PB91-243386
EPA/600/3-91/050
September 1991
DEVELOPMENT OF OZONE REACTIVITY SCALES
FOR VOLATILE ORGANIC COMPOUNDS
William P. L. Carter
Statewide Air Pollution Research Center
University of California
Riverside, CA 92521
Cooperative Agreement No. CR-814396-01-0
August 1991
Project Officer
Joseph J. Bufalini
Chemical Processes and Characterization Division
Atmospheric Research and Exposure Assessment Laboratory
Research Triangle Park, NC 27711
ATMOSPHERIC RESEARCH AND EXPOSURE ASSESSMENT LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U. S. ENVIRONMENTAL PROTECTION AGENCY
RESEARCH TRIANGLE PARK, NC 27711
-------
TECHNICAL REPORT DATA .
(fleetr rtod luumetioni on iht rtvmt btfort eompltn
I. REPORT NO.
EPA/600/3-91/050
a.
PB91-243386
4. Ti~i.E AND SUBTITLE
DEVELOPMENT OF OZONE REACTIVITY SCALES FOR VOLATILE
ORGANIC COMPOUNDS
». REPORT PATE
September 1931
*. PERFORMING ORGANIZATION CODE
7. AUTHOR!*)
William P. L. Carter
I. PERFORMING ORGANIZATION REPORT NO.
io. PROGRAM CLEMENT NO.
Aim P/R /£
8. PERFORMING ORGANIZATION NAME AND ADDRESS
Statewide Air Pollution Research Center
University of California
Riverside, CA 92521
mi
.CO
VI. CONTRACT/GRANT NO.
CR-814396-01
12. SPONSORING AGENCY NAME AND ADDRESS
13. TYPE OF REPORT AND PERIOD COVERED
Atmospheric Research and Exposure Assessment Laboratory
Office of Research and Development
U.S. Environmental Protection Agency
Research Triangle Park. North Carolina 27711
CODE
EPA/600/09
15. SUPPLEMENTARY NOTES
16. ABSTRACT
Methods for developing a numerical scale ranking reactivities of volatile
organic compounds (VOCs) towards ozone formation were investigated. Effects of small
VOC additions on ozone formation (incremental reactivities) were calculated for 140
types of VOCs in model scenarios representing a variety of single-day pollution
episodes. Relative reactivities determined from effects of the VOCs on maximum ozone
concentrations (ozone yields) varied widely among the scenarios, but relative
reactivities determined from effects on integrated ozone levels were less variable.
A "maximum reactivity" scale was derived from ozone yield reactivities in scenarios
where N0x inputs were adjusted so the VOCs had the greatest effect on ozone, and a
"maximum ozone" scale was derived from scenarios where N0x inputs gave maximum ozone
concentrations. These scales gave different relative reactivities for many VOCs,
particularly aromatics. Several '"multi-scenario" scales were derived from the ozone
yield and integrated ozone reactivities in the unadjusted scenarios. .The maximum
ozone scale was more consistent with averages of ratios of ozone yield reactivities,
but the maximum reactivity scale was more consistent with ratios of integrated ozone
reactivities and also corresponded best to multi-scenario scales developed to minimize
the total error in ozone predictions. Information concerning effects of NO levels,
of the composition of base case VOC emissions, and of other scenario conditions on
reactivities was also obtained.
7.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.lDENTlFIERS/OPEN ENDED TERMS
COS AT i Field'Croup
1i. DISTRIBUTION STATEMENT
RELEASE TO PUBLIC
IB. SECURITY CLASS (Ttuj KtfKtrtJ
UNCLASSIFIED
21. NO. OF PAGES
127
JO SECURITY CLASS (T*up*tr>
UNCLASSIFIED
22. PRICE
EPA
-------
NOTICE
The information in this document has been funded by the United States
Environmental Protection Agency under Cooperative Agreement No. CR-814396-
01-0 to the University of California at Riverside. It has been subject to
the Agency's peer and administrative review, and has been approved for
publication as an EPA document. Mention of trade names or commercial
products does not constitute endorsement or recommendation for use.
11
-------
ABSTRACT
Methods for developing a numerical scale ranking reactivities of
volatile organic compounds (VOCs) towards ozone formation were
investigated. Effects of small VOC additions on ozone formation
(incremental reactivities) were calculated for 140 types of VOCs in model
scenarios representing a variety of single-day pollution episodes.
Relative reactivities determined from effects of the VOCs on maximum ozone
concentrations (ozone yields) varied widely among the scenarios, but
relative reactivities determined from effects on integrated ozone levels
were less variable. A "maximum reactivity" scale was derived from ozone
yield reactivities in scenarios where NOV inputs were adjusted so the VOCs
A
had the greatest effect on ozone, and a "maximum ozone" scale was derived
from scenarios where NOV inputs gave maximum ozone concentrations. These
A
scales gave different relative reactivities for many VOCs, particularly
aromatics. Several "multi-scenario" scales were derived from the ozone
yield and integrated ozone reactivities in the unadjusted scenarios. The
maximum ozone scale was more consistent with averages of ratios of ozone
yield reactivities, but the maximum reactivity scale was more consistent
with ratios of integrated ozone reactivities and also corresponded best to
multi-scenario scales developed to minimize the total error in ozone
predictions. Information concerning effects of NO levels, of the
composition of base case VOC emissions, and of other scenario conditions
on reactivities was also obtained. Although case-by-case analysis is the
best method to determine effects of VOC emission changes on ozone
formation, the maximum reactivity scale may be appropriate when a single
VOC ranking must be used.
111
-------
ACKNOWLEDGEMENTS
The author wishes to thank Drs. Michael W. Gery and Gary Z. Whitten
for providing computer-readable input files for the scenarios employed in
this study, and Dr Roger Atkinson, Mr. Bart Croes, and Mr. Alvin Lowi,
Jr., for helpful discussions. He also thanks Dr. Roger Atkinson, Dr
Nelson Kelly, Mr. James Killus, Dr. Harvey Jeffries, Dr. Marcia Dodge, Dr.
Basil Dimitriades, Dr. Joseph Bufalini, Dr. Alvin Gordon and others for
reviewing drafts of this report and providing helpful comments. However,
the opinions and conclusions given herein are entirely those of the
author.
The author acknowledges additional support for related research
provided through contracts with the California South Coast Air Quality
Management District, the California Air Resources Board (Contract No.
A932-094), and a consulting contract for the Western Liquid Gas
Association.
IV
-------
TABLE OF CONTENTS
Page
I. INTRODUCTION 1
11. METHODS 9
A. Pollution Scenarios Used for Reactivity Assessment 9
1. Representative Pollution Episodes 9
2. Representation of Base Case ROG Compositions 15
3 . Types of Scenarios Used 19
a. Base Case Scenarios 19
b. Maximum Reactivity Scenarios and Maximum Ozone
Scenarios 20
c. Averaged Conditions Scenarios 22
d. Multi-Day Scenarios 23
B. Chemical Mechanism Used for Representative VOCs 24
1 . Description of Mechanism 24
2. Uncertainties in Mechanisms for Individual VOCs 34
C. Incremental Reactivity in an Individual Scenario 35
1. Ozone Yield and Integrated Ozone Reactivities 37
2. Separate Estimates of Reactivity Components 38
a. Estimation of Kinetic Reactivities 39
b. Estimation of Mechanistic Reactivities 40
D. Derivation of Generalized or Multi-Scenario
Reactivity Scales 41
1. Derivation of the Maximum Reactivity (MaxRct)
and the Maximum Ozone Reactivity (MaxOo)
Generalized Scales . . 42
2. Multi-Scenario (Base Case) Relative Reactivity
Scales 44
a. Average Ratio Method 45
b. Minimum Least Squares Error (Least Squares Fit)
Method 45
III. RESULTS AND DISCUSSION 49
A. General and Multi-Scenario Reactivity Scales ...49
1 MaxRct and MaxOo Reactivity Scales 49
a. Comparison of Kinetic Reactivities 54
b. Comparison of Mechanistic Reactivities 54
c. Comparison of Incremental Reactivities 55
-------
TABLE OF CONTENTS
(continued)
2. Base Case Relative Reactivity Scales 57
a. Effect of Derivation Method on Ozone
Yield Reactivities 64
b. Comparison of Case Ozone Yield Reactivities
with MaxRct and MaxO, 64
c. Integrated Ozone Reactivities 65
d. Comparison of Integrated Ozone Reactivities with
MaxRct and MaxOj 67
3. Comparison of Incremental Reactivities with the OH
Radical Rate Constant Scale 69
B. Dependence of Reactivities on Scenario Conditions 71
1. Dependence of Reactivity Scales on NO 71
a. NO -Limited Relative Reactivities 72
b. Effect of NOX on Integrated Ozone Reactivities 74
2. Effect of Variation of Non-N0x-Related Scenario
Conditions 75
3. Effect of Variation of the Base ROG Mixture 79
IV. SUMMARY AND CONCLUSIONS 85
V . REFERENCES 92
APPENDIX
A UPDATES TO THE CHEMICAL MECHANISM A- 1
B ESTIMATION OF MECHANISTIC REACTIVITIES USING
"PURE MECHANISM" SPECIES B-1
VI
-------
LIST OF TABLES
Table
Number Title
Summary of the Major Characteristics of the EKMA
Scenarios Used for Reactivity Assessment 11
Composition of the Mixtures used to Represent Base Case
ROG Emissions and Aloft ROG Pollutants in the Scenarios 17
Summary of VOC Species and Ozone Reactivity Estimates for
the Maximum Reactivity ("MaxRct") and the Maximum Ozone
Reactivity ("MaxO^" ) Scales 26
Summary of Relative Reactivities for Selected VOC
Species in the MaxRct, MaxOo, and the Base Case
Relative Reactivity Scales 60
Standard Deviations of Averages of Kinetic, Mechanistic,
Incremental, and Relative Reactivities of Selected
VOCs for the Maximum Reactivity and the Maximum
Ozone Scenarios 77
VII
-------
LIST OF FIGURES
Figure
Number Title
Diagram of the derivation of the generalized and the
multi-scenario reactivity scales .............................. 43
Distribution plots of maximum ozone, integrated ozone, the
IntOH parameter, and the base ROG mixture reactivity for the
representative maximum reactivity, maximum ozone, and
base case scenarios ........................................... 50
Distribution plots of kinetic, mechanistic, incremental, and
relative reactivities of CO in the representative maximum
reactivity, maximum ozone, and base case scenarios ............ 51
Distribution plots of kinetic, mechanistic, incremental,
and relative reactivities of n-butane in the representative
maximum reactivity, maximum ozone, and base case scenarios. .. .52
Distribution plots of kinetic, mechanistic, incremental, and
relative reactivities of toluene in the representative
maximum reactivity, maximum ozone, and base case scenarios. .. .53
Plots of mechanistic reactivities for the VOCs in the
MaxOo scale against their mechanistic reactivities in
the MaxRct scale .............................................. 56
7 Plots of incremental reactivities for the VOCs in the o
scale against their incremental reactivities in the
MaxRct scale .................................................. 58
8 Plots of incremental reactivities of selected VOCs against the
reactivity of the base case ROG mixture in the same scenario
for each representative base case scenario .................... 62
9 Plots of base case relative ozone yield reactivities for the
VOCs against their MaxRct or MaxOo reactivities ............... 66
10 Plots of base case relative integrated ozone reactivities
for the VOCs against their MaxRct or MaxOo reactivities ....... 68
11 Plots of OH radical rate constants (per mass basis) for the
VOCs against their MaxRct reactivities ........................ 70
12 Plots of relative ozone yield reactivities and relative
integrated ozone reactivities of selected VOCs against the
ROG/NOV ratio in the averaged conditions scenarios ............ 73
X
13 Effects of variation of the composition of the base ROG
mixture on incremental reactivities in the maximum
reactivity averaged condition scenario ........................ 82
Vlll
-------
LIST OF FIGURES
Figure
Number Title
14 Effects of variation of the composition of the base ROG
mixture on incremental reactivities in the maximum ozone
averaged condition scenario 83
IX
-------
I . INTRODUCTION
The formation of photochemical ozone continues to be a complex
problem in many urban areas. Ozone is not emitted directly into the
atmosphere; it is formed by the interaction of volatile organic compounds
(VOCs) with oxides of nitrogen (NO ) in sunlight. Effective ozone control
strategies must focus on both. However, in urban areas where NOX sources
are abundant and difficult to control, VOC control is the strategy
receiving the most attention, at least for the near term. Since reduction
of total VOC emissions from most of the major sources is difficult and
costly, strategies are being examined that involve changing the chemical
nature of VOC emissions so that they have less of a tendency to promote
ozone formation, i.e., so that they are less "reactive". Two example
strategies are the conversion of motor vehicles to alternative fuels and
the substitution of currently used solvents with less reactive
compounds. The development and assessment of such strategies require a
means to quantify and compare the ozone formation reactivities of the many
types of VOCs which can be emitted.
There are a number of ways to quantify the reactivities of VOCs.
Many of the reactivity scales used previously are based on the amounts of
ozone formed when the VOC is irradiated in N0x-air mixtures in
environmental chambers (e.g., Wilson and Doyle, 1970; Altshuller and
Bufalini, 1971; Laity et al., 1973). However, individual VOCs are almost
never emitted into the atmosphere in the absence of other reactive organic
compounds, and thus such experiments do not represent atmospheric
conditions. In addition, chamber effects are known to significantly
affect the results of such experiments (Bufalini et al., 1977; Joshi et
al. , 1982; Carter et al. , 1982), particularly if the compound reacts
relatively slowly or has radical sinks in its mechanism (Carter et al.,
1986a; Carter and Lurmann, 1990, 1991). Because of this, single organic-
NO -air experiments are no longer being considered as a viable means for
quantifying reactivity. An alternative measure of reactivity is the rate
at which the VOC reacts with hydroxyl radicals (e.g., Darnall et al. ,
1976; CARB, 1989), since for most compounds this is the main factor which
determines its atmospheric lifetime (Atkinson, 1989, 1990). The
advantages of this reactivity scale are that it is universal and that OH
-------
radical rate constants are known or can be estimated for most of the major
types of VOCs which are emitted (Atkinson, 1987, 1989, 1990). However,
it does not account for the significant differences in VOC reaction
mechanisms (e.g., see Gery et al. 1988; Atkinson, 1989; Carter, 1990),
which can affect how much ozone is formed once the VOC reacts (Carter and
Atkinson, 1987, 1989).
The most direct measure of ozone reactivity of a VOC is the change in
ozone levels caused by changing the emissions of the VOC in an actual air
pollution episode. Reactivity measured in this way takes into account not
only the effects of all aspects of the organic's reaction mechanism, but
also any effects of the environment into which it is emitted which
influence how much ozone its emissions cause. This cannot be measured
experimentally (other than by actually changing emissions in real airsheds
and measuring the effect on air quality), but can be estimated by computer
airshed models, provided that the models have an adequate representation
both of the conditions of the episode and of the kinetics and mechanisms
of the VOC reactions that affect ozone formation. This approach has been
employed in a number of modeling studies of the effects of VOC emission
changes on ozone formation (e.g., Bufalini and Dodge, 1983; Dodge, 1984;
Hough and Derwent, 1987; Carter and Atkinson, 1989; Chang and Rudy, 1990),
and it is the approach which is used in this work. Although obviously the
results are no more valid than the model of the chemical reactions or the
air pollution episode being considered, modeling provides the potential
for the most realistic and flexible means to assess the many factors which
affect VOC reactivity and for the development of VOC reactivity scales.
In general, the effect of changing the emissions of a given VOC on
ozone formation in a pollution episode will depend on the magnitude of the
emission change and on whether the VOC is being added to, subtracted from,
or replacing a portion of the base case (i.e., present day) emissions. If
the purpose of the calculation is to assess a particular proposed VOC
emission change, the amount of VOC(s) added, subtracted, or substituted
will be determined by the specific strategy being considered. However,
for general reactivity assessment purposes, the amount added, subtracted,
or substituted is essentially arbitrary. To avoid the dependence on this
arbitrary parameter, and for other reasons as discussed below, we have
proposed use of "incremental reactivity" as the means to quantify ozone
-------
impacts of VOCs (Carter and Atkinson, 1987). This is defined as the
change on ozone caused by adding an arbitrarily small amount of the VOC to
the emissions, divided by the amount of VOC added. (This can also be
called the "local sensitivity" of ozone to the VOC, or the derivative of
ozone with respect to emissions of the VOC.) In addition to removing the
dependence of the VOC impact on the amount of VOC added, the approach has
the further advantage that incremental reactivities of mixtures can be
obtained by linear summation of the incremental reactivities of their
components. This arises mathematically from the fact that incremental
reactivities are derivatives of a continuous (though nonlinear)
function. This has obvious advantages in the assessment of reactivities
of complex mixtures, such as vehicle exhausts (e.g., see Lowi and Carter,
1990).
Mote that since incremental reactivities measure the effects of
adding small amounts of VOCs to the emissions, they do not necessarily
predict the effects of large changes in emissions, as might occur, for
example, if all the motor vehicles in an airshed were converted to another
type of fuel. Effects of such large changes of emissions would obviously
have to be examined on a case-by-case basis. However, Chang and Rudy
(1990) found that incremental reactivities give good approximations to
effects on ozone of alternative fuel substitution scenarios involving
changing 30? of the total VOC emissions. This suggests that incremental
reactivities may be useful for estimating effects of proposed large
changes in emissions, at least for screening or initial assessment
purposes. Furthermore, many practical control strategies involve VOC
sources which make only relatively small contributions to the total
emissions, and even those involving large sources are almost always
implemented on a relatively gradual basis. Incremental reactivities will
predict the direction of an initial ozone trend, which results when a
control strategy is phased in, and in most cases should also give a good
approximation of the result once the control strategy is completely
implemented.
Incremental reactivities of VOCs have been investigated
experimentally (Carter and Atkinson, 1987) and in a number of computer
modeling studies (Bufalini and Dodge, 1983; Dodge, 1984; Hough and
Derwent, 1987; Carter 1989a, 1989b; Weir et al., 1988; Carter and
-------
Atkinson, 1989; Chang and Rudy, 1990). As expected, an organic compound's
atmospheric reaction mechanism was found to be important in affecting its
incremental reactivity. The most important factor is the rate at which
the VOC reacts in the atmosphere, but other aspects of the mechanism are
also important and cannot be ignored. The reactions of some compounds can
cause the formation of 10 or more additional molecules of ozone per carbon
atom reacted (either directly or through its effects on reactions of other
compounds), while reactions of others cause almost no ozone formation, or
actually cause ozone formation to be reduced (e.g., see Carter and
Atkinson, 1989). The predictions that VOCs have variable effects on ozone
formation, even after differences in reaction rates are taken into
account, and that some have negative effects on ozone formation under some
conditions, have been verified experimentally (e.g., Carter Atkinson,
1987).
These modeling studies also indicate that incremental reactivities of
VOCs can significantly depend on the environmental conditions where the
VOC is emitted (e.g., Dodge, 1984; Carter and Atkinson, 1989). The most
important environmental factor is the availability of NOV in the system,
X
which is most conveniently measured by the ratio of total emissions of
reactive organic gases (ROG) to NOV. In general, VOCs are found to have
A
the highest effects on ozone formation under relatively high NOV
A
conditions (i.e., low ROG/NOV ratios) and to have much lower, in some
A
cases even negative, reactivities under conditions where NO is limited
(high ROG/NO ratios). This is because under relatively high NO
conditions the amount of ozone formed is determined by the levels of
radicals formed from the reactions of the VOCs, while under lower NO,,
A
conditions it is the availability of NO , which must be present in order
for ozone to be formed, which limits ozone formation. Other aspects of
the environment in which the VOC is emitted, such as nature of the other
organics emitted into the airshed (Weir et al. 1988), the amount of
dilution occurring (Carter and Atkinson 1989), etc., can also be important
in affecting VOC reactivities, though investigations of these aspects are
more limited.
The fact that VOC reactivities (i.e., incremental reactivities)
depend on environmental conditions means that no single scale can predict
incremental reactivities, or even ratios of incremental reactivities,
-------
under all conditions. This obviously complicates the development and use
of VOC reactivity scales for regulatory and control strategy assessment
purposes. In view of this, those concerned with developing effective VOC
control strategies for reducing ozone are faced with the following
options.
(1) VOC reactivity can be ignored completely, with VOCs being
regulated on a per-mass (or per-carbon) basis entirely. This is not
strictly speaking the present policy, since some obviously unreactive VOCs
are exempted from regulation as ozone precursors. However, other than
exemptions for a few VOCs, regulation of VOCs on a per-mass or per-carbon
basis is largely the present regulatory approach. This has the obvious
advantage of simplicity and ease of enforcement. It also addresses to
some extent the fact that some VOCs may have other adverse air quality
impacts besides enhancing ozone. However, ignoring reactivity differences
among non-exempt compounds will result in opportunities for cost-effective
ozone control strategies to be missed, and in some cases this approach can
therefore be counter-productive.
(2) Control strategies involving VOC substitutions could be examined
on a case-by-case basis only. Thus, the concept of a "reactivity scale"
would not be used in developing ozone control strategies. This is
probably the best option from a scientific point of view, and it is
clearly the most appropriate when large and costly strategies are being
considered. This approach has often been employed in assessing the
effects of using alternative fuels (e.g., see Russell et al. 1989) and has
the potential for the most accurate representation of the specific cases
which are examined. However, because of the large costs of running
realistic airshed models, it is not presently practical for screening
purposes when a large number of alternative options need to be considered,
or for assessing relative impacts of small and varied VOC sources.
(3) If reactivity scales are to be used for regulations, the most
scientifically justifiable approach would be to have a separate reactivity
scale for each ozone pollution episode. In general this means not only
separate scales for each air basin, but also separate scales for each type
of meteorological condition where ozone exceedences occur in a basin.
However, this is not practical at the present time, if only because there
are acceptable airshed models for only a limited number of episodes. In
-------
addition, it is unclear how this approach could be implemented in
regulations applied on a statewide or national level.
(4) Reactivity scales derived for conditions of a selected "worst
case" or a "typical" ozone pollution episode could be assumed to have
general applicability. This permits use of reactivity scales for
screening purposes, for assessment of multiple options, and for assessment
of small and varied VOC sources. However, the results may not be
applicable to other air basins or meteorological conditions, and if the
wrong type of episode is chosen, non-optimal or counterproductive ozone
control strategies may result. The selection of the episode used for this
purpose is clearly critical. However, in practice the approach has been
to select ozone pollution episodes according to the availability of data
needed for modeling purposes, not whether the episode is optimal for use
in deriving reactivity scales for general use.
(5) A refinement of the above option is to use a generalized
reactivity scale derived such that its use will have the practical effect
of resulting in the greatest overall air quality improvements for the full
distribution of ozone pollution episodes. By definition, this would be
the optimum generalized reactivity scale for use in those applications
where detailed case-by-case examination of control strategies is not
practical. Note that such a scale would not necessarily represent the
exact conditions of any particular episode. How such a scale is derived
would depend, at least to some extent, on how "overall air quality" is
quantified and on the distribution of episodes of interest.
Although there have been a number of studies of VOC reactivities in
specific or idealized episodes, as well as several investigations of how
reactivities depend on environmental conditions (see references above),
until recently there has been relatively little work aimed specifically at
determining what constitutes an optimal VOC reactivity scale for assessing
ozone control strategies which will be applied to a variety of
conditions. In what appears to be a first effort in this regard, we
recently proposed the use of a "maximum incremental reactivity" criterion
as the basis for deriving a general reactivity scale (Weir et al., 1988;
Carter, 1989a,b; Lowi and Carter, 1990). This scale was based on using
the highest calculated incremental reactivity for each VOC for a set of
idealized one- and two-day scenarios. The rationale for this approach is
-------
that by definition it reflects conditions where VOC control is the most
effective for controlling ozone. It obviously makes more sense than
ranking VOCs according to conditions where changing the VOCs has little
effect on ozone, as is the case when ozone formation is NO -limited.
A
(Note that VOC reactivity is relevant only for conditions where VOC
control is an appropriate ozone reduction strategy. If ozone is NOX-
limited, NOX control is the only appropriate strategy for reducing ozone,
and thus VOC reactivity is irrelevant.) In light of this rationale, the
California Air Resources Board has recently approved using the maximum
incremental reactivity approach for deriving reactivity adjustment factors
in emissions standards for vehicles using alternative or reformulated
fuels (CARB, 1990).
However, the extent to which use of a maximum incremental reactivity
scale represents an optimum approach for assessing VOC controls for a wide
range of conditions has not been adequately assessed. Indeed, this
approach can be criticized on several grounds. As indicated above, the
reactivities of VOCs depend primarily on the ROG/NO ratio in the
atmosphere where they are emitted, and maximum VOC reactivities occur at
relatively low ROG/NO ratios. Many if not most airsheds are believed to
be characterized by ROG/NO ratios which are significantly higher than
those yielding maximum reactivity (e.g., see Bauges, 1986). In addition,
ROG/NOV ratios giving maximum ozone concentrations are always higher than
A.
those where VOCs have their maximum reactivities. Since the air quality
standard for ozone is based on peak ozone concentrations, a reactivity
scale based on conditions yielding maximum ozone concentrations may seem
to be more appropriate for assessing ozone control strategies. However,
the advantages of using maximum ozone rather than maximum reactivity as
the basis for deriving a general reactivity scale or even whether there
are significant practical differences between the two approaches have
not been investigated.
An alternative approach for deriving a reactivity scale for general
use would be to determine the distribution of all airshed conditions where
ozone formation is a problem, and then somehow combine or average the
reactivities calculated for a set of scenarios representing this
distribution. This approach would require a comprehensive analysis of the
distribution of chemical conditions in airsheds which might affect VOC
-------
reactivities. The distribution of chemical conditions in airsheds is
highly uncertain. In addition, the best approach must be identified for
determining a combined or average reactivity scale from varying
reactivities calculated for the individual episodes. Because reactivities
(including relative reactivities) depend significantly on the relative
availability of NOX in the system (Dodge, 1984; Carter and Atkinson,
1989), the reactivity scale would be particularly sensitive to the
distribution of relative NOX levels, which is highly uncertain (e.g., see
Bauges, 1986). It may not be appropriate to include in the average those
conditions where ozone formation is NOV limited, and where VOC control is
X
therefore ineffective as an ozone control strategy. This also has not
been adequately investigated.
In this study, we report the results of an investigation of
alternative approaches for deriving a generalized VOC reactivity scale for
assessment of ozone control strategies. A set of 62 idealized single day
ozone pollution scenarios, representing 12 different urban areas
throughout the United States, was taken as a representative distribution
of ozone pollution episodes, and several different types of reactivity
scales were derived based on the scenarios. The predictions of relative
reactivities are summarized and compared, and the potential advantages and
disadvantages of each as a basis for assessing ozone control strategies
affecting the set of scenarios are discussed. A limited sensitivity study
concerning how airshed conditions affect reactivity was also conducted,
primarily to assess the range of uncertainties in such reactivity
scales. Although the set of scenarios employed is highly idealized and
may not accurately represent any real pollution episodes, it represents a
widely varying set of conditions and as such is useful for investigating
the problem of developing generalized reactivity scales.
-------
II. METHODS
A. Pollution Scenarios Used for Reactivity Assessment
1. Representative Pollution Episodes
The development of a comprehensive set of idealized pollution
scenarios representing the true distribution of ozone pollution conditions
requires an analysis of the range of conditions in airsheds where ozone
formation is a problem. Such an analysis is beyond the scope of this
study. However, an extensive set of idealized pollution scenarios has
already been developed for the purpose of city-specific EKMA analyses
(e.g., see Gipson et al., 1981; EPA, 1984) of ozone formation in various
areas of the United States. The EKMA approach involves use of single-cell
box models of one-day urban ozone pollution episodes for estimating the
amount of ROG reduction needed to reduce ozone to achieve the national air
quality standard (Dodge, 1977). Although EKMA models use only a single
cell formulation and thus cannot represent realistic pollution episodes in
great detail, they can represent dynamic injection of pollutants, time-
varying changes of inversion heights with entrainment of pollutants from
aloft as the inversion height increases throughout the day, and time-
varying rates of photolysis reactions and temperatures. Thus these models
can represent a wide range of chemical conditions which may affect
predictions of effects of ROG control on ozone formation. These chemical
conditions are the same as those affecting predictions of VOC
reactivity. Therefore, at least to the extent they are suitable for their
intended purpose, the existing EKMA scenarios should also be suitable for
assessing methods to develop reactivity scales encompassing a wide range
of conditions.
The set of EKMA scenarios (or EKMA city-days) used as the
starting point in this study were those used by Gery et al. (1987) in
their study of effects of changing UV radiation (due to stratospheric
ozone depletion) on ozone formation in various urban areas, and those used
by Whitten (1988) to study the effects of changes in ozone due to changes
in vehicle fuel composition. These in turn were taken from EKMA analyses
used as parts of State Implementation Plans for reducing ozone in various
areas throughout the United States. The scenarios represented 62 episodes
(city-days) in 12 different urban areas. [Five of the city-days from
-------
Whitten (1988) represented episodes which were also on the set from Gery
et al. (1987). However, the input conditions are sufficiently different
that they are treated as separate scenarios for the purpose of this
study.] The major characteristics of these city-days are listed in Table
1, which shows that a wide variety of conditions are represented. For
example, the maximum calculated ozone ranged from 76 ppb (well below the
federal ozone standard) up to almost 0.37 ppm; the maximum inversion
height ranged from 400 to 3300 meters; and the initial plus emitted ROG
--)
pollutant input ranged from less than 2 to more than 50 mmol m .
