EPA-R2-73-276
December 1972
Environmental Protection Technology Series
I
«^° ***,
::::x:::::x;:::::::::::::::::;:::::::-:- >
-------
EPA-R2-73-276
CHARACTERISTICS
AND
PHOTOCHEMICAL REACTIVITY
OF EMISSIONS
by
B. Dimitriades, B.H. Eccleston,
G.P. Strum, andC.J. Raible
U.S. Bureau of Mines
Bartlesville Energy Research Center
Fuels Combustion Research Projects
Interagency Agreement No. EPA-IAG-0138(D)
Program Element No. 1A1010
EPA Project Officer: John Moran
Chemistry and Physics Laboratory
National Environmental Research Center
Research Triangle Park, North Carolina 27711
Prepared for
OFFICE OF RESEARCH AND MONITORING
U.S. ENVIRONMENTAL PROTECTION AGENCY
WASHINGTON, D.C. 20460
December 1972
-------
This report has been reviewed by the Environmental Protection Agency and
approved for publication. Approval does not signify that the contents
necessarily reflect the views and policies of the Agency, nor does
mention of trade names or commercial products constitute endorsement
or recommendation for use.
11
-------
iii
TABLE OF CONTENTS
Page
List of Figures , iv
List of Tables ..... v
Part I 1
Abstract 1
Introduction 2
Experimental Design and Procedures. . * 5
Selection of Test Fuels 5
Tests Using Full-Boiling Range
Gasolines 8
Results - Discussion 12
Tests Using Full-Boiling Range
Gasolines 12
Smog Chamber Program 17
Influence of Engine Air-Fuel
Ratio on Correlation Results 22
Automotive Combustion of Simple
Hydrocarbon Fuels 24
Summary and Conclusions 26
References 29
Appendix A 64
Part II 75
Introduction 75
Experimental Procedures and Results 76
References 80
Part III 81
Introduction 81
Odor Determination 81
Odorant Dilution Systems 82
-------
iv
Page
Odorant Reaction and Aging 82
Analytical Methods Development • 83
Results and Discussion 84
Odor Presentation 84
Unstable and Reactive Odorants 87
Analytical Methods Development 90
Summary • 90
ILLUSTRATIONS
Fig,.
1. Correlation of calculated exhaust reactivities with
polyalkylbenzene levels in fuel 53
2. Correlation of calculated exhaust reactivities with
olefin levels in fuel 54
3. Correlation of observed exhaust reactivities with
total aromatic levels in fuel 55
4. Correlation of exhaust reactivities — Fuel Aromatic
Relationships From Two Bureau of Mines Studies 56
5. Correlation of observed exhaust reactivities with
polyalkylbenzene levels in fuel 57
6. Correlation of observed and calculated (Jackson scale)
exhaust RNO reactivities 58
7. Correlation of observed exhaust %02 reactivities with
calculated reactivities by Jackson and Bureau of Mines
CQ
scales •
8. Correlation of observed exhaust reactivities with calcu-
lated "fuel" reactivities 60
9. Calculated RN02 reactivities of exhaust from simple hydro-
carbon fuels used under varied A/F conditions 61
10. Calculated RN02 reactivities of exhaust from gasoline
fuels used under varied air-fuel ratio conditions. ... 62
11. Correlation of observed and calculated RN02 reactivities
of exhaust from simple hydrocarbon fuels 63
A-l. Average emission levels for each car and each fuel .... 71
A-2. Variation of exhaust mass emission parameters with various
fuel composition parameters. ....
-------
TABLES
Page
1. Experimental fuels used in the gasoline fuel test program . . 31
2. Experimental fuels used in the simple hydrocarbon fuel
testing program 34
3. Test automobiles 35
4. Mass emissions and molar specific reactivity of exhaust in
first phase of fleet testing program 36
5. Mass emissions and molar specific reactivity of exhaust in
second phase of fleet testing program' 39
6. Correlation of exhaust composition with fuel composition
for car No. 5 41
7. Correlation of exhaust composition with fuel composition,
car dummy variables, and air fuel ratio 42
8. Smog chamber reactivities of exhaust from fuels 1-21 43
9. Smog chamber reactivities of exhaust from fuels 22-25 .... 48
10. Bureau of Mines rate-of-N02-formation reactivity scale. ... 49
11. Composition and smog chamber reactivities of exhaust from
simple hydrocarbon fuels 50
12. Summary of smog chamber reactivities of exhaust from
simple fuels 52
13. Comparative exhaust hydrocarbons composition from Onan
engine and as determined for four automotive engines.... 77
14. Exhaust odor intensity as influenced by sampling method
(DI odor units, exhaust dilution 100:1) 86
15. Efficiencies of reagent scrubbing in removal of tests
compounds . 86
16. Effects of selective chemical reagents on exhaust odor
intensity (DI odor units, exhaust dilution 100:1) 89
A-l. Summary of mass emission data for five cars and ten fuels . . 67
A-2. List of fuel physical and compositional 'pr°Perti.es used in
correlations 68
A-3. Correlation coefficients for exhaust mass emission levels
and fuel physical and compositional properties 69
A-4. Summary of regression analysis results 70
-------
THE ASSOCIATION OF AUTOMOTIVE FUEL COMPOSITION
WITH EXHAUST REACTIVITY
by
Basil Dimitriades1, B, H. Eccleston2, G. P. Sturm, Jr.3, and C. J. Raible1
ABSTRACT
The association of automotive fuel composition with exhaust reactivity
was studied in an experimental program that involved testing with different
r
automotive engines and with gasolines of varied composition. Fuel composi-
tion was determined by gas chromatography. Exhaust reactivity was both
estimated from detailed composition data and determined directly using
a smog chamber. Results showed clearly the exhaust reactivity to increase
with increasing levels of polyalkylbenzenes in the fuel. No other syste-
matic patterns of high significance were detected in the association of
exhaust reactivity with several broadly defined groups of fuel components.
Evidence suggested, however, that such lack of relatability of exhaust
reactivity and fuel composition was a result of inappropriate definition
of fuel composition. For the purposes of this study, had it been possible,
fuel composition should have been defined and expressed in terms of com-
ponent groups such that the potential for exhaust reactivity would be the
1 Scientist administrator with the Environmental Protection Agency at
Research Triangle Park, N. C. Formerly with the Bartlesville Energy
Research Center, Bartlesville, Oklahoma.
2 Project leader, Bartlesville Energy Research Center, Bureau of Mines,
Bartlesville, Okla.
3 Research chemist, Bartlesville Energy Research Center, Bureau of Mines,
Bartlesville, Okla.
4 Group leader, Bartlesville Energy Research Center, Bureau of Mines,
Bartlesville, Okla.
-------
same within each group and different from group to group. Classification
of fuel components in terms of the paraffins-olefins-aromatics groups does
not meet the latter requirements. For appropriate classification of fuel
components more information is needed on the combustion of hydrocarbons
in the multicylinder internal combustion engine. Statistical analysis
of the mass emissions data showed significant car and fuel effects on
hydrocarbon, carbon monoxide, nitric oxide, total aldehydes, and formal-
dehyde emission levels and on total photochemical reactivity. Car-fuel
interactions were not significant at the 90 pet level. Correlations
were found between mass emission parameters and fuel composition.
INTRODUCTION
Presently and undoubtedly for a large part of the decade of the 70*s,
a major portion of efforts in air pollution abatement has been and will
continue to be directed to reducing emissions from automobiles. Of the
objectionable automotive emissions, the hydrocarbons are prominent be-
cause of their principal role in photochemical smog formation. As a
result, several control methods for hydrocarbons have been devised, some
of which have already been put to use, others being further developed for
use in the future. In most of these methods the approach is to use engine
modification or accessory devices that would decrease the amount of hydro-
carbon unburned an the primary combustion or convert the hydrocarbon emis-
sion into harmless product. This "device" or engine modification approach
to emission control appears to have the potential for virtually complete
elimination of the hydrocarbon emissions. A drawback, however, is the
problem of retrofitting the automobiles already on the road.
-------
Adjustment of fuel formula also could effect emission reduction
either through reduction of emission levels or through emission modifi-
cation resulting in less polluting material. While this "fuel optimi-
zation" approach may have only a modest potential for emission reduction
per automobile, it has the advantage over the "device" approach that
its application automatically encompasses all automobiles on the road.
The direction of the hydrocarbon control effort took a decisive
turn with the recent setting of hydrocarbon emission standards for the
5/
1975 automobiles (12).— The stringency of these standards almost
_5_/ Underlined numbers in parentheses refer to the list of references
at the end of this report.
eliminated for 1975 and 1976 model automobiles all control options except
ones in which catalytic exhaust conversion devices are used. Such de-
vices as have been developed, however, do not function effectively with
fuels containing lead antiknock additives because of the poisoning effect
of lead on the catalyst. Therefore, an inevitable part of the control
measure necessary to meet the 1975 automobile standards is the reformu-
lation of gasoline to provide the requisite octane quality without the
use of lead antiknock additives. The extent of reformulation required
appears to have been lessened by a reduction in compression ratio that
- was put into effect to lower the octane requirement of the engines of
1971 and later automobiles. However, this engine modification has also
affected engine efficiency adversely, and one should reasonably expect
that in the years to come the tendency will be toward higher engine com-
pression ratio and hence toward gasolines made of higher octane quality
hydrocarbons.
-------
Besides the octane related change in fuel formula, additional fuel
reformulation might become necessary as a result of new findings on the
harmful effects of emissions. The present inventory of such effects is
not necessarily complete. Chemical analysis of hydrocarbon combustion
products, however advanced its methods may be, has not yet yielded
identification of each and every component of the complex mixture of
exhaust emissions. For example, with the exception of the lead additive,
virtually no information exists on the fate of fuel additives in the
combustion process. Evidence of heretofore unknown objectionable exhaust
components or effects arising from the use of additives might raise a need
for new or revised control techniques including further fuel reformulation.
It would therefore appear that, based on present projection of control
needs, modification of gasoline fuel composition is to be anticipated.
While the principal objective of such modification is to alter the com-
bustion characteristics of the basic fuel hydrocarbon mixture, it is also
desirable that the fuel change be in the direction toward minimum pollution
from emissions. Achievement of the latter would require reliable compre-
hensive information on the association of fuel composition with pollution
potential of emissions in order to provide some guidelines for fuel
reformulation.
Existing information on the association of fuel composition with
pollution potential of emissions lacks needed detail and at points is
conflicting. Earlier studies conducted in this laboratory (3_, 11) sug-
gested that fuel aromatics may be associated with organic emissions of
higher photochemical smog potential or reactivity. Results from other
studies (6^ 15) suggested that such an effect of fuel aromatics is either
small or none. All these previous studies of the fuel composition factor
-------
had the common inadequacy in that fuel composition was defined and expressed
in terms of relative levels of broadly defined fuel component groups, namely,
paraffins, olefins, aromatics. These classifications disregard possible
effects of differing composition within each of these groups. Lacking
evidence to the contrary, one cannot neglect the possibility that such
effects may be as strong or stronger than those effects associated with
each of the parent groups.
This report covers work done at the Bartlesville Energy Research
Center in cooperation with the Environmental Protection Agency to further
explore the association of fuel composition with pollution character of
emissions. In view of the importance of the current issue regarding the
effect of fuel aromatics on emission reactivity, this research was de-
signed, in part, to provide information directly relevant to this issue.
Further, attempts were made to determine more precisely the fuel/emission
association by expressing fuel composition in terms of fuel component
groups more narrowly defined than in the previous studies.
EXPERIMENTAL DESIGN AND PROCEDURES
Selection of Test Fuels
Ideally, in a study of fuel composition effects, test fuels would
be selected so that the level of each of the 200 or so fuel components
varies from fuel to fuel independently of the levels of other components.
For example, for 200 components and two levels per component, a factorial
experiment would require that about 2^ test fuels be used to determine
main and interaction effects of the 200 components. If these effects vary
with automobile, then additional tests will be required to determine the
automobile effect. These requirements are clearly impractical, and some
-------
simplification of the objectives must be made if th'e study is to entail
a manageable effort.
Simplification can be accomplished by classifying the fuel components
into a number of groups and by designing the study so as to determine the
relative effects of these component groups—rather than of the 200 individual,
components. The success of such a simplification depends on whether the
fuel component classification is consistent with the study's objective.
Thus, if the objective is to relate fuel composition with pollution charac-
teristics of fuel, then fuel composition ideally should be expressed in
terms of component groups such that pollution characteristics would differ
from group to group but would be similar within each group. The use of
any other grouping scheme will result in a rating of the component groups
used, based on their contributions to emission pollution, but will not re-
veal the full potential for fuel composition to influence pollution potential.
Available evidence suggests that fuel aromatics contribute to exhaust
hydrocarbons mainly via unburned fuel and contribute only to a minor ex-
tent via combustion reactions that result in aromatic addition-products
(9) . This suggests that the reactivity of those exhaust hydrocarbons that
originate from the fuel aromatics is determined mainly by the intrinsic
reactivities of the aromatic hydrocarbons in the fuel. These reactivities
are known to differ widely in value (8), and in a manner that justifies
classification 6f fuel aromatics into three groups, in the following order
of reactivity: polyalkylbenzenes > monoalkylbenzenes > benzene.
Relative to the aromatic fuel components, in typical automotive engine
combustion the fuel paraffins are more readily burned or fragmented—fuel
olefins even more easily oxidized. Therefore, the contribution of fuel
paraffins and olefins to exhaust organics consists mainly of combustion
-------
products rather than unfaurned fuel. Further, combustion products from
fuel paraffins and olefins are distinctly different than those from
aromatics in that they contain higher levels of reactive olefins and
aldehydes and lower levels of unreactive acetylenes. Therefore, fuel
paraffins, olefins, and presumably naphthenes also, should be differen-
tiated from aromatics, as having distinctly different combustion and
pollution characteristics.
For further classification of the fuel aliphatics, the supporting
x
evidence is sketchy. Such evidence and reasonable speculation suggest:
(a) Relative to straight chain fuel hydrocarbons, branched paraffins and
olefins are more likely to yield exhaust olefins substituted at the double
bond; such olefins are generally much more reactive than the straight chain
ones, (b) Relative to branched olefins, branched paraffins yield higher
levels of exhaust olefins (7). Finally, it is also likely that fuel ali-
phatics differ significantly in pollution characteristics even within each
of the groups of branched paraffins, branched olefins, straight chain
paraffins, and straight chain olefins, depending on molecular size and
structure. While there is no experimental evidence directly applicable
to this point, it has been shown that at least in pyrolysis of hydro-
carbons, the composition of final products differs greatly with structure
differing within the group of branched paraffins (14).
Thus the available information justifies classification of fuel com-
ponents on the basis of their contribution to exhaust pollution charac-
teristics in at least the following detail: Benzene, monoalkylbenzenes,
polyalkylbenzenes, olefins, branched paraffins, and straight chain paraf-
fins. Further breakdown or an entirely different classification scheme
based on molecular structure, may be shown to be more appropriate if and
-------
when more information is generated on degradation patterns in automotive
combustion of hydrocarbons.
The preceding considerations led to the design of 10 test fuels (fuels
12 to 21, table 1) that contained varying levels of benzene, monoalkylben-
zenes, polyalkylbenzenes, n-paraffins, isoparaffins, cycloparaffins, n-
butane, and olefins. Fuels listed as 1 to 9 in table 1 were test fuels
used in a separate study of commercial gasolines (92-94-RON) marketed
for use with the 1971 model year automobiles. Data from that fuel-evalua-
tion program were found to be relevant to the objectives of the fuel com-
position effect study; therefore, data from both studies were pooled.
Fuels 10 and 11 (table 1) were commercial unleaded premium gasolines
with unusually high levels of aromatics. These two fuels were also included
in order to provide additional data on the effect of fuel aromatics on
emission reactivity. Fuels 22 to 25 (table 1) were those used in an earlier
Bureau of Mines study (J3, 11) in which fuel aromatics were shown to have
a strong effect on emission reactivity; these fuels were reused in this
study in order that additional information could be added to that from
which previous conclusions were drawn.
In addition to the full-boiling range gasolines, a few simple hydro-
carbon fuels (table 2) were also used in experiments in the small two-
cylinder engine. The purpose of this work, as discussed later, was to
gain information on the degree to which pollution characteristics of the
various specific fuel hydrocarbons vary.
Tests Using Full-Boiling Range Gasolines
The gasolines were studied using 11 automobiles (table 3) selected
to represent typical, high production, late models.
-------
Engines of automobiles in the test program were adjusted according to
manufacturer's specifications. In the first phase of this program, auto-
mobiles 1 through 9 were used in tests with eight fuels (fuels 1, 4-10,
table 1). Each automobile was operated at 85° F ambient temperature
following a 22.9 minute engine duty cycle (Federal cycle for 1972 auto-
mobiles and later (13)). Limited testing was also conducted at 45° and
20° F ambient temperatures. Samples of exhaust in mixture with air were
drawn throughout the engine test as prescribed in Federal test procedures
for 1972 automobiles (13). '
In the second phase of the program, automobiles 5, 6, 7, 9, and 10
were used to test fuels 12 through 21. Each automobile was operated at
75° F ambient temperature following the same engine cycle as in the first
phase. Exhaust was sampled as prescribed in Federal test procedures for
1975 automobiles (12); that is, three samples were collected to provide
sample of (1) exhaust in the cold start part of the cycle, (2) in the
remaining stabilized part, and (3) in the hot start part, respectively.
In both phases of this study, exhaust was analyzed for individual hydro-
carbons (4), total hydrocarbon by flame ionization detection (FID), total
aldehydes (10), formaldehyde (1), oxides of nitrogen (NOX) by nondispersive
infrared (NDIR), carbon monoxide (CO) by NDIR, and carbon dioxide (C02)
by NDIR. The data from these tests were used to associate fuel composi-
tion with exhaust hydrocarbon composition (without reference to experi-
mentally observed reactivity or smog potential),
Note that while the two phases of the program had somewhat different
specific objectives and experimental conditions, their respective data,
being relevant to the issue of fuel/emission association, were pooled to-
gether. For such pooling, analytical data from the three exhaust samples
-------
10
per test in the second phase were combined mathematically to yield data
comparable to those obtained in the first phase1. Also, the difference
in test temperature between the two phases (85° F versus 75° F) was judged
to have no significant effect on the fuel/emission association and was,
therefore, ignored.
