy-duty two-stroke highway engines between 1981 and 1997 was approximately
178,000. DDC sales of 2-stroke engines ended in 1997. For the 2002 fleet, approximately
59,000 are still in use, representing approximately 1% of the 2002 fleet.
6. Matheaus, Andrew C., T.W. Ryan HI, R. Mason, G. Neely, R. Sobotowski, "Gaseous
Emissions from a Caterpillar 3176 (with EGR) Using a Matrix of Diesel Fuel (Phase 2)," Final
Report under EPA contract 68-C-98-169, September 1999.
7. Nandi, M., D.C. Jacobs, F.J. Liotta, H.S. Kesling, "The performance of a peroxide-based
cetane improvement additive in different diesel fuels," SAE paper no. 942019
8. Starr, M.E., "Influence on transient emissions at various injection timings, using cetane
improvers, bio-diesel, and low aromatic fuels," SAE paper no. 972904
9. Schwab, S.D., G.H. Guinther, T.J. Henly, K.T. Miller, "The effects of 2-ethylhexyl nitrate and
di-tertiary-butyl peroxide on the exhaust emissions from a heavy-duty diesel engine," SAE paper
no. 1999-01-1478
10. Based on inventory descriptions given in Chapter II, Section B, "Regulatory Impact
Analysis: Heavy-Duty Engine and Vehicle Standards and Highway Diesel Fuel Sulfur Control
Requirements," December 2000, EPA420-R-00-026
11. Letter from Edward H. Murphy, American Petroleum Institute, to Leah Weiss, Ozone
Transport Commission, "Re: API Comments on OTC Diesel Cetane Model Rule," September 26,
2000
38
------- |