No claim is made as to the accuracy of these model scenarios in
simulating the episodes they are designed to represent, though clearly
they were developed with an attempt to make their input data and
predictions as consistent as possible with the available (though generally
limited) data. However, even if they are not accurate in representing
their particular episodes, they represent the modelers' best efforts to
represent, as accurately as possible given the available data and the
limitations of the formulation of the EKMA model, the wide range of
conditions occurring in urban areas throughout the United States. This
variability, more than the accuracy of any particular scenario in
representing a specific episode, makes them useful for the objectives of
this study.
The input data for these scenarios were provided to the author
by Gery (1989) and Whitten (1989) in the form of input files to the OZIPM
(Gipson, 1984; Hogo and Gery, 1988) computer program. Since the available
versions of the OZIPM program are not compatible with the detailed
mechanism used in this study, these inputs were converted for use in the
software employed in this study. This involved some approximations to the
representations used by the OZIPM program. For example, in the
simulations reported here, the photolysis rates were calculated by using
the ground-level solar actinic fluxes given by Peterson (1976) for his
"best estimate" surface albedos, which are somewhat different from those
used by the OZIPM program. To partially correct for this, all photolysis
rates were adjusted upwards by a factor of 1.093 to yield the same N02
photolysis rate as calculated by OZIPM4 for direct overhead sun. The
effects of these differences on the model predictions should be minor -
10
-------
Table 1. Summary of the Major Characteristics of the Scenarios Used for Reactivity Assessment. (Scenarios with a
with the ROG/NOX ratios were used to derive the base case, MaxRct or MaxCK reactivity scales.)
shown
ID
[a]
G-MA1
G-MA2
G-MA3
G-MA4
W-MA4
G-MA5
G-MA6
G-MA7
G-NA1
G-NA2
G-NA3
G-NA4
G-NA5
G-NA6
G-NA7
G-NA8
W-NA8
G-NA9
G-NY1
G-NY2
City, Site, and Date
Massachusetts
Nashua, 5/30/78
Nashua, 7/21/78
Nashua, 8/13/78
Nashua, 8/15/78
(Same as G-MA4)
Nashua, 8/16/78
Lexington, 6/29/79
Lexington, 7/10/79
Nashville
Nashville, 7/2/80
Health Center, 7/25/80
Hendersonville, 7/25/80
Health Center, 8/1/80
Hendersonville, 8/1/80
Health Center, 8/10/80
Hendersonville, B/10/80
Hendersonville, 9/11/81
(Same as G-NA8)
Hendersonville, 9/12/81
New York
Stratford, 6/24/80
Middletown, 6/24/80
Base .
Oomax
(ppb)
127
107
117
125
131
109
143
1 10
88
86
97
76
87
120
131
85
137
80
218
226
--- ROG/NOX |b]---
Base MaxRct MaxCK
[c] [d] [er
"6.1
"11.7
6.3
6.1
6.4
8.5
6.0
7-4
14.6
"19.1
14.7
18.9
14.5
10.6
9.1
"11.0
11.0
12.9
"12.8
10.8
"6.0
3.5
"7.0
6.0
6.0
4.0
4.0
"3.5
6.0
"5.0
5.0
6.0
6.0
"5.0
6.0
8.0
"6.0
8.0
"6.0
"6.0
8.0
"5.0
"10.0
8.0
10.0
6.0
7.0
5.0
"8.0
8.0
8.0
10.0
10.0
8.0
8.0
"2.0
8.0
12.0
"8.0
"8.0
Light
If]
240
219
238
227
227
223
255
237
230
223
223
220
220
215
215
188
184
187
238
238
Inv. Height
(m)
Init. Final
250
250
250
250
250
250
449
360
250
250
250
250
250
250
250
250
250
250
300
300
571
1850
427
580
536
893
867
1704
1845
1845
1845
1845
1845
1845
1845
1845
1845
1845
1561
1561
p
- Input flux (mmol/nr),
(% input emitted) [g] -
._. HC --- - Base NOX-
5.45
3.12
5.00
5.26
5.13
3.05
4.60
3.74
2.81
1.87
2.57
1 .96
2.67
4.58
6.00
4.41
4.41
3.13
34.24
32.41
(40?)
(25?)
(35?)
(38?)
(38?)
(23?)
(24?)
(25?)
( 7?)
( 9?)
(23?)
( 9?)
(23?)
( 9?)
(23?)
(23?)
(23?)
(23?)
(28?)
(24?)
0.99 (65?)
0.49 (32?)
0.85 (59?)
0.95 (64?)
0.95 (64?)
0.48 (30?)
0.88 (52?)
0.80 (58?)
0.59 ( 7?)
0.40 ( 9?)
0.57 (24?)
0.41 ( 9?)
0.59 (24?)
0.98 ( 9?)
1.31 (24?)
0.93 (24?)
0.93 (24?)
0.69 (24?)
2.78 (46?)
3.15 (53?)
Aloft
-- HC ---
(ppb) (?)
40
40
40
40
80
40
40
40
90
90
90
90
90
90
90
90
90
90
40
40
( 9?)
(45?)
( 5?)
( 9?)
(15?)
(26?)
(13?)
(37?)
(67?)
(76?)
(69?)
(75?)
(69?)
(56?)
(49?)
(57?)
(57?)
(65?)
( 6?)
( 6?)
[h]
°3 N0x
- (ppb) -
33
81
24
56
60
60
56
63
15
35
35
13
13
42
42
i)
100
6
80
80
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
1
1
(continued)
-------
Table 1. (continued) - 2
ID
[a]
G-PH1
G-PH2
G-PH4
G-PH5
G-PH6
G-PH7
G-PH8
G-PX1
G-PX2
G-PX3
G-PX1.
W-PX1
G-PX5
G-PG1
G-PG2
G-PG3
G-PG1
G-PG5
G-PG6
G-PG?
G-PG8
City, Site, and Date
Philadelphia
Philadelphia, 7/13/79
Philadelphia, 7/19/79
Trenton, 6/15/81
Philadelphia, 6/16/81
Philadelphia, 8/19/82
Roxborough, 6/27/83
Roxborough, 6/27/83
Phoenix
Phoenix, 7/5/81
Phoenix, 8/29/81
Phoenix, 8/31/81
Phoenix, 9/11/81
(Same as G-PX1)
Glendale, 9/11/81
Seattle
Graham, 8/7/81
Sumner, 8/7/81
Graham, 8/10/81
Sumner, 8/10/81
Firwood, 8/10/81
Graham, 8/11/81
Sammamish, 8/11/81
Sumner, 8/11/81
Base.
Csmax
(ppb)
233
151
151
208
131
198
173
115
137
129
176
160
154
139
141
156
157
156
135
132
136
ROG/NOX (b]-~
Base MaxRct MaxCU
[c] [d] [e]J
6.8
6.1
8.3
8.2
"8.0
6.3
"5.7
12.5
"13.8
11.9
"11.8
11.6
11.6
5.5
5-5
5.6
5.6
5.6
5.6
5.6
5.6
6.0 8.0
"8.0 10.0
"7.0 "10.0
"6.0 10.0
6.0 8.0
6.0 "8.0
6.0 8.0
"6.0 "10.0
"7-0 10.0
"7.0 10.0
7.0 "12.0
6.0 10.0
7.0 10.0
"8.0 "10.0
8.0 10.0
7.0 10.0
7.0 10.0
7.0 10.0
7.0 "10.0
7.0 10.0
7.0 10.0
Light
[f]
236
243
211
239
210
242
212
196
152
133
152
150
152
227
227
222
222
222
217
217
217
Inv. Height
(m)
Init. Final
250
250
250
250
250
250
250
530
250
250
250
250
250
250
250
250
250
250
250
250
250
1135
871
1682
1211
1659
996
996
2100
3300
3300
2400
2400
2100
629
629
681
681
681
818
818
818
- Input flux (mmol/m ),
(? input emitted) (g) -
._- HC --- - Base NOV-
A
18.23
50.23
16.37
11.35
10.88
17.58
19.11
17. 16
21.32
21.16
30.10
21.31
21 .91
27.18
26.82
22.02
21.75
22.02
21.92
23.11
21.66
(49?)
(81?)
(43?)
(38?)
(15?)
(47?)
(52»
(50?)
(46?)
(52?)
(38?)
(22?)
(14?)
(39?)
(39?)
(33?)
(32?)
(33?)
(35?)
(39?)
(34?)
2.89 (56?)
8.38 (85?)
3.02 (58?)
2.83 (47?)
1.72 (26?)
3.26 (61?)
3.60 (65?)
1.61 (65?)
2.13 (66?)
2.23 (57?)
2.87 (52?)
2.29 (40?)
2.20 (38?)
5.03 (39?)
4.97 (39?)
4.08 (33?)
4.03 (32?)
4.08 (33?)
4.06 (35?)
4.34 (39?)
4.01 (34?)
Aloft
HC
(ppb) (?)
40 ( 7?)
40 ( 2?)
150 (35?)
225 (38?)
50 (21?)
100 (15?)
40 ( 6?)
40 (15?)
40 (17?)
40 (19?)
40 (10?)
25 ( 8?)
10 (11?)
35 ( 2?)
35 ( 2?)
35 ( 3?)
35 ( 3?)
35 ( 3?)
35 ( 1?)
35 ( 1?)
35 ( 1?)
[h]
°3
105
32
10
10
60
65
65
25
50
62
40
10
10
0
0
0
0
0
0
0
0
NOX
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
(continued)
-------
Table 1. (continued) - 3
ID
[a]
G-TL2
G-TL3
G-TL4
G-TL5
G-TL6
G-TL7
G-TL8
W-TL8
G-TL9
G-TLA
G-WA1
G-WA2
G-WA3
G-WA4
W-HA4
G-WA5
City, Site, and Date
Tulsa
Site 137, 7/16/81
Site 127, 7/16/81
Site 137, 8/6/81
Site 137, 6/29/82
Site 137, 8/6/82
Site 127, 8/23/82
Site 137, 7/26/83
(Same as G-TL8)
Site 127, 8/27/83
Site 137, 8/28/83
Washington
Takoma, 7/16/80
Takoma, 7/17/80
Takoma, 7/21/80
Takoma, 8/7/80
(Same as G-WA4)
Takoma, 8/29/80
Base.
Csmax
(ppb)
112
118
119
119
117
89
113
117
106
104
117
111
138
164
167
108
ROG/NOX lb]-
Base MaxRct MaxCs
[c] [d] (ef
30.2
27.7
29.6
29.7
29.5
30.9
"30.5
12.0
"29.3
31.6
7.9
"12.5
7.9
7.2
6.9
"6.0
6.0
"6.0
6.0
6.0
6.0
"6.0
5.0
6.0
6.0
«6.0
"5.0
5.0
5.0
6.0
"6.0
"6.0
8.0
8.0
8.0
"8.0
8.0
"8.0
8.0
8.0
8.0
8.0
"7.0
7.0
8.0
"10.0
8.0
10.0
Light
[f]
211
211
202
216
202
187
206
193
181
184
191
190
189
198
198
163
Inv. Height
(m)
Init. Final
250
250
250
250
250
250
250
250
250
250
160
160
160
420
420
150
2189
2189
1980
1913
1980
3187
2359
2359
3026
2752
2128
2128
2128
1658
1658
2229
- Input flux (mmol/m ),
(% input emitted) [g] -
--- HC --- - Base NOX-
18.08 (11%)
19.11 (16%)
17.68 ( 9%)
17.98 (105)
17.81 ( 9%)
17.26 ( 7?)
17.46 ( 8%)
17.46 ( &%)
18.28 (12%)
17.75 ( 9?)
6.81 (28?)
6.68 (26?)
10.77 (50%)
16.22 (43?)
16.22 (43?)
16.61 (57%)
0.70 (10?)
0.80 (21?)
0.69 ( 9?)
0.70 ( 9?)
0.70 ( 9?)
0.71 (11?)
0.68 ( 8?)
1.65 ( 8?)
0.78 (19?)
0.69 ( 8?)
1.26 (40?)
0.79 (37?)
1.77 (71?)
2.52 (57?)
2.52 (57?)
3.34 (73?)
Aloft
-- HC
(ppb) (?)
40 (15?)
40 (14?)
40 (14?)
40 (13?)
40 (14?)
40 (22?)
40 (16?)
40 (16?)
40 (20?)
40 (19?)
40 (32?)
40 (32?)
40 (23?)
40 (11?)
25 ( 7?)
40 (17?)
[h]
°3 N0x
- (ppb) -
46
46
51
62
51
47
55
55
74
72
65
65
65
56
60
65
0
0
0
0
0
0
0
1
0
0
0
0
0
0
0
0
Chicago
W-CH1 Kenosha, 7/11/79 178
St. Louis
W-SL1 St. Louis, 6/21/86 165
Los Angeles
W-LA1 Glendora, 8/24/85 367
"7.4 "6.0 "8.0 233 250 880 7 43 (49?) 1.35 (58?)
»8.3 "4.0 "8.0 215 250 1700 13.16 (23?) 1.80 (44?)
"8.8 "6.0 "8.0 234 250 700 21.87 (30?) 2.70 (54?)
100 (26?) 60 0
30 (12?) 50 0
100 ( 8?) 100 0
(continued)
-------
Table 1. (continued) - 14
ID City, Site, and Date
[a]
Base.
(Kmax
(ppb)
ROG/NOX [b]---
Base MaxRct MaxOo
[c] [d] [eT
Light
[f]
Inv. Height
(m)
Init. Final
o
- Input flux (mmol/m ),
(% input emitted) [g] -
_._ HC --- - Base NO -
Aloft
-- HC
(ppb) (?)
[h]
°3 N0x
- tppb) -
San Francisco
W-SF1 San Francisco, 9/30/80 206
W-SF2 San Francisco, 7/12/81 210
"12.8 "10.0 "16.0 175 89 100 9.17 (71?) 0.91 (21?) 250 (26?) 61 1
"10.1 "8.0*10.0 265 89 100 6.22(71?) 0.78(21?) 170(26?) 61 1
"Averaged Conditions" Scenarios (see text)
A-MxR
A-MxO
(Maximum Reactivity)
(Maximum Ozone)
113
176
5.1
8.0
5.1
5.1
8.0
8.0
235
235
151
151
1500
1500
16.00 (33?)
16.00 (33?)
2.96 (11?)
2.00 (11?)
75
75
(17?)
(17?)
50
50
0
0
[cj
Id]
[e]
If]
[g]
[h]
(1988). See these references
Notes:
[a] Scenarios with prefix "G- from Gery et al. (1987); those with "H-" are from Whitten et al.
for more complete description of input data for these scenarios.
(b] ROG/NOX = Ratio of final hydrocarbon and NO concentration at the end of the simulation if no chemical reaction occurs.
Includes hydrocarbons and (where applicable) NO entrained from aloft.
ROG/NOX for base case scenarios. Scenarios with an "*" in this column were used to derive the base case relative
reactivity scales.
ROG/NO,, for the maximum reactivity scenarios. Scenarios with an "*" in this column were used to derive the maximum
for the maximum reactivity scenarios.
reactivity (MaxRct) scale.
ROG/NO for the maximum ozone scenarios.
(MaxOn) reactivity scale.
"Light" is integrated N02 photolysis rate for the simulation.
Total initial and emitte
Scenarios with an '""' in this column were used to derive the maximum ozone
(It is a unitless quantity.)
hydrocarbon and NOX input (excluding species entrained from aloft) for the base case scenarios.
Units are millimoles NO or millimoles carbon HC input into the cell per square meter. The value in parentheses is the
percentage of the molar input which is present initially. Note that the total NOX inputs for the maximum reactivity and
the maximum ozone scenarios may be different from these values, though the fraction present initially would be the same.
Hydrocarbon (in ppbC) and On and NOX (in ppb) present in the aloft air mass which is entrained into the ground-level
(simulated) air mass when the inversion height raises. The values in parentheses are the percentages of the total
hydrocarbon present at the end of the simulation (assuming no chemical reaction) which are due to HC species entrained
from aloft. The fraction of NOX input due to NOX aloft in the base case simulations ranges from 0? to 5?.
-------
Perhaps more significantly, the city-specific EKMA scenarios
used in this study were developed by using various versions of the Carbon
Bond gas-phase chemical mechanism (see Gery et al. 1988 and references
therein), which is different in some respects from the mechanism employed
in this study (Carter, 1990 see Section II.B, below). However, despite
the differences in the chemical mechanisms and the other modifications of
the input files required to use different software systems to simulate the
scenarios, the simulations of maximum ozone in the scenarios using the
mechanism and software employed in this study were found not to be
significantly different (agreeing to within +10?) from those calculated by
using the Carbon Bond IV mechanism (Gery et al. 1988) and the OZIPP
software and input files. They also were not significantly different from
those reported by Gery et al. (198?) using an earlier version of the
Carbon Bond mechanism.
2. Representation of Base Case ROG Compositions
The composition assumed for the base case ROG emissions has been
shown to affect predictions of reactivity (Carter and Atkinson, 1989; Weir
et al. 1988), and thus it is important that this be represented as
accurately as possible. However, for most urban areas the composition of
the ROG emissions is highly uncertain, and no attempt was made to
represent the city-specific ROG compositions in the development of the
EKMA scenarios discussed above. Instead, in the manner of Gery et al.
(1987) and Whitten (1988), a standard default representation of the ROG
composition, which is based on air quality data in a wide variety of areas
(e.g., EPA 1984, Killus and Whitten, 198*; Lurmann et a.. 1987; Jeffries
et al. 1989), was developed and used for all scenarios. A recent
comprehensive analysis of detailed YCC speciation data was carried out by
Jeffries et ai. (1989), who analyzed and summarized results of 860 ground-
level and 160 aloft measurements mace in various urban areas throughout
the United States. (Of these 773 and 53 samples, respectively, were
considered to be useable for analysis.') With these data, they developed
an "all-city average" FiOG composition profile to represent typical ground-
level ROG emissions, and a separate "aloft" ROG profile to represent
compositions of such species aloft. These compositions are considered to
be appropriate for use in airshed modeling of areas where reliable HOG
speciation data are not available, and were thus used in all scenarios
15
-------
employed in this study. Note that the all-city average mixture is
referred to as the "base case ROG" throughout the this report.
The all-city average (base case ROG) and aloft ROG compositions
reported by Jeffries et al. (1989) were given in terms of the lumped
species used in the Carbon Bond IV (Gery et al., 1988) and the Lurmann et
al. (1987) condensed mecha;. .^ms. These are somewhat different than those
used in the mechanism employed in this study. The mechanism employed in
this study uses variable lumped species whose mechanistic parameters are
adjusted for the compound or mixture they represent (Carter, 1990). This
requires knowledge of the detailed compositions of the mixture being
represented. For use with this mechanism, the detailed compositions
associated with all-city average and the aloft mixtures derived by
Jeffries et al. (1989) were obtained from Jeffries (1988). The detailed
compositions of these two surrogates are given in Table 2.
It should be noted that the ground-level measurements analyzed
by Jeffries et al. (1989) did not include aldehyde measurements. In
deriving the all-city average mixture, Jeffries et al. (1989) assumed 5%
aldehyde composition assigned to the all-city average mixture is based on
previous EPA-recommended EKMA defaults (e.g, EPA 1984), which were derived
based on highly scattered and uncertain data (see, for example, Lurmann et
al. 1987). However, more recent analysis of Los Angeles South Coast Air
Quality Study (SCAQS) data indicates that this total aldehyde fraction may
not be inappropriate, though the relative contribution of the higher
molecular weight aldehydes appears to be larger than assumed by Jeffries
et al. (1989) (Lurmann, private communication, 1991). The composition of
the aldehyde fraction is probably the greatest uncertainty in this
mixture.
Although most reactivity calculations in this work used the all-
city average mixture given in Table 2, a limited number of calculations
were carried out to assess the effects of uncertainties or variations in
the composition of the base case ROG mixture on reactivities of selected
individual VOCs. The composition of the all-city average was used as the
starting point for all the variations examined. These variations, and
their effects, are discussed later in this report.
16
-------
Table 2. Composition of the Mixtures Used to Represent Base Case
ROG Emissions and Aloft ROG Pollutants in the Scenarios
VOC ID
ETHANE
PROPANE
N-C4
N-C5
N-C6
N-C7
N-C8
N-C9
N-C10
ISO-CM
ISO-C5
2-ME-C5
3-ME-C5
BR-C6
23-DMB
22-DMB
BR-C7
BR-C8
BR-C9
BR-C10
BR-C11
BR-C12
CYCC5
CYC-C6
CYCC6
CYC-C7
CYC-C8
ETHENE
PROPENE
1 -BUTENE
CM-OLE 1
1-PENTEN
C5-OLE1
1-HEXENE
C6-OLE1
C7-OLE1
C8-OLE1
C9-OLE1
C10-OLE1
C11-OLE1
2M-1-BUT
C-2-BUTE
Description
Ethane
Propane
n-Butane
n-Pentane
n-Hexane
n-Heptane
n-Octane
n-Nonane
n-Decane
Isobutane
Iso-Pentane
2-Methyl Pentane
3-Methylpentane
Branched C6 Alkanes
2,3-Dimethyl Butane
2,2-Dimethyl Butane
Branched C7 Alkanes
Branched C8 Alkanes
Branched C9 Alkanes
Branched C10 Alkanes
Branched C11 alkanes
Branched C12 Alkanes
Cyclopentane
C6 Cycloalkanes
Cyclohexane
C7 Cycloalkanes
C8 Cycloalkanes
Total Alkanes (ppbC)
Ethene
Propene
1-Butene
CM Terminal Alkanes
1-Pentene
C5 Terminal Alkanes
1-Hexene
C6 Terminal Alkanes
C7 Terminal Alkanes
C8 Terminal Alkanes
C9 Terminal Alkanes
C10 Terminal Alkanes
C11 Terminal Alkanes
2-Methyl- 1-Butene
cis-2-Butene
(ppb VOC/ppmC
Emissions
20.90
15.59
16.53
6.87
2.97
1.13
0.60
O.M6
0.64
6.97
1M.22
3.85
2.76
0.90
1.02
0.39
5.22
4.94
1.95
1.MO
0.56
0.13
0.6M
1.68
0.85
0.8M
0.10
522.91
19-29
M.96
2.51
1.74
0.55
1.09
0.21
0.7M
1.16
O.MO
0.72
0.03
O.M1
0.83
0.80
mixture)
Aloft
115.31
50.97
25.03
7.59
2.16
0.38
13.43
12.31
2.60
1.41
0.61
2.74
0.96
2.43
0.91
0.48
729.94
15.55
2.70
17
-------
Table 2 (continued) - 2
VOC ID
T-2-BUTE
2M-2-BUT
C5-OLE2
C6-OLE2
C7-OLE2
C8-OLE2
13-BUTDE
ISOPRENE
A-PINENE
BENZENE
TOLUENE
C2-BENZ
N-C3-BEN
I-C3-BEN
C9-BEN1
C10-BEN1
C11-BEN1
0-XYLENE
P-XYLENE
M-XYLENE
C9-BEN2
C10-BEN2
C11-BEN2
C12-BEN2
123-TMB
C9-BEN3
C10-BEN3
C11-BEN3
C12-BEN3
C10-BEN4
ACETYLEN
FORMALD
ACETALD
INERT
Description
trans-2-Butene
2-Methyl-2-Butene
C5 Terminal Alkenes
C6 Terminal Alkenes
C7 Terminal Alkenes
C8 Terminal Alkenes
1 , 3-Butadiene
Isoprene
a-Pinene
Total Alkenes (ppbC)
Benzene
Toluene
Ethyl Benzene
n-Propyl Benzene
Isopropyl Benzene
C9 Monosubstituted Benzenes
C10 Monosubstituted Benzenes
C11 Monosubstituted Benzenes
o-Xylene
p-Xylene
m-Xylene
C9 Disubstituted Benzenes
C10 Disubstituted Benzenes
C11 Disubstituted Benzenes
C12 Disubstituted Benzenes
1,2,3-Trimethyl Benzene
C9 Trisubstituted Benzenes
C10 Trisubstituted Benzenes
C11 Trisubstituted Benzenes
C12 Trisubstituted Benzenes
C10 Tetrasubstituted Benzenes
Total Aromatics (ppbC)
Acetylene
Formaldehyde
Acetaldehyde
Total Aldehydes (ppbC)
Unreactive Organics
(ppb VOC/ppmC
Emissions
1.03
0.11
2.01
0.94
0.10
0.02
0.67
0.68
0.61
147.72
3-46
8.27
1.24
0.29
0.14
0.33
1.28
0.08
1.55
1.95
1.95
2.28
1.64
0.74
0.19
2.91
0.10
0.91
0.74
0.19
0.18
248.34
11.50
20.00
15.00
50.00
8.02
mixture)
Aloft
3.63
57.37
4.34
5.20
0.94
2.17
0.77
0.77
99.65
9.34
74.86
10.75
96.37
18
-------
3. Types of Scenarios Used
Although the reactivity scales developed in this work are all
based on models for the representative pollution episodes discussed above,
in some cases these were modified for the purpose of developing special
types of reactivity scales, or for the purpose of sensitivity studies.
The terms "base case", "maximum reactivity", "maximum ozone", and
"averaged conditions" are used to refer to the different types of
scenarios employed in this work. These are discussed below.
a. Base Case Scenarios
The term "base case scenario" refers to the scenario as
obtained from Gery et al. (198?) or Whitten (1988), which is not modified
in any way except as required for compatibility with our software and
chemical mechanism. The conditions of these scenarios are as given in
Table 1 . Although as indicated above it is highly uncertain whether these
are accurate representations of any real air pollution episodes, for the
purpose of this work it is assumed that they represent a realistic
distribution of ozone pollution episodes. Reactivities (both relative and
absolute) of individual VOCs varied considerably among these base case
scenarios. To avoid giving undue weight to those urban areas for which
Gery et al. (198?) have many episodes, representative subsets of the
scenarios listed in Table 1 were used to derive the multi-scenario
reactivity scales. With one exception no more than two episodes were used
from each urban area. For urban areas with more than two episodes, the
episodes chosen were those with the smallest and the largest change of
ozone with respect to changing NOX emissions. The one exception is Tulsa,
where the W-TL8 episode used by Whitten (1988) had a significantly lower
ROG/NO ratio than the representation of the same episode (G-TL8) used by
A
Gery et al. (1987). Became of this difference, the W-TL8 was treated in
effect as if it represented a different urban area, and thus three "Tulsa"
episodes were used. On the other hand, only one "New York" episode was
used, because the two were very similar. A total of 21 episodes were
used, with the specific episodes being indicated in Table 1 (by the
asterisks preceding the ROG/NOX ratios). The incremental reactivities
calculated for these 21 representative scenarios were used to derive the
various base case multi-scenario relative reactivity scales discussed in
Section II.D.2.
19
-------
b. Maximum Reactivity Scenarios and Maximum Ozone Scenarios
The main reason for the variability of the reactivities in
the base case scenarios is the variability of the relative availability of
NOX (e.g., the ROG/NOX ratio) in the scenarios. An alternative approach
to deriving reactivities for scenarios with highly disparate ROG/NOX
ratios is to consider NOX availability separately from all other airshed
characteristics which affect reactivity. This can be done by deriving
separate reactivity scales, where each represents a specified and
consistent condition of NOV availability, but where each scale represents
X
(in effect) averages of other airshed conditions. For this purpose, two
sets of modified scenarios were developed, each with the NOX inputs
adjusted to yield specified sets of reactivity characteristics.
The "Maximum reactivity" scenarios consist of the base case
scenarios with the initial and emitted NO input adjusted to achieve
maximum incremental reactivities. The ratios of initial to emitted NOV,
A
the levels of NOV aloft (in the few scenarios where this is applicable),
A
the total ROG emissions input, and all other aspects of these scenarios
were not varied. For each scenario, the initial+emitted NO input which
corresponds to maximum reactivity was determined by calculating the
reactivity of the base case ROG mixture as a function of NOX input, and
the NOX input which yielded the highest reactivity of this mixture was
used to represent maximum VOC reactivity conditions. Incremental
reactivities for toluene, formaldehyde, propene and CO were also
calculated as a function of NO input, and the conditions yielding highest
reactivities for these compounds were similar, but not always exactly the
same, to those yielding the highest base case ROG reactivity. However,
for consistency of conditions for the various VOCs, the base ROG
reactivity was used as the criterion to define maximum reactivity
conditions. (The difference in the maximum reactivity of these VOCs and
their reactivities under maximum base ROG reactivity conditions was minor
VOC reactivities are not highly sensitive to NOX inputs when NO is
near maximum reactivity conditions. Indeed, one characteristic of maximum
reactivity conditions is that the derivative of incremental reactivities
with respect to NO inputs is zero.)
20
-------
In most cases, the MOV inputs giving maximum base ROG
A
reactivity were higher than those in the base case scenarios. The
exceptions were several of the "Massachusetts" and "Philadelphia"
scenarios, and all of the "Seattle" scenarios.
The "Maximum ozone" scenarios consist of the base case
scenarios with the initial and emitted NOV input adjusted to achieve
X
maximum peak ozone concentration in the calculations. As with the maximum
reactivity scenarios, the other scenario conditions were not varied. The
NOX inputs yielding maximum ozone tend to be near the mid-range of the
distribution of NOV inputs in the base case scenarios listed on Table 1.
A
The ROG/NCL ratios in the maximum reactivity and the maximum
A
ozone scenarios are given in Table 1, where they can be compared to those
of the base case scenarios. Note that the ROG/NO ratios yielding maximum
A
peak ozone levels are always higher than those yielding maximum base ROG
incremental reactivity. While the ratios for the maximum reactivity or
maximum ozone scenarios are much less variable than those for the base
case scenarios, it is clearly not correct to say that a single ROG/NO
ratio always corresponds to a given type of NOV availability. As shown in
A
Table 1, the ROG/NOV ratios yielding maximum incremental reactivities and
A
maximum ozone formation can vary significantly from scenario to scenario,
depending on other conditions (see also Carter and Atkinson, 1989). As an
example of why this is the case, consider the influence of light
intensity. Since the rate at which NOV is consumed would increase with
A
light intensity, an ROG/NOV ratio which is in the NO -limited regime at a
A A
high sunlight intensity might not be in that regime if the sunlight
intensity were reduced. Therefore, the ROG/NO ratio itself is not
considered adequate for quantifying NOX availability when a variety of
scenarios are being considered. Similar considerations apply for other
scenario conditions which affect the rate at which the overall
photooxidation processes occur.