This experimental program that involved the eleven automobiles was
expected to provide data suitable and adequate to establish the associa-
tion of fuel composition with exhaust hydrocarbon composition, but not
with exhaust hydrocarbon reactivity. The latter limitation was because
the exhaust samples from this program contained relatively high levels
of NOX that prohibited experimental measurement of exhaust hydrocarbon
reactivity in photochemical smog chambers (2). For such measurement
to be feasible, the hydrocarbon/NOx ratio in exhaust must be higher than
3:1 (ppmC:ppm) (2); such a condition can be realized only when the auto-
motive engine operates under richer than normal air-fuel (A/F) conditions.
Alternative methods for estimating exhaust hydrocarbon reactivity from
exhaust hydrocarbon composition have been shown to be much less reliable
than direct experimental measurement (3). This consideration led to de-
signing a separate program to study the association of fuel composition
with exhaust hydrocarbon reactivity.
Smog Chamber Tests Using Full-Boiling Range Gasolines
In the smog chamber program the experimental fuels (fuels 1-25,
table 1) were studied using an automobile that was operated fuel-rich
in order to yield exhaust with high hydrocarbon-to-NOx ratio as required
for smog chamber examination. Exhaust samples in this program were
analyzed as in fleet testing and were also examined in a smog chamber
for photochemical reactivity. For reactivity measurement, exhaust at the
-------
11
initial levels of 6 ppmC hydrocarbon, 0.9 ppm NO, and 0.1 ppm N02 was
irradiated in a smog chamber following procedures described in the litera-
ture (2).
Because the test automobile in the smog chamber program was operated
fuel-rich, the question was raised whether this perturbation in an engine
operation condition would invalidate the findings of the smog chamber
program. To investigate this, additional testing was conducted in which
one automobile and one small two-cylinder engine were operated under varied
A/F conditions, using gasolines as well as thetsimple hydrocarbon fuels.
Testing of Simple Hydrocarbon Fuels
Results from the gasoline testing program described in the preceding
sections indicated a need for information on the automotive combustion
characteristics of specific hydrocarbon types. More specifically, such
information was needed to serve the following purposes:
A. To demonstrate the degree to which reactivity of exhaust from
various hydrocarbons varies with hydrocarbon structure.
B. To provide guidelines for classifying hydrocarbons into groups
such that the potential for exhaust reactivity would be different
from group to group but uniform within each group.
The experimental program that was run was undertaken and designed with
the first purpose in mind; to serve the second purpose, a much larger
research effort would be necessary.
Fourteen prominent fuel compounds (table 2) were chosen to represent
several classes of hydrocarbons in gasoline fuels. The fuels were run in
a two-cylinder water-cooled engine (Onan) at three A/F equivalence ratios—
0.75, 1.0, 1.1. For all tests, the engine was operated at 1,800 rpm
-------
12
(1,800 rated speed) and at one-fifth full load. Exhaust was sampled and
analyzed for individual hydrocarbons by gas liquid chromatography (GLC),
total hydrocarbon, formaldehyde, total aldehydes, NOX, CO, and C^. Ex-
haust samples from the rich operation mode were also irradiated in the smog
chamber for measurement of the photochemical reactivity. Duplicated
chamber runs were conducted with initial exhaust level at 6.0 ppmC hydro-
carbon, 0.9 ppm NO, and 0.1 ppm NO-.
RESULTS - DISCUSSION
Tests Using Full-Boiling Range Gasolines
Results from tests with the eleven automobiles are given in tables
4 and 5. The tabulated data are for emission rates of exhaust hydrocarbon,
CO, NOX, total aldehydes, and for hydrocarbon reactivities calculated from
hydrocarbon compositions.
These results were analyzed statistically for correlations between
fuel composition and exhaust emissions. Such correlations, if present,
were to be interpreted to provide a measure of the relative contributions
of fuel hydrocarbons to exhaust emissions. Correlations with exhaust
emission rates are described and discussed in appendix A. The remainder
of this section deals with the correlation of fuel composition with con-
centrations of reactive organics (hydrocarbons and aldehydes) in exhaust.
The exhaust parameters used in this correlation analysis included
calculated reactivities, formaldehyde, total aldehydes, and several ole-
fins and aromatics that are known to contribute significantly to exhaust
reactivity. Varied schemes and combinations of hydrocarbon class and
subclass division and groupings were used as parameters of fuel composition.
Thus, in one attempt exhaust composition was correlated with fuel composition
-------
13
expressed in terms of paraffin, olefin, and aromatic. In other correla-
tion models, fuel composition was expressed in terms of more narrowly
defined component groups. Correlation analysis was applied upon data from
each automobile separately. Using such data, the fuel/exhaust correlations
were not obscured by car effects or car/fuel interaction effects. Such
effects, while undesirable in the correlation analysis, are nevertheless
of interest, and were explored in a separate statistical analysis.
The correlation models that were used and results from the correlation
analysis are illustrated— by information summarized in table 6 for car 5.
6/ Correlations were obtained from statistical analysis of data from
each car examined individually. Results were similar, car-to-car,
and the following discussion is to be construed as representing
the typical case.
Degree of correlation is represented by the value of R in table 6, a
high R value representing a high degree of correlation. From the data
of table 6, apparently better fuel/exhaust correlations are generally
obtained when fuel composition is expressed in terms of a more detailed
classification of the fuel components. For example, breaking the aromatic
portion of the fuel into benzene, monoalkylbenzenes, and polyalkylbenzenes
(model II) gave 'a fuel/exhaust correlation that was improved over that
obtained when the total of aromatics is used as one fuel component (model
I). This suggests that the source of a reactive exhaust organic is asso-
ciated with a narrowly defined group or even an individual fuel component
rather than with a broadly defined class of fuel hydrocarbons. This is
more clearly illustrated by the correlation results for exhaust toluene
-------
and xylenes (see table 6). Levels of toluene and xylenes in exhaust were
found to correlate much better with fuel composition when fuel aromatics
were broken into benzene, monoalkylbenzenes (mostly toluene) and poly-
alkylbenzenes (mostly xylenes), with R values increasing from 0.66 and
by
0.41 to 0.94 and 0.80, respectively, for car 5. Evidently this was so
because exhaust toluene and xylenes originate almost entirely from the
toluene and xylenes (in fuel) and not from all fuel aromatics. In line
with this reasoning, the rather poor correlations of exhaust olefins with
fuel composition might be due to inappropriate grouping of fuel components.
If so, then better correlation could be obtained if fuel components were
classified into groups such that yield in exhaust olefins would be uniform
within each group. Another difficulty in the correlations for exhaust
olefins is the rather narrow ranges in yields as indicated by the standard
deviations about the means as seen in table 6. For example, the standard
deviation for ethylene is about 20 pet of the mean value as compared to
about 60 pet for toluene. Also, for low yield components such as 2-methyl-
butene-2, the experimental error in the concentration measurement may be
important.
The main conclusion reached in consideration of the statistical data
thus far is that better correlations of fuel composition with exhaust
composition and calculated reactivity are obtained when fuel composition
is expressed in more detail than is provided by the simple classification
into paraffins, olefins, and aromatics. This statistical conclusion is
interpreted to mean that pollu-tion characteristics of fuel components vary
considerably within the class of aromatics, as predicted by theory, and
possibly within the paraffin and olefin classes also. Further interpretation
of these statistical results—e.g., to identify or quantitatively associate
-------
15
source of an exhaust hydrocarbon with a specific group of fuel components—
is not valid. Qualitative associations, however, might have significance.
Thus, we examined the associations between calculated (Jackson) exhaust
reactivity and the levels, in fuel, of aromatics, polyalkylbenzenes, olefins,
paraffins, and isoparaffins, Systematic patterns were obtained only in
the reactivity-versus-fuel polyalkylbenzenes and reactivity-versus-fuel
olefin plots (figures 1 and 2), and even then the dependencies implied,
FIGURE 1. - Correlation of Calculated Exhaust Reactivities With
Polyalkylbenzene Levels in Fuel.
FIGURE 2. - Correlation of Calculated Exhaust Reactivities With
Olefin Levels in Fuel.
although real, were nevertheless weak. Such failure to uncover strong
dependence of exhaust reactivity on a specific group of fuel components
may have been caused by the fact that the fuel component groups that were
considered were inappropriately defined.
Since the analysis discussed thus far involved data from one automobile
only, the question was raised whether the conclusions reached are valid
for other automobiles also. It was also questioned whether whatever car
effect exists is caused solely by air/fuel variation from car to car. To
answer these questions, a statistical analysis was applied upon the data
obtained in the second phase of the fleet testing program when A/F data
were taken for all test cars. The statistical method used was similar
to the one used in the fuel/exhaust correlation. Specifically, exhaust
composition and calculated reactivity were again correlated with the fuel
composition, except that this time data from five automobiles were pooled
-------
16
and the correlation models used were:
I. (Exh. Component) = F (fuel composition)
II. (Exh. Component). = F (fuel composition) -f a ^ + a c + a c_
•f a.c.
4 4
III. (Exh. Component). = F (fuel composition) + a.. (A/F)
where
F (fuel composition): Function of fuel composition
a^c^j a^0?' aj?°3' a4°4: Terms representing contributions of cars
6, 7, 9, and 10, to component i in the
exhaust relative to that from car 5.
a. (A/F) : Term representing ''contribution'1 of
A/F to exhaust component i.
The experimental data were fitted to these statistical models and
results were used to determine whether inclusion of the car or A/F vari-
ables improved the fuel/exhaust correlation. Results, summarized in table
7, show that inclusion of the car variables improved correlation substan-
tially, suggesting, thus, presence of a strong car effect. Use of the
A/F variable in lieu of the car variables had a much smaller effect, sug-
gesting, thus, that the car effect was not a result of variation in average-
over-the-engine-test A/F.
Presence of "car effect" is interpreted to mean that the same fuel
may yield different exhaust hydrocarbon mixtures in different automobiles.
This raises the question, then, whether a change in fuel composition would
have different effects on the exhaust from different automobiles. Such
a nonuniform fuel effect, in the language of statistics, is referred to
as car/fuel interaction effect and is represented by the car-fuel product
terms in the following statistical model:
-------
17
(Exh. component). = a. x Fuel- -4- .... + c- x Car.. + .... + k- Fuel- x
car. + ....
Statistical analysis of the data from the first phase of the multiple-
vehicle program showed that the car/fuel interaction effect was small,
suggesting, that a change in fuel composition would have equal effects
on exhaust from different automobiles. This finding is useful in that
it suggests that in studying the effect of fuel composition on hydrocarbon
emissions, use of a large number of test automobiles may be unnecessary.
t
Smog Chamber Program
Results from the smog chamber program are given in table 8. The most
significant observations from these results are that the exhaust samples
tested did not show large differences in reactivity in spite of the great
diversity in fuel composition. In fact, from plots of exhaust reactivity
against fuel aromatics (figure 3)— , the dependence of reactivity on fuel
FIGURE 3. - Correlation of Observed Exhaust Reactivity With Total
Aromatic Levels in. Fuel.
TJ Exhaust reactivity in this and following figures is expressed in units
of molar propylene equivalents; that is, in units of reactivity of
an equimolar concentration of propylene.
aromatics did not seem to be as strong as previous studies in this laboratory
had suggested (figure 4, dark circles). In view of the importance of the
FIGURE 4. - Correlation of Exhaust Reactivities With Total Aromatic
Levels in Fuel. (Previous Bureau of Mines Study.)
-------
18
issue regarding the effect of fuel aromatics on exhaust reactivity, the
data from this program were scrutinized in searching for information relevant
to this issue. Thus, we explored the possibility, suggested by the multiple-
car tests that exhaust reactivity depends on levels of some, but not the
total of all, fuel aromatics. Results showed that exhaust reactivity,
indeed, appears to be more strongly dependent on polyalkylbenzenes (figure
5), than on total aromatics in fuel. This finding, consistent with the
FIGURE 5. - Correlation of Observed Exhaust Reactivities With Polyalkyl-
benzene Levels in Fuel.
fleet testing results from the multi-vehicle tests probably establishes
the relatively high pollution potential of polyalkylbenzenes in fuel. However,
the effect of the polyalkylbenzenes on exhaust reactivity does not alone
explain the unusually large differences in exhaust reactivity that were
observed for some of the fuels in the earlier Bureau of Mines study (3).
In attempting to understand the differences in results between this
and the earlier Bureau of Mines studies regarding effect of fuel aromatics
on exhaust reactivity, we explored the possibility that experimental error
may account for part or all of the difference in question. To this end,
we fitted the data from this study to the exhaust reactivity—exhaust com-
position correlation equation developed in the earlier study. Figure 6
FIGURE 6. - Correlation of Observed and Calculated (Jackson Scale)
Exhaust R^ Reactivities.
shows the curve depicting that correlation equation along with the scatter
-------
10
boundaries and the correlation data (dots) obtained in the present study.
The excellent agreement between the correlation points of the two studies
suggests that the quality of experimental data (smog chamber reactivity
and chromatographic analysis data on exhaust hydrocarbons) was about the
same for the two studies. This would discredit experimental error as the
sole cause of the difference in results between the earlier and the present
studies. Finally, to determine whether differences in test conditions
can account for these differences, the fuels of the earlier Bureau of Mines
study (fuels 22-25, table 1) were tested again; this time using the experi-
t
mental automobile and conditions of the present study. Retesting included
chromatographic analysis of fuels since the present fuel samples were taken
from a new batch that might differ somewhat in composition from the fuel
batch used in the earlier study. Results of the reactivity observations
given in table 9, did show a trend of increasing reactivity with increasing
levels of fuel aromatics (see figure 4, crosses) but this trend was not
as strong as that that was observed in the earlier study (figure 4, dots).
It was, therefore, deduced that the differences in results between this
and the earlier fuel studies were caused, partly at least, by differences
in test conditions.
In attempts to associate exhaust reactivity with nonaromatic fuel
components, reactivities were plotted versus total olefins, internal olefins,
n-paraffins, and isoparaffins in fuel for constant total aromatic or total
polyalkylbenzenes. With one exception, no systematic patterns could be
observed in these plots, suggesting that composition differences within
the groups of olefins, ii-paraffins, and isoparaffins, are probably causing
some differences in exhaust reactivity. The exception was the apparent
correlation between peroxypropionyl nitrate (PPN) yields in the irradiated
-------
exhaust and the level of oleflns In the fuel. Note that peroxyacetyl
nitrate (PAN) yields did not show the same trend. This might suggest that
PPN results mainly from fuel olefins surviving in the exhaust, rather than
from olefins that result from the combustion process and yield primarily
PAN.
An alternative approach to relating exhaust reactivity with fuel com-
position is through intermediate correlation with exhaust composition.
Application of this approach was illustrated in the American Petroleum
Institute study (of the effect of fuel composition on exhaust reactivity)
(3). Briefly, the approach entails statistical treatment of exhaust com-
position and exhaust reactivity data to develop a. reactivity-composition
correlation equation. Using such an equation upon the data from the
multiple-vehicle tests, exhaust reactivities are obtained from exhaust
compositions and are subsequently correlated with fuel composition. In
the present study this approach did not appear promising at first, primarily
because exhaust reactivity did not correlate well with exhaust composition.
Part A of figure 7 depicts this correlation using as measure of exhaust
FIGURE 7. - Correlation of Observed Exhaust R^ Reactivities With
Calculated Reactivities by Jackson and Bureau of Mines
Scales.
composition the "linear summation reactivity" value obtained from exhaust
composition and Jackson specific reactivity data (8). Part B of figure 7
depicts the same correlation, except that here linear summation reactlvitj'
values were calculated from Bureau of Mines specific reactivities as shown
in table 10, rather than Jackson reactivities. Use of the specific
-------
21
reactivities shown in this table did improve the correlation but not to
the point of providing a usable correlation equation. This lack of strong
correlation in the data of this study was attributed to the narrow range
within which the correlation variables (exhaust reactivity and exhaust
composition) were included.
The analysis of the smog chamber data thus far points to the same
conclusion that was reached in tests with the eleven automobiles, namely:
Differences in reactivity among exhausts from different fuels are not caused
solely by differences in relative levels of paraffins, olefins, and aro-
matics in fuel; differences in composition within each of these hydrocarbon
classes are also a factor.
This conclusion led to the idea of attempting to correlate exhaust
reactivity with fuel composition expressed in terms of individual component —
rather than component group — data. The concept here entails development
of a "fuel reactivity scale", that is, a rating system by which each fuel
component would be rated according to its contribution to exhaust reactiv-
ity. Given those ratings or "specific reactivities" (rj[) , and the mole
fractions of the individual fuel components (x.^) , exhaust reactivity
, calc, ld then be caicuiated using the equation
exh
exh
The success of this method for correlating exhaust reactivity with
fuel composition would be measured by the extent of agreement between the
calculated (RC3?;C) and observed (R°b^) exhaust reactivity values.
exh £xn
In order to use this correlation method upon the data of this study,
first specific reactivity values had to be assigned to the various fuel
components. Several alternative sets of such assignments were made based
-------
22
on speculation and on the limited information available regarding auto-
motive combustion of hydrocarbons. The set of assignments that gave the
best predictions of exhaust rate-of-N02-formation reactivity was as follows:
Except for paraffins, each fuel hydrocarbon was given a specific reactivity
value equal to that assigned by Jackson (8); paraffins were given reactivity
values equal to twice the corresponding Jackson value. The rationale of
these assignments is based on the fact that reactive exhaust organics con-
sist mainly of unburned fuel and of products from cracking of the aliphatic
fuel components. Following these specific reactivity assignments, exhaust
reactivity was calculated, and resultant values were compared to those
for observed exhaust reactivity. Results are shown in figure 8. The
FIGURE 8. - Correlation of Observed Exhaust Reactivities With Calculated
"Fuel" Reactivities.
correlation appears to be good despite the speculative character of the
specific reactivity assignments made. This verifies the advantage of ex-
pressing fuel composition in terms of individual component data rather
than in terms of broadly defined fuel component groups. Interestingly,
exhaust reactivity did not correlate as well with exhaust composition (see
Part A ol figure 7) as it did with fuel composition (figure 8). This is
probably due to "relatively large experimental errors in the chromatographic
measurement of exhaust composition; error in chromatographic measurement
of fuel composition is much smaller.
