United States Air and Radiation EPA420-S-02-012
Environmental Protection June 2002
Agency
vxEPA The Effect of Cetane Number
Increases Due to Additives on
NOx Emissions from
Heavy-Duty Highway Engines
Draft Technical Report
> Printed on Recycled Paper
-------
EPA420-S-02-012
June 2002
of to
on
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.
-------
Table of Contents
I. Introduction 1
A. Background 1
B. Cetane number as an emission control strategy 1
II. Analytical approach 6
A. Database preparation 6
B. Summary of analysis 8
IE. Conclusions 11
A. S AS modeling results 11
B. Application to the in-use fleet 14
C. Some predicted NOx impacts of additized cetane 16
Appendix 18
References 23
-------
I. Introduction
A. Background
The control of diesel fuel properties as a means for reducing emissions of regulated
pollutants continues to be of interest to State air quality managers. 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 voiced concern about the NOx
emission reduction benefits claimed by Texas. Because of this concern, 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 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.
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 and/or modifying 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 I.B-
l3.
Figure I.B-1
Colinearities for natural cetane number
0.81 0.82 0.83 0.84 0.85 0.81
Specific gravity
S 25
0.83 0.84 0.85
Specific gravity
20 25 30 35 40 45
Total aromatics, vol% by FIA
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. A value close to 1
indicates that no correlation exists. A value higher than 5 would indicate a moderately strong
correlation, while values approaching 10 indicate very strong correlations. A listing of variance
inflation factors is shown in Table I.B-1 for in-use survey data.
-------
Table IB-1
Variance inflation factors for in-use survey data
Natural cetane
Additized cetanea
Sulfur
Aromatics
T10
T50
T90
Specific gravity
7.1
1.1
1.3
10.0
4.8
11.1
4.2
16.7
a "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 other fuel properties. In
addition, the high variance inflation factors 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 decorrelate fuel properties from each another, the
values in Table I.B-1, based on a survey of in-use fuels, may be misleading. We therefore
repeated the calculation of variance inflation factors for the Staff Discussion Document model
database. The result was that the variance inflation factors were indeed lower than the values in
Table I.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 I.B-2. The
matrix is necessarily identical on either side of the diagonal.
-------
Table I.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
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 I.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 I.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 LA.
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, inherent 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 used 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 instead of natural cetane. Additized cetane is largely uncorrelated
with other diesel fuel properties, as shown by the low variance inflation factor in Table I.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 properties of
the additives themselves are not so extreme that physical properties of diesel fuel such as
distillation properties or specific gravity are affected.
There is good reason to believe that additized cetane and natural cetane describe similar,
or at least overlapping, 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
collinear 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.
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 number with NOx emissions is an appropriate means
for providing inventory impact information to States who are considering cetane control
programs. The remainder of this technical report describes the analyses we conducted to
investigate additized cetane effects on NOx emissions.
-------
II. 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 LA). That database was composed of
three portions covering engine characteristics, emission measurements, and fuel properties. The
fuel properties included values for CETANE_DIF which represented the increase in cetane
number resulting from the addition of a cetane improver additive to conventional diesel fuel. 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. 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 I.B-2 were low,
the data subset used for our analysis produced correlation coefficients that were even lower, as
shown in Table n.A-1. A listing of the fuels that were retained for this analysis is given in the
Appendix.
Table H.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. Figures n.A-1 through H.A-3 show the distribution
of cetane values in the database. Note that for Figure II.A-3, none of the base fuels (cetane
difference = 0) were included.
-------
Figure n. A-1
Additized cetane versus natural cetane measurements
CD
:•! 15
T3
CO
CO
'>
8 10-
c
•£
1
CD <> -
c °
"CD
O
0
•
% '
• »« •
• •
•
n • _ • •
4 • • • •
• _ gl • _ |
• * I •
• • "
• 1 • V m
I i •
• • • •• mam • ••_• •
35 40 45 50 55 60
Natural cetane
Figure ll.A-2
Natural cetane distribution
Figure II. A-3
Additized cetane distribution
37 39 41 43 45 47 49 51 53 55 57 59
Natural cetane
6 8 10 12 14 16 18 20
Cetane difference
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. 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. 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.
B. Summary of analysis
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 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 II.B-
1. Standardization removes the scale differences between fuel terms, and also reduces some of
the colinearity between first and second-order terms.
Table II.B-1
Means and standard deviations used for standardizing independent variables
Cetane difference
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
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. 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 in.B.