As indicated in the footnotes to Table 1, only selected
subsets of the maximum reactivity and maximum ozone scenarios were used to
derive the respective reactivity scales. This was done primarily to
reduce the amount of computation required to derive these scales, and
because once NOV inputs were adjusted to achieve maximum reactivity or
X
maximum ozone conditions, several urban areas had scenarios which were
21
-------
quite similar. At least two scenarios were chosen for each urban area
which had more than one scenario, primarily to represent the range of
incremental reactivities of the base case ROG surrogate. The set used for
the maximum reactivity was different from that used for the maximum ozone
scale, and both were different from those used for base case scales. This
was to maximize the variety of scenario conditions which were used to
develop the scales. The specific scenarios used are indicated by the
asterisks preceding the ROG/NCL ratios on Table 1.
A
c. Averaged Conditions Scenarios
A limited number of sensitivity calculations were carried
out to assess effects of variations of selected scenario input parameters
on calculations of incremental reactivities. For this purpose, a series
of "averaged conditions" scenarios were developed based on the sets of
scenarios discussed above. The conditions of these scenarios (for two
different NO input levels) are included on Table 1. The light intensity,
maximum inversion height, VOC emission amounts, aloft pollutant levels,
etc., were approximate averages of those of the specific city-days used in
this study, for which each of the 12 urban areas was weighed equally. The
mixture used to represent the base case ROG emissions was the same as used
for the other scenarios in this study, except for those calculations where
this was explicitly varied. The NOX inputs were varied according to the
type of sensitivity calculations which were carried out. However, the
ratio of initial to emitted NOX was held constant at the level indicated
in Table 1, even in the calculations where NOX was varied. These averaged
conditions scenarios were used for sensitivity analyses only; they were
not used in the development of the various reactivity scales discussed in
this work.
Except for the calculations where NOX inputs were
systematically varied, most sensitivity calculations used the "averaged
conditions, maximum reactivity" (A-MxR) and the "averaged conditions,
maximum ozone" (A-MxO) scenarios. In the A-MxR scenario the NO input was
adjusted to yield maximum reactivity of the base case ROG mixture, and in
the A-MxO scenario the NOX input was adjusted to yield maximum peak ozone
concentrations. Sensitivities calculated for these two scenarios thus
approximate sensitivities for maximum reactivity and for maximum ozone
conditions, respectively. It was also found that incremental reactivities
22
-------
of the VOCs for the A-MxR and A-MxO scenarios corresponded quite well (to
within ~\Q% for most VOCs) with those derived from the full set of maximum
reactivity or maximum ozone scenarios.
d. Multi-Day Scenarios
We did not include any multi-day, rural, or regional-scale
ozone scenarios in this study. The EKMA scenarios used in this study do
take multi-day effects into account in the sense that they include some
levels of background and aloft ozone, VOCs, and (in some cases) NOX in the
simulations. These represent pollutants left over from previous days or
transported from downwind. However, they do not take into account effects
of changing VOC emissions on ozone formation on days subsequent to the
emissions of the VOCs. This is not because these are considered to be
unimportant: the most severe ozone episodes are multi-day in nature, and
rural and regional-scale ozone is a serious problem in many parts of the
United States and the world. Indeed, a two-day ozone episode was included
in our previous incremental reactivity modeling study (Carter and
Atkinson, 1989) and in our initial development of the proposed maximum
incremental reactivity scale (Carter, 1989b; Lowi and Carter, 1990).
However, it was decided not to include multi-day ozone episodes in this
study both for both practical and more fundamental reasons.
The practical reasons involved both considerations of
computer constraints and the effort required to develop a set of multi-day
episodes representing a variety of conditions. Because many VOCs and
scenarios need to be examined when investigating development of reactivity
scales to be applied to a variety of VOCs and conditions, many computer
calculations are required. Because of computer constraints, this means
limiting ourselves to trajectory models where only one reactive cell is
simulated. It is uncertain whether multi-day effects can be appropriately
simulated with one-cell models, and thus reactivities calculated for them
may not be represent realistic conditions. The two-day scenario used in
our previous studies used a two-cell model where the aloft layer was
simulated separately [Carter and Atkinson, 1989] and took 3-5 times more
computer time than the EKMA scenarios.) Furthermore, unlike the case with
simple EKMA models, there is no available set of multi-day scenarios
representing a variety of urban areas. Establishing such a set would be a
major effort, which was beyond the scope of the present study.
23
-------
More fundamentally, it can be argued that although multi-day
or regional-scale ozone episodes are of major concern, they are probably
not particularly relevant to the assessment of effects of VOC controls.
Most modeling results indicate that ozone formation in multi-day episodes
or on the regional scale is far less sensitive to VOC changes than it is
to changes in NOV emissions, which is consistent with the fact that NO,, is
A A.
removed from the atmosphere more rapidly than VOCs and their reactive
products are. This is also consistent with results of multi-day
environmental chamber experiments carried out at our laboratory (Carter et
al. , 1984, 1986b), where ozone levels on second and subsequent days are
much less sensitive to changes in total ROG inputs (Carter et al., 1984)
or effects of methanol substitution (Carter et al., 1986b) than they are
on the first day. This means that the main benefits of VOC substitution
strategies and thus the main reason that VOC reactivity scales should
be of interest to those concerned with ozone control strategies will
concern ozone formation on the same day that the VOC is emitted.
Obviously, if major VOC substitution strategies are being considered for
implementation, their effects on multi-day and regional episode should be
included in the assessment. However, since the effects on the first day
are so much greater, use of only single-day scenarios may be sufficient
when multiple options are assessed for screening purposes.
B. Chemical Mechanism Used for Representative VOCs
1. Description of Mechanism
The chemical mechanism used to represent the reactions of the
base case ROG and NO emissions, and to calculate the incremental
reactivities of the representative VOCs, was developed by the author and
is documented in detail elsewhere (Carter 1990). (Updates of the
mechanisms for several important VOCs based on more recent data are
documented in Appendix A.) This mechanism contains rate constant and
product yield assignments for over 140 separate "detailed model species",
making it the most detailed of all current state-of-the-science mechanisms
in terms of the number of organic species which can be separately
represented. Because of the number of different VOC species it can
represent, it is particularly suitable for use in reactivity assessment
calculations. The individual types of VOCs which can be represented in
24
-------
this mechanism, and for which reactivities were calculated in this work,
are listed in Table 3. The table also gives codes indicating the level of
uncertainty of the mechanism used for each type of VOC, and the extent to
which the mechanism for each VOC was tested against environmental chamber
data.
The mechanism was developed by using the results of recent
evaluations of relevant laboratory data (primarily Atkinson and Carter,
1984; NASA, 1987; Atkinson, 1989, 1990; and Atkinson et al., 1989) and
results of environmental chamber experiments carried out at our
laboratories and the University of North Carolina (Carter et al. 1986a;
Lurmann et al. , 1987; Carter, 1988; Carter and Lurmann, 1991). The
evaluated kinetic and mechanistic data were used wherever possible, but
for most VOCs insufficient data are available and estimates had to be
made. Environmental chamber data were used to test estimated mechanisms
for the approximately 20 VOCs for which such data are available, and in
some cases (particularly for the aromatic hydrocarbons) uncertain portions
of the mechanism had to be adjusted to attain acceptable fits of model
predictions to the experiments. These approximately 20 VOCs include
representatives of most of the major classes of organic compounds which
are emitted. The mechanisms for the other (approximately 120) VOCs were
estimated by referring to those for chemically similar compounds for which
data were available.
A total of over 500 environmental chamber experiments were used
to test the mechanisms of the -20 representative VOCs, both singly and in
mixtures (Carter et al. 1986a; Lurmann et al, 1987; Carter, 1988; Carter
and Lurmann, 1991)- The observed ozone yields could be simulated to
within ±30$ for over 60% of the runs modeled, and to within ±50$ for over
80$ of the runs. This is as good as presently can be expected with
current mechanisms, given uncertainties in characterizations of chamber
effects, combined with uncertainties in the mechanisms themselves (Carter
et al. 1986a; Gery et al. 1988; Carter and Lurmann, 1990, 1991).
However, it should be recognized that for many VOCs the
atmospheric reaction mechanisms are highly uncertain. The available
environmental chamber data are sufficient for testing mechanisms for only
a very small subset of the VOCs listed on Table 3, and for most compounds
the only aspect of the mechanism for which there are data is the rate
25
-------
Table 3. Summary of VOC Species and Ozone Reactivity Estimates for the Maximum Reactivity ("MaxRct") and the Maximum Ozone
Reactivity ("MaxOg") Scales.
Model
Code
la]
VOC Description
kOH
[b]
(ppm
-1
Kinetic React. Mech. React. Incremental Reactivity
(fract. react) (mol Og/mol C) (mol Oj/mol C) (gm Oj/gm VOC)
min~') MaxRct MaxO,
MaxRct MaxO-,
MaxRct MaxOo
MaxRct
--Codes--
Rep. Unc.
[c] [d]
CO
ro
Carbon Monoxide
Alkanes
3.5E+2 0.028 0.043
0.84
0.45
0.024 0.019
0.040 0.033
METHANE
ETHANE
PROPANE
N-C4
N-C5
N-C6
N-C7
N-C8
N-C9
N-C10
N-C11
N-C12
N-C13
N-C14
N-C15
ISO-C4
C4C5
BR-C5
ISO-C5
NEO-C5
2-ME-C5
3-ME-C5
BR-C6
23-DMB
Methane
Ethane
Propane
n-Butane
n-Pentane
n-Hexane
n-Heptane
n-Octane
n-Nonane
n-Decane
n-Undecane
n-Dodecane
n-Tridecane
n-Tetradecane
n-Pentadecane
Isobutane
Lumped C4-C5 Alkanes
Branched C5 Alkanes
Isopentane
Neopentane
2-Methyl Pentane
3-Methylpentane
Branched C6 Alkanes
2,3-Dimethyl Butane
1.3E+1
4.0E+2
1.8E+3
3.7E+3
5.8E+3
7.9E+3
9.9E+3
1 .2E+4
1.4E+4
1.6E+4
1.8E+4
2.0E+4
2.2E+4
2.4E+4
2.6E+4
3.5E+3
4.7E+3
5.9E+3
5.9E+3
1.1E+3
7.9E+3
8.5E+3
7-9E+3
8.0E+3
0.0010
0.032
0.134
0.26
0.37
0.46
0.54
0.60
0.66
0.70
0.74
0.78
0.80
0.83
0.85
0.24
0.31
0.37
0.37
0.086
0.46
0.48
0.46
0.47
0.0016
0.049
0.20
0.37
0.51
0.61
0.69
0.75
0.80
0.83
0.86
0.88
0.90
0.92
0.93
0.35
0.44
0.51
0.51
0.130
0.61
0.64
0.61
0.62
3.4
1.42
0.74
0.73
0.52
0.40
0.28
0.19
0. 148
0. 119
0. 101
0.088
0.078
0.070
0.064
1.03
0.66
0.70
0.70
0.65
0.58
0.58
0.58
0.47
1.6
0.61
0.34
0.34
0.27
0.21
0.15
0.107
0.084
0.069
0.059
0.051
0.046
0.042
0.038
0.48
0.32
0.34
0.34
0.23
0.26
0.29
0.26
0.26
0.0036
0.045
0. 100
0. 19
0.19
0.18
0.149
0. 116
0.097
0.084
0.075
0.068
0.063
0.058
0.054
0.25
0.21
0.26
0.26
0.055
0.27
0.28
0.27
0.22
0.0025
0.030
0.069
0.124
0.135
0.131
0.105
0.081
0.067
0.057
0.050
0.045
0.041
0.038
0.036
0.17
0. 140
0.17
0.17
0.029
0.16
0.19
0.16
0.16
0.0108
0.145
0.33
0.62
0.64
0.61
0.50
0.39
0.33
0.28
0.25
0.23
0.21
0.20
0.18
0.83
0.76
0.86
0.86
0.18
0.90
0.9^
0.90
0.73
0.0075
0.096
0.23
0.41
0.45
0.44
0.35
0.27
0.23
0.19
0.17
0.15
0. 140
o. 130
0.121
0.56
0.51
0.57
0.57
0.097
0.54
0.62
0.54
0.53
3
2
3
2
3
3
3
2
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
2
2
2
1
5
5
5
5
5
7
8
8
8
8
8
7
7
7
7
7
7
7
7
5
(continued)
-------
Table 3. (continued) - 2
rv>
Model
Code
[a]
22-DMB
C6PLUS
24-DM-C5
3-ME-C6
4-ME-C6
BR-C7
23-DM-C5
ISO-C8
4-ME-C7
BR-C8
BR-C9
4-ET-C7
BR-C10
4-PR-C7
BR-C11
BR-C12
BR-C13
BR-C14
BR-C15
CYCC5
ME-CYCC5
CYC-C6
CYCC6
CYC-C7
ME-CYCC6
ET-CYCC6
CYC-C8
CYC-C9
VOC Description
2,2-Dimethyl Butane
Lumped C6+ Alkanes
2,4-Dlmethyl Pentane
3-Methyl Hexane
4-Methyl Hexane
Branched C7 Alkanes
2,3-Dimethyl Pentane
Isoctane
4-Methyl Heptane
Branched C8 Alkanes
Branched C9 Alkanes
4-Ethyl Heptane
Branched C10 Alkanes
34-Propyl Heptane
Branched C11 alkanes
Branched C12 Alkanes
Branched C13 Alkanes
Branched C14 Alkanes
Branched C15 Alkanes
Cyclopentane
Methylcyclopentane
C6 Cycloalkanes
Cyclohexane
C7 Cycloalkanes
Methyl cyclohexane
Ethyl Cyclohexane
C8 Cycloalkanes
C9 Cycloalkanes
kOH
(pprl-1
min"1)
2.7E+3
9.0E+3
1.0E+4
1.0E+4
1 .OE+4
1 .OE+4
1 . 1E+4-"
6.9E+3
1.3E+4
1 .3E+4
1.5E+4
1.5E+4
1.7E+4
1 . 7E+4
2. 1E+4
2.3E+4
2.5E+4
2.7E+4
2.9E+4
8.2E+3
1.0E+4
1.2E+4
1 .2E+4
1 .5E+4
1 .5E+4
1 .8E+4
1 .8E+4
2.0E+4
Kinetic React.
(Tract, react)
MaxRct
0.19
0.51
0.54
0.56
0.56
0.56
0.56
0.42
0.62
0.62
0.69
0.69
0.73
0.73
0.79
0.81
0.84
0.86
0.87
0.48
0.55
0.61
0.61
0.68
0.68
0.74
0.74
0.78
MaxO,
0.29
0.66
0.69
0.71
0.71
0.71
0.71
0.56
0.76
0.76
0.82
0.82
0.85
0.85
0.89
0.91
0.92
0.93
0.94
0.63
0.71
0.76
0.76
0.81
0.81
0.86
0.86
0.89
Mech. React.
(mol Oo/mol C)
MaxRct
0.62
0.40
0.58
0.46
0.46
0.46
0.51
0.49
0.35
0.35
0.30
0.30
0.25
0.25
0.29
0.29
0.22
0.17
0. 16
0.96
0.89
0.40
0.40
0.51
0.51
0.54
0.54
0.60
MaxOo
0.27
0.20
0.25
0.22
0.22
0.22
0.26
0.21
0. 17
0. 17
0. 145
0.145
0.120
0. 120
0. 16
0.15
0. 120
0.099
0.089
0.45
0.43
0.19
0. 19
0.24
0.24
0.25
0.25
0.28
1 n.
(mol 0?,
MaxRct
0. 121
0.20
0.32
0.25
0.25
0.25
0.29
0.20
0.22
0.22
0.21
0.21
0.18
0. 18
0.23
0.23
0.19
0.149
0.137
0.46
0.49
0.25
0.25
0.34
0.34
0.40
0.40
0.47
;remental
/mol C)
MaxOo
0.077
0.133
0.18
0.16
0. 16
0. 16
0. 18
0. 116
0.129
0.129
0.119
0.119
0. 102
0. 102
0.139
0.138
0.111
0.092
0.084
0.28
0.30
0.143
0.143
0. 19
0. 19
0.22
0.22
0.25
Reactivi
(gm 03/
MaxRct
0.40
0.70
1 .06
0.85
0.85
0.85
0.96
0.69
0.73
0.73
0.70
0.70
0.62
0.62
0.77
0.79
0.63
0.50
0.46
1.6
1.7
0.84
0.84
1.18
1.18
1.36
1.36
1.6
ty
gm VOC)
Max03
0.26
0.45
0.59
0.52
0.52
0.52
0.62
0.39
0.43
0.43
0.40
0.40
0.35
0.35
0.47
0.47
0.37
0.31
0.29
0.96
1.03
0.49
0.49
0.66
0.66
0.74
0.74
0.86
Cod
Rep.
[c]
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
les
Unc.
[d]
7
7
7
7
7
7
7
7
7
7
7
7
8
8
8
8
8
8
8
7
7
7
7
7
5
7
8
8
(continued)
-------
Table 3. (continued) - 3
ro
CD
Model
Code
[a]
CYC-C10
CYC-C11
CYC-C12
CYC-C13
CYC-C14
CYC-C15
ETHENE
PROPENE
1-BUTENE
1-PENTEN
3M-1-BUT
1-HEXENE
C6-OLE1
C7-OLE1
C8-OLE1
C9-OLE1
C10-OLE1
C1 1-OLE1
C12-OLE1
C13-OLE1
C14-OLE1
C15-OLE1
ISOBUTEN
2M-1-BUT
VOC Description
C10 Cycloalkanes
C11 Cycloalkanes
C12 Cycloalkanes
C13 Cycloalkanes
C14 Cycloalkanes
C15 Cycloalkanes
Alkenes
Ethene
Propene
1-Butene
1-Pentene
3-Methyl-1-Butene
1-Hexene
C6 Terminal Alkenes
C7 Terminal Alkenes
C8 Terminal Alkenes
C9 Terminal Alkenes
C10 Terminal Alkenea
C11 Terminal Alkenes
C12 Terminal Alkenes
C13 Terminal Alkenes
C14 Terminal Alkenes
C15 Terminal Alkenes
Isobutene
2-Methyl-1-Butene
kOH
[b]
( PP"T '
min'1)
2.3E+4
2.6E+4
2.9E+4
3.1E+4
3.3E+4
3.5E+4
1.2E+4
3.8E+4
4.6E+4
4.6E+4
4.6E+4
5.4E+4
5.4E+4
5.4E+4
5.4E+4
5.4E+4
5.4E+4
5.4E+4
5.4E+4
5.4E+4
5.4E+4
5.4E+4
7.5E+4
8.8E+4
Kinetic React.
(fract. react)
MaxRct
0.82
0.85
0.87
0.88
0.90
0.91
0.67
0.92
0.95
0.95
0.95
0.96
0.96
0.96
0.96
0.96
0.96
0.96
0.96
0.96
0.96
0.96
0.98
0.98
MaxOn
0.91
0.93
0.94
0.95
0.95
0.96
0.80
0.97
0.98
0.98
0.98
0.98
0.98
0.98
0.98
0.98
0.98
0.98
0.98
0.98
0.98
0.98
0.99
0.99
Meoh. React.
(mol O^/mol C)
MaxRct
0.48
0.44
0.42
0.33
0.31
0.29
2.3
2.1
1.9
1.31
1.31
0.91
0.91
0.71
0.55
0.45
0.38
0.34
0.30
0.27
0.25
0.23
1.24
1.09
MaxOj
0.24
0.22
0.22
0.17
0. 149
0.145
0.95
0.85
0.72
0.49
0.49
0.34
0.34
0.26
0.20
0.16
0.137
0.119
0.107
0.095
0.088
0.081
0.50
0.45
T n
i ri
(mol 0^
MaxRct
0.39
0.37
0.37
0.29
0.27
0.27
1.6
1.9
1.8
1.23
1.23
0.87
6.87
0.69
0.52
0.43
0.37
0.32
0.29
0.26
0.24
0.22
1.21
1.07
icremental
/mol C)
MaxOo
0.22
0.20
0.20
0.16
0. 142
0.138
0.77
0.82
0.70
0.48
0.48
0.33
0.33
0.26
0.19
0.16
0.135
0.117
0.105
0.094
0.087
0.080
0.50
0.45
Reactivi
(gm Oj/
4-V
ty
'gm VOC)
MaxRct MaxOg
1.33
1.26
1.25
0.99
0.94
0.91
5.3
6.6
6.1
4.2
4.2
3.0
3.0
2.3
1.8
1.48
1.27
1.11
1 .00
0.90
0.83
0.77
4.1
3.7
0.74
0.70
0.69
0.54
0.49
0.47
2.6
2.8
2.1
1.6
1.6
1.14
1.14
0.88
0.67
0.54
0.46
0.40
0.36
0.32
0.30
0.27
1.7
1.5
Cod
Rep.
[c]
3
3
3
3
3
3
2
2
2
2
4
2
4
2
2
2
2
2
2
2
2
2
2
2
les
Unc.
[d]
8
8
8
8
8
8
1
4
4
7
7
4
7
8
8
8
8
8
8
8
8
8
5
7
(continued)
-------
Table 3. (continued) - 4
rv>
Model
Code
[a]
T-2-BUTE
C-2-BUTE
2M-2-BUT
C5-OLE2
23M2-BUT
C6-OLE2
C7-OLE2
C8-OLE2
C9-OLE2
C10-OLE2
C11-OLE2
C12-OLE2
C13-OLE2
C14-OLE2
C15-OLE2
13-BUTDE
ISOPRENE
CYC-PNTE
CYC-HEXE
A-PINENE
B-PINENE
BENZENE
TOLUENE
C2-BENZ
VOC Description
trans-2-Butene
cis-2-Butene
2-Methyl-2-Butene
C5 Internal Alkenes
2,3-Dimethyl-2-Butene
C6 Internal Alkenes
C7 Internal Alkenes
C8 Internal Alkenes
C9 Internal Alkenes
C10 Internal Alkenes
C11 Internal Alkenes
C12 Internal Alkenes
C13 Internal Alkenes
C14 Internal Alkenes
C15 Internal Alkenes
1 ,3-Butadiene
Isoprene
Cyclopentene
Cyclohexene
a-Pinene
b-Pinene
Aromatic Hydrocarbons
Benzene
Toluene
Ethyl Benzene
kOH
(^m-1
min'1)
9.2E+4
9.2E+4
1-3E+5
9.2E+4
1.6E+5
9.2E+4
9.2E+4
9.2E+4
9.2E+4
9.2E+4
9.2E+4
9.2E+4
9.2E+4
9.2E+4
9.2E+4
9.7E+4
1.5E+5
9.7E+4
9.8E+4
7.8E+4
7.8E+4
1.9E+3
8.7E+3
1.0E+4
Kinetic React.
(fract. react)
MaxRct
0.98
0.98
0.99
0.98
0.99
0.98
0.98
0.98
0.98
0.98
0.98
0.98
0.99
0.98
0.98
0.99
0.99
0.99
0.99
0.98
0.98
0.141
0.49
0.55
MaxOo
0.99
0.99
0.99
0.99
1.00
0.99
0.99
0.99
0.99
0.99
0.99
0.99
0.99
0.99
0.99
0.99
1.00
0.99
0.99
0.99
0.99
0.21
0.64
0.71
Mech. React.
(tnol Oo/mol C)
MaxRct
2.1
2.1
1.4S
1.8
1.07
1.5
1.28
1.07
0.92
0.82
0.73
0.67
0.61
0.57
0.53
2.2
1.8
1.13
0.9^
0.55
0.55
0.54
1.00
0.95
MaxOj
0.82
0.82
0.53
0.67
0.36
0.56
0.46
0.38
0.33
0.29
0.26
0.24
0.22
0.20
0. 19
0.85
0.70
0.40
0.36
0.21
0.21
0.111
0.17
0.19
Incremental
(mol 03/mol C)
MaxRct
2.1
2.1
1.44
1.8
1.07
1.5
1.26
1.05
0.91
0.80
0.72
0.66
0.60
0.56
0.52
2.2
1.8
1.12
0.93
0.54
0.54
0.075
0.49
0.53
Reactivi
(gm Cy
f V -
ty
'gm VOC)
MaxOo MaxRct MaxO-j
3 J
0.81
0.81
0.52
0.66
0.35
0.55
0.45
0.38
0.33
0.29
0.26
0.23
0.21
0.20
0. 19
0.84
0.70
0.40
0.36
0.21
0.21
0.023
0. 106
0.132
7.2
7.2
4.9
6.1
3.6
5.2
4.3
3.6
3.1
2.8
2.5
2.3
2.1
1-9
1.8
7.6
6.4
3.9
3.3
1.9
1.9
0.28
1.8
1-9
2.8
2.8
1.8
2.3
1.21
1.9
1.6
1.29
1.11
0.98
0.88
0.80
0.73
0.68
0.63
3.0
2.5
1.40
1.26
0.73
0.73
0.086
0.39
0.48
Coc
Rep.
[c]
2
4
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
4
2
2
3
les
Unc.
[d]
5
5
7
7
7
8
8
8
8
8
8
8
8
8
8
8
6
8
8
5
8
4
4
7
(continued)
-------
Table 3. (continued) - 5
oo
o
Model
Code
[a]
N-C3-BEN
1-C3-BEN
S-C4-BEN
C10-BEN1
C11-BEN1
C12-BEN1
C13-BEN1
0-XYLENE
P-XYLENE
M-XYLENE
C9-BEN2
C10-BEN2
C11-BEN2
C12-BEN2
135-TMB
123-TMB
124-TMB
C10-BEN3
C1 1-BEN3
C12-BEN3
TETRALIN
NAPHTHAL
ME-NAPH
23-DMN
VOC Description
n-Propyl Benzene
Isopropyl Benzene
sec -Butyl benzene
CIO Monosub. Benzenes
C11 Monosub. Benzenes
C12 Monosub. Benzenes
C13 Monosub. Benzenes
o-Xylene
p-Xylene
m-Xylene
C9 Disub. Benzenes
C10 Disub. Benzenes
C11 Disub. Benzenes
C12 Disub. Benzenes
1 ,3,5-Trimethyl Benzene
1 ,2,3-Trimethyl Benzene
1 ,2,4-Trimethyl Benzene
C10 Trisub. Benzenes
C11 Trisub. Benzenes
C12 Trisub. Benzenes
Tetrahydronaphthalene
Naphthalene
Methyl Naphthalenes
2,3-Dimethyl Naphth.
kOH
[b]
(ppm
min~ )
8.8E-f3
9.5E+3
8.8E+3
8.7E+3
8.7E+3
8.7E+3
8.7E+3
2.0E+4
2.1E+4
3.5E+4
3.5E+4
3.5E+4
3.5E+4
3.5E+4
8.4E+4
4.8E+4
4.8E+4
8.4E+4
8.4E+4
8.4E+4
5.0E+4
3.2E+4
7.6E+4
1. 1E+5
Kinetic React.
(fract. react)
MaxRct
0.50
0.52
0.50
0.49
0.49
0.49
0.49
0.77
0.79
0.91
0.91
0.91
0.91
0.91
0.98
0.95
0.95
0.98
0.98
0.98
0.96
0.89
0.98
0.99
Max03
0.65
0.68
0.65
0.64
0.64
0.64
0.64
0.88
0.89
0.96
0.96
0.96
0.96
0.96
0.99
0.97
0.97
0.99
0.99
0.99
0.98
0.95
0.99
0.99
Mech.
(mol 0
MaxRct
0.83
0.84
0.75
0.70
0.64
0.58
0.54
.6
.6
.8
.6
.41
.28
.18
2.1
1.9
1.9
1.9
1.7
1.6
0.20
0.25
0.64
1.01
React.
3/mol C)
MaxOj
0.16
0. 16
0. 142
0. 116
0. 105
0.096
0.089
0.46
0.46
0.52
0.46
0.42
0.38
0.35
0.67
0.57
0.57
0.60
0.54
0.50
0.030
0.0120
0. 145
0.27
i ~
(mol 0,
MaxRct
0.41
0.44
0.37
0.35
0.31
0.29
0.27
1.27
1.30
.6
.42
.28
.16
.07
2.1
1.8
1.8
1.8
1.7
1.5
0.20
0.22
0.63
1.00
cremental
/mol C)
Max03
0.102
0. 110
0.092
0.075
0.068
0.062
0.057
0.40
0.41
0.50
0.44
0.40
0.36
0.33
0.66
0.55
0.55
0.59
0.54
0.49
0.030
0.0109
0. 144
0.27
Reacti\
(gm 0-
j\ f v ---
j/gm VOC)
MaxRct MaxO^
.48
.6
.33
.24
. 12
.02
0.94
4.6
4.7
5.8
5.1
4.6
4.1
3-8
7.4
6.3
6.3
6.6
6.0
5.5
0.71
0.84
2.3
3.7
0.37
0.39
0.33
0.27
0.24
0.22
0.20
1.46
1.48
1.8
1.6
1.43
1.29
1.18
2.4
2.0
2.0
2.1
1.9
1.8
0. 108
0.041
0.53
0.99
Cod
Rep.
fc]
3
3
3
5
5
5
5
3
3
2
4
4
4
4
2
3
3
6
6
6
2
2
3
2
es--
Unc.
[d]
7
7
7
8
8
8
8
4
7
4
8
8
8
8
4
7
7
8
8
8
5
5
8
5
(continued)
-------
Table 3. (continued) - 6
Model VOC Description kOH
Code [b]
t a ] ( ppm~ ]
rain"1)
Kinetic React.