Influence of Engine Air-Fuel Ratio on Correlation Results
Operating the test automobile used in the smog chamber program with
fuel-enriched A/F raised questions regarding validity of results and
-------
prompted a brief investigation of the effect of A/F on reactivity of ex-
haust emissions. Data for this investigation were obtained both in the
simple hydrocarbon fuel testing program of this study and in another study
in which a 1963 Chevrolet engine, 283 cu-in-displacement (CID), was operated
under varied A/F conditions using gasoline fuels of various compositions.
The data from the simple fuel tests (discussed also in the following section
of this report), plotted in figure 9, showed the A/F to have a nonuniform
FIGURE 9. - Calculated R^ Reactivities of Exhaust From Simple Hydro-
carbon Fuels Used Under Varied A/F Conditions.
effect on exhaust reactivity. Thus, with the paraffin fuels, the A/F ef-
fect was strong in the fuel-rich region and weak in the fuel-lean region,
and its magnitude varied only moderately from fuel to fuel. With the aro-
matic fuels, the A/F effect was generally weak throughout the A/F range
used. With the olefin fuels, the A/F effect was strong, and its magnitude
varied considerably from fuel to fuel. These data suggest that with gaso-
lines the A/F effect may vary in magnitude from fuel to fuel depending
on aromatic and olefin content.
More direct information on the effect of A/F on gasoline exhaust re-
activity was provided by the tests with the Chevrolet engine. Results
from these tests' are shown in figure 10. These results indicate that
FIGURE 10. - Calculated R^ Reactivities of Exhaust From oasulint- Fuels
Used Under Varied Air-Fuel Ratio Conditions.
-------
24
although changes in A/F affect exhaust reactivity, the changes affect the
exhaust from different fuels in the same manner, preserving the relative
order of their reactivities and therefore, preserving relative fuel effects
on exhaust reactivity. This finding, in turn, suggests that fuel enrich-
ment of the A/F, necessary for acceptable HC/NOX in exhausts produced for
the smog chamber study, does not invalidate the conclusions from the study
regarding the effect of fuel composition on exhaust reactivity.
Automotive Combustion of Simple Hydrocarbon Fuels
Results from the simple hydrocarbon fuel tests are given in table
11. Observed exhaust reactivity data are summarized in table 12. Note
that these data represent specific reactivity of exhaust, that is, reactivity
per mole of exhaust hydrocarbon, and they should be looked at solely as
indicative of the nature of the exhaust organics that originate from the
various fuel components. Data for the n-heptane and hexene-1 fuels could
not be obtained directly because these hydrocarbons, having extremely low
octane number, could not be used alone. The exhaust reactivity values
listed in table 12 for these two hydrocarbons represent only gross estimates.
They were obtained from comparison of data from the heptane and hexene-
1 mixtures with aromatics versus data from the aromatics alone.
Regarding the main objective of these tests, the data of table 12
show that fuel hydrocarbons of varying structure yield exhaust of widely
varying reactivity. The data support deductions drawn from the other seg-
ments of this study that for purposes of predicting exhaust reactivity,
the simple classification of fuel hydrocarbons into paraffins, olefins,
and aromatics is clearly inadequate. Further, the data confirms that
structural differences within each of the constituent groups have equal
-------
25
or greater effect on exhaust reactivity than do the differences group-
to-group. Although the present data are not sufficient to permit a fully
adequate classification of fuel hydrocarbons, they nevertheless are sug-
gestive of certain trends. Thus, normal paraffins appear to yield exhaust
of relatively low reactivity, whereas cycloparaff in fuel yields exhaust
considerably more reactive. Isoparaffin fuels may very considerably in
(exhaust) reactivity, evidently, depending on their molecular structures.
Olefins of various molecular structure seem to yield exhaust of widely
varying R..- -reactivity but of almost constant ozgne-yield reactivity.
Aromatic fuels of differing aromatic make-up also yield exhaust of widely
varied reactivity, the polyalkylbenzenes appearing to yield exhaust that
is considerably more reactive than the exhaust produced from monoalkylbenzenes
and benzene. The data from the tests with the n-heptane mixtures (table
12) suggest also that the pollution potentials of the various hydrocarbons
are preserved when these hydrocarbons are used in mixture with others;
evidently there are no strong synergistic effects in this respect.
Another interesting observation from these data pertains to the degree
of agreement between observed and calculated reactivity values for exhaust
from simple fuels. Figure 11 illustrates the degree of agreement between
FIGURE 11. - Correlation of Observed and Calculated R Reactivities
of Exhaust From Simple Hydrocarbon Fuels
observed R^ -reactivity of exhaust and reactivity calculated from GLC-
composition of exhaust and the Jackson reactivity scale (8). Observed
reactivity values are considerably higher than the calculated ones. Note
that the calculated reactivity value does not include the reactivity
-------
26
contribution of formaldehyde and a portion of the contributions of the
heavier aldehydes. This is because the flame ionization detector used
in the GLC method does not respond to formaldehyde and has relatively ]ow
response to the heavier aldehydes (5). Nevertheless, levels of aldehydes,
as measured by the chromatropic acid and MBTH methods, are too low {sco
table 11) to affect noticeably the difference between observed and calcu-
lated reactivities.
These differences between observed and calculated reactivities have
been noticed previously (J3); however, the present data provide a somewhat
more precise definition of this problem. Thus, for exhaust from aromatic
fuels and for relatively more reactive exhaust the difference between ob-
served and calculated reactivities is larger than for the rest of the ex-
haust samples of this study. This explains the large amount of scatter
observed in correlations of observed and calculated reactivities of ex-
haust from fuels of varying aromatic contents (e.g. see figure 7). The
cause of this problem is not known. Nevertheless, based on experiences
in this laboratory, these investigators speculated that the higher-than-
expected reactivity of exhaust is caused, partly at least, by reactive
exhaust organics that have been totally missed or misidentifled by sampling
and the chromatographic method of exhaust analysis (5). The data from
the present study underscore once again the unreliability of the "calculation"
method as it is often used to estimate reactivity of exhaust emissions.
SUMMARY AND CONCLUSIONS
In efforts to associate automotive fuel composition with reactivity
of exhaust emissions, a number of fuels of varied composition were tested
using several automobiles operated with standard engine adjustments. The
-------
27
one investigative approach, exhaust reactivity, was calculated using a
reactivity scale and exhaust composition data. Such calculated exhaust
reactivity was then statistically correlated with fuel composition expressed
using each of several different component classification schemes. Results
showed that correlations were improved when compositional detail was added
to the simple classification as paraffins, olefins, and aromatics. A
similar conclusion was reached when levels of selected reactive exhaust
components were statistically correlated with fuel composition. Further,
from the fuel component groups considered, the group of polyalkylbenzenes
was found to be the one that correlated best with calculated exhaust re-
activity. The broader group of aromatics showed a weak correlation.
By an alternate investigative approach, fuel composition was correlated
to exhaust reactivity that, instead of having been calculated, was measured
experimentally using a smog chamber. For these smog chamber measurements
the experimental fuels were tested using one automobile that was operated
fuel-rich in order to yield exhaust with sufficiently high HC/NOX ratio.
Evidence was obtained in parallel to show that such operation of the test
engine did not invalidate the findings from this smog chamber program.
Findings in the smog chamber tests were essentially in agreement with those
of the multiple-car tests. The failure to observe a strong effect of aro-
matics on exhaust reactivity contradicted indications from a previous study
in this laboratory. In order to resolve or reduce the uncertainty posed
by the differences, fuels used in the earlier study were retested under
conditions of the present study. Results from this retesting did not pro-
vide a satisfactory explanation of the contradiction and the difference in
question was attributed to differences in experimental conditions between
the two studies.
-------
28
The polyalkylbenzenes were the only fuel component group that were
indicated by the correlations positively to have an influence on exhaust
reactivity. Other components are, nonetheless, believed to have such an
effect. Lack of strong correlation between exhaust reactivity and other
fuel component groups, such as paraffins, isoparaffins, olefins, and in-
ternal olefins, was attributed to inappropriate definition of these groups.
It is now indicated that to associate composition and exhaust reactivity,
fuel composition should be expressed either (1) in terms of individual
components or (2) in terms of component groups such that the potential
for exhaust reactivity would be the same within each group and different
from group-to-group. Classification of fuel components in terms of the
paraffins-olefins-aromatics groups apparently did not meet the latter re-
quirement. This was verified in tests with single component hydrocarbon
fuels that showed the potential for exhaust reactivity varies widely within
each of the fuel component groups of paraffins, olefins, and aromatics.
For appropriate classification of fuel components considerably more infor-
mation is needed on the fuel-derived products of combustion from automotive
engines. Given such information prediction of exhaust reactivity can be
made using either (1) a correlation derived from experimental data, or (2)
by the application of product-yield reactivity values to the fuel constit-
uents—individually and/or suitably grouped. The latter approach (2) was
tested using experimental data generated in the study. Using some gross
assumptions, fuel components were assigned reactivity ratings and exhaust
reactivity predictions were made based on these fuel component reactivities
and mole fractions. Such predicted exhaust reactivity values were found
to correlate well with those for observed reactivity.
-------
29
REFERENCES
1. Altshuller, A. P., D. L. Miller, and S. F. Sleva. Determination of
Formaldehyde in Gas Mixtures by the Chromatropic Acid Method. Anal.
Chera., v. 33, 1961, pp. 621-625.
2. Dimitriades, B. On the Function of Hydrocarbon and Nitrogen Oxides
in Photochemical Smog Formation. BuMines Rept. of Investigations
No. 7433, September 1970, 37 pp.
3. Dimitriades, B., B. H. Eccleston, and R. W. Hum. An Evaluation of
the Fuel Factor Through Direct Measurement of Photochemical Reactivity
of Emissions. J. APCA, v. 20, No. 3, March 1970, pp. 150-160.
4. Dimitriades, B. and D. E. Seizinger. A Procedure for Routine Use
in Chromatographic Analysis of Automotive Hydrocarbon Emissions.
Environmental Sci. & Technol., v. 5, No. 3, March 1971, pp. 223-229.
5. Dimitriades, B. and T. C. Wesson. Reactivities of Exhaust Aldehydes.
BuMines Rept. of Investigations No. 7527, May 1971, 18 pp.
6. Dishart, K. T. Exhaust Hydrocarbon Composition: Its Relation to
Gasoline Composition. Presented at 35th Midyear Meeting, American
Petroleum Institute Division of Refining, Houston, Texas, May 14,
1970.
7. Fleming, R, D. Effect of Fuel Composition on Exhaust Emissions From
a Spark-Ignition Engine. BuMines Rept. of Investigations No. 7423,
September 1970, 37 pp.
8. Jackson, M. W. Effects of Some Engine Variables and Control Systems
on Composition and Reactivity of Exhaust Hydrocarbons. SAE Vehicle
Emissions, Part II, v. 12 (selected paper, 1963-66), New York, 1967,
pp. 241-267.
-------
30
9. Ninomiya, J. S. and B. Biggers. Effect of Toluene Content in Fuel
on Aromatic Emissions in Exhaust. J. APCA, v. 20, No.9, September
1970, pp. 609-611,
10. Sawicki, E. T., W. Stanley, and W. Elbert. The 3-Methyl-2-Benzo-
thiazolene Hydrozone Test. Anal. Chem.,' v. 38, No. 1, January 1961,
pp. 93-96.
11. Sturm, G. P., Jr., and B. Dimitriades. Reactivity of Emissions From
Leaded and Lead-Free Fuels. Preprints, Symposia of ACS Division of
Petroleum Chemistry, Inc. (Los Angeles, Calif., March 28-April 2,
1971), v. 16, No. 2, March 1971, pp. E115-E119.
12. U.S. Department of Health, Education, and Welfare. Control of Air
Pollution From New Motor Vehicles and New Motor Vehicle Engines
(Part 1201). 35 FR 128, July 2, 1971, pp. 12,657-12,664.
13. U.S. Department of Health, Education, and Welfare. Control of -Air
Pollution From New Motor Vehicles and New Motor Vehicle Engines
(Part 85). 35 FR 219, November 10, 1970, pp. 17,228-17,313.
14. Walker, J. Q. and J. B. Maynard. Analysis of Vapor Phase Pyrolysis
Products of the Four Trimethylpentane Isomers. Anal. Chem., v. 43,
No. 12, October 1971, pp. 1,548-1,557.
15. Wigg, Eric. E., Jr., Raymond J. Campion, and Wm. Lewis Petersen.
The Effect of Fuel Hydrocarbon Composition on Exhaust Hydrocarbon and
Oxygenate Emissions. SAE Paper 720251, Automotive Engineering Congress,
Detroit, Michigan, January 10-14, 1972, 13 pp.
-------
TABLE 1. - Experimental fuels used in the gasoline fuel test program
Specifications
RVP
Distillation, °F:
IBP . .
End
GLC composition, mole percent:
Paraffin.
Olefin
I
Regular
10.4
.725
0
I4i
88
122
210
304
374
67.7
9.0
23.3
8.8
12.1
21.4
41.4
4.8
2
Regular
9.1
.747
0
^91
96
134
234
340
404
60.5
7.6
31.9
10,0
20.6
16.3
38.9
5.3
3
Regular
11.3
.746
.35
±'94
100
136
240
376
418
69.1
6.7
24.2
3.3
20.3
21.5
37.6
10,0
Fuel N
4
Regular
10.2
.747
.34
Hi
106
126
220
309
435
58.8
13.0
28.2
6.8
20.2
18.1
34.9
5.8
3.
5
Regular
9.3
.731
1.90
i/94
105
125
199
324
415
64.2
10.2
25.6
6.6
18.3
22.9
36.1
5.1
6
Regular
8.3
.725
1.80
i/94
92
128
198
308
392
67.0
6,2
26.8
9.1
15.3
19.5
42.4
5.1
7
Special
10.1
.707
0
92
88
122
218
300
382
81.0
4.5
14.6
4.3
8.2
21.3
58.0
1.6
8
Special
11.5
.755
0
97
88
110
215
308
378
48.2
6.6
45.2
14.3
23.9
23.6
22.2
2.4
I/ Estimates of RON from fuel specified.
-------
TABLE 1. - Experimental fuels used in the gasoline fuel test program (Cont'd)
Specifications
RVP
Distillation, °F:
IBP
End
GLC composition,. mole percent:
9
Special
10.0
.738
0
95
86
118
230
308
366
56.8
7.4
35.8
11.1
19.8
19.8
35.0
1.9
10
Premium
12.8
.755
0
^100
86
120
230
326
364
50.6
1.2
48.2
33.8
14.2
14.8
35.3
c
11
Premium
10.1
.771
0
4oo
92
140
240
318
378
45. S
2.1
52.0
28.5
23.0
11.4
34.0
.5
Fuel N
12
Special
8.4
.731
0
92
108
136
186
284
380
57.7
9.7
32.6
12.6
12.8
20.0
34.7
3.0
o.
13
Special
7.7
.731
0
92
100
138
186
254
382
56.6
9.8
33.7
30.3
3.1
19.3
34.2
3.0
14
Special
7.9
.731
0
93
102
134
202
292
378
59-0
10.1
31.0
2.4
28.3
20.2
35.7
3.1
15
Special
7.7
.762
0
95
106
146
226
282
350
44.7
4,9
50.4
24.6
25.6
23.1
18.9
2.7
16
Special
7.5
.712
0
94
108
136
168
246
288
71.4
0
28.6
12.0
9.7
17.7
52.6
1.2
I/ Estimates of RON from fuel specified.
-------
TABLE 1. - Experimental fuels used in the gasoline fuel test program (Cont'd)
Specifications
Grade .
RVP
Specific gravity
TEL, ml/gal ....
Research octane No,
Distillation, °F:
IBP
10 percent . . .
50 percent . . .
90 percent . . .
End
GLC composition,
mole percent:
Paraffin ....
Olefin
Aromatic ....
Monoalkylbenzenes .
Polyalkylbenzenes .
n-Paraffins ....
Isoparaffins . . .
Cycloparaf fins . ,
17
Special
8.7
.748
0
89
105
135
213
324
392
43.5
19.3
37.2
13.6
16.2
21.6
17.6
4.2
18
Special
7.7
.745
0
96
103
140
200
267
294
51.5
0
48.5
23.4
18.7
22.3
26.3
2.9
19
Special
8.1
.751
0
90
100
136
188
288
370
62.1
8.7
29.2
12.5
11.9
20-0
10.3
31.9
20
Special
14-0 •
.723
0
91
90
112
186
294
382
59-0
9.2
31.8
12.6
12.8
31.8
24.2
3.1
Fuel I
21
Special
8.2
.761
0
93
107
138
222
291
376
43.2 .
9.2
47.7
13.8
27-0
24.2
15.4
3.6
Jo.
22
Premium
10.0
.728
2.6
100
88
117
221
308
381
61. -4
11.1
27.6
20.3
7-0
23.1
37.3
1-0
23
Regular
9.9
.738
2-0
95
90
120
212
332
377
62.3
15.9
21.9
6.6
14.9
13.1
39.9
9.3
24
Premium
11-0
.769
0
102
87
117
238
315
370
39.9
10.7
50-0
19.6
27.8
2.3
36.1
.9
25
Regular
10.1
.744
0
98
88
115
204
313
369
48.5
17.2
34.3
10.1
22-0
5.8
40.2
2.6
-------
34
TABLE 2. - Experimental fuels used in the simple
hydrocarbon fuel testing program
H-Butane
2,2,4-TrimethyIpentane
2,3,4-Trimethylpentane
Me thyIcyclopentane
Butene-1
cis and trans Hexenes - 2 and 3
2,4,4-Trimethylpentene-l
2,4,4-Trimethylpentene-2
Benzene
Toluene
m-Xylene
sec-Butylbenzene
Hexene-1 + benzene
n-Heptane + benzene
n-Heptane + toluene
n-Heptane + m-xylene
n-Heptane + 2,4,4-trimethylpentene-2
-------
TABLE 3. - Test automobiles
Vehicle
No.
1
2
3
4
5
6
7
8
9
10
11
Make
Chevrolet
Pontiac
Rambler
Oldsmobile
Chevrolet
Plymouth
Ford
Chevrolet
Ford
Plymouth
Plymouth
Model
Impa la
Catalina
Ambassador
Cutlass
Impa la
Fury
Fair lane
Biscayne
Custom
Fury I
Valiant
Year
1968
1968
1969
1969
1970
1970
1970
1971
1971
1971
1969
Engine
displacement ,
CID
327
400
290
350
350
t
318
351
350
351
383
225
Compression
ratio
8.8
8.6
9.0
10.3
9.0
8.8
9.5
8.5
9.0
8.7
8.4
NOTE: Vehicles 1 through 4 were equipped to meet the California
emission standards of 1968-1969-
Vehicles 5 through 10 were equipped with evaporative control
devices.
