EPA-AA-TEB-EF-90-4
                       Technical Report
                    Fuel Volatility Effects
                     on Exhaust Emissions
                              By
                          Celia Shih
                          August 1990
                            NOTICE

Technical  Reports  do   not   necessarily  represent  final  EPA
decisions or position.   They  are intended to  present  technical
analysis  of  issues using  data  which  are currently  available.
The purpose in the release of such reports is  to  facilitate the
exchange  of  technical information and  to inform the.  public  of
technical developments which may form the basis for  a  final EPA
decision, position or regulatory action.

                   Test  and Evaluation Branch
             Emission Control Technology Division
                   Office of Mobile Sources
                  Office of Air and Radiation
             U.S.  Environmental  Protection Agency

-------
1.0  BACKGROUND


     In MOBILE4,  the effects of  fuel  volatility, expressed  as
Reid Vapor  Pressure (RVP) in pounds per square  inch (psi),  on
vehicles'  exhaust emissions  were  modeled using data  from EPA's
Emission Factor  Programs  (EFP)  conducted in Ann  Arbor and test
programs  performed  by Automotive   Testing  Laboratories,  Inc.
(ATL)  at  East  Liberty,   Ohio.   The  results  of  these  test
programs were  summarized  in two reports  [1,2].   The American
Petroleum  Institute (API)  submitted  some  of  their  test  data
(see  discussions in Data Sources   section  below)  before  the
release of  MOBILE4.   These data were  reviewed  and found not to
be contradictory to the  MOBILE4's  assumptions.   The algorithm
used  in MOBILE4  has been documented  [3],  and  is summarized in
the following paragraphs:

     (1)   When  ambient temperatures are less  than 41°F,  it  is
assumed  that  there  is  no   RVP  effect on  vehicles'  exhaust
emissions.   This assumption  is applied  to  all model  years  and
all gasoline-powered vehicle types.

     (2)   There  were  no data  to characterize the  RVP effects
on  exhaust  emissions  for  fuel  volatilities  less  than  the
certification  level  of   9.0  psi,   especially   with  ambient
temperatures below   75°F.   The  MOBILE4  model  has assumed  that
the effects of low  RVP fuels on exhaust emissions were the same
as those of 9.0 psi  fuel  (i.e.,  no effect).

     (3)   Vehicles  tested at  ambient  temperatures of 75°F  and
up  have  shown   exhaust   emissions   increases  when  the  fuel
volatility  is  higher than the  certification  level of 9.0  psi.
Therefore,   various  RVP  adjustment   factors   are   used   for
different vehicle types  and model  year groups.   For example,
adjustments are  made to  exhaust  HC  and CO composite emissions
for   1970-79   light-duty   gasoline-powered  vehicles  (LDGVs),
1970-80 LDGTls,  and 1979-80  LDGT2s when the ambient  temperature
is higher than 40°F.  A different set of correction  factors  is
used  to  adjust exhaust HC and  CO composite emissions for 1980+
LDGVs,  1981+  LDGT1S/LDGT2S,   and  1985+   HDGVs   at  ambient
temperatures between 41°F and  75°F.  For  ambient temperatures
higher  than   75°F,   and  model  years  1980+   LDGVs,   1981+
LDGTls/LDGT2s,  and  1985+  HDGVs,  a combined  temperature and fuel
volatility  correction  factor is used  for  each bag  of  the  FTP
exhaust emissions.