(fract. react)
MaxRct MaxO-i
Mech. React.
(mol 03/mol C)
MaxRct MaxO-j
Incremental
(mol 03/mol C)
MaxRct MaxO-j
(gra 03/gm VOC)
MaxRct MaxO^
Rep. Unc.
[c] [d]
Alkynes
ACETYLEN Acetylene
FORMALD
ACETALD
PROPALD
Aldehydes
Formaldehyde
Acetaldehyde
C3 Aldehydes
GLYOXAL Glyoxal
MEGLYOX Methyl Glyoxal
1.1E+3 0.088 0.1314
Alcohols, Ethers, Esters, etc.
MEOH Methanol
ETOH Ethanol
N-C3-OH n-Propyl Alcohol
I-C3-OH Isopropyl Alcohol
I-C4-OH Isobutyl Alcohol
N-C4-OH n-Butyl Alcohol
T-C4-OH t-Butyl Alcohol
ET-GLYCL Ethylene Glycol
PR-GLYCL Propylene Glycol
ME-O-ME Dimethyl Ether
MTBE Methyl t-Butyl Ether
ETBE Ethyl t-Butyl Ether
1.11
0.53
0.098 0.071
0.36
0.26
1.4E+3
4.8E+3
7.8E+3
7.6E+3
1 .4E+4
1.2E+4
1.7E+3
1.1E+4
1 .8E+4
4.4E+3
4.2E+3
1.1E+4
0.105
0.32
0.146
O.H5
0.66
0.61
0. 125
0.58
0.73
0.30
0.28
0.57
0.16
0.44
0.61
0.60
0.79
0.75
0.19
0.73
0.85
0.42
0.140
0.72
2.5
1.17
1.17
0.35
O.UU
1.01
0.89
1;22
0.68
0.91
0.56
0.85
0.93
0.42
0.140
0.18
0.21
0.37
O.UO
0.50
0.29
0.54
0.30
0.1J1
0.26
0.37
0.54
0.16
0.29
0.62
0.111
0.71
0.50
0.27
0.16
O.H9
0.1H7
0.19
0.24
0.109
0.17
0.28
0.075
'0.36
0.25
0.23
0.119
0.30
0.39
0.77
1.29
0.38
0.75
1.6
0.29
1.10
0.94
0.56
0.43
1.37
0.22
0.39
0.59
0.26
0.44
0.73
0. 19
0.56
0.47
0.47
0.33
0.84
2
2
3
3
3
3
3
3
3
3
2
2
1
1
7
7
7
7
7
7
7
7
4
7
1.4E+4
2.3E+4
2.9E+4
1.7E+4
2.5E+4
0.94
0.83
0.91
1.00
1.00
0.97
0.92
0.95
1.00
1.00
4.2
2.1
2.0
1.22
5.9
1.30
0.77
0.70
0.41
1.9
3.9
1.7
1.8
1.22
5.9
1.26
0.70
0.66
0.41
1.9
6.2
3-8
4.6
2.0
11.7
2.0
1.5
1.6
0.68
3.8
1
1
1
1
1
1
il
5
3
3
(continued)
-------
Table 3. (continued) - 7
Model VOC Description kOH
Code [b]
[a] (pprrT1
min"1)
Kinetic
(fract.
MaxRct
React.
react)
Max03
Mech. React.
(mol 03/mol C)
MaxRct Max03
Incremental
(mol 03/mol C)
MaxRct Mai
KUn
(gm 03/gm VOC)
MaxRct Max03
-fnr1
Rep.
[c]
es
Unc.
[d]
OO
r\J
Ketones
ACETONE Acetone 3.4E+2 0.043 0.058 3.6 0.95 0.16 0.055 0.39 0.136 1 5
MEK C4 Ketones 1.7E+3 0.16 0.22 1.8 0.53 0.28 0.116 0.75 0.31 1 5
Aromatic Oxygenates
BENZALD Benzaldehyde 1.9E+4 0.83 0.95 -0.20 -0.31 -0.17 -0.29 -0.54 -0.93 1 5
PHENOL Phenol 3.9E+4 1.00 1.00 0.26 -0.17 0.26 -0.17 0.79 -0.53 1 7
CRESOL Alkyl Phenols 6.2E+4 1.00 1.00 0.51 -0.23 0.51 -0.23 1.6 -0.72 1 5
Mixtures Used in Reactivity Scenarios [e]
ALLCITY5
ALOFT
Base ROG Mixture
Aloft ROG Mixture
0.
0,
.56
,34
0.
0.
.65
,42
0.70
0.60
0.28
0.24
2,
1.
.3
.8
0
0
.91
.74
Notes:
(a] Detailed model species name used in the mechanism (Carter, 1990).
(b] Hydroxyl radical rate constant for T=300 K used in the mechanism (Carter, 1990).
(c] Codes for mechanism representation and reactivity calculation method are as follows:
1. Explicitly represented in mechanism. Reactivity calculated directly.
2. Mechanistic parameters explicitly assigned. Reactivity calculated directly,
3. Mechanistic parameters explicitly assigned. Kinetic reactivities derived from the dependence of kOH on
fraction reacted. Mechanistic reactivities derived from "pure mechanism species" reactivities as discussed
in Appendix B.
4. Assumed to have the same per-molecule reactivity as the model species listed above.
5. Assumed to have the same per-molecule reactivity as toluene.
6. Assumed to have the same per-molecule as 1,3,5-triroethyl benzene.
(continued)
-------
Table 3. (continued) - 8
OJ
(d] Mechanism uncertainty codes are as follows. "Tested" means mechanism tested by model simulations of chamber data.
1. Least uncertain; tested.
2. Probably not uncertain, but not tested.
3. Laboratory data available for reactions in the mechanism, but mechanism not tested.
1. Uncertain portions adjusted or parameterized to fit chamber data.
5. Uncertain, and only limited or uncertain data available to test it.
6. Mechanism not optimized to fit existing chamber data.
7. Mechanism estimated and not tested.
8. Estimated mechanism is highly uncertain, and not tested.
[e] Compositions given in Table 2. Reactivities determined by linear summation of reactivities of components.
-------
constant for their initial reactions in the atmosphere. The mechanisms
used for the reactions of the radicals subsequently formed, which are
responsible for ozone formation, are estimates based on analogies with the
few compounds whose mechanisms are known in more detail. Nevertheless, the
mechanism incorporates our best present estimates for the reaction
mechanisms of the wide variety of VOCs which are emitted into the
atmosphere.
2. Uncertainties in Mechanisms for Individual VOCs
The levels of uncertainty in the reaction mechanism obviously
must be taken into account when the results of model calculations of
reactivities are used to assess ozone control strategies. To aid in such
assessments, all the VOCs listed on Table 3 have been categorized in a
six-class grouping according to the (1) author's qualitative assessment of
the level of our present knowledge of their atmospheric reaction
mechanisms and (2) the availability of environmental chamber data suitable
for testing these mechanisms. These groupings thus should give a rough
indication of the degree of uncertainty in predicting their reactivities
in any given scenario. These groupings are as follows.
Group 1 consists of compounds for which we believe we understand
at least the most important of the fundamental processes by which the VOC
promotes ozone formation, and whose mechanisms have been tested at least
to some extent with environmental chamber data. Examples include n-
butane, ethylene, formaldehyde, methanol, and ethanol. There are still
some uncertainties for most of the compounds in this group, but new data
are not expected to have major impacts on predictions of their
reactivity.
Group 2 consists of compounds for which we believe we understand
the major reactions involving ozone formation reasonably well from
fundamental knowledge, but whose mechanisms have not been tested against
chamber data. Examples include the C^-Co alkanes.
Group 3 consists of compounds for which we believe we understand
the major reactions involving ozone formation reasonably well based on
fundamental knowledge, and based on laboratory data concerning at least
some of their elementary reactions, but whose mechanisms have not been
tested against chamber data. Examples include glyoxal and methylglyoxal.
-------
Group 4 consists of compounds where there are major gaps in our
knowledge in their mechanisms involved in ozone formation, but for which
available environmental chamber data has allowed us to develop adjusted or
parameterized mechanisms which can adequately simulate ozone formation and
other observations in these experiments. Examples include several
representative aromatic compounds, propene, 1-butene and 1-hexene.
Group 5 consists of compounds which have uncertain mechanisms
and for which there are only limited or unreliable chamber data available
for testing or adjusting their mechanisms. Examples include several Cc+
alkanes, propionaldehyde, acetone, methyl ethyl ketone, and acetylene.
Group 6 consists of compounds for which there are probably
enough chamber data to place the compound in Group 3, but for which the
current mechanism does not have sufficient detail to simulate these data
as well as it could. Examples include isoprene and acrolein.
Groups 7 and 8 consist of compounds whose mechanisms are
uncertain and for which no adequate chamber data are available.
Mechanisms for these have been estimated by analogy or extrapolation from
mechanisms developed for other compounds. Two categories are used to
express the author's qualitative estimate concerning the relative degree
of uncertainty in the estimated mechanism, group 8 being more uncertain.
Examples of compounds in group 7 include the Co-Ch alcohols and the higher
(C
-------
Ozone Formed in the Ozone Formed in the
Incremental Scenario with the - Base Case Scenario
Reactivity VOC Added (Test Case)
of a VOC in = (I)
a Scenario Amount of VOC Added
in the Test Case
The "ozone formed" quantity is either the maximum ozone concentration
calculated for the scenario or the integrated ozone concentration
throughout the simulation; this depends on whether ozone yield or
integrated ozone reactivities are being calculated (see subsection 1,
below). In either case, the "amount of VOC added" quantity is the
concentration of the VOC which would occur at the time of the ozone
maximum if the VOC did not undergo chemical reaction.
To avoid dependence of the incremental reactivity on the amount of
VOC added, the incremental reactivity is defined as the limit as the
amount of VOC added approaches zero. This also permits incremental
reactivities of VOC mixtures to be calculated by linear summations of the
reactivities of their components. The validity of determining incremental
reactivities of mixtures by linear summation of incremental reactivities
of their components follows mathematically from the facts that (1)
incremental reactivities are derivatives, and (2) the solutions of the
differential equations describing the kinetics of the chemical
transformations in airshed models are continuous functions.
(In practice, the amount of VOC added in the test calculation was
such that the amount of VOC reacted in the simulation was equivalent to
o
emissions of 0.01 mmol m of reacting VOC. Test calculations showed that
this is well within the linear range where incremental reactivities are
independent of the amount of VOC added. However, if the amount of VOC
added in the simulations is too small, the numerical errors in the
simulation may become non-negligible in the reactivity determinations.
The algorithm for determining the amount of VOC added in the calculations
is such that numerical errors in the simulation translate directly into
absolute uncertainties in moles of ozone formed per mole of VOC reacted,
i.e., in mechanistic reactivities [see discussion of "reactivity
components" below, in subsection 2, for the definition of "mechanistic
reactivity"]. The stepwise numerical error tolerance parameter used in
the simulations was set such that the uncertainties in the mechanistic
36
-------
reactivities due to numerical errors were less than 0.05 mol of 0^ per
mole of VOC reacted. This is small compared to the magnitudes of
mechanistic reactivities of most VOCs. VOC mechanistic reactivities that
are comparable to this value tend to be very sensitive to scenario
conditions.)
1. Ozone Yield and Integrated Ozone Reactivities
Two measures of incremental reactivity are used in this work,
depending on how the "ozone formed" quantity in Equation (I) is
measured. These are the "ozone yield reactivity" and the "integrated
ozone reactivity," and they are discussed below.
Ozone Yield Reactivities are incremental reactivities calculated
by Equation (I), where the amount of ozone formed is measured by the
maximum ozone concentration. If both ozone formed and VOC added are given
in molar units, ozone yield reactivities can be thought of as unitless
quantities, reflecting numbers of molecules of ozone formed per molecule
of VOC added to the emissions. For this reason, ozone yield reactivities
are considered to have the best correspondence to the fundamental chemical
processes which occur on the molecular level. Note that the way the
amount of VOC added is quantified in Equation (I) corrects for the effect
of dilution on the ozone concentration, since dilution reduces both the
ozone yield in the numerator and the concentrations of the unreacting VOC
in the denominator by the same factor. The maximum reactivity and maximum
ozone reactivity scales (described in Section II.D.1) and two of the four
base case relative reactivity scales (Section II.D.2) are derived by using
ozone yield reactivities.
Integrated Ozone Reactivities are incremental reactivities
calculating by Equation (I) where amount of ozone formed is measured by
the ozone concentration integrated over time throughout the simulation.
Note that reactivities defined in this manner cannot be expressed as
unitless quantities and in general are dependent on the amount of time in
the scenario. For this reason, integrated ozone reactivities are not
considered to be as fundamental in a chemical sense as ozone yield
reactivities. However, integrated ozone levels may have a closer
correspondence to exposure of the population or the environment to ozone,
and thus integrated ozone reactivities may be more useful in some types of
cost-benefit analyses. Two of the four base case relative reactivity
37
-------
scales (Section II.D.2 are derived by using integrated ozone
reactivities.
In the subsequent discussion, if the terms ''reactivity",
"incremental reactivity", "maximum reactivity", "maximum ozone
reactivity", and "base case relative reactivity" are used without
qualifier, they will refer to ozone yield reactivities. This is the
measure of reactivity which has been used in the previous recent studies
of incremental reactivity (e.g., Dodge, 1984; Carter and Atkinson, 1987;
1989; Carter, 1989b; Chang and Rudy, 1990; Lowi and Carter, 1990). If
integrated ozone reactivities are discussed, they will be referenced
explicitly as such. Note that in this work integrated ozone reactivities
are used in only two of the four base case reactivity scales (see Section
D.2).
2. Separate Estimates of Reactivity Components
For reactivity estimation purposes, and to examine in more
detail how environmental and mechanistic factors affect the various
reactivity scales, it is useful to consider separately the two major
components of VOC reactivity. As discussed previously (Carter and
Atkinson, 1989), incremental reactivities can be thought of as being
products of two factors, which we have designated as "kinetic" and
"mechanistic" reactivities. The kinetic reactivity is defined the
fraction of the emitted VOC which undergoes chemical reaction in the
pollution scenario being considered,
Kinetic Fraction VOC Reacted
Reactivity = Reacted (II'
VOC Emitted
and the mechanistic reactivity is the amount of ozone formed relative to
the amount of VOC which reacts in that scenario.
Mechanistic Ozone Formed
Reactivity = (III)
VOC Reacted
The product of these two quantities then gives the overall incremental
reactivity.
38
-------
Incremental
Reactivity
Ozone Formed
VOC Emitted
Kinetic
= Reactivity
VOC Reacted
VOC Emitted
X
V
Mechanistic
Reactivity
Ozone Formed
VOC Reacted
(IV)
These two components of incremental reactivity, each of which is affected
by different aspects both of the VOC reaction mechanism and the conditions
of the airshed scenario, are often more straightforward to estimate than
overall incremental reactivity. Considering these two components of
reactivity separately can also provide a basis for making incremental
reactivity estimates for VOCs without having to calculate them directly in
airshed models. This is discussed below.
a. Estimation of Kinetic Reactivities
The kinetic reactivity of a VOC is defined as the fraction
of the emitted VOC which undergoes chemical reaction in the air pollution
episode. It depends on how rapidly the VOC reacts in the atmosphere and
on conditions of the episodes such as overall light intensity and radical
levels, but not on the other aspects of their reaction mechanism. As
shown in footnotes to Table 3, for many VOCs the kinetic reactivities used
in deriving the reactivity scales were determined by direct calculation.
However, if a VOC reacts only with OH radicals, its kinetic reactivity is
a function only of kOH, its OH radical rate constant, and characteristics
of the scenario which affect OH radical levels. Therefore, if one knows
the dependence of the kinetic reactivity on kOH for a given scenario, the
kinetic reactivity can be determined for any VOC (which reacts only with
OH radicals) from its OH rate constant. As indicated on Table 3, this
approach was used for a number of VOCs whose reactivities were not
calculated directly.
The dependence of kinetic reactivity on kOH is determined
for a given scenario by calculating fractions reacted for model species
with variable values of kOH (and which react only with OH). This was
determined in this work for each of the scenarios used to derive the
reactivity scales. Note that this dependence of kinetic reactivity on kOH
for a given scenario can be approximated by the empirical relation (Carter
and Atkinson, 1989):
39
-------
Kinetic Fraction - kOH x IntOH
Reactivity Reacted ( 1 - e ) (V)
where IntOH is a scenario-dependent parameter which reflects primarily the
overall integrated OH radical levels of the scenario. This approximation
can estimate kinetic reactivities to within 10? for most scenarios and can
calculate them exactly if all VOCs are present at the start of the
simulation. Although Equation (V) was not used to estimate kinetic
reactivities in this work (since the kinetic reactivity was directly
calculated as a function of kOH for each scenario, and thus use of an
approximation was unnecessary), it is useful for discussion purposes,
since a single scenario-dependent parameter, IntOH, gives a good
indication of how the kinetic reactivity is related to kOH for the
scenario. The scenario-to-scenario variability of IntOH will have a
direct correspondence to the scenario-to-scenario variability of the
kinetic reactivity of any VOC which reacts only with OH radicals.
b. Estimation of Mechanistic Reactivities
Mechanistic reactivities are a measure of the amount of
ozone formation caused by the reaction of a given amount of the VOC and
are not directly dependent on how rapidly the compound reacts. They are
determined by the nature of the VOC reaction mechanism, such as number of
conversions of NO to N02 which occur during its oxidation process, whether
the reactions enhance or inhibit radical or NO levels, and the
reactivities of the products they form. They are also strongly affected
by the conditions of the scenario such as the ROG/NOV ratio and other
A
factors which affect the overall efficiency of ozone formation (Carter and
Atkinson, 1989). Depending on the conditions of the scenario and the
nature of the reaction mechanisms, mechanistic reactivities can range from
negative values (indicating the VOC's reactions actually reduce overall
ozone formation) to values as high as over 10 moles of ozone formed per
mole of carbon reacted (Carter and Atkinson, 1989).
Note that extremely high or negative mechanistic
reactivities are almost always due to the effect of the VOC reactions on
the overall photooxidation process, rather than to the direct formation or
removal of ozone by the VOC reactions. For example, if VOC has an
unusually high mechanistic reactivity, it is usually because its reactions
40
-------
tend to enhance overall radical levels, causing more of the other VOCs
present to react and thus form ozone. In such cases, only a fraction of
the additional ozone formation caused by adding the VOC is directly due to
the reactions of the added VOC itself; most of the additional ozone is due
to reactions of other VOCs which would not react if the test VOC were not
added. Likewise, if a VOC has a negative mechanistic reactivity, it is
not usually because its reactions directly remove ozone, it is either
because its reactions tend to reduce radical levels (and thus cause less
overall reaction of all VOCs), or because its reactions tend to reduce the
overall efficiency of ozone formation from all reacting VOCs, as would be
the case if they removed NOX in scenarios where ozone formation is NOX-
limited.
The mechanistic reactivities of a VOC in a given scenario
can be calculated directly by carrying out computer model simulations of
the VOC's incremental reactivity (using Equation I) and fraction reacted
(kinetic reactivity), and dividing the incremental reactivity by the
kinetic reactivity. However, for VOCs whose mechanism involves only
reaction with OH radicals, an alternative method, which does not require
explicit reactivity calculations for each VOC, can be used. For this
purpose, mechanistic reactivities calculated for the "pure mechanism
species" can be used. The basis for this approach is discussed in
Appendix B. Equivalent results are obtained by either method. Footnotes
to Table 3 indicate those VOCs where this method was used.
D. Derivation of Generalized or Multi-Scenario Reactivity Scales
The focus of this work is to examine possible methods for developing
reactivity scales for assessing VOC control strategies for a variety of
conditions. Two general types of approaches are employed: (1) developing
generalized scales for specified sets of chemical conditions (e.g.,
maximum reactivity or maximum ozone) which might be most appropriate for
general reactivity assessment purposes, and (2) developing multi-scenario
scales using various methods to combine or average reactivities for a set
of base case scenarios which are assumed (for the purpose of this study)
to represent a realistic distribution of conditions. A diagram outlining
the derivation and interrelationships between the generalized and multi-
-------
scenario reactivity scales discussed in this section is given in Figure
1. The methods used to derive these scales are discussed in more detail
in the following sections.
1. Derivation of the Maximum Reactivity (MaxRct) and the
Maximum Ozone Reactivity (MaxOp) Generalized Scales
As indicated above, two general reactivity scales were developed
to represent specified sets of chemical conditions with respect to NOX
availability. These are (1) the "maximum reactivity" scale (designated by
the abbreviation "MaxRct" in the subsequent discussion), representing NOX
conditions yielding maximum VOC reactivities, and (2) the "maximum ozone"
reactivity scale (designated "MaxOo"), representing NOX conditions
yielding maximum peak ozone concentrations. The derivation of these two
scales are discussed in this section.
The scenarios used to derive the maximum reactivity (MaxRct) and
the maximum ozone reactivity (MaxOo) scales are described in Section
II.A.S.b. Briefly, the NOX inputs in the base case scenarios (whose
conditions are summarized in Table 1) were adjusted to yield either
maximum reactivity of the base case ROG emissions or maximum peak ozone
concentrations, and all the other inputs were kept the same. Note that
since only the NOX inputs were adjusted, these two scales represent
maximum reactivity or maximum ozone conditions only with respect to NOX
inputs they represent averaged conditions for all other aspects.
Therefore, these are not truly "general" reactivity scales, since the
distribution of the other airshed conditions will affect the results to
some extent. However, since these conditions have smaller effects on
overall reactivities, the ranges of reactivities being averaged or
combined should in principle be less than for the base case reactivity
scales, and thus the results should be less sensitive to the specific set
of scenarios employed.
The MaxRct and MaxO^ reactivity scales were derived by
separately averaging the kinetic reactivities and the mechanistic
reactivities for their respective types of scenarios, and then using the
product of these averages as the incremental reactivity. The kinetic
reactivities and mechanistic reactivities were computed as discussed in
Section II.C.2. Although the resulting incremental reactivities in the
MaxRct or MaxOo scales turn out to be essentially the same as those
-------
Aaalfsis of Airsheds
Bt[?DC]
lai. Read
Scenarios
(i=l,2,.. )
1
lodel
Calc.
!
idj. lOi to
,
lai. IS[B.!OG]
B!YOC]i
Base Case
Scenarios
(j=U,...)
1 1
lodel lodel
Calc. Calc.
1 1
IE[B.ROG]j
IRIlB.ROG],
Uerage Image
1
laiRct
n[Yoc]
L
t
1
lailct
ffi[TOC]
liltiplj < i
1
laiSct
IE[IOC]
Lin
IR
Idj. lOi to laiimu Oi
^
> Scenarios
lai. Oi (k=l,2,...)
1
lode!
t Calc.
fOC]j
IR][TOC],
I /
O[YOC]i
1 I
K[TOC]i
1
\ 4 * Rstio Jierage l?erage
/
/
/
i 1
M[fOC],
lillTOC],
laid
O[70C]
i
laiOj
O[70C]
/ ' 1 !
e Fit Image 1 > Inltiplj < (
1
t.Sq. Fit
RRITOC]
RRI[70C]
lig.
J 1
Ratio
RR[YOC]
RRI[TOC]
laid
IR[YOC]
lailct Scale
Base Case itlatire Scales
laid} Scale
lOilTIOI:
IOC iij organic ctntponnd. B.ROG liitnre used to represent base case EOG emissions.
n[TDC]i, ffi[YOC]i Incremental, kinetic, or mechanistic leactivitj of the YOC for scenario i.
Incremental, kinetic, or mechanistic resctinty of the YOC in the general scale.
Incremental reactiiitj of the base IOC for scenario i.
Integrated oione reactmtj of the YOC or the base EOG for scenario i.
RR[?DC], IE[VOC),/IE[B.ROG]i ielatiie rejctiiitj of the YOC for scenario i.
1II[YOC], IRl[TOC]i/I!l[B.ROG], lelatire integrated oione reactifitj for scenario i.
RR[TOC], RRI[YOC] Relatife reactijitj or relatiie integrated oione reactiritj in the general scale.
II[TOC], n[ioc],
IIJB.IOG]!
Figure 1. Diagram of the derivation of the generalized and multi-
scenario reactivity scales.
-------
derived by directly averaging the incremental reactivities in the
individual scenarios, this more indirect approach is preferred because it
provides a means to obtain a kinetic reactivity and a mechanistic
reactivity scale for these two sets of chemical conditions. These are
useful, for example, in an analysis of the factors causing the differences
between these two scales, as is discussed later in this report. The
scenario characteristics affecting kinetic reactivities are different than
those affecting mechanistic reactivities, since kinetic reactivities are
determined primarily by overall radical levels, while mechanistic
reactivities are determined by more complex chemical factors involving how
the reactions cause ozone formation. Therefore, it can be argued that
kinetic reactivities and mechanistic reactivities are somewhat more
j
fundamental in a chemical sense than the product of the two.
The specific scenarios used in deriving the MaxRct and MaxOo
reactivities are indicated in Table 1 . To avoid giving undue weight for
those urban areas where Gery et al. (1988) had many episodes, averages
were first calculated for each urban area, and then the averages for the
urban area were averaged to obtain the kinetic or mechanistic reactivity
values used in the scale.
2. Multi-Scenario (Base Case) Relative Reactivity Scales
Incremental reactivities calculated for the conditions of the
base case scenarios listed in Table 1 were used to derive the multi-
scenario scales in this work. Because the incremental reactivities of a
VOC varies widely in magnitude from scenario to scenario (primarily
because of the variabilities of ROG/NOX ratios), it is considered to be
more appropriate that multi-scenario reactivity scales be derived from
ratios of incremental reactivities rather than from absolute incremental
reactivities. For this purpose, we use the ratio of the incremental
reactivity of the VOC relative to that base case ROG mixture, and the term
"relative reactivity" is used throughout this report to designate this
ratio.
The incremental reactivities for the 21 representative base case
scenarios (i.e., the scenarios where NOX inputs were not varied, as
discussed above) were used to derive four different base case relative
reactivity scales. As indicated above, two different measures of "ozone
formed" in Equation (I) ozone yield and integrated ozone were
-------
employed. In addition, two different methods were used to derive single
multi-scenario scales from the varying incremental reactivities for
individual scenarios. These two derivation methods, designated the
"average ratio" or the "least squares fit" methods, are discussed
below.
a. Average Ratio Method
The most obvious way to derive a multi-scenario relative
reactivity scale is simply to average the relative reactivities for the
various scenarios. Since this is an average of ratios of incremental
reactivities (the incremental reactivity of the VOCs / the incremental
reactivity of the base ROG), it is called the "average ratio" method.
This method has the advantages of simplicity and the fact that it
minimizes errors in predictions of ratios of incremental reactivities
throughout the set of scenarios. However, it has the disadvantage that
the scenarios where VOC reactivities are very low, i.e., where VOC changes
have only small effects on ozone formation, are given the same weight as
scenarios where VOCs have high reactivities, and thus where ozone is much
more sensitive to VOC changes.
b. Minimum Least Squares Error (Least Squares Fit) Method
An alternative approach to derive a single relative
reactivity scale from the varying reactivities in the individual scenarios
is to use a method designed to minimize the total error in ozone
predictions throughout the entire set of scenarios which would result from
use of the single scale. Such an approach would necessarily give greater
weight to those scenarios which are sensitive to VOC changes than to those
which are less sensitive to VOC changes. The derivation of a method which
implements this approach is discussed below.
If the sum of squares criterion is used to measure the total
error in ozone predictions resulting from the use of a single relative
reactivity scale to predict the effect of adding a given amount a VOC to
emissions in a set of scenarios, the mathematical problem of deriving a
scale which minimizes this error can be expressed as follows (where the
notation is generally consistent with that used in Figure 1).
(1) Let IRfVOC], and IRtB.ROGU be incremental reactivities
1~ h
of a given VOC and the base ROG mixture for the j scenario in a set of
no scenarios, where 1 <_ j <_ n^.
-------
(2) Let RR[VOC] be the single relative reactivity value for
the VOC which will be used to predict ozone changes caused by adding a
given (arbitrarily small) amount of the VOC, D(VOC), to all of the
scenarios.
(3) Since the relative reactivity is the ratio of
incremental reactivity of the VOC to the incremental reactivity of the
base ROG mixture, the use of RRfVOC] as the relative reactivity for the
VOC for all scenarios means that the predicted incremental reactivity of
the VOC in scenario j, IRpred[VOC] , , is
IRPred[VOC] j = RR[VOC] x IR[B.ROG]j
(4) If the amount of the VOC added to all scenarios,
D(VOC), is sufficiently small, then the predicted change in ozone caused
by adding it to scenario j, D[OoPred].:, is
D[03Pred]j IRPred[VOC]j x D(VOC)
Therefore,
D[03pred]j = RR[VOC] x IR[B.ROG]j x D(VOC)
(5) Likewise, under the same conditions, the actual change
in ozone caused by adding D(VOC) of the VOC to scenario j, DtO^L, is
D[03]j - IR[VOC]j x D(VOC)
(6) The problem is therefore to determine the value of
RR[VOC] which minimizes sum of squares error in the ozone predictions,
i.e., to minimize the quantity
Sum of no
Squares = I { D[03]j - D[03Pred]j }2
Error j=1
nS
= X { IRtVOC], x D(VOC) - RR[VOC] x IR[B.ROG]i x D(VOC)
J
-------
ns
D(VOC)2 x I { IR[VOC]1 - RR[VOC] x IR[B.ROG]i }2
J=1 J J
(7) Since D(VOC) is a constant (since we are predicting the
effect of adding the same amount of the VOC to all scenarios), this is the
same as minimizing the quantity
ns
I { IRtVOC], - RR[VOC] x IRtB.ROG], }2
J J
(8) The solution to the above problem (obtained by setting
the derivative of the above with respect to RR[VOC] to zero, and solving
for RR[VOC]) is
ns
I { IR[VOC]j x IR[B.ROG], }
RR(VOC) =
ns
I { IR[B.ROG]i
J=1
Note that the above is exactly the same problem as finding
the slope of the least squares line, forced through zero, which fits a
plot of IR[VOC]j (on the "y" axis) against IR[B.ROG]j (on the "x" axis)
for all the n^ scenarios. Examples of such plots are shown
in the Results section.