-------
TABLE 4. - Mass emissions and molar_sp_eclflc reactivity of exhaust in first phase
of fleet testing program
(Each value is average of three tests)
Vehicle
No.
Fuel No.
1 1 4 | 5 | 6 | 7 | 8 | 9
10
Mean
Std.
dev. I/
Coeff.
var.
HYDROCARBON, g/mile
1
2
3
4
5
6
7
8
9
Mean
Std.dev.j^
Coeff.
var.
4.02
3.91
3.77
2.84
2.53
5.70
3.52
2.51
3.13
3.55
.26
7.3
4.16
' 3.71
3.96
2.88
2.55
4.48
3.20
2.36
3.43
3.41
.23
6.7
4.15
4.09
3.68
3.03
2.38
4.82
3.42
2.62
3.85
3.56
.24
6.7
4.05
3.89
3.61
2.99
2.41
4.57
3.61
2.21
3.83
3.45
.19
5.5
4.04
3.90
3.76
2.91
2.40
5.12
3.43
2.46
3.58
3.51
.21
6.0
3.98
3.51
3.81
3.38
2.59
5.23
3.29
2.40
3.94
3.57
.21
5.9
3.86
3.36
3.73
3.04
2.54
5.22
3.46
2.10
3.37
3.41
.26
7.6
3.84
3.96
4.14
3.24
2.42
8.09
3.54
3.17
3.81
4.02
.29
7.2
4.02
3.79
3.81
3.04
2.48
5.40
3.43
2.48
3.62
0.20
.17
.19
.22
.29
.36
.17
.20
.20
5.0
4,6
5.1
7.1
11.6
6.7
4.9
7.9
5.5
CARBON MONOXIDE, g/mile
1
2
3
4
5
6
7
8
9
Mean
Std.dev.L
Coeff.
var.
37
38
40
48
42
71
34
30
30
41.2
' 4.3
10,4
34
32
39
47
35
55
29
28
35
37.0
3.0
8.1
32
34
35
44
28
54
32
26
34
35.5
4.5
12.6
32
32
36
54
32
48
33
24
33
36.2
4.0
11.0
29
29
34
53
38
59
34
34
37
38.5
4.2
10.8
26
27
40
56
48
70
34
35
49
42.8
6.3
14.6
25
21
34
43
45
62
33
28
34
36.2
6.5
17.8
30
34
42
57
36
106
38
53
47
49.7
5.0
10.0
30.8
30.7
37.6
50.2
38.0
65.7
33.2
32.4
37.4
4.0
2. .8
4.4
4.7
8.0
5.3
3.2
3.6
3.8
13.0
9,2
11.6
9.3
21.0
8.0
9.8
11.2
10.3
I/ Standard deviation calculated from the chree replicates on each car fuel combination.
ON
-------
TABLE 4. - Mass emissions and molar specific reactivity of exhaust in first phase
of^fleet testing program (cont'd)
(Each value is average of three tests)
Vehicle
No.
1 1 <
» 1 5 |
Fuel
6 1
No.
7 1
8 1
9 | 10
Mean
Std.
dev. lj
Coeff
var.
NITRIC OXIDE, as NO2, g/mile
1
2
3
4
5
6
7
8
9
Mean
Std. dev. I/
Coeff.
var.
6.7
10. 0
5.6
5.0
2.6
5.1
3.7
3.6
4.1
5.2
.5
9.1
. 8.2
10.8
5.8
5.7
3.0
5.5
3.7
3.5
4.9
5.7
.5
8.5
6.7
11.5
5.9
5.4
2.7
5.6
3.6
3.4
5.3
5.6
.5
9.3
7.3
10.1
6.2
4.9
2.6
5.8
3.3
3.3
5.2
5.3
.6
11.0
7.9
13.0
7.1
4.6
2.2
5.0
3.1
3.3
4.0
5.6
.7
11.8
6.5
8.6
5.5
4.3
2.6
5.5
3.9
3.7
4.9
5.1
.7
13.0
8.4
9.8
6.5
5-. 5
2.5
5.2
3.4
3.5
4.6
5.5
.5
9.0
7.4
11.4
5.8
5.6
2.6
4.5
3.6
3.7
4.5
5.5
.4
7.3
7.4
10.7
6.1
5.1
2.6
5.3
3.5
3.5
4.7
0.7
1.1
.5
.3
.2
.3
.3
.2
.5
9.8
10.1
7.9
5.1
8.9
5.4
7.9
6.1
10.7
TOTAL ALDEHYDES, g/mile
1
2
3
4
5
6
7
8
9
Mean
Std. dev. !/
Coeff.
var.
0.18
.22
.16
.15
.11
.19
.18
.19
.21
.18
.01
5.6
0.18
.22
.15
.16
.12
.18
.20
.16
.24
.18
.02
11.1
0.16
.22
.15
.15
.14
.16
.16
.17
.25
.17
.03
17.6
0.14
.21
.15
.14
.12
.18
.20
.14
.25
.17
.02
11.8
0.18
.25
.15
.16
.13
.20
.23
.20
.22
.20
.03
15.0
0.21
.27
.22
.19
.10
.15
.16
.14
.18
.18
.02
11.1
0.18
.22
.14
.16
.12
.19
.19
.18
.21
.18
.02
11.1
OU5
.20
.14
.15
.11
.15
.18
.13
.16
.15
.02
13.3
0.17
.23
.16
.16
.12
.17
.19
.16
.21
0.02
..02
.02
.01
.01
.02
.02
.02
.03
8.9
7.0
10.3
7.9
12.4
10.4
11.8
10.6
16.1
JL/ Standard deviation calculated from the three replicates on each car fuel combination,
-------
of fleet testing program (cont'd)
Vehicle
No.
(Each value is
, Fuel No.
1 | 4 | 5 | 6 | 7
average of three tests)
Std. Coeff.
1 8 | 9 | 10 Mean d«,w. If var.
EXHAUST HYDROCARBON REACTIVITIES, ETHYLENE EQUIVALENT, g/mile
1
2
3
4
5
6
7
8
9
Mean
Std.dev.l/
Coeffc
/var.
2.16
2.17
2.02
1.59
1.39
2.94
1.94
1.44
1.81
1.76
.09
4.3
2.30
2.12
2.13
1.65
1.41
2.47
1.83
1.46
1.91
1.93
.13
6.2
2.31
2.38
2.06
1.72
1.37
2.73
2.03
1.57
2.28
2.05
.12
5.8
2.11
2.09
1.85
1.54
1.30
2.41
1.99
1.25
2.13
1.84
.10
5.4
2.17
1.98
2.03
1.77
1.29
2.58
1.84
1.46
2.02
1.90
.10
5.1
1.80
1.63
1.81
1.40
1.18
2.42
1.62
1.16
1.90
1.66
.09
5.4
2.05
2.03
1.91
1.40
1.31
2.51
1.75
1.09
1.79
1.76
.13
7.6
1.65
1.77
1.74
1.42
1.08
3.17
1.60
1.47
1.72
1.73
.09
5.1
2.09
2.02
1.94
1.56
1.29
2.65
1.83
1,36
1.95
0.09
.09
.09
.10
.14
.11
.08
.11
.12
4.1
4.7
4.7
6.6
10.7
4.1
4.5
8.3
6.0
iy Standard deviation calculated from the three replicates on each car fuel combination.
00
-------
TABLE 5. - Mass emissions and molar specific reactivity of exhaust
in second phase of fleet testing program
[Each value is an average of three tests, time weighted, according
to the cold start transient and stabilized portions of the
Federal test procedure for 1975 automobiles.1
Vehicle
No.
Fuel No.
12
13
14
15
16
17
18
19
20
21
Mean
Std .!/
dev.
Coeff.
var.
EXHAUST HYDROCARBON, g/mile
5
6
7
9
10
Mean
Std,devJ/
Coeff.
var.
5
6
7
9
10
Mean
Std. dev.V
Coeff.
var.
5
6
7
9
3.56
4.41
3.07
2.87
3.16
3.41
.29
8.4
71
143
40
39
42
47-1
3.4
7.3
3. .48
5.10
3.26
2.67
3.30
3.55
.29
8.0
71
46
43
35
45
48-2
10.2
4.5
6.9
5.0
6.1
10 6.1
Mean
Std. devi'
Coeff.
var.
5.7
.2
3,3
4.1
6.8
4.7
6.1
5.8
5.5
.3
6.0
3.84
4.74
3.07
2.88
3.28
3.56
.33
9.3
3.74
4.86
3.37
3.00
3.20
3.63
.21
5.9
72
46
40
38
42
47.4
4.2
8.9
70
46
44
36
41
48.5
3.4
7.2
3.72
5.02
3.75
3.12
3.37
3.80
.14
3.7
3.54
4.10
2.83
2.78
2.88
3.23
.20
6,1
3.66
4.89
3.54
3.08
3.43
3.72
.23
6.2
3.61
4.61
3.03
2.72
2.99
3.39
.13
3.8
3.45
4.18
2.97
2.73
3.08
3.28
.18
5.4
3.48
4.22
3.24
2.96
3.00
3.38
.07
2.0
3.61
4.61
3.21
2.88
3.17
0.21
.30
.20
.17
.18
5.7
6.6
6.1
5.9
5.6
EXHAUST CARBON MONOXIDE, g/mile
66
44
44
41
42
48.8
2.0
4.1
65
44
38
36
39
44.4
3.5
7.8
65
46
43
42
46
48.4
7.3
74
48.
44
40
44
6.5
EXHAUST NITRIC OXIDE, as NO?
4.2
6.5
4.7
6.1
5.7
5.4
.1
2.6
4.9
7.2
5.6
6.7
6.5
6.2
.3
4.0 j
4.1
6.8
5.0
5.8
5.5
5.4
.2
4.3
5.0
7.3
5.3
6.9
6.7
6.3
.3
4.1
4.8
7.0
5.5
6,2
6.1
5.9
.3
5.5
4.6
7.2
5.1
6.1
6.3
5.9
.3
4.5
62
44
40.
40
47
4.4
9.4
62
45.
44
W
42
1.2
2.7
67.9
45.3
42.0
38.7
43.1
5 1
2 4
2.9
3.4
2.9
R/mile
4.6
6.8
5.3
6.0
5.7
5.7
.1
1.6
4.9
7.2
5.6
6.6
6.6
6.2
.3
4.7
4,6
7.0
5.2
6.3
6.1
.3
.1
.3
.3
.2
7.5
5.2
6.9
8.8
6.8
5.7
1.7
5.4
4.8
3.4
I/Standard deviation calculated from the three replicates on each car fuel combination.
-------
TABLE 5. - Mass emissions and molar specific reactivity of exhaust
in second phase of fleet testing program (Continued)
[Each value is an average of three tests, time weighted, according
to the cold start transient and stabilized portions of the
Federal test procedure for 1975 automobiles.]
Vehicle
No.
5
6
7
9
10
Mean
Std. dev.l/
Coeff.
var.
Fuel No.
12
0.11
.18
.16
.18
.16
.16
.01
8.7
13
14
15
16
17
18
19
EXHAUST TOTAL ALDEHYDES, g/mile
0.11
, .17
.14
.18
.15
.15
.02
10.1
0.10
.19
.14
.17
.17
.15
.01
9.4
0.10
.16
.15
.17
.14
.14
.01
9.1
0.13
.17
.19
.19
.17
.17
.02
9.4
0.12
.17
.14
.17
.16
.15
.02
10.0
0.13
.16
.15
.16
.14
.15
.02
11.0
0.11
.17
.13
.16
.14
.14
.01
9.2
20
0.13
.17
.15
.17
.15
.15
.02
11.7
21
0.09
.15
.15
.15
.14
.14
.01
9.5
Mean
0.11
.17
.15
.17
.15
.15
Std. I/
dev.
0.01
.02
.01
.01
.01
Coeff.
var.
13.3
8.9
8.9
8.7
9.8
JACKSON MOLAR SPECIFIC REACTIVITY, ethylene equivalents 2/
5 0.87 0.86 0.97 0.86 0.87 0.95 0.83 0.87 0.90 0.91 0.89 0.03 J.!>
6 .96 .92 1.03 .90 .93 1.00 .83 .91 .95 .96 .94 .03 2.7
7 .95 .89 1.03 .86 .85 .99 .85 .91 .91 .90 .91 .03 3.7
9 1.02 1.04 1.11 1.00 .94 1.06 .90 .94 1.02 .96 1.00 .03 3.0
10 .95 .92 1.05 .94 .93 1.00 .86 .93 .91 .97 .95 .03 3.5
Mean .95 .92 1.04 .92 .90 1.00 .86 .91 .94 .94
Std. dev.l/ .04 .02 .02 .02 .03 .03 .04 .04 .02 .02
Coeff.
var A 2 2.6 2.3 2.6 3.2 3.4 4.6 4.2 2.1 2.6
I/ Standard deviation calculated from the three replicates on each
2/ Molar reactivity of ethylene = 1.
-------
TABLE 6. - Correlation of exhaust composition with fuel composition for car NO. 5
Exhaust composi-
tion parameter
(Y)
Ethylene
2-Methyl-2-butene
R
Jackson
R
EPA
Total aldehydes
Total ppm hydrocarbon ....
Total reactive
grams (ethylene equivalents) .
Mean
(Y)
0 187
.0728
.0177
.0047
0596
.0324
8610
2 467
0347
OS9S
88.8
11.61
Standard
CSY)
0 032
0116
0076
.0019
0359
.0194*
0554
122
0099
01 47
15.5
2.25
Model
i y
0 50
68
46
.002
66
.41
LU
12
n
1 ^
.02
.15
Rso^
Model
II 2f
0 52
f.Q
46
.073
Qf,
80
S7
^^
1 7
1 A
.09
.18
./
Model
III 21
fl 5?
fiq
47
.007
fifi
48
A5
1 ";
1 q
1 c:
.07
.21
Model
IV 21
0 S T
£Q
47
.008
Q5
g4
Q7
i 7
.11
.21
_!./ R0/v = The square of the multiple correlation coefficient.
sQ
21 Model I: Y = a (mole fraction [MFl paraffins) + a. (MF olefins) +
a^ (MF aromatics)
Model II: Y = a. (paraffins) + a? (olefins) + a. (benzene) + a, (monoalkyl-
benzenes) 4- a_ (polyalkylbenzenes)
Model III: Y = a. (ri-paraffins) + a_ (isoparaffins) + a (olefins) +
a, (aromatics)
Model IV: Y = a. _(n_-paraffins) + a (isoparaffins) + a. (olefins) +
a, (benzene) + a (monoalkylbenzenes) + a, (polyalkylbenzenes)
-------
42
TABLE 7. - Correlation of exhaust composition with fuel composition,
car dummy variables, and air-fuel ratio
[Data from second phase of fleet testing program]
Exhaust composition
parameter
Model 1^
Model II
Model
Ethylene 0.41
Propylene .28
iso + 1-Butenes .35
2-Methyl-2-butene .19
Toluene .44
m + £-Xylenes .42
RJackson •26
REPA
Formaldehyde .04
Total aldehydes .007
Total ppm hydrocarbon .07
Total reactive
grams (ethylene equivalents) . .
0.85
.59
.71
.33
.47
.44
.72
.71
.73
.50
.84
.75
0.45
.40
.46
.31
.44
.42
.35
.24
.39
.30
.18
.16
_!/ The square of multiple correlation cofficient.
2J Model I: Y = a^ (paraffins) + a2 (olefins) + a3 (aromatics)
Model II: Y = a (paraffins) + &<^ (olefins) + a-j (aromatics) +
a4 (car6) + a5 (car?) + a6 (carg) + a? (car10)
Model III:Y = a (paraffins) + ao (olefins) + 83 (aromatics) +
4 (A/F)
-------
TABLE 8. - Smog chamber reactivities of exhaust from fuels 1-21
1
No.
Initial chamber
concentrations
HC
X
ppmC
1
2
3
4
5
6.02
6.06
5.88
5.90
6.07
6.06
5.96
5.87
6.05
6.07
6.08
6.09
5.92
6.09
6.00
6.27
6.07
6.20
6.06
6.05
6.11
6.03
6.04
6.06
6.01
6.03
6.04
6.05
6.06
NO
a\j
i '
i
ppm
HC
NO
X
Reactivity data
R
N02,
ppb/min
Maximum, ppm
°3
PAN
PPN
HCHO
, .Dosages, ppm x min
0 —
PAN
PPN
HCHO
6 ppmC : 1 ppm NOX
0.102
.105
.111
.099
.089
.107
.102
.106
.101
.107
.103
.105
.103
.098
.099
.109
.107
.110
.105
.106
.095
.095
.097
.089
.092
.095
,108
.114
.103
6.05
6.10
5.54
5.84
6.04
6.05
5.92
5.77
5.99
5.97
5.86
6.01
5.84
5.98
5.90
6.30
5.90
5.96
6.10
6.03
6.05
6.25
5.93
6.14
5.87
5.98
5.94
6.58
6.04
13.01
12.88
14.28
13.73
11.86
14.85
15.41
15.04
11.97
12.49
11.66
12.32
13.45
14.30
12.90
13.27
15.39
14.93
13.12
11.84
13.25
12.30
11.81
10.23
11.01
11.35
11.92
10.77
11.51
0.92
.90
.97
.85
1.03
.96
.98
.83
.91
.98
.98
1.04
.94
.90
1.02
.94
.96
.87
1.08
.96
.97
.87
.90
.86
.89
.86
.86
.92
.88
0.094
.099
.129
.102
.101
.117
.118
.113
.089
.086
.073
.077
.114
.117
.083
.122
.120
.121
.116
.114
.102
.123
.097
.089
.074
.106
.091
.100
,087
0.017
.017
.024
.019
.015
.021
.022
.020
.021
.019
.013
.015
.020
.022
.018
.025
.025
.026
.021
.024
.023
.029
.022
.024
.017
.026
.027
.028
.025
0.76
.81
.78
.84
.72
.88
.84
.74
.66
.68
.64
.96
.73
.87
.65
.72 -
.67
.75
.76
.81
.90
.81
.66
.65
.62
.66
.74
,72
.67
134
134
163
149
151
166
182
143
133
147
136
152
156
165
159
139
171
160
194
152
153
142
134
119
127
123
127
146
126
14.25
14.91
20.73
17.21
14.66
19.62
21.53
19.81
13.06
13.23
10.63
12.14
18.93
20.97
13.27
18.10
21.56
21.32
19.41
17.66
15.87
19.12
14.19
13.86
10.61
15.23
13. L7
15. A2
I2.r>7
2.60
2.69
3.63
3.22
2.10
3.53
3.86
3.34
3.04
3.07
1.93
2.31
3.46
3.91
2.72
3.80
4.38
4.51
3.49
3.75
3.43
4.32
3.35
3.43
3.39
3.84
3.97
4.20
*.5b
202
198
211
226
195
195
217
202
177
178
166
232
183
215
175
187
173
199
209
202
206
208
168
158
154
170
201
196
178
Measured with the chemiluminescence oxonc detector.