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 cetane
difference, 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 cetane difference 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
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 produced 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 cetane difference and natural cetane number.
Figure II.A-1 shows that there is good separation between cetane difference 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 n.B-2.
-------
Table II.B-2
Independent variables investigated
in correlation between cetane number and NOx
Cetane difference
Cetane difference2
Natural cetane
Natural cetane2
Cetane difference x 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 whose residual exceeded four standard deviations from the predicted
effect, removed them, and regenerated the model.
10
-------
III. Conclusions
A. SAS modeling results
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 HI. A-l by the
labels used in the database.
Table IE.A-l
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 in.A-2.
Table IH.A-2
SAS proc mix output for final model
Variable
Intercept
Cetane difference
Cetane difference2
Natural cetane
Cetane difference 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 n.B-1),
we converted the coefficients back into unstandardized form. The fixed effects portion of the
resulting model is shown below:
11
-------
ln(NOx, g/bhp-hr) = 1.79883 (1)
- 0.015151 x (cetane difference)
+ 0.000169 x (cetane difference)2
- 0.006014 x (natural cetane)
+ 0.000223 x (cetane difference) x (natural cetane)
We can convert this equation into one that provides a percent change in NOx emissions as a
function of cetane difference 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 (cetane difference) (2)
+ 0.000169 x (cetane difference)2
+ 0.000223 x (cetane difference) x (natural cetane) ] - 1 } x 100%
The predicted NOx impacts are shown graphically in Figure ffi.A-1.
12
-------
Figure IE. A-1
Predicted effect of cetane difference on NOx for all heavy-duty highway engines
except 2-strokes and those equipped with EGR
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.
However, there are certain conditions under which a "turnover" appears in the predicted effects.
For instance, when the natural cetane is 50, the slope of the cetane difference curve changes from
positive to negative at a cetane difference 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 cetane difference term that we used to represent nonlinear 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 cetane difference = 44.83 - 0.6598 x (natural cetane)
(3)
Thus for any values of the cetane difference that are larger than the value calculated from
equation (3), the predicted NOx impact should be the value calculated from equation (2) using
13
-------
the cetane difference 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. Application to the in-use fleet
As described in Section II.B, 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 II.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. Thus one possible approach to estimating fleet-wide NOx
effects of cetane improver additives is to use a weighted combination of equation (2),
representing engines without EGR, and the zero effect attributable to EGR-equipped engines.
Because the relative NOx inventories between EGR and non-EGR engines change over time,
these weighting factors would be dependent on calender 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 have EGR 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
have EGR in the future, this assumption assures that we are not overestimating NOx benefits of
cetane control via additives for future years, since EGR-equipped engines have been shown to
exhibit no discernable NOx response to changes in cetane number. From this information we
were able to estimate the fraction of the NOx inventory that derived from non-EGR-equipped
engines for any calender year. These fractions are given in Table ffi.B-1.
14
-------
Table HI.B-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 non-EGR-equipped 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 ni.B-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 (cetane difference)
+ 0.000169 x (cetane difference)2
+ 0.000223 x (cetane difference) >
(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 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 even 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. 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 VII.B.6 of our Staff Discussion Document regarding this type of extrapolation have not
yet been fully addressed.
15
-------
Finally, we discussed in Section IB 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 I.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
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 ni.B-1 representing engines without EGR
NATCET; = Initial value of natural cetane number
NATCETf = Final value of natural cetane number
C. Some predicted NOx impacts of additized cetane
The predicted NOx impact of a given change in cetane number is a function of both the
calender year (Table in.B-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 calender 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 in.C-1.
16
-------
Table IH.C-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
States always have separate NOx inventory estimates for highway and nonroad. We have
therefore made no attempt to combine the estimates for highway and nonroad in Table m.C-1
into a single estimate representing total diesel contributions to the NOx inventory.
17
-------
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
CARB-TOXIC SAE 922214 SAE 961974
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
II
II
II
II
II
II
II
II
II
II
II
II
II
II
II
CETANE NUM CETANE DIF BASE
51
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
18
-------
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
19
-------
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
20
-------
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
21
-------
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
1
1
1
1
1
1
1
1
1
_PHASE
_PHASE
_PHASE
_PHASE
_PHASE
_PHASE
_PHASE
_PHASE
_PHASE
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
22
-------
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 U, Section B, "Regulatory Impact
Analysis: Heavy-Duty Engine and Vehicle Standards and Highway Diesel Fuel Sulfur Control
Requirements," December 2000, EPA420-R-00-026
23
------- |