Since this method is designed to minimize the total error in
ozone predictions through the set of scenarios, it can be argued that the
resulting reactivity scale is a closer approximation to an "optimum"
reactivity scale for the set than that derived by using the average ratio
method. However, since it gives greater weight to scenarios where
reactivities are higher (specifically, where IR[B.ROG] is higher), its use
may result in less than optimal ozone control strategies in areas where
VOC reactivities are relatively low. Note that low VOC reactivities do
not imply low ozone levels; indeed, many N0x-limited, low-VOC-reactivity
scenarios have relatively high ozone levels. But even though the ozone
-------
levels may be higher in such low-reactivity scenarios, the change in
absolute ozone levels caused by changing VOC emissions is still relatively
low, and thus VOC controls are less relevant to solving the ozone problem
under such conditions.
-------
III. RESULTS AND DISCUSSION
Incremental reactivities were calculated for all the VOCs listed in
Table 3 for the conditions of the 21 base case, the 28 maximum reactivity,
and the 21 maximum ozone scenarios used to derive the reactivity scales.
In addition, integrated ozone reactivities were calculated for the 21 base
case scenarios. (The specific scenarios are those listed in Table 1 with
an asterick preceding the ROG/NOX ratio.) Plots showing examples of these
results are given in Figures 2-5. Complete sets of the results of the
reactivity calculations for all these scenarios are available in computer-
readable format from the author upon request. Figure 2 shows
distributions of the maximum ozone concentration, of the integrated ozone
concentrations, of the IntOH parameter, and of the incremental reactivity
of the mixture representing total ROG emissions. Figures 3-5 show
distributions of kinetic reactivities, mechanistic reactivities,
incremental reactivities, and relative reactivities (incremental
reactivities divided by incremental reactivities of the base ROG mixture)
for CO, n-butane, and toluene, respectively. These are typical of results
for most other VOCs, and show that the various measures of reactivity vary
from scenario to scenario, even after NOV inputs have been adjusted to
A
yield consistent chemical conditions (i.e., either maximum reactivity or
maximum ozone). These results, and the reactivity scales derived from
them, are discussed in the following sections.
A. General and Multi-Scenario Reactivity Scales
1. MaxRct and MaxO? Reactivity Scales
The results of the calculations for the maximum reactivity
(MaxRct) and the maximum ozone (MaxO?) scales are given in Table 3. This
includes, for each VOC or class of VOCs which is represented separately in
the Carter (1990) mechanism, the OH radical rate constant, and the
kinetic, mechanistic, and incremental reactivities for the VOCs in both
the MaxRct and the MaxOo scales. In addition, as discussed in Section
II.B.2, Table 3 gives codes indicating the level of uncertainty of the
mechanisms of the VOCs, and thus the relative degree of uncertainty of the
their calculated reactivities.
-------
Max Ret
Max 0
Base Case
8 12 16 0
Maximum
8 12 16
c
'E
i
E
o.
C
E
a.
a.
c
o
o
o
K)
O
ROG
Mixture
Reactivity
8 12 16 0 4 6 12 16 0 4
Number of City-Days
8 12 16
Figure 2. Distribution plots of maximum ozone, integrated ozone, the
IntOH parameter, and the base ROG mixture reactivity for
the representative maximum reactivity, maximum ozone, and
base case scenarios.
50
-------
CO
Max Ret
Max 0
Base Cose
Kinetic
Reactivity
0.00
4 8 12 16 0 4 8 12 16 0 4 8 12 16
.3
Mechanistic
Reactivity
4 8 12 16 0 4 8 12 16 0 4 8 12 16
0.000
'L
Incremental
Reactivity
0 4 8 12 16 0 4 6 12 16 0 4 6 12 16
0.21-
0.18-
0.15-
0.12-
0.06-
onn-
-
-
-
>j
yxx
3
^i^Wf^^
KXXxXX
:1
Relative
Reactivity
8 16 24 0 4 8 12160 4 8 1216
Number of CityDays
Figure 3. Distribution plots of kinetic, mechanistic, incremental, and
relative reactivities of CO in the representative maximum
reactivity, maximum ozone, and base case scenarios.
51
-------
Max Ret
Butane
Max 63 Base Case
Reactivity
0.00
o
.0
0.9
2, °-3"
0.0
:a
Mechanistic
Reactivity
0 4 8 12 16 0 4 8 12 16 0 4 8 12 16
O
S3
i_
o
o
0.3B
0.24-
1
ai2l
3.00-
3
Incremental
Reactivity
0 4 8 12160 4 6 12160 4 6 1216
0.8-
^ 0.6-
o
o 0.4-
0.2-
0.0-
:a
Relative
Reactivity
j
0 8 16 24 0 4 8 12 16 0 4
Number of CityDays
12 16
Figure 4. Distribution plots of kinetic, mechanistic, incremental, and
relative reactivities of n-butane in the representative maximum
reactivity, maximum ozone, and base case scenarios.
52
-------
Wax Ret
Toluene
Wax 0^ Base Case
Kinetic
Reactivity
0 4 8 12160 4 8 12160 4 8 1216
1.4*
Mechanistic
Reactivity
KXXXXXXH
0 4 8 12 16 0 4 8 12 16 0 4 8 12 16
0.84
-,12
Incremental
Reactivity
0 4 8 12 16 0 4 6 12 16 0 4 8 12 16
0.7-
^, 0.5-
o
o 0.3-
0.1-
Relative
Reactivity
0 8 16 24 0 4 8 12 16 0 4 8 12 16
Number of City-Days
Figure 5. Distribution plots of kinetic, mechanistic, incremental, and
relative reactivities of toluene in the representative maximum
reactivity, maximum ozone, and base case scenarios.
53
-------
As discussed in Section II.D.1, the incremental reactivities in
the MaxRct and MaxOo scales were derived by determining average kinetic
and mechanistic reactivities for all of the 12 urban areas, and then
multiplying the two average values to yield the incremental
reactivities. The incremental reactivities are given both in units of
moles of ozone formed per mole of volatile organic carbon emitted ("per-
carbon" units), and in grams of ozone formed per gram of VOC emitted ("per
mass" units). (Mechanistic reactivities are given only in per-carbon
units. Kinetic reactivities are unitless.) Note that although the units
of incremental reactivities affect relative reactivities of different VOCs
within a given scale, the units are irrelevant when comparing ratios of
reactivities of VOCs in different scales.
a. Comparison of Kinetic Reactivities
Table 3 and Figures 3-5 show that kinetic reactivities at
NOX levels yielding maximum ozone concentrations are higher than those for
NOX levels yielding maximum VOC reactivities. This is expected, since the
reaction of OH radicals with N02 is an important radical-terminating
process, and N02 levels are higher under maximum reactivity conditions.
Differences in overall radical levels among various scenarios can be
measured by the IntOH parameter, which, as discussed in Section II.C.2.a,
relates the OH radical rate constant for a VOC to the amount of the VOC
which reacts in the scenario (Equation V). As shown in Figure 2, the
IntOH values tend to be higher in the maximum ozone than in the maximum
reactivity scenarios. A higher IntOH value means that radical levels are
higher, and relatively more of a slowly reacting VOC will react in the
scenario. (If the compound reacts sufficiently slowly, the kinetic
reactivity is proportional to IntOH). The average IntOH for the maximum
reactivity scenarios (with each urban area weighed equally) is 82 ppt-min,
while the average IntOH for the maximum ozone scenarios is 128 ppt-min.
Thus on the average, a slowly reacting compound should have an
approximately 56$ higher kinetic reactivity under maximum ozone conditions
than it does under maximum reactivity conditions.
b. Comparison of Mechanistic Reactivities
Since (by definition) incremental reactivities are higher
under maximum reactivity conditions despite the lower kinetic
reactivities it is clearly the mechanistic reactivities which are the
54
-------
more important in influencing how overall incremental reactivities vary
with VOC/NCL. This is consistent with the results which have been
A
presented previously (Carter and Atkinson, 1989). However, if the
relative differences between the mechanistic reactivities under the two
sets of conditions were the same for all types of VOCs, then the
differences would have no practical significance in terms of developing
VOC reactivity scales.
Figure 6 shows plots of mechanistic reactivities of the VOCs
in the MaxOo scale conditions against those in the MaxRct scale. If the
ratios of mechanistic reactivities for maximum ozone and maximum
reactivity conditions were similar for all compounds, the points would all
be on approximately the same line, whose slope is the ratio of mechanistic
reactivities for the two sets of conditions. However, it can be seen that
although compounds of similar types tend to fall on the same line, the
lines are different for different classes of compounds. For example, the
ratio of mechanistic reactivities for maximum ozone relative to maximum
reactivity conditions is significantly lower for the carbonyl compounds
and the aromatics (especially naphthalene and tetralin), and somewhat
lower for the alkenes, than it is for the alkanes. Clearly, differences
in mechanistic reactivities for these two sets of conditions are not the
same for all VOCs.
c. Comparison of Incremental Reactivities
It is the overall incremental reactivity, not the
mechanistic reactivity, which is of practical interest in VOC reactivity
scales. For the most slowly reacting compounds, the higher radical levels
in maximum ozone conditions relative to maximum reactivity conditions in
part offsets the opposing effects of differences in mechanistic
reactivity. For example, despite the factor of 2 difference in
mechanistic reactivity, the incremental reactivity of methane in maximum
ozone conditions is only -30% lower than its reactivity in maximum
reactivity conditions. On the other hand, for rapidly reacting compounds
there is relatively little difference between kinetic reactivities (since
the compounds are almost completely reacted in any case), and thus the
differences in incremental reactivity will be determined entirely by
differences in mechanistic reactivities.
55
-------
-t->
'>
'^j
cj
D
0)
C£
u
-t-1
en
'E
0
u
eu
ro
0
X
D
^>
"D
0)
CJ
O
CD
Q^
CJ
O
U
E
CL
Q_
\
ro
O
E
CL
Q.
2.On
1.5-
1.0-
0.5-
0.0
-.5-
o Alkanes
D Alkenes
A Aromatics
A Acetylene
v Aldehydes
Ketones
Alcohols
Aromatic Aldehydes
CO
Alkanes
Aromatics
-.5 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0
MaxRct Mechanistic Reactivity
(ppm OT, / ppmC VOC Reacted)
Figure 6. Plots of mechanistic reactivities for the VOCs in the MaxOo
scale against their mechanistic reactivities in the MaxRct
scale. The lines show the least-squares fits, forced through
zero, to the results for the alkanes or the aromatics.
56
-------
Comparisons of incremental reactivities for these two scales
are shown in Figure 7, which gives plots of MaxRct incremental
reactivities against those for the MaxOo scale. The incremental
reactivities for CO, the alkanes, alcohols, ethers, glycols, and the
ketones have been multiplied by various factors so their magnitudes would
be comparable to those for the aromatics, alkenes, and aldehydes. Figure
7 in general has the same features as Figure 6: Under maximum ozone
conditions the reactivities of the aromatics, oxygenates, and alkenes,
relative to those for the alkanes and alcohols, are lower under maximum
ozone conditions than they are under maximum reactivity conditions. This
shows the importance in mechanistic reactivities in determining the
dependence of incremental reactivities on relative NOV levels. In
A
addition, the differences between the incremental reactivities of the
alkanes relative to those of the alkenes (as shown on Figure 7) are
greater than one would expect from their differences in relative
mechanistic reactivities as shown in Figure 6. This is because the higher
radical levels of the maximum ozone conditions tend to decrease the
differences in the reactivities for the relatively more slowly reacting
alkanes to a greater extent than they do for the relatively rapidly
reacting alkenes. This effect on alkane/alkene reactivity ratios is in
the same direction as the effect on the ratio of their mechanistic
reactivity, and it thus enhances the differences in their relative
reactivities between the two scales.
2. Base Case Relative Reactivity Scales
As discussed in Section II.D.1, (and illustrated in Figure 1)
four multi-scenario relative reactivity scales were derived from the
incremental reactivities calculated for the representative base case
scenarios. These four scales, and their derivations, are summarized as
follows (see also Figure 1).
Ozone yield, average ratio scale: Relative
reactivities of the VOCs in the scale are averages
of ratios of ozone-yield incremental reactivities of
the VOCs, relative to ozone-yield incremental
reactivities of the base ROG mixture, for each of
the 21 representative city-days.
57
-------
(J
o
en
O
E
en
0
D
0)
cr
ro
O
x
o
6-
5-
4-
3-
2-
1-
0-
-1
o Alkanes X 5
D Alkenes
A Aromatics
A Acetylene X 1 0
o Alcohols X 5
v Aldehydes
x
T
Alkanes
Ketones X 5
Aromatic Oxygenates
CO X 100
Aromatics
01 2345
MaxRct Reactivity (gm
gm VOC)
Figure 7- Plots of incremental reactivities for the VOCs in the MaxOo
scale against their incremental reactivities in the MaxRct
scale. The lines show the least-squares fits, forced through
zero, to the results for the alkanes or the aromatics.
58
-------
Ozone yield, least squares fit scale: Relative
reactivities of the VOCs are determined to minimize
the least squares error in predicted absolute ozone
change caused by adding a given (small) amount of
the VOC to the emissions in each of the 21
representative scenarios. As discussed in Section
II.D.I.b, this can be determined by plotting, for
each scenario, the incremental reactivity of the VOC
against that of the base ROG mixture, and taking the
slope of the least squares line, forced through
zero, as the multi-scenario relative reactivity.
Integrated ozone, average ratio scale: Relative
reactivities are averages of ratios of integrated
ozone incremental reactivities of the VOCs to
integrated ozone incremental reactivities of the
base ROG mixture for the scenarios.
Integrated ozone, least squares fit scale: Relative
reactivities are derived as indicated above for the
ozone yield, least squares fit scale, except that
integrated ozone incremental reactivities are
used.
Table 4 lists, for representative VOCs, the calculated relative
reactivities for these four scales. For comparison purposes, it also
gives the relative reactivities (VOC reactivity/base ROG reactivity)
derived from the MaxRct and MaxOo scales. The quantities in the
parentheses are the (one-sigma) standard deviations of the averages (for
the average ratio scales) or the standard deviations of the slopes (for
the least squares fit scales) and thus indicate the degree of scatter of
relative reactivities in the 21 city-days.
Examples of results of incremental reactivity calculations for
the individual representative city-days (base case scenarios) and for four
representative VOCs are shown on Figure 8. This figure consists of two
plots for each VOC those on the left-hand side are plots of ozone yield
reactivities, and those on the right are plots of integrated ozone
reactivities. Each point on the plots is the result of reactivity
59
-------
Table 4,
Summary of Relative Reactivities for Selected VOC Species in the MaxRct, MaxOo, and the
Base Case Relative Reactivity Scales. Standard deviations of average ratios or least
squares fits for the various scenarios are also shown.
VOC ID
[a]
CO
METHANE
ETHANE
PROPANE
N-C4
N-C6
N-C8
ISO-C8
N-C12
BR-C12
N-C15
ETHENE
PROPENE
T-2-BUTE
ISOBUTEN
1-HEXENE
C10-OLE1
C10-OLE2
BENZENE
TOLUENE
M-XYLENE
135-TMB
TETRALIN
NAPHTHAL
Tnf
MaxRct
0.034
0.0052
0.065
0.143
0.27
0.26
0. 17
0.29
0.097
0.34
0.078
2.2
2.8
3.0
1.7
1.25
0.53
1.15
0.108
0.71
2.3
2.9
0.28
0.32
jremental
MaxOo
0.070
0.0090
0.108
0.25
0.45
0.47
0.29
0.42
0.16
0.50
0.128
2.8
3.0
2.9
1.8
1.20
0.49
1.03
0.084
0.38
1.8
2.4
0.107
0.039
Reactivity (Per Carbon) Relative to Base Case ROG Mixture fb.
Base Case Ozone Yield - Base Case Integrated (
Avg. Ratio Line Fit Avg. Ratio Line
0.076
0.0094
0.116
0.25
0.45
0.43
0.18
0.35
0.026
0.46
0.0077
3.0
3.3
3.1
2.2
1.10
0.30
0.96
0.080
0.21
1.7
2.3
0.0021
-0.129
(63%)
(53%)
(4870
(457.)
(4070
(437.)
(0.25)
(29*)
(0.33)
(407.)
(0.31)
(3570
(247.)
(1770
(547.)
(427.)
(0.55)
(527.)
(647.)
(0.60)
(427.)
(317.)
(0.50)
(0.64)
0.043
0.0060
0.078
0.17
0.32
0.30
0.18
0.30
0.093
0.37
0.074
2.4
2.9
3.1
1.8
1.26
0.51
1.17
0.098
0.58
2.2
2.8
0.21
0.21
(1170
( 970
(1070
( 870
( 870
( 870
( 870
( 570
(1370
( 670
(1570
( 470
( 270
( 170
( 370
( 270
( 570
( 370
( 570
( 970
( 370
( 470
(1570
(2270
0.050
0.0061
0.072
0.17
0.29
0.29
0.120
0.24
-0.0020
0.33
-0.022
2.4
2.9
3.7
2.4
0.92
0.19
1.05
0.086
0.45
2.1
3.1
0.20
0.122
(4770
(3870
(3170
(3470
(2870
(3470
(7570
(1570
(0.149)
(2670
(0.15)
(2570
(1470
(2370
(3770
(4570
(0.44)
(2570
(20/0
(5070
(1870
(1270
(7270
(0.22)
0.036
0.0048
0.058
0.135
0.24
0.24
0. 140
0.24
0.062
0.30
0.043
2.2
2.7
3.3
2.0
1.13
0.41
1.17
0.090
0.59
2.3
3.2
0.27
0.25
1
j __
Dzone -
Fit
( 570
( 470
( 470
( 470
( 470
( 4%)
( 670
( 2%)
(1570
( 370
(2170
( 2%)
( U)
( W
( 2%)
( 370
( 670
( 370
( 470
( 470
( 170
( 270
( 670
( 870
[continued)
-------
Table 4. (continued) - 2
VOC ID
[a]
ACETYLEN
MEOH
ETOH
MTBE
FORMALD
ACETALD
ACETONE
MEK
CRESOL
BENZALD
Tr
MaxRct
0.140
0.37
0.53
0.23
5.6
2.5
0.22
0.41
0.73
-0.24
icremental Reactivity (Per
MaxOq Base Case
Avg. Ratio
0.26
0.53
0.67
0.43
4.5
2.5
0.20
0.42
-0.84
-1.06
0.27
0.53
0.67
0.48
4.9
2.7
0.19
0.41
-3.0
-2.0
(557=)
(3650
(357.)
(5870
(3970
(6770
(1770
(3370
(6.0)
(2.4)
Carbon) Relative to Base Case ROG Mixture [b
Ozone Yield - Base Case Integrated
Line Fit Avg. Ratio Line
0.17
0.40
0.57
0.28
5.2
2.6
0.21
0.41
0.082
-0.64
( 9%)
( 570
( 770
(1070
( 570
( 470
( 270
( 470
(0.32)
(25%)
0.18
0.38
0.44
0.32
7.3
1.9
0.17
0.30
-1.09
-1.27
(4170
(2270
(2270
(4770
(2170
(5770
(1770
(2970
(3.2)
(1.27)
0.140
0.35
0.44
0.24
6.7
2.2
0.19
0.33
0.38
-0.54
Ozone -
Fit
( 570
( 370
( 370
( 570
( 270
( 370
( 270
( 270
(3570
(1470
[a] See Table 3 for VOC descriptions.
[b] Quantities in parentheses are standard deviations of the averages or least squares fits.
Given as percentage unless standard deviation is greater than 907«.
-------
_>*
'>
o
O
cr
o
o
0.75-,
0.45-
0.15-
-.15
Ozone Yield
Reactivity
n-Octane
0.2-i
0.1 -I
.0.0
Propene
Integrated Ozone
Reactivity
0.0 0.6 \.2 1.8 2.4 0.0 0.2 0.4 0.6 0.8
(gm 03 / gm VOC) (ppm-min 03 / gm VOC)
Base ROG Reactivity
Figure 8. Plots of incremental reactivities of selected VOCs against
the reactivity of the base case ROG mixture in the same
scenario for each representative base case scenario.
* = VOC and base ROG reactivities in a scenario.
= Slope is MaxRct VOC/Base ROG reactivity ratio.
= Slope is MaxOo VOC/Base ROG reactivity ratio.
- - - - = Least-squares fit, forced through zero. (Slope is
least-squares fit relative reactivity.)
= Slope is average of ratios (average ratio relative
reactivity)
62
-------
calculations for one city-day, the x-value being the base ROG incremental
reactivity, and the y-value being the incremental reactivity of the VOC
(n-octane, propene, methanol, or toluene). Each line on the plots gives
the ratio of the VOC reactivity to the base ROG reactivity, and each
corresponds to a given reactivity scale. Thus each line starts at the
origin and has a slope set to the relative reactivities in one of the
general or multi-scenario reactivity scales. Each plot has four lines,
one each for the MaxRct, the MaxO,, the average ratio base case, or the
least squares fit base case reactivity scale. (Note that the MaxRct and
the MaxOn lines are the same on both plots for each VOC, while the average
ratio and the least-squares fit base case lines in general are different
for ozone yield or integrated ozone reactivities.) The differences in the
slopes of the lines thus show differences among the various scales in the
relative reactivities of the VOCs, and distances between the lines and the
points show how well (or poorly) the scales predict relative reactivities
for the individual city-days. Note that the line whose slope gives the
least squares fit base case relative reactivity is the same as the least
squares line, forced through zero, fitting the points for the individual
city-days.
If the relative reactivity of the VOC (the incremental
reactivity of the VOC divided by that of the base ROG) were nearly the
same for all scenarios, then all the points in the plots of Figure 8 would
fall very close to the same line. This is the case for propene,
especially for integrated ozone reactivities. Significant scatter in the
points means the relative reactivity of the VOC is highly variable from
scenario to scenario. This is the case for the ozone yield reactivities
of toluene, octane, and (to a somewhat lesser extent) methanol. The
magnitude of this scatter is also indicated by the standard deviation
quantities given in the tabulations of the base case relative reactivity
results in Table 4. It is interesting to note that the scatter tends to
be much less with integrated ozone relative reactivities than it is for
ozone yield relative reactivities, and that the former tend to correspond
reasonably well with MaxRct reactivities. This is discussed further
below.
63
-------
a. Effect of Derivation Method on Ozone Yield Reactivities
The data in Figure 8 and Table 4 indicate that for many
compounds the method of deriving the single reactivity scale from the
varying reactivities of the individual scenarios can have a significant
effect on the base case relative ozone yield reactivities. This is due to
variability of relative ozone yield reactivities from scenario to
scenario. If the scenarios with higher incremental reactivities tend to
have different reactivity ratios than those with lower incremental
reactivity for a given VOC, different relative reactivities would be
derived by each method.
As discussed above, the major reason for this variability in
relative reactivities is the variability of the NOX conditions (i.e., the
ROG/NOX ratio) for these scenarios. This causes not only differences in
absolute incremental reactivities, but in many cases differences in ratios
of reactivities as well. (Differences between relative reactivities in
the higher-NOv maximum reactivity conditions and relative reactivities in
A
lower-NOv maximum ozone conditions have already been discussed in Section
A
III.A.I.b. As discussed later in this report, the differences become even
greater if NO -limited scenarios are included [see also Carter and
Atkinson, 1989].) The scale derived by using the least squares fit is
most sensitive to the higher NOX scenarios which have higher VOC
reactivities, while the scale derived by using the line fit method weighs
results for each scenario equally, including scenarios where ozone is NOX
limited.
b. Comparison of Ozone Yield Reactivities with MaxRct and MaxOQ
Figure 9 shows comparisons of the ozone yield base case
reactivities with the MaxRct and MaxO, scales for the individual VOCs. As
in Figure 7, each point on the plots gives reactivities for a given VOC,
the y-value being the base case ozone-yield relative reactivity derived by
using either the average ratio or the least squares method, and the x-
value being either the MaxRct or MaxO? reactivity. (The four plots on the
figure give the four possible combinations of reactivity scales.) Also as
in Figure 7, reactivity values for selected classes of compounds have been
multiplied by constant factors to yield reactivities of comparable
magnitudes, for easier comparison of relative results. The lines on the
plots show least squares fits, forced through zero, to the data for the
-------
individual VOCs. These lines are included only to aid in showing how well
(or poorly) the reactivities in the various scales correspond to each
other. If the correspondences were perfect, all points would lie on the
same line.
Figure 9 shows that the base case ozone yield scale derived
by the least squares fit method corresponds reasonably well to the MaxRct
scale, while the base case scale derived by the average ratio method
corresponds better to the MaxO^ reactivities. This is expected, since the
least squares derivation method is most sensitive to scenarios near
maximum reactivity conditions. On the other hand, the average ROG/NOX
ratios for all the scenarios are closer to those for maximum ozone
conditions (see Table 1), so the average ratio of reactivities, with all
scenarios weighed equally, would be expected to be closer to maximum ozone
reactivities. Note, however, that there are fewer cases of systematic
differences among classes of compounds in the comparison between MaxRct
reactivities and least-squares fit base case reactivities than there are
in the comparison between MaxOo reactivities and the average ratio
values.
c. Integrated Ozone Reactivities
The data in Figure 8 and Table 4 show that relative
integrated ozone reactivities tend to be less sensitive to the conditions
of the base case scenarios than is the case for ozone yield
reactivities. This is indicated by the relatively low standard deviations
of the average ratios or slopes of fitted least squares lines given in
Table 4 for the integrated ozone multi-scenario reactivities compared to
those for ozone yield reactivities. It is also indicated by the fact that
the integrated ozone reactivities are less sensitive to which derivation
method (average ratio or least squares fit) is employed. There are still
cases in which relative reactivities vary with scenario, particularly for
the compounds with low or negative mechanistic reactivity. Such compounds
tend to have the most variable reactivities no matter what types of
scenarios are examined. However, even for those compounds the scenario-
to-scenario variability of integrated ozone reactivities is much less than
is the case for ozone yield reactivities.
65
-------
o
o
o
CD
AJkones X 5
Alkenes
Aromatics
Acetylene X 10
Alcohols. Ethers
Aldehydes
Ketones X 5
CO X 100
Boae Cose ROC
0 " 2 4 6 8
MoxRct Reactivity (gm Oj / gm VOC)
12345
MaxOj Reactivity (gm Oj / gm VOC)
Figure 9. Plots of base case relative ozone yield reactivities for the
VOCs against their MaxRct or MaxCh reactivities. The lines
show least-squares fits, forced through zero, to the points
for all VOCs.
66
-------
d. Comparison of Integrated Ozone Reactivities with MaxRct and
MaxOo
Figure 10 shows comparisons of the base case integrated
ozone reactivities with those in the MaxRct and MaxO? scales for the
individual VOCs. (Its format is exactly analogous to that of Figure 9,
discussed above.) Note that the MaxRct and MaxOo scales are derived from
ozone yield reactivities, so these plots are comparing integrated ozone
with ozone yield reactivities. These plots show that the base case
integrated ozone reactivities derived by using either the average ratio or
the least-squares fit method correspond reasonably well to the MaxRct
scale, though the correspondence is better for the least squares fit
reactivities. On the other hand, integrated ozone reactivities correspond
rather poorly with the MaxO, scale. This is despite the fact that for
most of the base case scenarios the NO levels are closer to those of
maximum ozone conditions than those yielding maximum incremental
reactivities.
These results suggest that the mechanistic factors that
influence VOC reactivities under maximum reactivity conditions are similar
to those which influence integrated ozone reactivities under a much wider
set of conditions. The major characteristic of maximum reactivity
conditions is that ozone is not NOX limited, and thus the final ozone
yields are determined primarily by how rapidly ozone is formed. Aspects
of the VOC reaction mechanisms which affect ozone formation rates, such as
those that involve initiating or inhibiting radical levels, therefore are
the major factors affecting reactivity under those conditions (Carter and
Atkinson, 1989). At lower NOX levels the availability of NOX becomes a
more important factor affecting ozone yields, and at sufficiently low NOX
the rate of ozone formation is not a major factor in affecting how much is
ultimately formed. However, the ozone formation rate will affect the
integrated ozone concentration under practically any set of conditions.
Even in scenarios where ozone is ultimately WOV limited, VOCs causing more
A
rapid ozone formation will result in earlier formation of ozone, and thus
higher integrated ozone levels. Thus the good correspondence between
relative integrated ozone reactivities and maximum reactivities is not
unexpected.
67
-------
0
HI
a:
o
o
o
C"
o
CD
D
C
2-
Alkones X 5
Alkenea
Aromotics
Acetylene X 10
Alcohols, Ethers
Aldehydes
Ketones X 5
CO X 100
Boae Cose ROC
2468
kdaxRct Reactivity (gm Oj / gm VOC)
2-
5-
012345
MaxOj Reactivity (gm 03 / gm VOC)
Figure 10. Plots of base case relative integrated ozone reactivities for
the VOCs against their MaxRct or MaxO^ reactivities. The
lines show least-squares fits, forced through zero, to the
points for all VOCs.
68
-------
3. Comparison of Incremental Reactivities with the OH Radical Rate
Constant Scale
For many VOCs reaction with OH radicals is the major process
affecting the rate at which they react in the atmosphere, and thus the
rate at which they can affect ozone formation. Because of this, it has
been suggested that this rate constant can serve as a basis for a VOC
reactivity scale (e.g., Darnall et al. 1976, CARB, 1989). This scale has
the obvious disadvantage that it ignores other aspects of VOC reaction
mechanisms which affect ozone formation, and it also does not take into
account the fact that if the compound reacts sufficiently rapidly, the
amount of VOC which reacts in the atmosphere, and thus contributes to
ozone formation, becomes independent of the rate at which it reacts.