-------
TABLE 8. - gmog chamber reactivities of exhaust from fuels 1-21 (Cont'd)
Fuel
No.
6
7
8
9
10
11
12
Initial chamber
concentrations
HCiS
ppmC
6.02
6.04
6.00
6.15
6.04
5.97
6.01
6.09
5.98
5.99
6.02
6.04
6.05
6.00
6.00
6.09
6.06
5.99
6.01
5.96
5.99
6.16
6.03
6.08
6.03
5.93
5.97
NO
V
ppm
HC
N0r
Reactivity data
RN00,
2
ppb/min
Maximum , ppm
o,i/
PAN
PPN
HCHO
Dosages , ppm x min
0)*
6 ppmC:l ppm NOX (cont'd)
0.101 '
.101
.104
.102
.088
.092
.094
.098
.096
.091
.096
.100
.103
.104
.101
.098
.096
.089
.086
.100
.087
.104
.106
.103
.102
.101
.109
5.90
5.86
5.88
6.18
6.00
5.93
6.18
6.04
5.88
5.94
5.99
5.83
6.02
6.02
5.93
6.05
6.13
6.07
5.62
6.03
5.73
6.14
5,88
6.02
6.03
5.94
5.92
11.87
11.58
12.36
12.38
11.09
11.11
11.04
11.17
12.47
10.96
11.52
11.54
10.42
11.57
11.94
11.86
10.26
10.37
10.60
9.44
9.13
11.44
12.70
12.58
11.5
11.8
11.4
1.04
1.01
.95
.84
.73
.70
.81
.87
.87
.89
.84
.88
.88
.89
.88
.91
.84
.78
.69
.72
.66
.94
.96
.85
.68
.74
.63
0.066
.081
.091
.101
.058
.058
.082
.079
.088
.091
.088
.098
.106
.088
.092
.103
.083
.060
.055
.072
.058
.116
.120
.122
.07
.07
.06
0.016
.014
.019
.021
.013
.012
.018
.016
.015
.016
.014
.018
.02
.018
.017
.02
.016
.009
.009
.010
.010
.013
.014
.013
.007
.007
.006
0.88
.79
.84
.84
.68
.65
.66
.76
.56
.61
.61
.58
.59
.85
.69
.78
.66
.63
.55
.58
.57
.74
.66
.65
.55
.59
.54
146
140
142
132
96
95
113
121
130
137
125
136
126
136
128
140
113
103
86
93
79
152
163
158
92
102
83
PAN ] PPN
10.15
12.74
13.46
15.38
7.23
8.77
10.94
10.87
12.92
14.99
13.51
15.20
15.62
13.75
13.50 ,
15.38
11.89
9.29
8.47
9.56
8.31
18.97
20.26
21.75
10.23
9.77
9.03
2.15
2.13
2.72
3.13
1.66
1.80
2.30
2.17
2.27
2.58
2.43
2.64
2.71
2.68
2.44
2.99
2.21
1.44
1.26
1.29
1.16
2.16
2.18
2.32
.97
1.05
.89
HCHO
209
194
224
210
188
187
182
217
147
157
159
153
159
203
174
195
176
153
141
141
154
160
157
183
152
164
159
I/ Measured with the chemiluminescence ozone detector.
-------
TABLE 8. - Smog chamber reactivities of exhaust from fuels 1-21 (Cont'd)
Fuel
No.
13
14
15
16
17
18
19
20
21
Initial chamber
concentrations
HCit
ppmC
6.01
5.95
6.03
5.97
6.05
6.00
5,98
6.00
6.05
6.08
6.05
6.05
6.00
5.99
6.00
6.01
6.07
6.02
NO,
V
ppm
0.091 '
.099
.100
.103
.099
i .104
.101
.104
.105
.103
.100
.105
.103
.103
.100
.103
.097
.100
HC
NO
X
6.40
5.88
5.59
5.83
5.81
5.89
5.93
5.95
5.92
6.12
5.85
5.92
6.00
5.90
5.98
5.93
5.80
5.99
Reactivity data
RN00,
2
ppb/min
Maximum . ppm
o,I/
PAN
PPN
HCHO
Dosages , ppm x rain
V
6 ppmC:l ppm NOX (cont'd)
10.3
9.7
10.9
12.5
11.2
10.8
10.5
10.3
11.5
11.3
10.6
9.5
10.3
11.0
11.2
10.5
10.2
10.5
0.60
.52
.80
.93
.81
.93
.64
.63
.68
.79
.67
.66
.61
.75
.64
.70
.69
.87
0.05
.04
.10
.11
.10
.09
.06
.05
.08
.07
.06
.05
.05
.05
.06
.05
.08
.08
0.006
.005
.008
.008
^ . 006
,006
.003
.003
.013
.010
.003^
.003
.006
.006
.006
.006
.007
.007
0.64
.72
.77
.67
.71
.48
.60
.58
.62
.79
.54
.45
.60
.52
.71-
.56
.75
.52
72
58
107
135
111
125
76
73
90
102
81
77
75
95
78
87
84
111
PAN | PPN
6.44
5.20
13.87
15.56
13.66
12.92
7.79
6.32
10.58
9,29
8.38
6.83
6.59
6.73
7.49
7.43
10.57
11.42
0.76
.65
1.07
1.18
.86
.75
.43
.34
1.61
1.39
738
.33
.75
.76
.80
.80
.98
.96
HCHO
191
185
200
182
185
119
176
154
169
164
149
117
160
141
185
159
167
141
I/Measured with the chemiluminescence ozone detector.
-------
TABLE 8. - Smog chamber reactivities of exhaustfrom fuels 1-21 (Cont'd)
Fuel
No.
Initial chamber
concentrations
ppmC
•v
ppm
HC
NO
x
Reactivity data
V,,
ppb/min
Maximum
°3~ PAN
, ppm
PPN
HCHO
Dosages, ppm x min
°3~ 1 PAN | PPN
HCHO
3 ppmC: 1/2 ppm NO.
x
1
2
3
4
5
6
11
3.00
3.02
3.00
2.96
3.07
3.04
3.03
3.04
3.06
3.02
3.00
3.03
3.04
3.03
3.02
3.02
3.04
3.04
3.00
3.01
2.93
2.99
3.02
3.03
3.06
2.98
3.02
3.00
0.050
.053
.045
.048
.048
.046
.048
.045
.050
.049
.045
.044
.046
.045
.049
.046
.046
.050
.047
.048
.045
.046
.047
.044
.050
.048
.047
.042
6.04
X6.01
"6.04
6.08
6.38
5.94
6.46
6.24
6.13
6.03
6.17
6.19
6.16
5.99
6.02
6.13
6.02
6.23
5.99
6.29
6.08
6.00
5.97
5.97
6.13
5.91
6.26
6.15
6.11
6.13
5.62
6.60
5.62
6.36
5.95
6.40
5.85
6.98
6.76
6.11
6.50
6.43
7.00
7.63
7.19
6.46
6.21
6.48
6.00
6.49
6.04
6.33
5.60
6.51
5.28
6.35
0.53
.68
.54
.63
.53
.61
.63
.65
.56
.68
.62
.60
.64
.62
.62
.69
.62
.61
.60
.68
.63
.62
.61
.61
.58
.66
.62
.63
0.041
.049
.040
.050
.040
.047
.051
.057
.040
.051
.049
.049
.046
.053
.051
.051
.047
.046
.046
.051
.045
.049
.043
.045
.043
.050
.045
.053
0.008
.009
.008
.010
.008
.009
.011
.012
.008
.010
.009
.009
.011
.013
.011
.011
.010
.011
.012
.013
.010
.010
.009
.010
.005
.006
.004
.006
0.41
.38
.45
.40
.33
.33
.47
.40
.45
.41
.42
.40
.43
.40
.41
.38
.56
.33
.35
.38
.42
.38
.37
.41
.28
.34
.39
.40
80
105
80
104
80
95
101
107
86
115
105
96
106
101
107
119
102
97
94
114
98
100
92
94
84
110
93
109
6.40
7.97
6.37
8.23
6.65
7.70
8.69
9.64
6.37
8.76
8.54
7.88
8.15
8.68
8.85
9.08
7.96
7.29
7.34
8.68
7.15
7.78
6.30
6.87
7.13
9.09
7.45
9.83
1.14
1.39
1.14
1.50
1.17
1.47
1.72
1.88
1.26
1.62
1.58
1.36
1.73
1.96
1.91
1.95
1.73
1.66
1.68
1.96
1.58
1.56
1.34
1.40
.77
.93
.85
1.02
106
106
112
112
89
84
113
108
97
119
109
109
100
101
108
95
114
80
104
108
117
109
105
115
73
95
100
105
J^/Measured with the cheroiluminescence ozone detector,
-------
TABLE 8. - Smog chamber reactivities of exhaust from fuels 1-21 (Cont'd)
Fuel
No
12
13
14
15
16
17
18
19
20
21
Initial chamber
concent- rat- ions
HC^,
ppmC
2.66
3.01
3.04
3.03
2.98
3.01
3.01
3.01
3.01
3.00
3.01
3.03
3.04
3.03
3.01
3.13
3.00
3.07
3.09
3.08
3.06
3.00
NO.
V
ppm
0.049
.052
.054
.053
.053
.055
.055
.059
.059
.051
.051
.050
.054
.049
.051
.055
.056
.061
.054
.054
.048
.055
HC
NO
X
5.40
5.96
5.73
6.17
5.96
6.02
5.89
6.15
5.98
5.93
5.93
6.05
6.08
5.98
6.04
6.12
5.90
6.18
6.19
5.94
6.04
5.78
Rparfivi tv data
^0 ,
ppb /min
Maximum, ppm
o3i/
PAN
PPN
HCHO
3 ppmC:l/2 ppm NO
X
6.55
6.19
6.38
6.19
6.58
5.56
6.71
6.40
6.32
6.15
6.97
6.12
6.21
6.04
6.94
6.99
6.74
6.33
5.42
6.35
6.52
5.76
0.38
.47
.42
.51
.52
.54
.56
.58
.50
.48
.53
.51
.52
.54
.55
.60
.52
.53
.46
.56
.56
.45
0.023
.031
.025
.037
.049
.040
.048
.045
.034
.032
.038
.036
.037
.040
.031
.042
.038
.038
.036
.044
.053
.039
0.002
.003
.003
.004
.004
.004
.003
.003
.002
.002
.006
.005
.002
.002
.004
.005
.004
.004
.004
.004
.005
.003
0.48
.37
.35
.43
.33
.45
.35
.34
.36
.33
.45
.46
.38
.39
.47
.49
.39
.43
.41
.41
.37
.37
Dosages, ppm x min
V
61
70
58
78
82
84
87
97
73
71
78
76
77
80
83
94
78
79
65
83
88
66
PAN 1 PPN
3.43
4.65
4.20
5.61
7.93
6.16
7.80
7.24
5.02
4.68
6.02
5.31
5.74
5.95
4.76 '
6.29
5.94
5.48
5.02
6.84
8.03
5.78
0.29
.44
.36
.52
.62
.56
.44
.45
.23
.21
.81
.68
.30
.29
.51
.66
.59
.54
.49
.57
.68
.49
HCHO
155
103
97
120
9J
122
100
99
98
109
127
117
109
104
130
124
110
112
112
107
103
97
I/Measured with the chemiluminescence ozone detector.
-------
TABLE 9. - Smog chamber reactivities of exhaust from fuels 22-25
Fuel
No.
Initial chamber
concentrations
HC.,
ppmC
«%>
ppm
HC
N0x
REACTIVITY DATA
RN02,
ppb/min
Maximum, ppm
'?
PAN
PPN
HCHO
Dosages , ppm x rain
'a1'
PAN 1 PPN
HCHO
6 ppmC:1 ppm NOX
22
23
24
25
5.96
6.05
6.01
6.02
6.04
6.04
6.05
5.96
0.099
.104
.103
.102
.104
.109
.100
.101
5.92
6.06
5.97
6.04
5.97
5.95
6.13
5.97
10.5
10.8
12.4
11.6
10.9
10.7
12.7
10.9
0.68
.62
.80
.75
.71
.77
.71
.72
0.07
.06
.09
.09
.10
.12
.08
.10
0.008
.008
.012
.012
.008
.010
.011
.011
0.72
.67
.75
.89
.67
.62
.68
.71
87
72
122
106
101
115
102
99
9.48
8.26
13.71
13.17
14.44
18.28
12.44
14.62
1.12
.98
1.74
1.19
1.15
1.54
1.58
1.69
189
177
192
204
174
166
190
204
I/ Measured with the chemiluminescence ozone detector.
00
-------
49
TABLE 10. Bureau of Mines rate-of-NC^-formation
reactivity scale l_l
Exhaust component
or group
Bureau of Mines specific
%K)2 reactivity in
propylene equivalents
Ethylene
Propylene
1-Alkenes
Int-Alkenes
Diolefins
C4+ Paraffins-/
Benzene
0 /
R-benzyls—'
R-nonbenzyIs —'
R^-benzenes —'
Formaldehyde
Aliphatic aldehydes
£-ToluaIdehyde
Aromatic aldehydes
0.47
1.00
.65
3.78
1.41
.18
.15
.49
.45
1.19
.79
1.41
5.98
.19
I/ Smog chamber experiments involving several
typical exhaust aldehydes as reported in
reference ,5 and reported results involving
19 typical exhaust hydrocarbons.
2/ Excludes methane, ethane, and propane.
3/ Includes toluene, ethylbenzene, n-propyl-
benzene, etc.
4_/ Includes iso-propylbenzene, t^-butyl
benzene, etc.
5_/ Includes the xylenes, mesitylene, and other
polysubstituted benzenes.
-------
TABLE 11. - Composition and smog chamber reactivities of exhaust from simple hydrocarbon fuels
Fuel
n-Butane
n-Heptane +
2,4,4-Trimethyl-
2,3,4-Trimethyl-
Methylcyclopentane .
Butene-1
Hexene-1 +
2,4,4-Trimethyl-
pentene-1 ....
Mixed 2- and 3-
2,4,4-Trimethyl-
pentene-2 ....
Toluene
Concentrations in raw
wet exhaust, ppm
Hydro-
carbon
1,058
1,369
879
925
960
1,260
1,122
1,177
913
999
1,085
1,400
960
1,099
1,089
1,371
1,134
1,227
1,090
1,234
723
939
703
755
857
Formalde-
hyde
14.3
15.2
12.3
12.3
8.1
17.1
17.8
14.4
8.8
11.2
20.4
19.5
16.2
12.1
17.7
20.5
23.4
19.1
14.0
23.1
8.5
10.1
7.3
7.1
4.5
Total
aldehydes
22.8
21.8
29.7
36.8
22.1
28.8
31.7
27.7
24.2
16.1
39.3
26.2
26.2
19.5
40.0
41.0
51.6
43.2
36.7
36.7
14.3
15.6
17.2
17.9
14.9
Jackson
reactivity,
propylene
equivalents
0.24
.23
.24
.26
.32
.25
.51
.46
.47
.44
.48
.44
.30
.28
.47
.46
.58
.53
.61
.57
.13
.12
.30
.28
.27
I/ Calculated using the EPA reactivity scale normalized to propylene
EPAi/
reactivity,
propylene
equivalents
0.22
.22
.23
.23
.30
.24
.34
.31
.27
.26
.51
.47
.33
.31
.48
.47
.47
.45
.40
.38
.06
.06
.31
.29
.29
reactivity =
Chamber reactivity,
propylene equivalents
RN02
0.43
.42
.59
.58
.53
.48
.78
.76
1.00
.94
.66
.57
.58
.54
.75
.83
.94
.89
1.86
1.78
.67
.66
.87
.87
.84
l.Q.
Maximum 0-
0.16
.18
.15
.16
.36
.29
.63
.61
.37
.33
.66
.59
.29
.33
.74
.73
.66
.58
.75
.71
.03
.02
.36
.35
.31
Ul
o
-------
TABLE 11 . - Composition and smog chamber reactivities of exhaust from simple hydrocarbon fuels
(continued)
Fuel
sec-Butylbenzene . .
n-Heptane + toluene
ii-Heptane 4- m-xylene
n-Heptane + 2,4,4-
trimethylpentene-2
Concentrations in raw
wet exhaust, ppro
Hydro-
carbon
736
862
587
806
989
987
1,520
Formalde-
hyde
12.3
12.7
5.9
3.7
13.9
8.1
20.5
Total
aldehydes
28.9
35.1
22.6
11.6
36.5
22.8
41.0
Jackson
reactivity,
propylene
equivalents
0.35
.32
.65
.64
.29
.42
.44
EPAi/
reactivity,
propylene
equivalents
0.29
.27
.54
.53
.30
.38
.34
Chamber reactivity,
propylene equivalents
RN02
0.84
.79
1.57
1.55
.67
.97
1.01
Maximum 0-
0.23
.23
1.33
1.39
.41
.92
.70
I/ Calculated using the EPA reactivity scale normalized to propylene reactivity * 1,0.
