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 City—Days
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  City—Days
                                                            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.

-------