Since incremental reactivities take all of these factors into account, it
can be argued that they form a more justifiable basis for deriving ozone
reactivity scales for VOCs. However, the OH rate constant scale has the
significant advantages that (1) OH radical rate constants are known or can
be estimated with a reasonable degree of reliability for most of the types
of VOCs emitted into the atmosphere (Atkinson, 1989) and (2) it can be
universally applied under all conditions. Thus if it can be shown to give
acceptable correlations with reactivities quantified by more comprehensive
methods, its use might be appropriate for some types of applications.
To show how well an OH radical reactivity scale would correlate
with one based on incremental reactivities, Figure 11 shows plots of
T=300 K OH radical rate constants (kOH) against incremental reactivities
in the MaxRct scale for all the VOCs represented in the model. Since the
incremental reactivities are in units of ozone formed per mass of VOC, the
OH radical rate constants are divided by their molecular weight to place
them on a comparable basis. Plots of OH radical rate constants against
MaxOo reactivities or the base case relative reactivity scales look
similar to Figure 11, and are not shown.
Figure 11 shows that although there is a correlation between kOH
and incremental reactivities for certain homologous classes of compounds,
such as the alkenes and some of the aromatics, the correlation between
different types of VOCs is extremely poor. In addition, there is a
negative correlation between kOH and incremental reactivity for the
homologous series of n-alkanes. Thus, although there are cases where the
69
-------
o
E
\
E
en
-4 '
5
TS
^b
>w
\
^
7
c
'E
T
F
t
Q.
Q.
T~
J_
O
*^s
2500-,
2000-
1500-
1000-
500-
-
o
a
A
A
O
7
Z
f
1
1
Alkones X 5
Alkenes
Aromatics
Acetylene X 10
Alcohols X 5 D a
Aldehydes a
Ketones X 5
Aromatic Oxygenates °
CO X 100 D »D °
D
Q
O
°° * D
D D 00 0 ^0 00 Oo
a°° tS>DD 0^> B 4, "
r^>^° .> v* =- .
* _n° 00
jF1 o o o . A *
"* o «0 / * ^
u A o o^^A \
~r r 1 i i i i i i i
01 23456789
MaxRct Reactivity (gm 03 / gm VOC)
Figure 11. Plots of OH radical rate constants (per mass basis) for the
VOCs against their MaxRct reactivities.
70
-------
OH radical rate constant correlates reasonably well with reactivity, in
general it does not appear to be a reliable predictor of the relative
effect of the emissions of a VOC on ozone formation in the atmosphere.
B. Dependence of Reactivities on Scenario Conditions
The main problem with the development of general or multi-scenario
reactivity scales is the dependence of VOC reactivities on scenario
conditions. Methods of developing single scales despite this problem are
discussed in the previous sections. However, regardless of how the single
scale is developed, the level of uncertainty of the scale, and thus its
potential for utility as an ozone control assessment tool, is influenced
by the sensitivity of its reactivities to scenario conditions. The
sensitivity of incremental reactivities to relative NOX levels (usually
measured by the ROG/NOV ratio) has been studied previously (e.g., see
A
Dodge, 1984; Carter and Atkinson, 1989) and is discussed extensively
above, and some additional information in this regard is discussed
below. However, other than some investigations of the effect of dilution
on reactivities (Chang and Rudy, 1990; Carter and Atkinson, 1989), and
some calculations showing differences in incremental reactivities when
different base case ROG mixtures are used (Carter and Atkinson, 1989; Weir
et al. 1988), there has been less information about the sensitivities of
reactivities to other, non-NO -related, scenario conditions. These are
A
also discussed in this section.
1. Dependence of Reactivity Scales on NOX
As indicated above, the dependence of incremental reactivities
on NCL has been investigated previously, and further information
X
concerning this is obtained in this work by comparing incremental
reactivities in the high-NOx MaxRct with the lower-NOx MaxO^ scale (see
Section III.A.I.b). However, to investigate this further, additional
calculations were carried out in this study to investigate (1) whether it
is possible to derive a third type of general scale for reactivities in
NO -limited conditions. (2) how N0v-limited VOC reactivities compare with
A *»
MaxOo reactivities, and (3) whether integrated ozone reactivities are as
sensitive to NOV conditions as are ozone yield reactivities. For the
X
purpose of this investigation, the NOX inputs (initial + emitted NOX) were
varied systematically for the "averaged conditions" scenarios discussed in
71
-------
Section II.A.S.c. The results of these calculations are discussed
below.
a. N0x-Limited Relative Reactivities
The MaxOo scale represents conditions which are just on the
borderline between the situation where ozone is ROG-sensitive and the
situation where ozone is limited by the availability of NOX (i.e., NOX-
limited conditions). The MaxRct scale represents conditions which are on
the high-NOx side of this "borderline", but reactivities for the low-NOx
side have not been considered thus far in this work. This is because it
is considered that VOC reactivities for NO -limited conditions are not
X
particularly relevant to ozone control strategies, because ozone is not
sensitive to VOC controls under such conditions. However, a number of the
base case scenarios listed on Table 1 clearly represent N0x-limited
conditions, because their ROG/NOX ratios are higher than those yielding
maximum ozone. Therefore, it is of interest to determine if it is
possible to derive a general VOC reactivity scale for N0x-limited
conditions, and, if so, how relative reactivities in this scale differ
from those in the "borderline NOX availability" MaxOo scale.
Other than its probable lack of relevance to ozone control
strategies, the main problem with deriving a general N0x-limited
reactivity scale is that there is not a unique condition of NOX
availability that defines an NO -limited scenario. "NO -limited" could be
A A
any level of NOX input which is less than that yielding maximum ozone
i.e., where decreasing NOX decreases ozone. This is in contrast with NOX
conditions of "maximum reactivity" or "maximum ozone", which each
correspond to a unique level of NOX input if all other scenario conditions
are specified. This means that it is not really possible to define a
"general" N0v-limited reactivity scale in a manner analogous to the
X
derivation of the MaxRct or MaxOo scales some arbitrary level of NOX
input would need to be specified. However, if it can be shown that ratios
of incremental reactivities (e.g., relative reactivities) are insensitive
to NOV inputs when restricted to the NO -limited regime, then this degree
X A
of arbitrariness would not be significant, and the concept of a general
NOY-limited reactivity scale might be meaningful.
A
72
-------
h-
o
tr
_j
<
cr
CJ
I cr
< uj
s ^
cr m
LJ \
II
LJ O
0.2'
0.16-
0.08-
0.00
I.On
0.8-
0.6-
0.4-
0.2-
0.0-
0.0-
LU
2
LJ
CC
CJ
CJ
O
-0.4-
-0.8
OZONE YIELD REACTIVITIES
INTEGRATED OZONE REACTIVITIES
CARBON MONOXIDE
N-BUTANE
0-4 N-OCTANE
-I 0 ETHANOL
0.8-
0.6-
0.4-
0.2-
0.0--
8
12 16 20 24
ETHENE
g MaxRct
6-
4-
2-
0
PROPENE
TOLUENE
ROG / NOX RATIO
Figure 12. Plots of relative ozone yield reactivities and relative
integrated ozone reactivities of selected VOCs against the
ROG/NOX ratio in the averaged conditions scenarios.
73
-------
To investigate this, incremental reactivities were
calculated for selected representative VOCs, and for the base case ROG
mixture, for the "averaged conditions" scenario with varying NOV inputs.
A
Relative reactivities were then determined by ratioing the incremental
reactivities of the VOCs to those of the base ROG mixture. The results
are shown on Figure 12 (solid lines), which gives plots of relative
reactivities of the representative VOCs against the ROG/NOV ratio. The
A
ratios corresponding to maximum reactivity and maximum ozone conditions
are indicated on the plots.
The plots in Figure 12 show that relative reactivities for
N0x-limited conditions can be significantly different than those for
maximum ozone conditions. In addition, it is also clear that there is no
unambiguous set of relative reactivities that corresponds to all
conditions that can be characterized as being NO limited. For example,
the incremental reactivities of toluene and n-octane, relative to that of
the base ROG mixture, continually decrease, and the relative reactivities
of CO, n-butane, ethene, and propene continually increase, as the NO
inputs are decreased under N0,.-limited conditions (i.e., as NOV inputs are
A A
decreased below that of maximum ozone.) Although there are some compounds
(e.g., ethanol) where relative reactivities do not change significantly as
NO varies under NO -limited conditions, it is clear that for a majority
of compounds this is not the case. Therefore, it is concluded that (1)
ratios of reactivities in the MaxOo scale are not necessarily good
approximations of those under lower NOX conditions, and (2) there is
really no such thing as a general reactivity scale for N0x-limited
conditions.
b. Effect of NO^ on Integrated Ozone Reactivities
- ' ~ A
The discussion in the above section concerned ozone yield
reactivities. One finding from our calculations of reactivities in the
base case scenarios is that relative integrated ozone reactivities tend to
be much less variable than relative ozone yield reactivities. This
explained by the fact that integrated ozone reactivities are expected to
be sensitive to the VOC's effect on rates of ozone formation under all
conditions, while ozone yield reactivities are sensitive to this only
under relatively high NOV conditions. To investigate the dependence of
A
integrated ozone reactivities on NOX conditions more systematically,
-------
relative integrated ozone reactivities were calculated for selected VOCs
as a function of NOX inputs for the "averaged conditions" scenarios. The
results are also shown on Figure 12 (as the dashed lines), where they can
be directly compared with relative ozone yield reactivities for the same
compounds.
Figure 12 shows that, as expected from the results of the
base case simulations, the integrated ozone reactivities are indeed much
less dependent on the NOX inputs than are the ozone yield reactivities.
However, it is also clear that they are not independent of NCL inputs, and
X
that the qualitative trends of the two measures of reactivity tend to
track one another as NOX varies. This is expected, since effects of VOCs
on maximum ozone concentrations would also be expected to influence, at
least to some extent, their effects on the integrated ozone levels.
However, in general, the dependence of integrated ozone reactivities on
NOX is relatively minor until NO inputs are reduced to levels where ozone
is NOX limited, and where changes in VOC emissions have only minor effects
on ozone formation.
It should also be noted from Figure 12 that the integrated
ozone and the ozone yield relative reactivities tend to be relatively
close to each other under maximum reactivity conditions. The two measures
of relative reactivity at maximum reactivity conditions agree within 20$
for ethanol and formaldehyde, and much more closely than that for the
other six representative compounds shown on the figure. This is
consistent with the result that relative base case integrated ozone
reactivities tend to correspond reasonably well with those in the MaxRct
scale. The two measures of reactivity tend to increasingly diverge from
each other as NO becomes more limited.
2. Effect of Variation of Non-NO^-Related Scenario Conditions
A ' -'- ..
A study of the dependence of reactivity on non-NO -related
A
airshed conditions is complicated by the fact that most airshed conditions
will affect the rate at which NOV is consumed during the day, which in
A
turn determines the level of NO which defines the boundary between NO -
A A
limited and VOC-sensitive conditions. For example, increasing the light
intensity or the amount of radical initiating species in the base case ROG
mixture increases radical levels in the scenario, which increases NO.,
X
consumption rate, and thus results in NO -limiting conditions occurring at
75
-------
higher NOV levels. Thus the main reason for differences in reactivities
A
calculated for different scenarios at the same ROG/NOV ratio is that the
X
given ratio can correspond to different conditions of relative NOV
X
availability in the different scenarios. Holding the ROG/NOX ratio fixed
therefore does not ensure consistent conditions of NOV availability.
X
However, the derivation of separate reactivity scales for
maximum reactivity or for maximum ozone conditions has the advantage that
it tends to factor out, at least to some extent, effects on reactivities
of variability of conditions of NOX availability. This is because in the
derivation of these scales the NOX inputs of each scenario are adjusted to
yield consistent conditions of NOX availability. Therefore, if there is
still variability in reactivities calculated among the various scenarios
where conditions of relative NOX availability are held fixed, this would
be due to effects of other, non-NO -related airshed conditions.
A
As shown in Table 1, there is considerable variability among the
scenarios in terms of amounts of dilution, total VOC inputs, integrated
light intensity, amounts of pollutants entrained from aloft, and relative
amounts of pollutants present initially or emitted during the day. As
shown in the distribution plots of Figure 2, this causes variabilities in
the maximum and integrated ozone concentrations, the value of the IntOH
parameter, and the reactivity of the base ROG mixture, even for the
scenarios with consistent conditions of NOV availability. The
X
variabilities in kinetic, mechanistic, incremental, and relative
reactivities of representative VOCs in the N0x-adjusted scenarios,
measured as (one-sigma) percent standard deviations of averages, are
summarized on Table 5, and Figures 3-5 show examples of distribution
plots of these values. Although the scenario-to-scenario variability for
the NO -adjusted scenarios is less than for the unadjusted, base case
scenarios, it can be seen that reactivities are indeed affected by the
variability of the other, non-NOx-related, airshed conditions.
In general, at least for the scenarios and urban areas used in
this study, the urban-area-averaged measures of reactivity (other than
relative reactivities) tend to have standard deviations in the range of
±15 - ±30% (see Table 5), and the extent of variability for maximum
reactivity and maximum ozone conditions tends to be similar. The
exceptions include the mechanistic reactivities for the Cg+ alkanes under
all conditions, and the mechanistic reactivities of the aromatics (other
76
-------
Table 5. Standard Deviations of Averages of Kinetic (KR), Mechanistic
(MR), Incremental (IR), and Relative (RR) Reactivities of
Selected VOCs for the Maximum Reactivity and the Maximum Ozone
Scenarios, [a]
VOC ID
[b]
CO
METHANE
ETHANE
N-C4
N-C8
ISO-C8
N-C12
BR-C12
N-C15
ETHENE
PROPENE
T-2-BUTE
ISOBUTEN
1 -HEXENE
C10-OLE1
C10-OLE2
BENZENE
TOLUENE
M-XY.LENE
135-TMB
TETRALIN
NAPHTHAL
ACETYLEN
MEOH
ETOH
MTBE
FORMALD
ACETALD
ACETONE
MEK
CRESOL
BENZALD
Maximum
KR
19?
19?
19?
17?
13?
15?
10?
9?
8?
10?
6?
3?
3?
4?
4?
3?
18?
14?
7?
3?
5?
7?
19?
19?
16?
17?
5?
9?
15?
16?
0?
10?
Reactivity Scenarios
MR IR RR
8?
1%
18%
17$
23%
20%
40?
22?
46?
13?
16?
17?
13?
20?
23?
20?
11?
11?
13?
14?
16?
14?
8?
7?
20?
9?
13?
21?
10$
18?
16$
20?
23?
24?
30?
29?
31?
29?
45?
28?
50?
20?
20?
19?
15?
22?
25?
21?
24?
22?
17?
16?
18?
19?
24?
23?
30?
23?
16?
25?
20?
26?
16?
24?
11?
10?
13?
11?
17?
12?
32?
12$
37?
6?
2?
3?
6?
6?
10$
8$
7$
5?
4$
7$
7$
4$
9$
9$
12$
9$
10$
8$
3$
8$
10$
23$
Maximum Ozone Scenarios
KR MR IR RR
21?
21?
21$
18?
11?
15?
8?
7$
6?
9?
4?
2?
2?
3$
3?
2?
19?
13?
5?
2?
3?
5$
20?
20?
17?
17?
3$
7?
18?
18?
0$
8$
10?
8$
19$
18?
23$
25?
46$
27$
54?
12?
20?
21?
11?
29$
39$
29$
45$
41$
20$
17?
75?
255$
8?
10$
29$
7$
15?
26$
18$
25$
33$
18$
22?
21?
35?
31?
31$
35$
50?
30?
57?
17?
22$
21$
12$
30?
40$
29?
56?
47?
21$
17$
76$
252?
22?
19?
41?
19?
16?
29?
22?
36?
33?
21?
18?
13$
16$
14$
22$
15$
37$
13?
43?
9?
2?
5?
13$
9$
18$
9$
39?
29?
8$
12$
62?
314$
13$
10?
20$
13$
20$
8?
5?
15?
51$
17$
[a] Standard deviations of averages of averages for each of the 12
urban areas. Each urban area weighed equally.
[b] See Table 3 for VOC descriptions.
77
-------
than di- or tri-alkylbenzenes) under maximum ozone conditions. These tend
to be low in magnitude and have high relative variabilities. In addition,
the higher molecular weight alkenes have higher variabilities in
mechanistic reactivities (with standard deviations on the order of ±30%)
under maximum ozone conditions.
The variability in the IntOH parameter, which reflects
integrated OH radical levels and thus fractions of VOCs which react, is
comparable to the variabilities in the mechanistic reactivities. Its
variability yields a corresponding variability in the kinetic reactivities
of the slowly reacting compounds, since from Equation (V) fractions
reacted become proportional to IntOH when the OH radical rate constant
becomes sufficiently small. As shown in Table 5, the variability in
kinetic reactivities is significantly less for the more rapidly reacting
compounds (including most of those which react with species other than OH
radicals), since the fraction reacted approaches the upper limit value of
unity. Thus for most reactive compounds, the variability of mechanistic
reactivity with airshed conditions is the more important factor that is
influencing variability, and thus uncertainties, of incremental
reactivities for maximum ozone or maximum reactivity conditions.
Table 5 also shows that this variability of reactivities in the
N0x-adjusted scenarios tends to be less when considering ratios of
incremental reactivities, such as relative reactivities. This is despite
the fact that relative reactivities in the non-NOv-adjusted, base case
A
scenarios can be quite variable (as shown, for example, in Figures 3 -
5). The relative reactivities also tend to be less variable under maximum
reactivity conditions than under maximum ozone conditions. As shown in
Table 5, the standard deviations of averages of relative reactivity in the
maximum reactivity scenarios are 12% or less for all VOCs except for high
molecular weight alkanes and special compounds such as benzaldehyde. The
percentage deviations tend to be higher for the maximum ozone scenarios,
especially for the aromatics. Note that the higher variability of
reactivities of aromatics under maximum ozone conditions causes higher
variability of relative reactivities of all VOCs under those conditions;
this is because of the contribution of aromatics to the reactivity of the
base case ROG surrogate, which is used as the standard for calculating
relative reactivities.
78
-------
These results indicate the approximate magnitude of the
sensitivity of the maximum reactivity or maximum ozone reactivity scales
to the scenario conditions which were varied in this study. It should be
recognized, however, that although they represent a wide range of
conditions, not all potentially important aspects of airshed scenarios
were varied. For example, they are all single box model scenarios, and it
is not clear to what extent reactivities may differ in more complex,
gridded airshed scenarios are used. Perhaps more significantly in terms
of direct chemical effects on reactivity, the same mixture of VOCs was
used to represent base case ROG emissions for all scenarios. The effects
on reactivity of varying the base case ROG are discussed in the following
section.
3. Effect of Variation of the Base ROG Mixture
As discussed in Section II.A.2, the "all-city average" mixture
derived by Jeffries et al. (1989) was used to represent base case ROG
emissions in the calculation of all the reactivity scales discussed in
this work. This was used because it represents our current "best
estimate" as to actual VOC emissions in urban areas in the United States,
and because data are insufficient to unambiguously derive separate
mixtures for each urban area or city-day. However, the aldehyde fraction
of this mixture is highly uncertain, and the composition of ROGs in
emissions inventories is significantly different from those derived from
analysis of air quality data. We have found that using emissions-derived
ROG mixtures can result in non-negligible differences in calculated
reactivities (Weir et al., 1988). These uncertainties are not reflected
in the variabilities in the reactivities calculated in this study.
To investigate the effects of uncertainties in the base ROG
composition on calculated reactivity scales, the composition of the base
case ROG mixture was systematically varied, and the effects of this
variation of incremental reactivities of selected VOCs under maximum
reactivity and maximum ozone conditions were determined. In all cases,
incremental reactivities were calculated for averaged conditions scenarios
under both maximum reactivity and maximum ozone conditions. Since, in
general, varying the composition of the ROG mixture will affect its
reactivity and therefore the ROG/NOV ratio at which maximum reactivity and
A
maximum ozone conditions occur, NOX inputs yielding maximum "base" ROG
79
-------
reactivity (i.e., maximum reactivity of the modified base ROG mixture) and
yielding maximum peak ozone were determined separately for each modified
ROG mixture. (If the ROG/NOX ratio were held constant in the comparison,
the effect of the change of the ROG on effective rates of NOX removal
would dominate all other effects, and the exercise would be equivalent to
evaluating the effects of varying effective NOX inputs. For example, if a
modification of the ROG mixture caused its reactivity to double, the
modification might be considered to be equivalent to doubling the
effective ROG/NOX ratio.) Adjusting NOX inputs to yield consistent
conditions of NOX availability for the different mixtures allows other,
perhaps less obvious, effects of variations of the base ROG mixture to be
examined.
The variations of the base ROG mixture which were examined in
this study are listed below. Note that the all-city average mixture given
in Table 2 was used as the starting point, and the composition of the
"aloft" mixture was not modified.
Variable Aldehydes. The aldehyde fraction was varied from 0% to
15% (as carbon) of the total mixture. (As shown in Table 2, the base ROG
mixture has 5% aldehydes.) The relative amounts of the individual
aldehydes (formaldehyde and acetaldehyde) within the aldehyde fraction,
and the relative amounts of the other VOCs within the non-aldehyde
fraction, were the same as in the standard mixture. When the aldehyde
fraction was varied, the inputs of each of the non-aldehyde species were
increased or decreased by an appropriate amount so the total number of
moles of ROG carbon was unchanged. The specific aldehyde fractions used
were 0%, 1%, 2%, 5% (the standard fraction), 1Q% and 15%, respectively.
The maximum reactivity and maximum ozone ROG/NOX ratio varied from 6.0 and
9.0 for the 0% aldehyde ROG mixture to 4.5 and 6.5 for the ROG mixture
with 15% aldehydes. (The lower maximum reactivity and maximum ozone
ROG/NO ratios for the higher aldehyde mixtures are consistent with the
X
fact that the reactivity, and thus the rate of NOX removal, will increase
with aldehyde levels.) The MaxRct reactivity of the 0% and the 15%
aldehyde mixtures was respectively 86% and 129? of that for the standard
mixture (on a per-carbon basis).
Low and High Aromatics. The standard mixture was modified by
cutting in half or doubling the total aromatic fraction of the mixture,
80
-------
and increasing or decreasing the alkanes by an amount such that the total
number of moles of ROG carbon was the same. The total amounts of
aldehydes and alkenes, and the relative amounts of individual aromatics
and alkanes within the aromatic or alkane fraction, were the same as in
the standard mixture. The low-aromatics mixture had a MaxRct reactivity
which was 14? lower than that of the standard mixture, while that of the
high-aromatics mixture was 29% higher. The maximum reactivity and maximum
ozone ROG/NOX ratios were respectively 6.0 and 8.0 for the low-aromatics
mixture, and 4.5 and 6.5 for the mixture with high aromatics.
Low and High Alkenes. The standard mixture was modified by
cutting in half or doubling the total alkene fraction of the mixture, and
increasing or decreasing the alkanes by an amount such that the total
number of moles of ROG carbon was the same. The total amounts of
aldehydes and aromatics, and the relative amounts of individual alkenes
and alkanes within the alkene or alkane fraction, were the same as in the
standard mixture. The low-alkenes mixture had a MaxRct reactivity which
was 13? lower than that of the standard mixture, while that of the high-
alkenes mixture was 26% higher - The maximum reactivity and maximum ozone
ROG/NO ratios were respectively 6.0 and 8.0 for the low-alkenes mixture,
and 4.5 and 6.5 for the mixture with high alkenes.
Figures 13 and 14 give plots showing effects of these changes on
incremental reactivities of selected VOCs for maximum reactivity and
maximum ozone conditions, respectively. The ratio of the MaxRct
reactivity of the mixture, relative to that of the unmodified base ROG, is
used on the x-axis to measure the extent of change of the mixture, as
indicated above. Increasing aldehydes, aromatics, or alkenes increases
the MaxRct reactivity of the overall mixture, and the extent of change of
the MaxRct reactivity provides a measure of the extent of change to the
mixture. However, note that the MaxRct reactivity of the base ROG is not,
by itself, a good predictor of its effect on VOC incremental
reactivities. The largest effect of those shown is clearly the effect of
changing the aldehyde fraction on formaldehyde reactivity particularly
when the aldehyde fraction is reduced below the default value of 5%.
Other than this, it is hard to make any generalizations concerning the
effects of changing the ROG mixture on VOC reactivity the qualitative
effects seem to vary considerably from case to case. On the other hand,
81
-------
FORMALDEHYDE
6-
4-
2-
0.15-1
0.10-
0.05-
N-OCTANE
0.00
o ALDEHYDES VARIED
o ALKENES VARIED
A AROMATICS VARIED
,
^
pi
CJ
LJ
or
CJ
0
^>
L?
O
CJ
~o
E
\
o
o
E
i i . i i
0.25-,
0.20-
0.15-
0.10-
0.05-
Onn-
N-BUTANE PROPENE
2.5-,
°^>*===^L^^ 2.0-
si
1.5-
1.0-
0.5-
r\ n
^»© *-^
" ' | j-i
B m;
TOLUENE
0.0-
0.8 0.9 1.0 1.1 1.2 1.3 0.8 0.9 1.0 1.1 1.2 1.3
BASE ROG MaxRct REACTIVITY
(RELATIVE TO DEFAULT)
Figure 13. Effects of variation of the composition of the base ROG
mixture on incremental reactivities in the maximum reactivity
averaged condition scenario. Incremental reactivities are
plotted against the ratio of the MaxRct reactivity of the
modified base case ROG mixture relative to the MaxRct
reactivity of the standard base case ROG mixture.
82
-------
>
I
o
<
LJ
on
o
o
o
o
>
o
"6
o
"5
2.5-,
2.0-
1.5-
1.0-
0.5-
0.0-
FORMALDEHYDE
0.15-1
0.12-
0.09-
0.06-
0.03-
0.00
N-BUTANE
N-OCTANE
0.00
o ALDEHYDES VARIED
D ALKENES VARIED
A AROMATICS VARIED
1.0-,
0.8-
0.6-
0.4-
0.2-
0.0-
PROPENE
TOLUENE
0.8 0.9 1.0 1.1 1.2 1.3
0.8 0.9 1.0 1.1 1.2 1.3
BASE ROG MaxRct REACTIVITY
(RELATIVE TO DEFAULT)
Figure 14. Effects of variation of the composition of the base ROG
mixture on incremental reactivities in the maximum ozone
averaged condition scenario. Incremental reactivities
are plotted against the ratio of the MaxRct reactivity of the
modified base case ROG mixture relative to the MaxRct
reactivity of the standard base case ROG mixture.
83
-------
except for the effect of reducing aldehyde content on formaldehyde
reactivity, the effects of these rather large variations in the base ROG
composition on the NO -adjusted incremental reactivities of these
A
representative VOCs can be considered to be relatively small.
-------
IV. SUMMARY AND CONCLUSIONS
A quantitative reactivity scale which compares the effects of
different types of VOCs on ozone formation would be a useful tool for
developing effective and flexible control strategies for ozone
formation. However, the development of such a scale has a number of
difficulties. These can be categorized into three major areas. The first
concerns the fact that the gas-phase chemical mechanisms by which VOCs
react in the atmosphere to form ozone are in many cases highly
uncertain. This results in uncertainties in the model predictions of the
reactivity of a VOC in any given scenario. The second concerns the fact
that the effects of VOCs on ozone formation their reactivities
depend significantly on the environment in which they are emitted. This
means that even if we are capable of reliably predicting the reactivity of
a set of VOCs in a set of scenarios, it is not obvious how these results
should be used in developing a single reactivity scale or even whether
use of a single reactivity scale has any scientific or regulatory
validity. The third concerns the fact that there are uncertainties in
conditions of airsheds and episodes where unacceptable levels of ozone are
formed. The uncertainties in conditions of a specific episode affect
predictions of VOC reactivities for that episode, and uncertainties in
distribution of conditions affect the development of appropriate methods
for aggregating scenario-specific reactivities into a generalized
reactivity scale.
The focus of this report has been primarily on the second of these
problems, that of deriving a single reactivity scale, given that
reactivities depend on environmental conditions. This has been studied by
deriving reactivity scales by using several different techniques, given a
single chemical mechanism and a single set of representative airshed
scenarios. The chemical mechanism employed is uncertain for many VOCs,
but it incorporates our current best estimate of their atmospheric
reactions, and represents most of the major types of species which need to
be incorporated in reactivity scales. The representative environmental
scenarios employed are even more uncertain, but they represent their
developers' best estimate of the conditions of a wide variety of
representative pollution episodes, given the limitations in available data
85
-------
and the constraints of the simplified physical formulation of model
used. This is sufficient for the purpose of at least an initial
evaluation of methods for deriving reactivity scales.
The difference in relative levels of NOV is the most important reason
A
why VOC reactivities vary from scenario to scenario. Relative levels NOX
are usually measured by the ROG/NOX ratio -- though this is an
oversimplification, the relative NOX level since is also affected by other
factors, particularly those involving rates of NOV removal. It can be
A
argued that the maximum reactivity (MaxRct) and the maximum ozone (MaxOo)
reactivity scales developed in this work represent respectively the high
and low limits for conditions of NOX availability which are appropriate
for defining a VOC reactivity scale. If NOV levels are significantly
A
higher than those of maximum reactivity conditions, then ozone inhibition
by NOX will prevent significant concentrations of ozone from being formed
in single day episodes. If NOV levels are lower than those favorable for
A
maximum peak ozone concentrations, then ozone formation is NCL limited,
A
and under those conditions changes in VOC emissions have only small
effects on ozone. Therefore, comparison of relative reactivities in the
MaxRct and the MaxOo scales gives an appropriate indication of the effect
of uncertainties in relative reactivities due to variations or
uncertainties in NOV conditions, within the range of conditions where
A
reactivity considerations are relevant to ozone control strategies.