-------
52
TABLE 12. - Summary of smog chamber reactivities of exhaust from simple fuels
Reactivity, propylene equivalents
Fuel
Maximum
n-Butane 0.4
ii-Heptane — .5
2,2,4-Trimethylpentane .5
2,3,4-Trimethylpentane .8
Methylcyclopentane 1.0
Butene-1 .6
Hexene-1 -/ .5
2»4,4-Trimethylpentene-l .8
Hexene-2,3 .9
2>4,4-Tritnethylpentene-2 1.8
Benzene .7
Toluene .9
sec-Butylbenzene .8
m-Xylene 1.6
n-Heptane + toluene .7
n-Heptane + m-xylene 1.0
n-Heptane + 2,4,4-trimethylpentene-2. . 1.0
0.2
I/ .4
.3
.6
.4
I/
,6
,6
.6
.7
0
.3
.2
1.3
.4
.9
.7
I/ Gross estimates obtained from comparison of data from the n-heptane-
mixtures with aromatics versus data from the aromatics alone.
-------
l.2r—
Data from car 6
1.0 —
.8 —
I
o
o?
I
I
•
X «
x
o
tu
1-
<
_l
o
«J
5 10 15 20 35
POLYALKYLBENZENES IN FUEL, mole percent
FIGURE I-Correlation of Calculated Exhaust Reactivities With Polyolkylbenzene Levels in Fuel.
30
-------
1.2
Data from car 6
tz
>
c
4)
o-
V
=> ^
2 t
O
UJ
O
1.0
.8
.6
t
•
I
.2
5 10 15 20
OLEFINS IN FUEL, mole percent
FIGURE 2-Correction of Calculated Exhaust Reactivities With Olefin Levels in Fuel.
20
-------
1.5
1.0
i-
o
UJ
a:
o
z
K
tn
3
<
X
UJ
o
UJ
-------
1.2
3
o-
4)
4>
1
O
0.
1.0
0.9
u
4
tu
a: 0.8
O
a:
w 0.7
i
x
LJ
0.6
m
o
0.5
• 1968 Bureau of Mines study (3)
X 1972 Bureau of Mines study
10
20 30
AROMATICS IN FUEL, mole percent
40
50
FIGURE 4.-Comparison of Exhaust Reactivities — Fuel Aromatic Relationships From Two Bureau of Mines
Studies.
-------
lOi—
9 —
8
.7
.6
UJ
01
O
z
(t
X
UJ
O
tu
to
(D
O
* Data for fuels 1-21
X Dato for fuels 22-25
1
20
25
5 10 15
POLYALKYLBENZENES IN FUEL, mole percent
FIGURE 5.-Correlation of Observed Exhaust Reactivities With Polyalkylbenzene Levels in Fuel
30
-------
58
3.5r—
3.0
I »
S 2.0
>
t
>
h-
U
S 1.5
or
2
DC
O
W
CD
o
0.5
x Correlation line from A.P.I, study
• Correlation points from present study
1
0.2
1.4
0.4 0.6 0.8 1.0 1.2
CALCULATED EXHAUST REACTIVITY (Rj), ethylene equivalents
FIGURE 6.- Correlation of Observed and Calculated (Jackson Scale) Exhaust
Reactivities.
1.6
-------
2.5
2 0
I 5
Ill
IE
<\J
O
Z I
d
111
UJ
OT
03
O
Par 1 A
* vl • •
Correlation coefficient =0.36
0.4 0.6 0.8 I .0 I .2 1.4
CALCULATED EXHAUST REACTIVITY (Rj), ethylene equivalents
* .9
C 8
I-
8
u
cc
(M
O
X
X
UJ
o
UJ
Port 6
UJ
(ft
1 •
«*••
• •
Correlation coefficient =0.52
'O O.I 0.2 0.3 04 0.5 06 O.T
CALCULATED EXHAUST RNC,2 REACTIVITY, p rooy lene equivalents
FIGURE 7.-Correlation of Observed Exhaust Rf^Og Reactivjties
With Calculated Reactivities by Jackson and Bureau
of Mines Scales.
-------
o
Ct
a.
to
cc
UJ
O.
1.6
1.4
1.2
O -5
< §•
LJ «
ir
w
CM §
z £
o: o
H tt
IO
X
X
LJ
o
LJ
>
CC
LU
irt
CD
O
.8
.6
0.10
1.00
uo
0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90
CALCULATED "FUEL"' RNQ2" REACTIVITY (RJ), propylene equivalents
FIGURE 8.-Correlation of Observed Exhaust Reactivities With Calculated "Fuel" Reactivities.
Cf-
o
-------
Air- fuel rolio (A/f)
• Rich (075)
O S'Oichiometric (I 01
X Leon (ID
FIGURE 9.-Calculated RN()2 Reactivities of Exhaust From Simple Hydrocarbon Fuels Used Under Varied A/F Conditions.
-------
10
20 30
AROMATICS IN FUEL, mole percent
40
50
FIGURE 10.-Calculated R[vjOo Reactivities of Exhaust From Gasoline Fuels Used
Under Varied Air-Fuel Ratio Conditions.
-------
2.0 r-
1.8 -
X Aro.rrvotic fuels
• Poroffin, olefin fuels
O Aliphatic / aromatic mixtures,
cycloparaf fins
I 0.2 0.4 0.6 0.8 1.0 1.2
Rj CALCULATED RNOZI propylene equivalents
FIGURE 11.-Correlation of Observed and Calculated R|sjp2
Reactivities of Exhaust From Simple Hydrocarbon
Fuels.
-------
64
APPENDIX A.—ANALYSIS OF EXHAUST MASS EMISSION DATA
The mass emission data for cars 5, 6, 7, 9, and 10 with fuels 12-
21 were subjected to an analysis of variance to determine the importance
of car effects, fuel effects, and car/fuel interactions on mass emission
of hydrocarbon (HC), carbon monoxide (CO), nitric oxide (NO) , formaldehyde
(HCHO),- and total aldehydes (TALD), and the effect upon total reactivity
according to the Jackson scale (TRJ). A large car effect, significant
at the 99.5 pet level, was found for each of the mass emission parameters.
A much smaller fuel effect was observed although the effect was still sig-
nificant at the 99.5 pet level for all parameters except CO which showed
the fuel effect at the 90 pet confidence level. Car/fuel interactions
were considered to be negligible as no effect was observed at a confidence
level of 90 pet for any of the mass emission parameters. The car and fuel
effects are depicted in figure A-l, where the bar for each car represents
FIGURE A-l. - Average emission levels for each car and each fuel.
an average over the ten fuels and the value for each fuel is an average
over the five cars. It is clear that the car differences are indeed larger
than the fuel differences for each of the mass emission parameters.
To further explore the fuel effects, the data for each fuel averaged
over the five cars, summarized in table A-l, were correlated with thirteen
fuel physical and chemical properties listed in table A-2. It should be
noted that these thirteen fuel parameters are not all independent of each
other. For example, specific gravity correlates well with 50 pet point
and aromatic content. Thus, a correlation between mass emission data and
-------
65
a fuel physical property may be the result of a relationship between the
mass emission data and some chemical property of the fuel which also re-
lates to the physical property. With this in mind, we turn our attention
to the correlation analysis results which are summarized in table A-3.
The HC mass emissions showed a high negative correlation with fuel 90 pet
point and mole percent olefin and a fair positive correlation with fuel
isoparaffin content and 10 pet point. Although the CO emissions did not
correlate well with any of the fuel parameters, fair positive correlations
were obtained with cycloparaffin content and paraffin content and fair
negative correlations with olefin content and 90 pet point were observed.
The NO emissions showed good positive correlations with specific gravity,
50 pet point, aromatic content and 90 pet point. Good negative correlations
were observed for NO emissions with paraffin and isoparaffin contents.
Formaldehyde and total aldehyde emissions correlated well with isoparaffin
content and specific gravity. Formaldehyde emissions also correlated fairly
well with aromatic content as did total aldehydes with cycloparaffin content.
Total reactivity by the Jackson scale (TRJ) did not correlate well with
any of the fuel parameters.
Regression models were chosen on the basis of the correlations dis-
cussed above. For those cases where fuel parameters correlated well with
each other, only one of the parameters was included in the model. Also,
in some of the models OLEF was replaced by PAR + AROM, and PAR was replaced
by OLEF + AROM in order to give positive correlations since by definition
OLEF + PAR + AROM = 100. Fuel parameters were deleted from the model when
their coefficients were not significantly different from zero at the 95
pet confidence level. Results of the regression analysis are summarized
in table A-4. Statistically significant models containing two independent
-------
66
variables (fuel parameters) were obtained for CO and total aldehyde mass
emissions. The models for the other four mass emission parameters reduced
to the simple linear cases when all insignificant parameters were deleted.
The final model for CO containing two fuel parameters (cycloparaffins and
olefins, model CO-I) resulted in a correlation much improved over the best
of the simple correlations (model CO-II, table A-4) as can be seen by the
squares of the correlation coefficients (0.798 versus 0.392). A much smaller
improvement resulted from the two parameter models for total aldehydes (0.915
versus 0.824). The best of the simple correlations (one parameter model)
for each of the six exhaust mass emission parameters are illustrated in
the graphs in figure A-2.
FIGURE A-2. - Variation of exhaust mass emission parameters with various
fuel composition parameters.
In brief, correlations between exhaust mass emissions and fuel com-
position were obtained in all cases. Those for formaldehyde and for total
reactivity-Jackson Scale may be questionable; the others appear to be fully
valid for the sets of fuels and vehicles tested. Although useful as in-
dicators, the correlations obtained in the study may not be reliable in
extension to a much broader population of fuels and engines. In this
respect it is noteworthy that only relatively small fuel effects were ob-
t
served despite the great diversity in composition in this set of fuels.
For the five cars, the car differences have a far greater impact upon exhaust
emission levels than do fuel differences.
-------
67
TABLE A-l. - Summary of mass emission data for
five cars and ten fuels
Fuel
12
13
14
15
16
17
18
19
20
21
Mass emission parameter
HC^
25.6
26.6
26.7
27.3
28.5
24.2
27.9
25.4
24.6
25.4
CO
353
362
355
357
358
333
363
374
350
350
NO
43.0
41.2
40.8
46.2
40.8
46.8
44.4
43.9
42.6
46.2
HCHO
0.677
.656
.708
.597
.700
.685
.590
.595
.633
.591
TALD
1.186
1.126
1.123
1.072
1.285
1.097
1.107
.946
1.124
1.077
TRJ
13.5
12.9
15.1
12.3
13.4
13.3
11.6
12.4
12.7
12.3
I/All values are averages over triplicate tests with each
~ of five cars expressed as grams/test except for TRJ
which is expressed as (grams ethylene)/test.
-------
63
TABLE A-2. - List of fuel physical and compositional
properties used in correlations
1. Reid Vapor Pressure (RVP)
2. Specific Gravity (Sp.Gr.)
3. 10 pet point in °F (10%P)
4. 50 pet point in °F (50%P)
5. 90 pet point in °F (907.P)
6. Mole pet paraffin (PAR)
7. Mole " olefin (OLEF)
8. Mole " aromatic (MOM)
9. Mole " mono a Iky 1 benzenes (MonoAlkBz)
10. Mole lf poly alky 1 benzenes (PolyAlkBz)
11. Mole " n-paraffins (n-Par)
12. Mole " isoparaffins (iso-Par)
13. Mole " cycloparaffins (cyclo-Par)
-------
61''
TABLE A-3, - Correlation coefficients^ for exhaust mass emission
levels and fuel physical and compositional properties
RVP
Sp.Gr
10%P
50%P
907.P
PAR
OLEF
AROM
MonoAlkBz . .
PolyAlkBz..
n-Par
iso-Par... .
eyclo-Par . .
HC
-0.557
- .241
.500
- .238
- .823
.373
- .856
.161
.281
.013
- .484
.612
- .258
CO
-0.291
- .033
.231
- .387
- .629
.509
- .626
- .142
.228
- .240
- .301
.052
.568
NO
-0.044
.856
.300
.797
.607
- .856
.313
.711
.199
.437
- .274
- .782
.098
HCHO
-0.051
- .703
- .180
- .433
- .002
.437
.290
- .654
- .467
- .213
- .393
.708
- .367
TALD
-0.016
- .746
- .108
- .484
- .471
.423
- .316
- .242
- .099
- .240
- .203
.908
- .733
TRJ
-0.077
- .483
- .142
- .208
.156
.348
.358
.603
- .647
.116
- .332
.495
- .189
-------
TABLE A-4. - Summary of regression analysis results
Exhaust mass
emission parameter
(dependent variable)
HC
CO
NO
HCHO
TALD
TRJ
Model
No.
.1
II
I
II
I
I
I
11
I
Fuel parameter
(independent
variable)
1/10% P
~" 90% P
iso-Par
(PAR + AROM)
eye Ip- Par
OLEF
PAR
(PAR 4 AROM)
(OLEF + AROM
90% P
iso-Par
AROM
iso-Par
iso-Par
cyclo-Par
iso-Par
AROM
MonoAlkBz
iso-Par
Estimate of
coefficient
2/ -
-0.0521
0.217
0.745
-1.328
1.189
\ 0.213
0.00270
0.00494
- .00332
0.00621
-0.0763
Standard error
of estimate of
coefficient
0.0127
0.0456
0.199
.327
0.532
0.046
0.00095
0.00089
.00122
0.00102
0.0318
t -value
for coef-
ficient
-4.10
4.76
3.76
4.06
2.23
4.61
2.84
5.56
-2.73
6.11
-2.40
Regression
constant
40.9
6.27
362
246
34.0
0.570
1.000
0.947
14.2
Square of
multiple
correlation
coefficient
0.678
0.739
0.798
0.384
0.726
0.502
0.915
0.824
0.418
1 /Abbreviation of parameters are listed in table A-l.
2/The dash indicates the parameter was dropped from the model because its coefficient was not significantly
different from zero at the 95 pet confidence level.
--J
o
-------
NO , grams/tatl
N *
O 0 O
4O
EXHAUST HYDROCARBON, grams /ten
— M 01
> 0 0 0
n a
iYDE ,gram*/t*»t
<
b> <
ut
3 .4
EXHAUST FORM
3 (0
—
—
—
—
—
^
$$;
s$
^
$s
s$
^
I
^
^
i
$$
$s
^
^
<§
^
^
^
$$
SS
^
I
I
^
^
1
^
^
^
^
^
^
^
^
1
^
^
rm
\
I
^
$$
§§
$§
§X
\^
^
•^
^
^
$§
:$$
^
^$
W
^
^
I
I
^
^
$^
^
^
W
^
^
^
^
•$$
•§$
^
W
NV
T*
L •
t* C
« ".
> f
fmmm
r'j
< "•
«v
* 1
1 (
i 1 <
1 (.
<• k
> *. i
A *
< >
J t
* t
-1".
* >
V +
* •* i
* j
^ * .
•»«. *
» i
* ^
r 4
; r J
' „ -t
* 4 *
> ^ *
« '
* *",
» 4 ,
* J
•* 1.
I.''
»r-"-
/'-
I '
»••
A >
» 1
A ,
to >
•^™
%*•«
, >
V *
r *
, L '
•1
J r
t. *
4*
y •*
v»-
t"«
. * .
b- r
, >
* V
* *
•» „
r >
*. *
* V
•;r
». * '
i k
7TT
» ^
•* r
* f-
^^»«
« b
^.'
t -*
^ <-
nr-
i w
* »
T *-
J 4
1. ••
l> *
T*
j *
w ••
< J
>
•* >
* *
,-„
i «-
•* ^ 4
' «
t <
r »
t j
< *-
* v
• »•
j
1 *.
i. (
" 4
A C
H *
'••*"
•» >
V *
^ >
* »
MhpK
t •»"
% *
-
•*•••
r *•
>**
^ >
> w
» 4-
*• t
t *
W *
' * •
< ^
> "•
V t
1 «'
^ »•
* r <
* >
I- -<
r «
> L
— T-
r -4
" t
A *•
> W
4. j
,__'
" •*
^ j
. t
1 1
*. 4
, ^ ^
» •
A *
4. *
* fr
;»"
«• *
r*^
> <.
, » "
»*!.
« «•
t v
"* *
t- |>
>^T *
•» *
r- i
^ k
* *
T*
. « •
^ j*
^
^ ^-
* *
••Y
* *«"
* i
i k
^ *
•* t.
«• j
» *
•", *
v .
4 t.
• «
. t fc
5 6 7 9 10
CAR
12 13 14 IS 16 17 18 19 20 21
FUEL
FIGURE A- I .-Average Emmision Levels for Each Car and Each Fuel.
-------
72
600
S 400
v,
•»
E
o zoo
o
—
—
1
i
s$
i
1
i
i
1
^
i
1
i
«
i
1
1
s?
i
1
r •
f
T
*
i ,
•f t
T"tr
4
„
*
*"
. % '
;V
*:<
> V
f b
,
< fc
J ,
• L
fc V
•
t
t> *
L. T
;/
*\
* * 4
V * 4
_
."*
^^^
T *
* t •
;-,
> ',
:l*
- 1.5
£
o
.5
—
1
I
I
I
I
€>
m
E
15
10
t-
o
< 5
ui
K
O
CO
—
I
I
I
I
I
S 6 7 9 10
CAR
12 13 14 15 16 17 18 19 20 21
FUEL
FIGURE A-I.-Average Emmislon Levels for Each Car and Each Fuel
(continued)
-------
I 40
1.30
1.20
10
I.OO
.90
16
380
10
J I
J I
20 30 40
ISO-PAR , mole percent
10 IS 20
MONOALKBZ , mol« percent
50
25
J
60
J
30
- 3SO
J40
3301
80
48 |—
I 44
40
85 90 95
PARtAROM ,mol« percent
J
100
J I
30 4O $0
OLEF+AROM, molt percent
J
60
.75.
.70
6
Z 65
.55
20 30 40
ISO-PAR, mole percent
SO 60
JO
26
•26
22
85 90 95
PAR*AROM , mole oerceni
FIGURE A-2-Variation of Exhaust Mass Emission Parameters With Various Fuel Composition
Parameters.
-------
75
Final report for the Bureau of Mines-EPA
cooperative research project
"Characteristics and Photochemical
Reactivity of Emissions"
PART II. TOXIC PRODUCTS FROM FUEL ADDITIVES
INTRODUCTION
Under provisions of legislature now in effect, the Environmental
Protection Agency (EPA) has a responsibility to control the use
of fuel additives wherein such use might have an adverse effect
in automotive emissions or emission control systems. One possible
regulatory course is to require any manufacturer or user of an
additive to provide data on additive effects as related to emis-
sions. These data would necessarily be obtained using a stand-
ardized test procedure specified or approved by EPA. No such
procedure now exists and this work was authorized by EPA as a
first effort toward developing one part of a test procedure to
measure additive effects on emission.
Much has been published on the combustion of hydrocarbons (_1) ,
but very little is known of the fate of gasoline additives in the
combustion process and, as indicated above, analytical procedures
for the additive-derived products are non-existent. Thus, this
research reported upon is a first effort in the experimental devel-
opment of the required analytical procedures. More specifically,
procedures were sought for the determination of toxic gaseous
products, if any, resulting from thermochemical reactions of fuel
additives in the combustion process. The general approach was
to operate an engine using additive-free fuel to obtain reference
data on exhaust hydrocarbon and oxygenate composition; this was
followed by tests in which the engine was operated using the same
fuel except with additive. The exhaust analyses included hydro-
carbon and oxygenates determination together with analysis for the
additive and for its direct combustion products in the gaseous
portion of the exhaust.
To place this report in proper context, it should be noted that
it covers essentially nothing more than the preparatory phase of
an experimental study with total effort not exceeding about one-
half man year. The work was done solely to provide guidance in
selection of analytical methods for consideration and was not
intended to provide definitive information on additive effects,
per se.
-------
76
EXPERIMENTAL PROCEDURES AND RESULTS
Preliminary experiments to identify products from onc> tucl addi-
tive were conducted using a single-cylinder research engine. The
fuels used were isooctane and isooctane with methyldLphenylphos-
phate (HDP) additive at a level of 3.624 grams per Ballon (corres-
ponding to two theories based on 2 cc TEL per gallon). Exhaust
samples were analyzed for phosphorous compounds with a Micro Tek
GC-2500R gas chromatograph equipped with a flame photometric
phosphorous detector (also from Micro Tek) and a 4-ft glass column
with 3 pet 0V 101 on Gas Chrom Q at a temperature of 160° C.
Earlier experiments had shown this column to be superior to a 6-ft
glass, 3 pet 0V 101 on Gas Chrom Q column. The three samples
analyzed for phosphorous were (1) exhaust gas prediluted with
nitrogen to prevent condensation of water, (2) condensate from
raw exhaust passed through a trap at 0° C, and (3) the exhaust
gas which passed through the 0° C trap.
No phosphorous compound was detected in any of the samples from
the pure isooctane runs. Also, no phosphorous compound was detected
in the gaseous samples from the tests using isooctane without additive;
however, the condensate chromatograms showed two, and occasionally
three, peaks. Two of these peaks matched the retention times of
the peaks obtained when a dilute solution of the additive itself
was chromatographed. Therefore, it was concluded that part of the
fuel additive escaped combustion and passed into the exhaust
unburned. This unburned additive in the exhaust was estimated,
from the chromatographic data, to be less than 1 pet of the
amount in the fuel.
In a subsequent run using pure isoctane fuel, phosphorous com-
pounds were detected in the condensate sample suggesting that
additive-derived products deposit in the engine-exhaust system
causing memory effects.
Because the combustion system of the single-cylinder research
engine is not representative of automotive equipment, a second
engine was selected for additional exploratory work. This engine,
an Onan, 2-cylinder, four cycle, 49.8 cubic-inch displacement (CID)
model was judged to provide a combustion environment closely
approaching that of an automotive engine in medium duty service.
After engine break-in, several tests were run using an additive-
free gasoline to compare hydrocarbon composition of exhaust from
this engine with composition of exhaust from four commercial
passenger automobiles. The Onan exhaust composition was found
to be somewhat different from that of the automobile exhaust as
shown in table 13, but the similarity was close enough that its
exhaust was believed suitable for purposes of analytical method
development in the context of a preparatory study.
-------
TABLE 13.- Comparative exhaust hydrocarbon composition
from Onan engine and as determined forfour
automotive engines I/
77
Compound
Paraffins
Olefins
Aromatic s
Methane
Acetylenes
HC composition^ mole pet
Onan
engine
34.4
16.6
36.6
18.2
13.4
Range for
four auto engines
25.7--33.0
27.0--35.6
25.0--31.0
11.1--18.9
9.9--13.7
If All tests made using identical fuels.
-------
78
Concurrent with the experiments using the single-cylinder and
Onan engines, work was begun on the procedures for oxygenate
analysis. The initial phase of this development involved prepa-
ration of two calibration blends of mixed oxygenates and prepa-
ration of porous polymer packed beds for collection of samples.
Familiarization with the analytical equipment and techniques was
accomplished using the calibration blends. Initial attempts to
analyze samples from the engine runs were not successful presumably
due to insufficient sample. This problem was not explored further
because of an interruption in the program and the subsequent deci-
sion to use a full-boiling range gasoline in an automobile engine,
thereby making detailed oxygenate analysis impossible with avail-
able techniques.
The work reported just above was done in the second quarter of
FY 72. Because funds were lacking to sustain an experimental
program throughout the last three quarters, the program was inter-
rupted to await availability of a test engine specified by the
EPA project monitor. The experimental program subsequently was
resumed in the last quarter using a 350-CID 1971 Chevrolet engine.
Prior to its use in the tests, the engine was partially dismantled
and the valves, cylinders, and pistons cleaned thoroughly. The
engine was reassembled on a dynamometer test stand with a completely
new exhaust system including new exhaust manifolds, muffler, and
exhaust pipe up to the sample port section. This section was
cleaned thoroughly using Skasol, water, and acetone. All sample
lines in the proportional sampler, used to obtain exhaust sample,
were cleaned or replaced. The engine was then run for several
hours using an additive-free gasoline and adjusted to manufac-
turer's specifications. The motor oil used was Havoline Super
Premium 10W40.
Concurrent with the engine preparation, analytical development
continued with evaluation of various chromatography columns and
temperature conditions for determination of two commonly used
phosphorous additives, tricresylphosphate (TCP) and MDP using
the phosphorous-specific flame photometric detector.
Analysis for TCP was difficult due in part to the extremely low
vapor pressure of TCP. Several different stainless steel columns
were used; these were:
1. A 12-ft x 0.010-inch column dynamically coated with
a 10 pet solution of SE-30 in methylene chloride.
2. A 6-ft x 0.010-inch uncoated column.
3. A 6-ft x 0.010-inch column coated with an 0.8 pet
solution of OV-17.
4. A 6-ft x 0.020-inch column coated with a 5 pet
solution of OV-17.
-------
79
5, A 6-ft x 0.020-inch column coaced with a 10 pet
solution of OV-17.
6. A 14-ft x 0.020-inch column coated with a 5 pet
OV-17 solution.
7. A 30-ft x 0.020-inch column coated with a 5 pet
OV-17 solution.
8. A 4-ft x 1/8-inch column with 4 pet HIEFF-BP on
60/80 mesh, acid washed, DMCS treated Chromosorb G.
Carrier gas flow and column temperature were varied for each
column to search for optimum sensitivity, peak shape, and
retention time. The best results for TCP left much to be desired.
With the 14-ft x 0.020-inch 5 pet OV-17 column at 270° C with a
carrier flow of 50 cc/min, a 0.5 u-1 injection of a 10 ppm solu-
tion of TCP in methanol produced a 3-inch peak with considerable
tailing at a retention time of 0.5 minute. Considering the instru-
ment noise and the poor quality of the peak, the limit of detec-
tion would be about 1 to 2 pprn TCP with these conditions.
Analysis for MDP showed much more promise as easily interpretable
peaks were obtained with several of the columns. With the 14-ft
x 0.020-inch, 5 pet OV-17 column for example, two peaks with
retention times of 0.25 and 1.3 min were obtained using a. column
temperature of 140° C and a carrier flow of 95 cc/min. The
response for the second peak was about 2.4 inch per ppm for a
1 til injection of a 100 ppm solution of MDP in methanol calculated
for the most sensitive instrument settings. Acceptable results
were obtained also with the 4 pet HIEFF-BP column at a temperature
of 220° C and a carrier flow of 80 cc/min. Response for the
second of two peaks (retention times 0.6 and 2.8 minutes) was
about 1.8-inch per ppm for a 1 (0,1 injection of the 100 ppm MDP
solution.
Prior to the time that tests were run using the 350-CID engine,
the project directors (EPA) directed that work with the phosphorous
additives be discontinued and effort directed toward methylcyclo-
pentadienyl manganese tricarbonyl and polybutene amine F-310.
Work was begun on analytical procedures for organic manganese
compound determination using gas chromatography and flame photom-
etry. No definitive results were obtained in the short time
that remained in FY 72.
-------
80
REFERENCES
For example see
(a) Fleming, R. D. Effect of Fuel Composition on Exhaust
Emissions From a Spark-Ignition Engine. Bureau of Mines
Report of Investigations 7423, 1970, 68 pp.
(b) Morris, W. E. and K. T. Dishart. The Influence of Vehicle
Emission Control Systems on the Relationship Between Gaso-
line and Vehicle Exhaust Hydrocarbon Composition. Effect
of Automotive Emission Requirements on Gasoline Character-
istics, ASTM STP 487. American Society for Testing and
Materials, 1971, pp. 63-101.
-------
81
Final report for the Bureau of Mines-EPA
cooperative research project
"Characteristics and Photochemical
Reactivity of Emissions1'
PART III. - DIESEL ODOR
INTRODUCTION
The initial objectives for work during the past year were isolation, frac-
tionation, and reconstitution of diesel odorants, particularly the unstable
and reactive materials. For convenience in exhaust handling and odor meas-
urements the materials to be examined were grouped as either those retained
in water scrubbed (0° C) samples or as those passing a water scrubber.
There were three major problems to be considered in carrying out this
program. They were:
1. Development of an odorant dilution and presentation system suit-
able for use with isolated exhaust components and synthetic
odorants.
2. Establishing methods for demonstrating reactivity and stability
characteristics of diesel exhaust odorants. These demonstrations
were based on odor measurements of exhaust that was treated for
removal of specific classes of compounds or for decomposition of
unstable materials.
3. Isolation and collection of unstable components. As far as possible
these materials were to be examined for odor contribution based on
samples obtained from exhaust.
In addition to work undertaken for the objectives above, a report was prepared
and published on previous work* This report covered a study to quantify the
contribution to diesel exhaust odor of: (1) carbonyl compounds, (2) phenols,
and (3) hydrocarbons I/.
EXPERIMENTAL
Odor Determination
2/
Odor evaluations were made by a trained panel using the Turk— odor rating
method. This was the procedure used in previous studies I/.
i/James W. Vogh.BuMines Rept. of Inv. No. 7632 (1972), 11 pp.
2/Amos Turk. P.H.S. Publ. No. 999-AP-32, Washington, D.C. (1967).
-------
82
Odorant Dilution Systems
Chamber Dilution
A metal and glass chamber that had been used in photochemical studies was
modified to permit the presentation of diluted exhaust and odorant for odor
evaluation. Samples (either gas or liquid) were introduced into the chamber
in amounts calculated to give the desired concentration or dilution ratio
on the basis of the total volume of the chamber. The diluted sample in
the chamber was forced out through a sniff port by inflating a Tedlar film
bag that was contained in the chamber. The chamber was found to have a
strong background odor. This odor approached the intensity of diluted
(100:1) diesel exhaust and it appeared to be too strong to eliminate its
contribution to the total odor by background odor measurements. Thorough
cleaning of the interior of the chamber eliminated the odor for a short
period. The rapid return of odor to its original level interfered with
use of the chamber as a dilution system and no further experimental work
was done using this system.
A second system for odorant dilution was b^sed on a large Teflon film bag
(about 200 liters volume) with a stainless-steel bellows pump (Metal
Bellows Corp., model MB-155) for delivery of the sample to a sniff tube.
The dilution air was introduced into the bag prior to and during the odorant
sample introduction in such manner as to promote rapid mixing. Flow rate
and duration of flow both for diluent air and for the odorant sample were
measured to establish the dilution ratio. The diluent air was passed through
activated charcoal at 0° C and through distilled water at 15° C before
entering the bag. At the end of introduction of the air and odorant sample,
the bag was kneaded about 20 seconds to complete the mixing. In some odor
evaluation work the bellows pump was replaced by a press composed of a ply-
wood sheet and sufficient weights to obtain the desired flow out the sniff
tube.
The third system for odorant dilution was a modification of the syringe
fed dilution device used in previous odorant studies \J. The changes per-
mitted the syringe delivery rate to be adjusted so that it could be used
with exhaust samples. Diluent air for the tests was passed through charcoal
and water as described above.
Odorant Reaction and Aging
Exhaust scrubbing and presentation of the treated exhaust was carried out
fay procedures used previously 3/. The procedure involved use of a liquid
lift scrubber at 0° C containing 15 ml of water or solution and an exhaust
or gas flow of 4.5 liters per minute. In addition to the special attention
given to water scrubbing of exhaust, certain reagent solutions were used.
They were:
1. Sodium sulfite solution. This solution was used at 1.0 molar
concentration. With this and the other reagent solutions odor
evaluations were made after passing exhaust through the solution
for four minutes.
2/J. W. Vogh. J. Air Poll. Control Assoc., 19., 773 (1969).
-------
83
2. Morpholine solution. This solution was prepared at a concentra-
tion of 1.7 molar morpholine and 0.9 normal acetic acid or benzoic
acid in water. In a few runs diethanolamine was used instead of
morpholine.
3. Potassium permanganate solution. This was used at concentra-
tion of 0.2 molar KM 0, with 0.33 molar H.SO..
n tt i. 'f
4. Hydroxylamine solution. Used at 0.5 molar NH-OH and 0.5 molar
NH_OH x HC1. This reagent has been examined previously 3_/ for
its effect on diesel odor.
All of these solutions were evaluated for their effectiveness in removing
selected components from gas streams. The evaluation method has been
described 3/. Trial components tested in gas blends were ketene, ethyl
hydroperoxide, propylene oxide, methyl acrylata, vinyl propionate, and
allyl propionate.
A series of runs were made for evaluation of odor changes in exhaust due
to aging at room temperature. This procedure was based on holding an
exhaust sample in a glass pipe section of the odor evaluation system for
a measured period before mixing into the dilution air stream. The glass
pipe was 2.6 cubic I.D. and had a retention volume of 2200 cc. It was
covered with aluminum foil to prevent entrance of light. Contents of
this glass pipe volume were drawn out by an air aspirator to be mixed
into the carrier stream. This was accomplished by use of a pair of three
way valves positioned so as to permit gas flow either through the glass
pipe or through a small bypass. If the contents of the glass pipe were
displaced from it in strictly plug flow it would have provided uniform
samples for about 29 seconds. However, measurement of a synthetic sample
out a side stream indicated that a uniform sample (that is, unmixed with
displacement air) could be obtained only for about 22-25 seconds. This
determination was accomplished by use of an organic vapor blend displaced
by clean air with the side stream led directly to the flame ionization
detector of a gas chromatograph. Because plug flow was not achieved,
odor measurements were limited to the 15 second period immediately after
the start of mixing the aged sample into the carrier gas stream.
Analytical Methods Development
Preliminary studies were carried out on detection of organic hydroperoxides
by gas chromatographic methods. Organic hydroperoxides were prepared for
use in development of the chromatographic method. The synthesis method
generally used was based on oxidation of the magnesium or zinc organo
metallic compounds 4.5/. Ethyl, n-propyl, isopropyl, allyl, and benzyl
hydroperoxides were prepared. Chromatographic columns used were either
the all glass packed columns with UCON 50-HB5100 or Carbowax 20M used for
oxime chromatography 6/ or glass capillaries coated with the same liquid
phases. Hydroperoxides in synthetic samples and in water through which
4/H. Hock and F. Ernst. Chem. Ber., 92., 2716-2723 (1959).
5_/H. E. Seyforth, J. Henkel, and A. Rieche. Angew. Chem., Intern. Ed.,
4, 1074 (1965).
6/J. W. Vogh. Anal Chem., 43, 1618 (1971).
-------
84
exhaust had been passed were isolated by extraction of the neutral materials
with an organic solvent such as pentane or benzene after the solution had
been made basic with sodium hydroxide. The hydroperoxides were then
recovered from the neutralized solution by extraction with ethyl ether.
This is a common procedure for hydroperoxide isolation and purification Tj •
RESULTS AND DISCUSSION
Odor Presentation
Earlier work in odor evaluation of synthetic odorants and isolated exhaust
fractions had been based on dilution by a carrier gas stream of samples
held in a gas syringe I/. Whr'le this procedure served well for the deter-
minations that were made in the previous work there were some doubts con-
cerning its suitability in study of other diesel exhaust odorants. The
primary difficulty appeared to be the loss of odorants in samples obtained
from exhaust trapped on porous polymer (Chromosorb 102) type beds. However,
information 8_/ obtained later indicated that diesel odorants could not be
recovered from the porous polymer by the methods used previously.
The chamber dilution method offered certain advantages such as accurate
and rapid dilution and large enough sample volume to allow repeated odor
evaluation of a single sample. However, the chamber had a strong background
odor that could not be removed by scouring and solvent washing. The odor
became more intense when the chamber air was humidified, possibly indicating
displacement of adsorbed odorants from the walls. Although some exhaust
samples were introduced into the chamber, no useful odor evaluation results
were obtained.
The odorant dilution system based on the Teflon film did not present any
problems in background odor. In odor evaluation measurements using n-propyl
propionate as a test material, the same odor intensities were observed at
the same dilution level in this system and in the dynamic flow system used
in direct exhaust odor evaluation. However, odor intensity of diluted
exhaust that had been passed through a scrubber was much lower than expected
in comparison with the same exhaust in the dynamic dilution system. Rough
estimates on odorant levels indicated that of the amounts originally present,
60-90 pet had disappeared by the time the diluted exhaust was pumped out
the sniff tube. The greatest losses occurred for exhaust produced at full
load operation. It was noted by the odor panelists that the air pumped out
of the Teflon bag by the stainless steel bellows pump was somewhat heated
(by compression and by heat transferred from the pump body). Since it seemed
possible that this heating might cause destruction of unstable odorants, the
pump was replaced with the large bag press previously described. However,
no improvement in exhaust odor level was found.
7/A. G. Davies. "Organic Peroxides," Butterworth & Co., London, 1961, pp. 114.
8/A. D. Little, Inc. Report, "Analysis of the Odorous Compounds in Diesel
~ Engine Exhaust," CRC Project CAPE-7-68, Report No. ADL 73686-1,
September 16, 1971, 22 pp.