Consistent with results of previous studies, it was found that the
NOX conditions significantly affected both absolute reactivities and
ratios of reactivities. Absolute reactivities are the most strongly
affected, but differences in ratios of reactivities are more significant
in practical VOC control strategy assessment applications. Ratios of
reactivities among the same chemical class of VOC are not strongly
affected by the ROG/NOX ratio (at least when it is varied between the
maximum reactivity and maximum ozone range), but ratios of reactivities of
different chemical classes of VOCs can be quite different. In general,
the reactivities of aromatics and alkenes relative to those of alkanes,
CO, and some alcohols are significantly lower under low-NOx, maximum-ozone
conditions than under relatively high-NOx, maximum-VOC reactivity
conditions. This makes a difference, for example, in predictions of the
ozone benefits of using reformulated gasolines which reduce vehicle
86
-------
emissions of alkenes and aromatics relative to those of alkanes. The
maximum ozone reactivity scale predicts much less benefit from using such
fuels than the maximum reactivity scale does. This has a practical effect
in assessing the relative advantages of reformulated gasolines as opposed
to methanol or other alternative fuels.
Since these two reactivity scales can in some cases give significant
differences in predictions of benefits of proposed VOC substitutions, it
can be questioned whether it is appropriate ever to use reactivity scales
in control strategy applications. It is obviously easy to conclude that
it is not, and to point to the results discussed above to support this
conclusion. It is more difficult, however, to propose a practical
alternative which is more scientifically justifiable for all cases where
alternative VOC substitution approaches need to be assessed. Obviously,
if major substitution strategies are being considered, they should be
evaluated as comprehensively as possible, and reliance on a single
reactivity scale alone would not be justifiable. A comprehensive
evaluation would include examining the effects of the proposed
substitution under as many airshed conditions as possible, and using
models which are as accurate as possible in representing those
conditions. However, this approach is not practical for initial
evaluations of cases where multiple options or types of emissions need to
be considered, or when effects of small and varied VOC sources are being
considered. In such cases, one has the option of either ignoring
reactivity altogether, or using a general reactivity scale which gives at
least some indication of relative ozone impacts.
While it may not be possible to develop a reactivity scale which will
be applicable to all conditions, it is probably not impossible to develop
one whose use will result in more effective ozone control strategies than
ignoring reactivity altogether. Despite its obvious problems, use of the
OH radical rate constant scale will probably result in better ozone
control strategies than treating all VOCs equally. At a minimum, it
provides a means for determining which VOCs are unreactive. However, it
is also reasonable to expect that a reactivity scale which takes other
aspects of a VOC's mechanism into account besides just its OH radical rate
constant would give even better results. The problem is therefore not to
develop a reactivity scale which is valid for all conditions (which is
87
-------
impossible); it is to determine what reactivity scale would give the best
overall results when it is used.
To obtain an indication of the characteristics of an optimum
reactivity scale for general use, incremental reactivities were calculated
for a wide variety of VOCs for a set of scenarios representing a wide
variety of airshed conditions. While not conclusive (since the degree to
which the set of scenarios employed represents a realistic distribution of
airshed conditions is highly uncertain), the results are highly suggestive
that a maximum reactivity scale such as the MaxRct scale may give a
good approximation to such an "optimum" scale. Although the relative
reactivities of the individual VOCs were highly variable from scenario to
scenario, and the average of ratios of (ozone yield) reactivities was not
well predicted by the MaxRct scale, the MaxRct scale was found to predict
reasonably closely to the reactivity scale derived to minimize total
predictions in absolute ozone changes caused by adding the VOCs to all the
scenarios. In other words, although the MaxRct scale may not have
performed well in predicting relative impacts in any single scenario, it
performed well in predicting total ozone impacts in all scenarios where
each scenario is weighed by the degree of ozone impact caused by changing
emissions of the VOCs.
The correspondence between the maximum reactivity scale and the
results of the individual scenario simulations was even better when
integrated ozone rather than peak ozone was used to measure the ozone
impacts of adding the VOCs. In that case, the MaxRct scale even gave a
reasonably good correspondence to the average of reactivity ratios for all
scenarios including those which are far from maximum reactivity
conditions. This is attributable to the fact that maximum reactivities
and integrated ozone reactivities are sensitive to the same aspects of the
VOC reaction mechanisms those which affect the rate at which ozone is
formed. Although ozone air quality standards are expressed in terms of
peak ozone levels, integrated ozone levels (or at least integrated ozone
levels above certain threshold levels) may be more relevant in terms of
impacts of exposure of the population, plants, and materials to ozone.
In this regard, it should be noted that Russell (1990), using a
complex gridded airshed model for a Southern California high-ozone episode
(see Russell et al., 1989), found that the MaxRct scale gave good
88
-------
predictions of relative effects of changing VOC emissions on calculations
of person-hours of exposure to outdoor ozone levels above the Federal
standard of 0.12 ppm. Although this needs to be further tested with
realistic airshed models for other airsheds and episodes, and with more
detailed ozone exposure models, it is highly suggestive that a maximum
reactivity scale developed with very simple airshed models (such as the
MaxRct scale) may have applicability to predictions of ozone exposure in
more realistic situations.
The maximum reactivity scale may not give a good indication of
effects of VOC changes on peak ozone levels in multi-day and long-range
transport episodes where ozone formation is NOV limited. Indeed, if the
A
episode is sufficiently NOX limited, even the MaxOo scale gives a poor
indication of effects of VOC changes on peak ozone. However, as discussed
above, NOX control is a much more effective strategy for reducing ozone
under those conditions, and VOC reactivity is largely irrelevant to the
development of effective strategies for reducing peak ozone. On the other
hand, the results of this study suggest that the MaxRct scale may give a
reasonably good indication of effects of VOCs on integrated ozone levels
in such episodes. This needs to be examined further, by using models for
multi-day and long-range transport episodes.
Because of these considerations, it is concluded that if one needs to
use a single generalized reactivity scale for assessing effects of VOCs on
ozone, a maximum reactivity scale may be the most appropriate for this
purpose, at least for the near term. It is not as sensitive to the
distribution of airshed scenarios used to derive it, as is the case for
scales optimized to fit a particular assumed distribution of scenarios
(e.g., the base case scales), and can be calculated by using relatively
simple airshed models without having to accurately simulate any particular
air pollution episode. While no single reactivity scale will perfectly
represent all conditions, the maximum reactivity scale seems to give a
reasonably close correspondence to relative ozone impacts of VOCs in a
large number of cases, particularly those where the impacts of VOCs are
the greatest, or those where impacts on integrated ozone levels are
considered. Although these conclusions are based on reactivities
calculated for highly simplified, single day, and perhaps in some cases
inaccurate scenarios, the scenarios employed are sufficiently varied that
89
-------
it is not unreasonable to expect that similar results would be obtained if
more detailed and accurate scenarios were employed. The results of
Russell (cited above) tend to support this expectation.
However, it is clear that further research is needed to reduce the
uncertainties in the derivation of VOC reactivity scales. The scenarios
employed in this study may not represent the best currently available
estimate of the distribution of ozone pollution conditions, and more work
in this area is clearly needed. A systematic study of the effects of
airshed conditions other than NOX availability on reactivity would aid in
this effort. It would indicate which scenario characteristics are the
most important to accurately represent when maximum reactivity is
calculated and would assist in estimates of uncertainties in reactivity
predictions caused by uncertainties in characterizing airshed
conditions. Some work in this area was carried out in conjunction with
this study and is described in this report, but more such work is
needed.
One aspect of scenario conditions which affects VOC reactivities is
the composition of the base case ROG emissions. The limited sensitivity
calculations carried out in this study indicate that predictions of
aldehyde reactivity are particularly sensitive to the aldehyde fraction
which is assumed in the total emissions, at least when the aldehyde
fraction is less than approximately 5% of the total ROG. The aldehyde
fraction is probably the most uncertain aspect of the composition of base
case ROG emissions, and this results in a corresponding uncertainty in
predictions of aldehyde reactivities. On the other hand, other than this
sensitivity of aldehyde reactivities to aldehyde emissions, VOC
reactivities were not dramatically affected by large changes in the base
ROG composition. This suggests that once the aldehyde emissions are
better characterized, the other uncertainties or variabilities in the
composition of base ROG emissions may not be as large a source of
uncertainty in reactivity scales as is generally believed. However, this
needs to be investigated further.
It should be recognized that regardless of which approach or set of
airshed conditions is used for developing a reactivity scale, the
reactivities for many VOCs will be uncertain because of uncertainties in
the chemical mechanism used to calculate them. Only for a limited number
90
-------
of VOCs are sufficient data available to test these mechanisms. Modeling
studies may give us ar indication of the magnitudes of the effects of
these uncertainties, but will not reduce them. To reduce these
uncertainties, experimental data are needed to test the mechanisms used to
derive the reactivity factors, or at a minimum to test their predictions
of maximum reactivity.
91
-------
V. REFERENCES
Altshuller, A. P., and J. J. Bufalini (1971): Environ. Sci. Technol. 5,
39, and references therein.
Atkinson, R., and W. P. L. Carter (1984): "Kinetics and Mechanisms of the
Gas-Phase Reactions of Ozone with Organic Compounds Under Atmospheric
Conditions," Chem. Rev. 1884, 437-470.
Atkinson, R. (1987): "A Structure-Activity Relationship for the
Estimation of Rate Constants for the Gas-Phase Reactions of OH
Radicals with Organic Compounds," Int. J. Chem. Kinet. _19, 799-828.
Atkinson, R. (1989): "Kinetics and Mechanisms of the Gas-Phase Reactions
of the Hydroxyl Radical with Organic Compounds," J. Phys. Chem. Ref.
Data, Monograph No. 1.
Atkinson, R., D. L. Baulch, R. A. Cox, R. F. Hampson, Jr., J. A. Kerr, and
J. Troe (1989): "Evaluated Kinetic and Photochemical Data for
Atmospheric Chemistry. Supplement III, J. Phys. Chem. Ref. Data ^8,
881-1097-
Atkinson, R. (1990): "Gas-Phase Tropospheric Chemistry of Organic
Compounds: A Review," Atmos. Environ., 24A, 1-24.
Bauges, K. (1986): "A Review of NMOC, NO and NMOC/NO Ratios Measured in
1984 and 1985," EPA-450/4-86-015, September.
Bufalini, J. J., S. L. Kopczinski, and M. C. Dodge (1972): Environ.
Lett., 3, 101.
Bufalini, J. J., T. A. Walter, and M. M. Bufalini (1977): Environ. Sci.
Technol. JJ, 1181-1185.
Bufalini, J. J. and M. C. Dodge (1983): Environ. Sci. Technol. V7: 308-
311.
CARB (1989): "Definition of A Low-Emission Motor Vehicle in Compliance
with The Mandates of Health and Safety Code Section 39037.05
(Assembly Bill 234, Leonard, 1987)," Report by Mobile Sources
Division, California Air Resources Board, El Monte, California. May
19.
CARB (1990): "Low-Emission Vehicles/Clean Fuels Technical Support
Document," Mobile Source Division, Research Division, Stationary
Source Division, and Technical Support Division, California Air
Resources Board, Sacramento, CA.
Carter, W. P. L., R. Atkinson, A. M. Winer, and J. N. Pitts, Jr. (1982):
Experimental Investigation of Chamber-Dependent Radical Sources,"
Int. J. Chem. Kinet. _14, 1071.
92
-------
Carter, W. P. L., M. C. Dodd, W. D. Long, and R. Atkinson. (1984):
Outdoor Chamber Study to Test Multi-Day Effects. Volume I: Results
and Discussion. Final report, EPA-600/3-84-115, November
Carter, W. P. L. , F. W. Lurmann, R. Atkinson, and A. C. Lloyd (1986a):
"Development and Testing of a Surrogate Species Chemical Reaction
Mechanism," EPA-600/3-86-031, August.
Carter, W. P. L., W. D. Long, L. N. Parker, and M. C. Dodd (1986b):
"Effects of Methanol Fuel Substitution on Multi-Day Air Pollution
Episodes," Final Report on California Air Resources Board Contract
No. A3-125-32, Statewide Air Pollution Research Center, University of
California, Riverside, CA, April.
Carter, W. P. L., and R. Atkinson (1987): "An Experimental Study of
Incremental Hydrocarbon Reactivity," Environ. Sci. Technol. 21, 670.
Carter, W. P. L. (1988): "Development and Implementation of an Up-to-Date
Photochemical Mechanism for Use in Airshed Modeling," Final Report
for California Air Resources Board Contract No. A5-122-32, Statewide
Air Pollution Research Center, University of California, Riverside,
CA, October.
Carter, W. P. L., and R. Atkinson (1989): "A Computer Modeling Study of
Incremental Hydrocarbon Reactivity," Environ. Sci. Technol. 23,
864-880.
Carter, W. P. L. (1989a): "Estimates of Photochemical Reactivities of
Solvent Species Used in Architectural Coatings," presented at the
82nd Annual Meeting of the Air and Waste Management Association,
Anaheim, CA, June 25-30.
Carter, W. P. L. (1989b): "Ozone Reactivity Analysis of Emissions from
Motor Vehicles," Draft Report to the Western Liquid Gas Association,
Statewide Air Pollution Research Center, University of California,
Riverside, CA, July.
Carter, W. P. L. (1990): "A Detailed Mechanism for the Gas-Phase
Atmospheric Reactions of Organic Compounds," Atmos. Environ., 24A
481-518
Carter, W. P. L., and F. W. Lurmann (1990): "Evaluation of the RADM Gas-
Phase Chemical Mechanism," Final Report, EPA-600/3-90-001.
Carter, W. P. L. , and F. W. Lurmann (1991): "Evaluation of a Detailed
Gas-Phase Atmospheric Reaction Mechanism Using Environmental Chamber
Data," Atmos. Environ., in press.
Chang, T. Y., and S. J. Rudy (1990): "Ozone-Forming Potential of Organic
Emissions from Alternative-Fueled Vehicles," Atmos. Environ., 24A,
2421-2430.
Darnall, K. R., A. C. Lloyd, A. M. Winer, J. N. Pitts, Jr. (1976):
Environ. Sci. Technol. 10, 692.
93
-------
Dodge, M. C. (1977): "Combined Use of Modeling Techniques and Smog
Chamber Data to Derive Ozone-Precursor Relationships" in Proc.
International Conference on Photochemical Oxidant and Its Control,
Research Triangle Park, North Carolina, EPA-600/3-80-028a.
Dodge, M. C. (1984): "Combined Effects of Organic Reactivity and NMHC/NOX
Ratio on Photochemical Oxidant Formation A Modeling Study," Atmos.
Environ. ]8, 1657.
"^
EPA (1984): "Guideline for Using the Carbon Bond Mechanism in City-
Specific EKMA," EPA-450/4-84-005, February.
Gery, M. W., R. D. Edmond, and G. Z. Whitten (1987): "Tropospheric
Ultraviolet Radiation. Assessment of Existing Data and Effects on
Ozone Formation," Final Report, EPA-600/3-87-047, October.
Gery, M. W., G. Z. Whitten, and J. P. Killus (1988): "Development and
Testing of the CBM-IV For Urban and Regional Modeling," EPA-600/3-88-
012, January.
Gery, M. W. (1989): Atmospheric Research Associates, Boston, MA, private
communication.
Gipson, G. L., W. P. Freas, R. A. Kelly, and E. L. Meyer (1981):
"Guideline for Use of City-Specific EKMA in Preparing Ozone SIPs,"
EPA-450/4-80-027, March.
Gipson, G. L. (1984): "Users Manual for OZIPM-2: Ozone Isopleth Plotting
With Optional Mechanisms/Version 2," EPA-450/4-84-024, August.
Hogo, H., and M. W. Gery (1988): "Guidelines for Using OZIPM-4 with CB-IV
or Optional Mechanisms. Volume 1. Description of the Ozone Isopleth
Plotting Package, Version 4." Report SYSAPP-88/001, EPA Contract No.
68-02-4136, Systems Application, Inc., San Rafael, CA.
Hough, A. M., and R. G. Derwent (1987): "Computer Modelling Studies of
the Distribution of Photochemical Ozone Production Between Different
Hydrocarbons," Atmos. Environ. 21_, 2015-2033.
Jeffries, H. E. (1988): Department of Environmental Science and
Engineering, School of Public Health, University of North Carolina,
Chapel Hill, NC, private communication.
Jeffries, H. E., K. G. Sexton, J. R. Arnold, and T. L. Kale (1989):
"Validation Testing of New Mechanisms with Outdoor Chamber Data.
Volume 2: Analysis of VOC Data for the CB4 and CAL Photochemical
Mechanisms," Final Report, EPA-600/3-89-010b.
Joshi, S. B., M. C. Dodge, and J. J. Bufalini (1982): Atmos. Environ. Jj>,
1301-1310.
Killus, J. P., and G. Z. Whitten (1984): "Technical Discussions Relating
to the Use of the Carbon Bond Mechanism in OZIPM/EKMA." EPA-450/4-
84-009, May.
94
-------
Laity, J. L., F. G. Burstain, and B. R. Appel (1973): "Photochemical Smog
and the Atmospheric Reactions of Solvents," in "Solvents Theory and
Practice," Tess, R. W., Ed., Adv. Chem. Series 124. 95.
Lurmann, F. W., W. P. L. Carter, and R. A. Coyner (1987): "A Surrogate
Species Chemical Reaction Mechanism for Urban-Scale Air Quality
Simulation Models. Volume I - Adaptation of the Mechanism," EPA-
600/3-87-0!4a.
Lowi, A. L., and W. P. L. Carter (1990): "A Method for Evaluating the
Atmospheric Ozone Impact of Actual Vehicle Emissions," presented at
the SAE International Congress and Exposition, Detroit, MI, February
26 - March 2.
NASA (1987): "Chemical Kinetics and Photochemical Data for Use in
Stratospheric Modeling. Evaluation Number 8," JPL Publication 87-
41, Jet Propulsion Laboratory, Pasadena, CA, September.
Peterson, J. T. (1976): "Calculated Actinic Fluxes (290 - 700 nm) for Air
Pollution Photochemistry Applications," EPA-600/4-76-025, June.
Russell, A., J. Harris, J. Milford, and D. St. Pierre (1989):
"Quantitative Estimate of the Air Quality Impacts of Methanol Fuel
Use," Final Report, California Air Resources Board Agreement No. A6-
048-32, Carnegie-Mellon University, Pittsburgh, PA., April.
Russell, A. G. (1990): "Air Quality Modeling of Alternative Fuel Use in
Los Angeles, CA: Sensitivity of Pollutant Formation to Individual
Pollutant Compounds," Presented at the AWMA 83rd Annual Meeting, June
24-29.
Weir, B. R., A. S. Rosenbaum, L. A. Gardner, G. Z. Whitten and W. Carter,
(1988): "Architectural Coatings in the South Coast Air Basin:
Survey, Reactivity, and Toxicity Evaluation", Final Report to the
South Coast Management District, SYSAPP-88/137, Systems Applications,
Inc., San Rafael, CA, December.
Whitten, G. Z. (1988): "Evaluation of the Impact of Ethanol Gasoline
Blends on Urban Ozone Formation," SYSAPP-88/029, Systems
Applications, Inc., San Rafael, CA.
Whitten, G. Z. (1989): Systems Applications, Inc., San Rafael, CA,
private communication.
Wilson, K. W., and G. J. Doyle (1970): "Investigation of Photochemical
Reactivities of Organic Solvents," Final Report, SRI Project PSU-
8029, Stanford Research Institute, Irvine CA, September
Wu, C. H., S. M. Japar, and H. Niki (1976): J. Environ. Sci. Health, A11,
191.
95
-------
APPENDIX A
UPDATES TO THE ATMOSPHERIC REACTION MECHANISM
The gas-phase chemical mechanism used for the calculations of
reactivities of most of the volatile organic compounds (VOCs) presented in
this study is documented by Carter (1990). That mechanism represents our
state of knowledge as of early 1989. Since that time, there have been
additional data concerning several aspects of the mechanism. For example,
new data (e.g., Rogers, 1990) indicate that the formaldehyde photolysis
absorption cross sections used in the Carter (1990) mechanism should be
reduced, and that the rate constant ratio for the reactions involved in
PAN formation should be modified somewhat (Tuazon et al. , 1991a). These
changes, which would require reevaluation of the mechanism against the
chamber data and re-calculation of reactivities for all of the VOCs, were
not implemented for this study. It is generally not practical to update a
mechanism used in an airshed model every time there is new information
concerning its components, since there is always a gap of time between its
development and application. The plan is for the mechanism to be updated
in late-1991, and these and other appropriate changes will be implemented
then.
However, a reassessment of existing data concerning the reactions of
ethanol, an extremely important VOC in current emissions and a major
component of some proposed alternative fuels, indicate that the represen-
tation for it in the Carter (1990) mechanism will slightly underestimate
its reactivity. In addition, the mechanism of Carter (1990) does not
include representation of the reactions of the potentially important fuel
additives methyl t-butyl ether (MTBE) and ethyl t-butyl ether (ETBE). New
data are available concerning the atmospheric reactions of MTBE (Japar et
al., 1990; Carter et al., 1990; Tuazon et al., 1991b), making it now one
of the better understood VOCs in terms of its mechanism and reactivity.
We have also found that the Carter (1990) mechanism represents dimethyl
ether, a possible alternative propellant in aerosol sprays, in a manner
which will probably overestimate its reactivity. Since reactivities of
these compounds are or may be important in assessments of effects of
alternative fuels or reformulations of some consumer products, it was
A-1
-------
determined that it is appropriate that their mechanisms be updated before
their calculated reactivities were presented in this report. These
updates are documented in this Appendix.
In Sections 1 through 4 of this Appendix, our best estimates as to
the overall reactions of ethanol, MTBE, ETBE, and dimethyl ether (DME) are
given. In Section 5, the methods used to represent their reactions in the
framework of the Carter (1990) mechanism is documented. This involves
developing methods to represent the reactions of the reactive products of
these VOCs by lumped organic product species already in the mechanism.
1. Ethanol
As part of an assessment of the state of knowledge of the relative
reactivities of methanol and ethanol, Atkinson (1989a) reviewed the
current data concerning the atmospheric chemistry of these compounds and
their major atmospheric oxidation products. The results of that review
were consistent with the Carter (1990) mechanism in its recommendations
for the methanol mechanism and the value of the OH radical rate constant
* O o
for ethanol (the recommended value for the latter being 6.18 x 10 T
e532/T Cm3 molecule"1 s~1), but not in its recommendation for the products
of the OH + ethanol reaction. The ethanol mechanism was updated to be
consistent with these new recommendations.
The mechanism of Carter (1990) assumes that the reaction proceeds
significantly only following OH reaction at the CH2 group, resulting in
the subsequent formation of H02 and acetaldehyde
°2
OH + CH3CH2OH > H20 + CH3CH(.)OH > H20 + H02 + CH3CHO (a)
while Atkinson (1989a) also considers the additional reactions,
°2
OH + CH3CH2OH --> H20 + .CH2CH2OH > H20 + .OOCH2CH2OH (b)
°2
> H20 + CH2CH20. > H20 + H02 + CH3CHO (c)
A-2
-------
and the possible subsequent reactions of the radical formed in reaction
(b),
.OOCH2CH2OH + NO > N02 + .OCH2CH2OH
°2
.OCH2CH2OH > HCHO + .CH2OH > 2 HCHO -t- H02 (b1)
°2
> H02 + HOCH2CHO (b2)
Although pathway (c) is equivalent to (a) in its overall effect, pathway
(b) can give rise to an additional NO-to-N02 conversion and formation of
different products.
Using primarily structure-reactivity estimates (Atkinson, 1987),
product data of Meier et al. (1985a,b), and kinetic data of Huess and
Tully (1988), Atkinson (1989a) estimated that pathways (a), (b), and (c)
occur Q5%, '\0%, and 5% of the time, respectively. The .CH2CH2OH radical
formed in pathway (b) is also formed in the reactions of OH radicals with
ethene, which has been studied by Niki et al. (1981). The results of Niki
et al. (1981) indicate that decomposition of the .OCH2CH2OH radical
(pathway b1) occurs 78 ± k% of the time under atmospheric conditions.
Based on these considerations, estimates of the overall reactions of
ethanol in NO -air systems are as follows:
A
°2
OH -t- CH3CH2OH > H20 + 0.9 CH3CHO + 0.9 H02 + 0.1 .OOCH2CH2OH
°2
.OOCH2CH2OH + NO > N02 +0.22 HOCH2CHO +1.56 HCHO + H02
or
A-3
-------
°2
CH3CH2OH + OH > 0.90 CH3CHO + 0.022 HOCH2CHO + 0.156 HCHO
-H20 - 0.10 NO + 0.10 N02 + H02
This was used to derive the ethanol mechanism used in the model, which is
given in Section 5.
This mechanism predicts ethanol has a slightly higher reactivity than
the mechanism of Carter (1990), which has only acetaldehyde + H02 forma-
tion (pathway a). The new mechanism gives a ]3% higher MaxRct reactivity
and an 18$ higher MaxO, reactivity, compared to the previous version. The
mechanistic reactivity of ethanol in environmental chamber experiments
under maximum reactivity conditions has recently been measured in our
laboratories, and the results, which will be presented in a subsequent
report which is in preparation, are more consistent with this slightly
more reactive mechanism.
2. Methyl t-Butyl Ether (MTBE)
MTBE is not represented in the mechanism of Carter (1990), but is
included among the VOCs used in this study because of its importance as a
motor vehicle fuel additive. As is the case with other saturated
compounds, ethers are expected to react significantly in the atmosphere
only with OH radicals. The rate constant for this reaction was measured
by Cox and Goldstone (1982) and Wallington et al. (1988), and based on
these data Atkinson (1989b) recommends a rate constant of (6.81 ± 2) x
10-18 T2 e(460 ± 112)/T Cm3 moiecule"1 s"1. The products formed in this
reaction in the presence of NOV under atmospheric conditions have recently
A
been determined in our laboratories to be: t-butyl formate, 76 ± 7%',
formaldehyde, 37*!??; methyl acetate, 17 ± 2%; and acetone, 2 ± 1? (Carter
et al., 1990; Tuazon et al., 1991b). These products account for 95 ± Q%
of the reacting carbon in the MTBE molecule. The 76 ± 1% t-butyl focmate
yield is consistent, within the uncertainty ranges, with the 60$ yield
observed by Japar et al. (1990) and the approximately 60$ yield estimated
with the structure-reactivity methods developed by Atkinson (1987).
As discussed in more detail elsewhere (Carter et al., 1990; Tuazon et
al. , 1991b), these products can be accounted for by the reaction scheme
shown in Figure A-1. Note that this mechanism predicts that the
A-H
-------
OH + CHj-0-i-CHj
CHa
NO
NO2
CH3
-O-CHz-O-C-CHs
CHj
HC-O-C-CHj + HOi
14 (76±7%)
CH3
HCHO + CHs-C-O.
CHj
CHj
NO
N02
02
HOj + HCHO
02
NO
CHsO. + CHj-C-CHj
N02
CHs -O-(j;-CH2 -O.
CH3
CH3
CHs-O-C. + HCHO
bb
V
- + CHj-fi-CHj
02
NO
N
> N02
/
CH3
CH3 -0-C-O.
CH3
CHj
-O-C-CHj +
CHj + CHj
02
NO
H02 + HCHO
(2*1%)
HO2 + HCHO
Figure A-1. Schematic of the OH radical-initiated reactions of MTBE in
the presence of NO , showing observed yields of the major
products.
A-5
-------
formaldehyde yield should be equal to twice the methyl acetate + acetone
yield, which indeed is observed. Note also that these results are
somewhat surprising in that they predict that 1,4-H shift isomerization
reactions such as
(CH3)3C-0-CH20. > .CH2(CH3)2C-0-CH2OH
are not important, and that the decomposition of radicals such as CHU-O-
C(CH3)2CH20. (shown on the figure) dominates over the competing reaction
with 02,
CH3-0-C(CH3)2CH20. + 02 > CH3-0-C(CH3)2CHO + H02
under atmospheric conditions. This is contrary to predictions of
estimation methods given by Carter and Atkinson (1985). However, the
observed products, and their yields, cannot be rationalized unless it is
assumed that the above two reactions are unimportant relative to those
shown in Figure A-1.
The product yields do not rule out the possibility that some organic
nitrate formation, via reactions such as
M
(CH3)3C-0-CH200. + NO --> (CH3)3C-0-CH2ON02
are not occurring at up to the 13/& level. Some IR bands which can be
attributed to the -ON02 were observed, but no quantitative information
could be obtained (Carter et al., 1990; Tuazon et al., 1991b). Such reac-
tions occur up to this level in analogous reactions in the photooxidations
of the higher alkanes (Carter and Atkinson, 1985). As discussed elsewhere
(Carter et al., 1990), results of recent measurements of mechanistic reac-
tivities of MTBE carried out in our laboratories are consistent with model
simulations if it is assumed that organic nitrate formation occurs between
0% and 5% of the time, 2% nitrate formation giving the best fit to the
data. (A report describing the MTBE reactivity experiments, which
includes the results of the ethanol reactivity experiments discussed
A-6
-------
above, is in preparation.) Therefore, we assume 2% organic nitrate
formation in the MTBE mechanism used in this study.