-------
85
At this point it was not established whether the loss of odor was due to
absorption in the bag or to chemical deterioration of the odorants during
the time required for sample introduction, mixing, and presentation for
odor rating. The syringe fed dilution device had the advantage of more
rapid sample handling and presentation. This sampling procedure was
examined in a series combined with exhaust aging measurements. Results are
shown in table 14. In this study, exhaust that had been passed through a
cold water scrubber was drawn into the syringe and then delivered into the
carrier stream. The odor values given under "direct" sampling refer to
exhaust passed directly into the dynamic dilution system and are taken as
the reference values for odor intensity measurements. Engine difficulties
prevented completion of this study under full load conditions. However,
results at no load operation indicate a serious loss of exhaust odorants
and would limit the use of this sampling method to qualitative odor evalua-
tion unless the loss of specific odorants could be determined.
It is possible that some of the loss in odor intensity with syringe sampling
may have been due to adsorption and conditioning problems. The adsorption
of odorants on the relatively limited interior surface of the syringe would
cause more serious loss at low concentrations found in exhaust than it would
in samples prepared from exhaust by concentration and isolation techniques.
At higher concentrations the adsorptive capacity of the syringe wall would
become saturated. The conditioning problem arises in the very rapid delivery
of exhaust from the syringe required to present the 100:1 dilution to the
panelists. This prevents the proper conditioning of sample lines up to the
sniff port. On the other hand, a more concentrated sample in the sample
permits slower delivery to maintain an equivalent 100:1 dilution on original
exhaust basis which permits longer conditioning.
Some evidence that one or both of these effects influence odor intensity
was found in evaluation of exhaust samples trapped at Dry Ice temperature
in stainless steel loops packed with clean glass wool. These traps are
less efficient than the Chromasorb 102 traps in that some of the more volatile
components pass the fiberglass trap. In an odor evaluation of exhaust of
the GMC3-53 at no load, 1,000 rpm the direct sampling by the dynamic dilution
procedure found 2.9 intensity units while the cold trapped material carried
through the syringe method showed 2.0 units. This is a considerably smaller
loss than found for the exhaust drawn directly into the syringe. Some of
the loss of odor of the cold trapped material may be caused by the inadequate
retention of high volatility material in the trap and by incomplete transfer
of low volatility material from the trap to the syringe.
The intent of the work in odorant dilution and presentation systems was
development of procedures suitable for isolated fractions of diesel exhaust
or for synthetic materials representing compound classes known to be present
in exhaust. For this purpose it was assumed that if the odor of an exhaust
sample was unchanged by the dilution and presentation method from that for
exhaust through the dynamic dilution apparatus, then the individual exhaust
components would not be lost in sampling and presentation.
-------
86
Table 14.- Exhaust odor intensity as influenced by sampling method
(PI odor units, exhaust dilution 100:1)
(GMC-353 engine)
Engine mode
No load, 1,000 RPM
Full load, 2,100 RPM
Sampling
Direct
3.7
<5.8)i'
Syringe
2.0
-
Delay (4 min)
2.9
(2.8)^
I/ Limited data
Table 15.- Efficiencies of reagent scrubbing in removal of test compounds
Test Compounds
Scrubber
solution
H20
Na2S°3
Morpholine
Hydroxylamine
Ketene
13
79
89
75
Ethyl hydro-
peroxide
92
94
nt
95
Propylene-
oxide
0
5
8
5
Methyl
aery late
2
29
24
5
Vinyl
propionate
15
11
8
49
Allyl
propionate
3
1
1
5
-------
87
Unstable and Reactive Odorants
Several reagent scrubbing solutions were examined to determine their selec-
tivity in removal of reactive exhaust components. The sodium sulfite solu-
tion has been used in determination of epoxides 9/, alpha, beta unsaturated
compounds 10/, and organic hydroperoxides ll/. The morpholine solution and
similar secondary amines such as diethanolamine (which was also used in
exhaust scrubbing) have been used in determination of aloha, beta unsaturated
compounds 12/, epoxides 12/, and carboxylic anhydrides |3/. ™e fac^°<\
with carboxyHc anhydrides can be considered as indicating probable reaction
with ketenes. Potassium permanganate is quite general in its action as a
strong oxidant and probably reacts with most oxygenates and many unsaturated
and aromatic hydrocarbons. Hydroxylamine is known best for its reaction with
carbonyl compounds but it may also be used for determination for ketenes 147 .
The effectiveness of these reagent solutions in removal of several trial com-
pounds from gas streams is shown in table 15. The values shown are the per-
cent removal of the test compound at 10 minutes scrubbing with the scrubber
containing 15 ml water or reagent solution and gas flow of 4.5 liters per
minute.
Previous experience with the hydroxylamine scrubber 3/ has shown it to have
only a minor effect on diesel exhaust odor when compared with the water
scrubbed exhaust. Because of this it is not likely that either ketene or
vinyl alcohol derivatives such as vinyl propionate are important diese.
odorants .
Since ethyl hydroperoxide was removed from the gas stream by water it is
not possible to reach a conclusion on odor contribution through reagent
scrubbing for this compound. However, the water solubility of ethyl
hydroperoxide should be greater than that of the higher molecular weight
aliphatic hydroperoxides. In this case, a reagent such as sodium sulfite
would be discriminating for the less water soluble hydroperoxides. Examina-
tion of the solutions subsequent to passing ethyl hydroperoxide through
them showed that the hydroperoxide could be recovered by extraction from
water but not from the sulfite or hydroxylamine solutions. Both of these
latter reagents have reducing action and sodium sulfite is known to form
alcohols from hydroperoxides ll/.
Methyl acrylate is the representative compound for alpha,beta unsaturated
compounds. It is evident that both the sulfite and morpholine have a low
level of effectiveness in removal of this compound class. However, if this
class is a major source of odor, some indication of its presence would be
given by use of these reagents. Also, methyl acrylate is one of the less
reactive members of this class and other compounds such as those with
carbonyl or nitrile groups conjugated to the unsaturated bond should be
removed to a greater extent. _ ^^^
9/J. D. Swan. Anal. Chem., 26, 878 (1954).
TO/F. E. Critchfield and J. B. Johnson. Anal. Chem., 28, 7J UiobJ.
11/A. G. Davies. "Organic Peroxides," Butterworth & Co., London, 1961,
pp. 189-191. OR
12/F. E. Critchfield, G. L. Funk, and J. B. Johnson. Anal. Chem., 28.,
13/J. B. Johnson and G. L. Funk. Anal. Chem., 27., 1464 (1955)
"14/W. M. Diggle and J. C. Gage. Analyst, 78, 473 (1953).
-------
Contrary to literature indications, none of the reagents are useful in
removing epoxides. Allyl propionate represents compounds with isolated
unsaturation and it was not expected that sulfite and morpholine would
react with it.
The other approach used in demonstration of the presence of reactive and
unstable odorants was carried out by aging the exhaust. Since the exhaust
was aged by only four minutes this method was capable of indicating
only the most unstable and reactive classes of compounds. Within this time
period ketene might be hydrolysed 15/ and hydroperoxides might react with
nitric oxide 16/. Other reactions such as rearrangements of strained com-
pounds might occur.
Results of odor measurements on scrubbed exhaust are given in table 16 and
on aged exhaust are given in table 14, None of the reagents were highly
effective in reduction of odor although permanganate was fairly effective
on exhaust of the GM3-53 engine. Permanganate is not specific in its
reactions and it is difficult to interpret its effect on odor. The effects
of morpholine and sulfite were rather variable and not very large and the
exhaust of the GM3-53 engine appeared to be more susceptible to these
reagents than exhaust of the Cummins NH-250 engine. Since both the effect
on odor and the behavior towards the trial compounds was similar for these
two reagents, it is not possible to reach any conclusions on the nature of
the odorants. However, both are consistent with the interpretation that
some of the odorants are alpha,beta unsaturated compounds. Many of the
hydroperoxides may be removed by water scrubbing. Part of the odor removed
by water scrubbing has been accounted for in carbonyls, phenols, and hydro-
carbons \] and it is possible that some of the odor not accounted for may
be hydroperoxides.
The reduction in odor by aging of the exhaust (table 14) is greater than
that found in reagent scrubbing. No specific class of compounds can be
identified by this procedure but odorants disappearing may include those
sensitive to oxygen or nitrogen oxides or those subject to internal rearrange-
ment. These types of compounds would probably not be affected by the
reagents used.
The loss of odor on aging suggests several problems. One is that it may
not be possible to isolate and identify these odorants by methods currently
used. Another is the comparability of exhaust odor measurements made under
methods of dynamic dilution and of odor chamber or room dilution. Related
to this is the ability to account for the total odor as the combined con-
tribution of identified odorants in fresh or stale exhaust. A further
problem is the meaningfulness of exhaust odor measurements as made by any
individual method in relation to diesel exhaust encountered publicly.
15/R. N. Lacey. Advances in Organic Chemistry, Methods and Results, v. II,
(ed. R. A. Raphael, E. C. Taylor, and H. Wynberg), Interscience Pub. Inc.,
New York, 1960, pp. 216.
16/D. Gray, E. Lissi, and J. Heicklen. J. Phys. Chem., 7j>, 1919 (1972).
-------
TABLE 16. - Effects of selective chemical reagents on
exhaust odor intensity (PI odor units,
exhaust dilution lOQij-l
Exhaust treatment, ,
scrubber solution—
H2Q
Morpholine
Sodium sulfite
Permanganate
GM 3-53 engine
No load
3.1
2.6
2.6
2.1
Half load
2.8
2.5
3.2
2.2
Full load
5.2
4.8
4.6
3.1
Cummins NH-250 engine
No load
4.2
4.1
3.6
2.8
Half load
2.7
1.8
2.6
3.2
Full load
3.8
3.6
3.7
3.7
II All solutions were tested at 0° C, 15 ml in a liquid lift scrubber. Exhaust flow was 4.5 liter
"" per minute. Solution composition specified in experimental section.
oo
ID
-------
90
Analytical Methods Development
Organic hydroperoxides are known to be important combustion products 17/
although their instability can make analysis difficult. Some have been
shown to be intensely irritating to the nose and to be powerful vesicants 187.
This suggests that they may be detected at low concentrations as part of
the diesel odor. The original interpretation of the effectiveness of the
sulfite reagent scrubber was that its primary action was removal of hydro-
peroxides or other oxidizing compounds. Later it was recognized that it
might also remove alpha,beta unsaturated compounds but this did not elimi-
nate the possibility that it might also be acting on oxidizing odorants.
Some preliminary work was begun in isolation and analysis of hydroperoxides.
The all-glass packed columns appeared to be satisfactory for aliphatic
hydroperoxides but not for benzyl hydroperoxide. However, at .the termination
of this project promising results were being obtained in chromatography
of benzyl hydroperoxide with a short glass capillary column and it is
probable that this could be developed to give satisfactory results with
both the aliphatic and aromatic compounds. A single exhaust sample
collected in a cold water scrubber was extracted and processed to show
trace amounts of the aliphatic hydroperoxides that had been examined as
pure compounds. No quantitative determination was attempted on this sample.
At the stage to which this study had progressed, the major problem was the
evaluation of the hydroperoxide collection method over a range of molecular
weights and classes of compounds and determination of losses during isolation
and processing. Following this, it is likely that the whole procedure could
be systematized to permit quantitative exhaust analysis and collection of
samples for odor evaluation.
SUMMARY
Three main accomplishments may be noted for the program carried out during
the year of this report. They are:
1. Development of procedures and equipment for evaluation of
odorants isolated from diesel exhaust or formed synthetically.
The need for this method developed in the processing of small
isolated samples. The dynamic blending method, used in previous
work, was suitable only for large volume continuously generated
samples. Experience gained in conversion of a photochemical
chamber to use in exhaust odor rating showed the difficulties
that may be encountered in alteration of odor and presence of
background odor. In particular, it showed the need for regular
17/A. FisTu Angew. Chem. Intern. Ed., 7_, 45 (1968).
18/S. Dykstra and H. S. Mosher. J. Amer. Chem. Soc., 72, 3475 (1957).
-------
91
background odor measurements. The bag dilution system proved
to be adequate for use with materials that were not strongly ab-
sorbed on the bag walls. However, there was a serious loss of
odor from exhaust samples in the bag. The syringe dilution and
presentation method was the most satisfactory procedure since
it showed the least loss of exhaust odor and presented no back-
ground odor problem. Odor loss in the syringe was sufficient
to present some problem with quantitative odor assessment, but
it should be possible to evaluate unstable and reactive materials
when allowance is made for these losses.
t
2. Demonstration of the existence of unstable and reactive diesel
odorants.
Past studies on the detailed composition of diesel odorants has
concentrated on compound classes that are stable in ordinary
collection and isolation processes. However, common experience
with diesel exhaust odor has indicated that it is unstable. In
addition, combustion products are known to include several
classes of unstable and chemically reactive compounds. Reagent
scrubbing of exhaust showed possible presence of several such
classes, including particularly hydroperoxides and alpha,beta
unsaturated compounds. Simple aging of exhaust also showed odor
change.
3. Initial development of analysis and isolation methods for organic
hydroperoxides.
Work was begun in analysis of those compound classes indicated as
chemically reactive odorants. Demonstration of suitable gas
chroraatographic procedures for hydroperoxides was accomplished.
-------
INSTRUCTIONS FOR COMPLETING FORM NTIS-35 (10-70) (Bibliographic Data Sheet based on COSATI
Guidelines to Format Standards for Scientific and Technical Reports Prepared by or for die Federal Government,
PB-180 600).
1. Report Number. Each individually bound report shall carry a unique alphanumeric designation selected by the performing
organization or provided by (he sponsoring organization. Use uppercase letters and Arabic numerals only. Examples
FASEB-NS-87 and FAA-RD-68-09.
2. Leave blank.
3. Recipient's Accession Number. Reserved for use by each report recipient.
4. Title and Subtitle. Title should indicate clearly and briefly the subject coverage of the report, and be displayed promi-
nently. Set subtitle, if used, in smaller type or otherwise subordinate it to main title. When a report is prepared in more
than one volume, repeat the primary title, add volume number and include subtitle for the specific volume.
5- Report Date. F.ach report shall carry a date indicating at least month and year. Indicate the basis on which it was selected
(e.g., date of issue, date of approval, date of preparation.
6. Performing Organization Code. Leave blank.
7. AuthoK*)- Give name(s) in conventional order (e.g., John R. Doe, or J.Robert Doe). List author's affiliation if it differs
from the performing organization.
8. Performing Orgoniration Report Number. Insert if performing organization wishes to assign this number.
9. Performing Orgonilotion Name and Address. Give name, street, city, state, and zip code. Lisc no more than two levels of
an organizational hierarchy. Display the name of the organization e*actly as it should appear in Government indexes such
as USGRDR-I.
10. Projec»/To*k/Work Unit Number. Use the project, task and work unit numbers under which the report was prepared.
11. Controet/Gront Number. Insert contract or grant number under which report was prepared.
12- Sponsoring Agency Name and Address. Include zip code.
13. Type of Report and Period Covered. Indicate interim, final, etc., and, if applicable, dates covered.
14. Sponsoring Agency Code. Leave blank.
15. Supplementary Not*«. Enter information not included elsewhere but useful, such as: Prepared in cooperation with . . .
Translation of ... Presented at conference of ... To be published in ... Supersedes . . . Supplements . . .
16. Abstract. Include a brief (200 words ot less) factual summary of the most significant information contained in the report.
If the report contains a significant bibliography or literature survey, mention it here.
17 Key Words and Document Analy«i«- ("}• Descriptors. Select from the Thesaurus of Engineering and Scientific Terms the
proper authorized terms that identify the major concept of the research and are sufficiently specific and precise to be used
as index entries for cataloging. ... .,
-------
BIBLIOGRAPHIC DATA
SHEET
1. Report No.
EPA-R2-73-276
3. Recipient's Accession No.
4. Title and Subtitle
Characteristics and Photochemical Reactivity of Emissions
5- Report Dace
December 1972
6.
7. Author(s)
8. Performing Organization Kept.
No.
9. Performing Organization Name and Address
U. S. Bureau of Mines
Fuels Combustion Research Projects
Ba r11es v i11e, Ok 1ahoma
10. Project/Task/Work Unit No.
11. Contract/Grant No.
Interagency Agreement
EPA-IAG-0138(D)
12. Sponsoring Organization Name and Address
ENVIRONMENTAL PROTECTION AGENCY
Chemistry and Physics Laboratory
Research Triangle Park, North Carolina 27711
13. Type of Report & Period
Covered
Final
14.
15. Supplementary Notes
16. Abstracts y^g repOrt describes three separate projects which relate to the study of the
characteristics of emissions. Part I studied the association of automotive fuel composi-
tion with exhaust reactivity. This involved experimental tests with different automotive
engines and with gasolines of varied composition. Fuel composition was determined by gas
chromatography. Exhaust reactivity was both estimated from detailed composition data and
determined directly using a smog chamber. Part II, Toxic Products from Fuel Additives,
studied procedures for the determination of any toxic gaseous products resulting from
thermochemical reactions of fuel additives in the combustion process. An engine was op-
erated using additive-free fuel to obtain reference data on exhaust hydrocarbon and oxy-
genate composition; followed by tests in which the engine was operated using the same
fuel except with additive. The exhaust analyses included determination of: hydrocarbon,
oxygenates, and the additive and its direct combustion products. Part III, Diesel Odor,
studied the following problems: development of an odorant dilution and presentation sys-
tem suitable for use with isolated exhaust components and synthetic odorants, establlsh-
ng methods for demonstrating the reactivity and stability of diesel exhaust odorants,
and isolation and collection of unstable components.
17. Key Vords and Document Analysis. 17o. Descriptors
Ai r pollution
Exhaust emissions
Photochemical reactions
Gasoline
Fuel additives
Combustion products
Hydrocarbons
Odors
Chemical analyses
17b- Identifiers/Open-Ended Terms
17c- COSATI Field/Group 135
18. Availability Statement
Unlimited
19. Security Class (This
Report)
UNCLASSIFIED
20. Security Class (This
Pag
ASSIF1ED
21. No. of Pages
96
22. Price
FORM NTIS-3S (REV. 3-72)
THIS FORM MAY BE REPRODUCED
USCOMM-DC I4I32-P72
------- |