If 2% organic nitrate formation from the reactions of NO with the
initially formed C5 peroxy radicals is assumed, and the product yields
given above and in Figure A-1 are multiplied by a factor of 1.03 to
account for the other 98$ of the reacted carbon, then the overall mech-
anism for the reactions of OH radicals with MTBE in N0x-air systems can be
summarized as follows:
°2
MTBE + OH > 0.78 (CH^C-O-CHO + 0.39 HCHO + 0.18 CH3-0-COCH3
-H20
+ 0.02 CH^COCH^ - 1.37 NO + 1.35 N02 +0.98 H02
+ 0.02 (C^ Organic Nitrates)
The representation of this in the framework of the Carter (1990) mechanism
is given in Section 5.
3. Ethyl t-Butyl Ether (ETBE)
ETBE is another potentially important fuel additive, and thus it is
also included among the VOCs whose reactivities are calculated in this
study. Like MTBE, the only significant atmospheric removal process for
ETBE is believed to be reaction with OH radicals. Its room temperature
(around 298 K) rate constant has been measured recently in several
laboratories (Wallington et al., 1988, 1989; Bennett and Kerr, 1989), and
the measured values range from 5.6 to 8.8 x 10 cm^ molecule" s .
These are reasonably consistent with the value estimated by the group
additivity method of Atkinson (1987). The average of the measured values
is 7.5 x 10" "^ cm-3 molecule"1 s~\ and that is used in the mechanism. The
temperature dependence of this rate constant is estimated to be small and
is ignored.
The major initial reactions of OH radicals with ETBE are probably the
following:
A-7
-------
OH + ETBE > CH3CH(.)-0-C(CH3)3 (a)
OH + ETBE > CH3CH2-0-C(CH3)2CH2. (b)
OH + ETBE > .CH2CH2-0-C(CH3)3 (c)
Structure-reactivity estimates (Atkinson, 1987) predict that pathways (a),
(b), and (c) occur, respectively, 90?, 8?, and 2% of the time. Because
pathway (c) is estimated to be so minor, it is ignored. In the only
available product study of this reaction, Wallington and Japar (1991)
observed that t-butyl formate (CH^-O-CtCH^) is formed in a 76 ± 6?
yield. This can be rationalized only if it is assumed that the radical
formed in pathway (a) reacts as follows:
°2
CH3CH(.)-0-C(CH3)3 > CH3CH(00.)-0-C(CH3)3
CH3CH(00.)-0-C(CH3)3 + NO > N02 + CH3CH(0.)-0-C(CH3)3
CH3CH(0.)-0-C(CH3)3 > CH3. + HCO-0-C(CH3)3
02 NO 02
CH3. > --> > N02 + H02 + HCHO
Further evidence for the facility of the decomposition shown above for
CH3CH(0.)-0-C(CH3)3 is the observed formation of 92 ± 6? yields of methyl
formate in the simpler diethyl ether system from the same study of
Wallington and Japar (1991). If this is assumed to be the major product
formed following pathway (a), then the results of Wallington and Japar
(1991) indicate that pathway (a) occurs approximately 80? of the time.
This is reasonably consistent with the structure-activity estimates
indicated above, and 80? occurrence therefore is assumed in the mechanism
used in this study.
If pathway (c) is ignored and (a) is assumed to occur approximately
80? of the time, then pathway (b) is assumed to occur approximately 20? of
A-8
-------
the time. There are no data concerning the subsequent reactions of the
CH3CH2-0-C(CH3)CH200. radical formed in this reaction, but it is reason-
able to expect that the reactions will be entirely analogous to those
shown in Figure A-1 following reaction at the t-butyl group in the MTBE
system. Thus, the reactions in the ETBE system would be as follows:
02 NO
CH3CH2-0-C(CH3)2-CH2. > > N02 + CH3CH2-0-C(CH3)2-CH20.
CH3CH2-0-C(CH3)2-CH20. > CH3CH2-0-C(.)(CH3)2 + HCHO
02 NO
CH3CH2-0-C(.)(CH3)2 > > N02 + CH
CH3CH2-0-C(CH3)2-0. > CH3CH20. + C
> CH3CH2-0-COCH3 + CH3. (b2)
02 NO 02
CH3. > > > N02 + H02 + HCHO
If we assume that (1) the ratio of rate constants for the two competing
decomposition reactions for CH3CH2-0-C(CHo)2-0. is the same as the ratio
for the analogous decompositions of the CH3-0-C(CHo)2-0. radical formed in
the MTBE system, and (2) the observed formation of acetone in the MTBE
system is due entirely to the reaction of the CHn-0-C(CHo)2-0. radical (as
opposed to being formed following reaction at the methoxy group in MTBE),
then the acetone/methyl acetate yield ratio in the MTBE system (Carter et
al., 1990; Tuazon et al., 1991b) would imply that kbi/(kb1+kb2) =0.1.
This would imply an overall yield of this process of only 2%, which means
it probably can be ignored. If at least some of the acetone in the MTBE
system is formed from the other process shown in Figure A-1, then pathway
(b1) may be less important than this.
As discussed above with MTBE, organic nitrate formation from the
reaction of the peroxy radicals with NO may be occurring to some extent.
In the case of the n-alkanes, it is known that the relative importance of
A-9
-------
this nitrate-forming reaction increases with the size of the molecule
(Carter and Atkinson, 1985). Therefore, if we assume an overall 2%
organic nitrate yield in the MTBE-NOV system, it is reasonable to assume
A
that this is a lower limit to the organic nitrate yield from ETBE. No
data are available to indicate what this yield would be. For the purpose
of this mechanism, we estimate (somewhat arbitrarily) a 3% yield of
organic nitrates from ETBE. This is based on the expectation that the
additional carbon causes a slightly higher nitrate yield, but not one
which is twice as high. Since nitrate formation is a radical-terminating
process, assumptions concerning its importance can significantly affect
predictions of reactivity. This assumption obviously needs to be tested
experimentally.
If 3% organic nitrate formation from the initially formed peroxy
radicals is assumed, and pathways (a) and (b) are assumed to occur
approximately 80% and 20%, respectively, of the time when nitrates are not
formed, and if pathway (b1) is neglected, then the overall reactions of
ETBE in NO air systems can be summarized as follows.
°2
ETBE + OH > 0.78 (CHoKC-O-CHO +1.16 HCHO + 0.19 CH3CH2-0-COCH3
-H20
- 2.16 NO + 2.13 N02 + 0.97 H02
+ 0.03 (Cg Organic Nitrates)
The representation of this in the framework of the mechanism used in this
study is given in Section 5.
4. Dimethyl Ether (DME)
Dimethyl ether is expected to react significantly in the atmosphere
only with OH radicals, as other ethers do. The available data concerning
its rate constant was reviewed by Atkinson (1989b), who recommends the
Arrhenius expression (1.04 ± 0.02) x 10~11 e~(372 ± 39)/T cm3 molecule'1
s~1 for this reaction. Japar et al. (1990) observed that the only
significant product in this reaction is methyl formate. The formation of
this product is expected, since this mechanism is exactly analogous to the
A-10
-------
observed formation of, for example, t-butyl formate in the MTBE system.
Therefore, the overall process for this reaction can be written as:
OH + CH3-0-CH3 > CH3-0-CHO - NO + N02 + H02
The dimethyl ether mechanism of Carter (1990) is based on the assump-
tion that this indeed is the overall process which occurs. However, the
Carter (1990) mechanism does not include a species which explicitly
represents methyl formate, and the model species "RCHO," or the lumped
higher molecular weight aldehyde, was used to represent this product.
But, the mechanism for the model species "RCHO" is based on the mechanism
for propionaldehyde, which reacts in the atmosphere much more rapidly than
does methyl formate (Atkinson, 1989b). In view of this, it is now
believed that representing methyl formate by "RCHO" would result in a
significant overestimation of the reactivity of DME. A more appropriate
representation of methyl formate, which is used in the revised mechanism,
is discussed in the following section.
5. Representation of the VOCs in the Carter (1990) Mechanism
Although the Carter (1990) mechanism is unique among other current
mechanisms in that it can represent the reactions of over 100 VOCs
separately, as any mechanism designed for use in airshed models, it does
not represent explicitly all the intermediate processes and products which
are involved. Instead, a number of "lumping" procedures are used to keep
the mechanism to a manageable size and complexity. This is done by two
types of lumping: (1) by lumping complex sequences of radical reactions
into single steps by using a set of radical "operators" designed to
represent the net effects of these reactions on radical levels and NOX
conversions, and (2) by using a limited number of lumped product compounds
to represent the multitude of organic product species which can be formed
from the VOCs. Both of these lumping approaches need to be used in
representing the reactions of the four VOCs in this mechanism. These are
discussed separately below, and that discussion is followed by a summary
of the specific reactions used in the model for each of the four VOCs.
A-11
-------
Radical Operators. The reactions of OH radicals with VOCs in the
Carter (1990) mechanism are represented by a single lumped reaction,
giving the net overall effect of the reactions in terms of final organic
product yields, and yields of several "chemical operators" which represent
the net effects of the intermediate radicals on the rest of the system.
In the case of the VOCs discussed here, the following radicals and
chemical operators are used:
H02- - Represents the formation of H02 in processes which do not
convert NO to N02, as occurs (for example) in the
acetaldehyde-forming pathways in the ethanol mechanism. It
is not given as the product if NOV reactions are involved in
A
sequences of reactions which ultimately give rise to F^;
instead, operators such as those listed below are used.
R02-R. - Represents peroxy radicals which react to convert NO to N02
and then form H02 and other products (i.e., H02 - NO + N02)
in the presence of NOV. When NOV levels are low, this
A A
"radical" is consumed by other processes, as is the case
with the other operators which are involved in NOX
reactions.
R202. - Represents extra NO-to-N02 conversions due to multi-step
mechanisms (i.e., -NO + N02) in the presence of NOX.
R02-N. - Represents those peroxy radicals which react with NO in the
presence of NOX to form organic nitrates. All organic
nitrates (except for PAN and its analogues and other
peroxynitrates, and C^-Co organic nitrates which are treated
as inert) are represented by the C5 species "RNOg," which
reacts like a pentyl nitrate.
In terms of these radicals and operators, the major oxidation
pathways of ethanol, MTBE, ETBE and DME, which were discussed above, can
be written as follows:
A-12
-------
OH + Ethanol > 0.90 CHgCHO + 0.022 HOCH2CHO + 0.156 HCHO
+0.10 R02-R. + 0.9 H02.
OH + MTBE > 0.78 (CH^C-O-CHO +0.39 HCHO + 0.18 CH3-0-COCH3
+0.02 CH3COCH3 +0.37 R202. +0.98 R02-R.
+0.02 R02-N.
OH + ETBE > 0.78 (CH^C-O-CHO +1.16 HCHO
+0.19 CH3CH2-0-COCH3 +1.16 R202. +0.97 R02-R.
+ 0.03 R02-N. + 0.03 "Lost Carbon"
OH + DME > CH3-0-CHO + R02-R.
(where the "lost carbon" in the ETBE reaction is due to the fact that the
organic nitrate species formed in the model from R02-N. has only five
carbons, while those formed from ETBE have six).
Representation of Reactive Products. The above reactions do not
completely indicate how to represent these VOCs in the mechanism, because
the mechanism does not explicitly represent the reactions of all of these
products. In particular, although the mechanism explicitly represents
formaldehyde (by HCHO), acetaldehyde (by CCHO) and acetone (by ACET), it
does not represent glycolaldehyde (HOCH2CHO), methyl formate (CH3~0-CHO),
t-butyl formate [(CH3)3C-0-CHO], methyl acetate (CH3-0-COCH3), or ethyl
acetate (CH3CH2-0-COCHo). The mechanism currently represents glycolalde-
hyde (which is also formed in the OH + ethene system) by the model species
CCHO (acetaldehyde), but the other products have not been given an
established representation. Although these products are of relatively low
reactivity, their reactions are not negligible and should not be ignored.
In addition, because of its relatively low reactivity, ACET (acetone) has
been removed from the condensed version of this mechanism (Carter and
Lurmann, 1990), and because of this and its low yield it was decided not
to use it in representing the MTBE mechanism. The representations used
for acetone and the other products in the mechanisms of MTBE and the other
VOCs are discussed below.
The Carter (1990) mechanisms represent non-aldehyde oxygenated
products such as ketones and esters with the generalized ketone species
A-13
-------
"MEK," whose mechanism was derived to represent the reactions of methyl
ethyl ketone. The problem then becomes determining how many moles of MEK
would have to be assumed to be formed to have a similar effect on the
overall photooxidation process as the formation of one mole of an organic
product such as (for example) methyl acetate. In the subsequent dis-
cussion, this is referred to as the "MEK reactivity weighing factor" for
the product. A new procedure was developed for the derivation of this
factor, based in part on the results of the work discussed in the main
body of this report.
The impact of the reactions of an organic compound on the photooxida-
tion process can be measured in a number of ways, and clearly the relative
impacts of different organic compounds would depend on how an impact is
quantified. However, in terms of development of ozone reactivity scales,
obviously the effect on ozone formation is the most important single
consideration. If we adopt the approximation that the relative impact of
the formation of a particular organic reaction product is approximately
proportional to its incremental reactivity, then the MEK reactivity weigh-
ing factor would just be the ratio of the incremental reactivity of the
product to the incremental reactivity of MEK.
As discussed in the main body of this report, the incremental reac-
tivity of a compound depends on the conditions of the airshed simulation
as well as on the nature of the compound's reaction mechanism. However,
for the purposes of deriving a fixed-parameter approximate mechanism, we
assume that the ratios of incremental reactivities in the "maximum reac-
tivity" (MaxRct) and the "maximum ozone" (MaxOo) reactivity scales are
suitable for estimating the range of appropriate values of these ratios
for a variety of conditions. The MaxRct and MaxOo incremental reactivi-
ties for MEK, acetone, and the various esters of interest are listed in
Table A-1, along with the ratios of their incremental reactivities to
those of MEK. The incremental reactivities of the esters were not
calculated explicitly, but were estimated as discussed below.
For estimation purposes, it is often useful to consider the incre-
mental reactivity of a compound as a product of two factors: its "kinetic
reactivity" and its "mechanistic reactivity." (See Section II.C.2 in the
main report for a discussion of these factors.) The former is defined as
the fraction of the emitted compound which undergoes reaction in the
A-14
-------
Table A-1. Ozone Reactivity Values Used in Estimates of MEK Weighing
Factors for Selected Organic Products
Product
[a]
KR(prod) [b] MR(prod) [c] IR(prod) [d] MEK Factor [e]
MaxRct Max03 MaxRct Max03
MaxRct Max03 MaxRct Max03 Used
MEK
0.16 0.22
7.2 2.1
1.15 0.47 1.0 1.0
Acetone
Me. Form.
tBu.Form.
Me.Acet.
Et.Acet.
0.044
0.027
0.065
0.040
0.18
0
0
0
0
0
.058
.042
.100
.062
.26
10.8
7.2
7.2
7.2
7.2
2.9
2.1
2.1
2.1
2.1
0
0
0
0
1
.48
.19
.47
-29
.26
0.17
0.09
0.21
0.13
0.55
0.42
0.17
0.41
0.25
1.10
0.36
0.19
0.47
0.28
1.16
0.40
0.18
0.45
0.27
1.13
Notes:
[a] MEK - methyl ethyl ketone; Me.Form Methyl Formate; tBu.Form.
t-Butyl Formate; Me.Acet Methyl Acetate; Et.Acet. Ethyl
Acetate.
[b] KR(prod) = Kinetic reactivity (Average fraction reacted) of
the organic product in the maximum reactivity (MaxRct) or maximum
ozone (MaxOo) scenarios. Calculated directly for MEK and acetone.
For the esters, the kinetic reactivity was estimated using the
equation: KR(prod) = (1 - e-IntOH x kOH), where IntOH -- 8.2 x
10~5 ppm-min for MaxRct and IntOH = 1.28 x 10~H ppm-min for
MaxOo, and the kOH values (in units of 10^ ppm~ min~ ) used
are as follows: 3.33 for Me.Form.; 8.22 for tBu.Form.; 4.99 for
Me.Acet.; and 23.5 for Et.Acet. See text for sources of kOH.
[c] MR(prod) - Mechanistic reactivity of the product in units of moles
ozone per mole VOC reacted for the MaxRct and the Max03 scales. The
esters are assumed to have the same molar mechanistic reactivity as
MEK. (Note that the mechanistic reactivities given in Table 3 of the
main report are in units of moles ozone per mole carbon VOC.
Therefore, the values shown here for acetone and MEK are higher by
factors of 3 and 4, respectively.)
[d] IR(prod) = Incremental reactivity of the product in units of moles
ozone per mole VOC added for the MaxRct and Max03 scales. Computed
by multiplying kinetic and mechanistic reactivities.
[e] MEK factor = MEK weighing factor from the ratios of incremental
reactivities (ozone per mole units) relative to MEK in the
MaxRct and the Max03 scales. The "Used" column gives the
factors used in the mechanism.
A-15
-------
airshed simulation, and the latter is thus the amount of additional ozone
formed caused by adding the compound to the emissions, divided by the
amount which reacts. Note that the kinetic reactivity depends only on how
rapidly the compound reacts, while the mechanistic reactivity depends on
the types of radicals which are formed once it reacts and on the reactivi-
ties of the products formed. These are estimated separately, as discussed
below.
For the purpose of estimating MEK reactivity weighing factors, we
assume that the mechanistic reactivities for the various esters are
similar to those of MEK. This is equivalent to assuming that similar
types of radicals and products are formed once these compounds react. As
shown in Table A-1, the mechanistic reactivities for MEK are 7.2 and 2.1
moles of ozone per mole of MEK reacted for the maximum reactivity and the
maximum ozone conditions, respectively, so these values are also used for
the four esters.
If the esters are all assumed to have the same mechanistic reactivi-
ties as MEK, the MEK weighing factor would be affected only by differences
in their kinetic reactivities i.e., differences in how rapidly they
react in the atmosphere. Esters do not absorb light at wavelengths
important in the troposphere (Calvert and Pitts, 1966) and are believed to
be consumed in the atmosphere primarily by reaction with OH radicals. For
compounds which react only with hydroxyl radicals, the kinetic reactivi-
ties can be estimated by using the empirical relationship
Kinetic - kOH x IntOH
Reactivity = 1 - e
where kOH is the OH radical rate constant, and IntOH is a scenario
dependent parameter which is determined primarily by the overall OH
radical levels. The average IntOH values for the maximum reactivity and
the maximum ozone scenarios used to derive the MaxRct and the MaxO^ scales
are 82 ppt-min and 128 ppt-min, respectively. Given these, and the OH
radical rate constants for the esters, their MaxRct and Max03 kinetic
reactivities can be readily determined.
The measured or estimated 296-300 K OH radical rate constants for
_ 1 O O _ 1
these esters are as follows: methyl formate, 2.27 x 10 J cm0 molecule
A-16
-------
s~1 (Wallington et al., 1988); t-butyl formate, 5.6 x 10~13 cm3 molecule'1
s~ [estimated by using the group additivity method of Atkinson (1987)];
methyl acetate, 3-4 x 10~13 cm3 molecule"1 s~1 (Wallington et al., 1988);
and ethyl acetate, 1.6 x 10~12 cm3 molecule'1 s~1 (Atkinson, 1989b).
These values were used to derive the kinetic reactivity estimates shown
for these compounds in Table A-1.
These estimated kinetic and mechanistic reactivities were combined
estimates to yield estimates of incremental reactivities for these
compounds which are shown in the table. The table also shows the MEK
weighing factors derived from ratios of these incremental reactivities to
those of MEK. The weighing factors for acetone and the esters are
approximately the same for both maximum reactivity and maximum ozone
conditions. (In the case of the esters, this is due in part to assuming
these have the same mechanistic reactivities as MEK.) The weighing
factors given in the "Used" column are roughly averages of those for the
maximum reactivity and the maximum ozone conditions, and they are the
factors actually used in the mechanism.
Summary. By using the radical operators discussed above, and using
acetaldehyde (CCHO) to represent the glycolaldehyde formed from ethanol
and MEK to represent acetone and the esters formed from the ethers, the
overall reactions of these four VOCs can be written as follows. The OH
radical rate constants are given above and also in Table 3 in the main
body of the report.
OH + Ethanol > 0.922 CCHO + 0.156 HCHO + 0.10 R02-R. + 0.90 H02.
OH + MTBE > 0.41 MEK + 0.39 HCHO + 2.87 -C + 0.37 R202.
+ 0.98 R02-R. + 0.02 R02-N.
OH + ETBE > 0.57 MEK + 1.16 HCHO + 2.56 -C + 1.16 R202.
+ 0.97 R02-R. + 0.03 R02-N.
OH + DME > 0.18 MEK + 1.28 -C + R02-R.
(The species "-C" is used to designate a carbon which is "lost" because
the model species used to represent the reactive products has fewer
A-17
-------
carbons than the actual product species. It is assumed to be inert in
this mechanism, and it is included only for carbon accounting purposes.)
These ethanol and DME mechanisms supersede the mechanisms (but not the OH
radical rate constants) given for these compounds by Carter (1990). MTBE
and ETBE were not included in the mechanism of Carter (1990).
6. References
Atkinson, R. (1987): A Structure-Activity Relationship for the Estimation
of Rate Constants for the Gas-Phase Reactions of OH Radicals with
Organic Compounds. Int. J. Chem. Kinet., _19, 799-828.
Atkinson, R. (1989a): Atmospheric Gas-Phase Chemistry of Methanol and
Ethanol and Their Organic Degradation Products: A Review.
Unpublished report, University of California, Riverside, CA.
Atkinson, R. (1989b): Kinetics and Mechanisms of the Gas-Phase Reactions
of the Hydroxyl Radical with Organic Compounds. J. Phys. Chem. Ref.
Data, Monograph No. 1.
Bennett, P. J. and J. A. Kerr (1989): J. Atmos. Chem., 8, 87.
Calvert, J. G. and J. N. Pitts, Jr. (1966): Photochemistry, John Wiley
and Sons, New York.
Carter, W. P. L. and R. Atkinson (1985): Atmospheric Chemistry of
Alkanes. J. Atmos. Chem., 3, 377-405.
Carter, W. P. L. (1990): A Detailed Mechanism for the Gas-Phase
Atmospheric Reactions of Organic Compounds. Atmos. Environ., 24A,
481-518.
Carter, W. P. L. and F. W. Lurmann (1990): Development of a Condensed
Mechanism for Urban Airshed Modeling. Interim Progress Report for
EPA Cooperative Agreement No. CR-815699-01-0, Atmospheric Research
and Exposure Assessment Laboratory, Research Triangle Park, NC.
Carter, W. P. L., E. C. Tuazon, and S. M. Aschmann (1990): Investigation
of the Atmospheric Chemistry of Methyl t-Butyl Ether (MTBE). Draft
Report to the Coordinating Research Council, Inc, for the Automotive
Emissions Cooperative Research Program, Atlanta, GA, October.
Cox, R. A. and A. Goldstone (1982): Proceedings of the 2nd European
Symposium on the Physico Chemical Behavior of Atmospheric Pollutants,
D. Riedel Publishing Co., Dordrecht, Holland, pp. 112-119.
Huess, W. P. and F. P. Tully (1988): Chem. Phys. Lett., 152, 183.
A-18
-------
Japar, S. M., T. J. Wallington, J. F. 0. Richert, and J. C. Ball (1990):
The Atmospheric Chemistry of Oxygenated Fuel Additives: t-Butyl
Alcohol, Dimethyl Ether and Methyl t-Butyl Ether. Int. J. Chem.
Kinet., 22, 1257.
Meier, U. , H. H. Grotheer, G. Reikert, and Th. Just (1985a): Ber
Bunsenges Phys. Chem., 89, 325.
Meier, U., H. H. Grotheer, G. Reikert, and Th. Just (1985b): Chem. Phys.
Lett., 115. 221.
Niki, H., P- D. Maker, C. M. Savage, and L. P. Brietenbach (1981):
Chem. Phys. Lett., 80, 499.
Rogers, J. D. (1990): Ultraviolet Absorption Cross Sections and
Atmospheric Photodissociation Rate Constants of Formaldehyde.
J. Phys. Chem., 90, 4011-4015.
Tuazon, E. C., W. P. L. Carter, and R. Atkinson (1991a): Thermal
Decomposition of Peroxyacetyl Nitrate and Reactions of Acetyl Peroxy
Radicals with NO and N02 Over the Temperature Range 283-313 K.
J. Phys. Chem., 95, 2434.
Tuazon, E. C., W. P. L. Carter, S. M. Aschmann, and R. Atkinson (1991b):
Products of the Gas-Phase Reaction of Methyl tert-Butyl Ether with
the OH Radical in the Presence of NOX. Int. J. Chem. Kinet., in
press.
Wallington, T. J., R. Liu, P. Dagaut, and M. J. Kurylo (1988):
Int. J. Chem. Kinet., 20, 41-49.
Wallington, T. J., J. M. Andino, L. M. Skewes, W. 0. Siegl, and S. M.
Japar (1989): Int. J. Chem. Kinet., 2J_, 993-1001.
Wallington, T. J. and S. M. Japar (1991): Atmospheric Chemistry of
Diethylether and Ethyl-t-butylether. Environ. Sci. Technol., 25,
410-414.
A-19
-------
APPENDIX B
ESTIMATION OF MECHANISTIC REACTIVITIES
USING "PURE MECHANISM" SPECIES
If the mechanism used to represent the atmospheric reactions of a
volatile organic compound (VOC) involves only reaction with hydroxyl
radicals, its mechanistic reactivity can be estimated by using calculated
mechanistic reactivities for "pure mechanism species," as discussed
below. Although this requires more calculations to determine the
mechanistic reactivity of a single VOC, this approach can result in a
significant savings of computer time when calculations of reactivities of
the large numbers of alkane, aromatic, and alcohol species in this
mechanism are required.
The structure of the mechanism used in this study is such that any VOC
which reacts only with OH radicals can be represented as follows:
ki
OH + VOC(i) > ru R1
where R-j, R2, etc., are radicals or radical "operators" (e.g., H02, R02-R.,
R02-N., etc.); r^, r2i, etc., are the yields of these operators used in
the mechanism for VOC(i); P-p P2, etc., are non-radical product species
(e.g., formaldehyde, cresol, etc.); and pu, p2i, etc., are ultimate yields
of these products from VOC(i). It can be readily shown that the
differential equations describing the effects of reactions of this VOC on
the chemical transformations are exactly the same as those describing a
mixture of hypothetical "pure mechanism" compounds P(Rj,ki) and P(Pj,ki),
where: (1) "j" is an index over all radical and product species in the
mechanism; (2) the amounts of P(Rj,ki) and P(Pj,ki) in the mixture are r^
and p.-, respectively; (3) the mechanism for the reactions of the "pure-
J
radical-producing" species P(Rk^) is
OH
+ P(Rj)ki) ;
B-1
-------
(4) the mechanism for the "pure-product-producing" species P(P-,k-) is
J
OH +
> OH + PJ;
and (5) ^ is the OH radical rate constant for VOC(i).
Note that only the reactions of the pure-radical-producing species
P(Rj,ki) affect OH radical levels. For this mixture to have the same
effect as an OH sink that the VOC does, the relationship
= 1
must hold. This is the case for all VOCs in the mechanism, because it
follows from the principle of conservation of radicals.
Since the reactions of this mixture of "pure mechanism species" has
the same effect on the chemical transformations, it would obviously have
the same incremental reactivity as VOC(i). Since the incremental reac-
tivities of mixtures are additive, it follows that:
IR[VOC(i)] = I r.^ IRtPUj,^)] + Z p^
j J
where IR[ species] is the incremental reactivity of the VOC or pure
mechanism species. Since the incremental reactivity is the product of the
kinetic reactivity (KR) and the mechanistic reactivity (MR), and since the
VOC and all its pure mechanism species have the same kinetic reactivity
(since they all react with the same rate constant) it follows that
IR[VOC(i)] = KR[VOC(i)] x { L rji MR[P(RJ)ki)] *
j
Z Pji MR[P(Pj,ki)] };
j
therefore,
B-2
-------
MR[VOC(i)] =
To estimate the mechanistic reactivities by this method, the reac-
tivities of the pure radical or pure product-forming species for all
radical species, radical operators, or reactive products in the mechanism
which are formed from VOCs which react with OH radicals are calculated as a
function of OH rate constant. Once these are calculated, the mechanistic
reactivity of any species in the mechanism which reacts only with OH
radicals can be readily estimated. (In practice these were calculated for
kOH 3 x 1
-------
VH O -JH
O fcj g
£ 8 g
*3 d
0) S ««
^ a
O) U
a^S d g
^ n^ p
2 ^S u
d »H «M
^ < a
(L) ^ "-i
§ g
Reproduced by NTIS
National Technical Information Service
U.S. Department of Commerce
Springfield, VA 22161
This report was printed specifically for your
order from our collection of more than 2 million
technical reports.
For economy and efficiency, NTIS does not maintain stock of its vast
collection of technical reports. Rather, most documents are printed for
each order. Your copy is the best possible reproduction available from
our master archive. If you have any questions concerning this document
or any order you placed with NTIS, please call our Customer Services
Department at (703)487-4660.
Always think of NTIS when you want:
Access to the technical, scientific, and engineering results generated
by the ongoing multibillion dollar R&D program of the U.S. Government.
R&D results from Japan, West Germany, Great Britain, and some 20
other countries, most of it reported in English.
NTIS also operates two centers that can provide you with valuable
information:
The Federal Computer Products Center - offers software and
datafiles produced by Federal agencies.
The Center for the Utilization of Federal Technology - gives you
access to the best of Federal technologies and laboratory resources.
For more information about NTIS, send for our FREE NTIS Products
and Services Catalog which describes how you can access this U.S. and
foreign Government technology. Call (703)487-4650 or send this
sheet to NTIS, U.S. Department of Commerce, Springfield, VA 22161.
Ask for catalog, PR-827.
Name
Address.
Telephone
- Your Source to U.S. and Foreign Government
Research and Technology.
------- |