     The  purpose  of  this  report  is  to   assemble  all  data
available  as   of May   1,  1990  to  re-evaluate  these  MOBILE4
assumptions, especially  in  the  areas of  low  RVP  fuels  (less
than 9.0 psi)  at temperatures less than 75°F.
                              -1-

-------
2.0  DATA SOURCES
2.1  Pre-1980 Vehicles

     (1)    In  1975,  Chevron  Research  [4]  tested  a  total  of
seven  vehicles   (model   years  1973  to   1976)  at   two  fuel
volatility  levels  and  at  two  ambient  temperatures  using  a
randomized test sequence.  The  two  fuels were the equivalent of
8.5 and 6.5 psi RVP*.   The two ambient temperatures were 75 and
55°F,  with   relative  humidities   of  42   and  73   percent,
respectively.  One of the test  vehicles  was tested twice,  once
with  the  original  carburetor  and  once  with  a  replacement
carburetor.  All  other vehicles  were tested  in  "as-received"
condition  after  their  ignition  and  emission control  systems
were checked  and  failed components  repaired.   Out of  32 tests
("8" vehicles  x 2 fuel RVPs  x 2 temperatures), 20 were run in
replicates with the  average emissions  listed  in their  report.
Descriptions of the  test vehicles  are included in Table Al of
the  Appendix  to  this   report.    Note   that  they   were  all
California vehicles.

     (2)    In  1979, Exxon Research  and  Engineering Company [5]
tested a total  of eight  1974  through 1977 model year vehicles
at six levels  of  fuel  RVP and at the FTP ambient temperature of
75°F.  The test fuels  used  were 11.8,  9.4,  9.1, 8.8,  6.8,  and
6.6  psi   RVP.   There  were  two  9.1 psi  fuels  used   in  the
program:    one  being  a  "high   octane  Indolene"    used  for
certification,  and one  being  a 50-50 blend of 6.5 and 11.8 psi
fuels, with  many more  tests  on  the  latter  fuel   than  the
former.  At  each  fuel  volatility  level,  most of the vehicles
were tested two or three times.   Among the  test  fleet,  three of
them  were  California  vehicles,  and two  of  them  had  their
engines  overhauled   previously.    Descriptions  of   the  test
vehicles  are given in Table A2 of the Appendix.

     (3)    In  1988,  the  California  Air Resources Board (GARB)
included  one  model  year  1978  vehicle  in   their   five-car
evaluation  project  [6]  with  two  fuel  volatilities  at  FTP
ambient temperature of  75°F.   The two test fuels were  8.6 and
7.2  psi   RVP.   Each  vehicle  was  tested  twice at  each  fuel
volatility level.   Description of this test vehicle  is included
in Table A3 of the Appendix.
*    With 11.8 and 8.3 Front End Volatility  Index  (FEVI)  rating
     volatilities,  respectively.   This  FEVI rating  volatility
     is defined as [RVP + 0.13 * (% evaporated at 158°F)].
                              —2 —

-------
2.2. Post-1980 Vehicles

     (1)    In  1988,   Chevron  Research   conducted  two   test
programs  in an effort to evaluate the effects of  reduced RVP on
exhaust  emissions   [7,8].   The  first  portion of  the  program
involved  a total  of fourteen vehicles tested  during two phases,
each comparing  exhaust emissions with  two different RVP fuels
[4].   In  Phase  I,  11.4 and  8.1 psi  fuels  were used  with an
ambient  temperature of  43°F.    In  Phase  II,  8.4  and  6.1  psi
fuels  were   used  with   an  ambient   temperature  of  55°F.
Descriptions of  the test vehicles  are  included  in  Table Bl of
the Appendix.  Note that all were California vehicles.

     (2)    Four of  the  test vehicles  in the  1988 GARB  five-car
evaluation project  (discussed in  A.3  above)  were  1981+  model
year vehicles, as described in Table A3  of the Appendix.

     (3)    In 1987,  ATL,  under the sponsorship of API,  tested
six  vehicles  at  four  fuel   volatility   levels  and  at  three
ambient temperatures [9] using a randomized  test  sequence.   The
four fuels were  10.5,   9.0, 8.0,  and 6.5 psi in  fuel  RVP.   The
ambient temperatures were 42,  55, and 80°F.   Out  of  72  tests (6
vehicles  x 4  fuels  x 3 temperatures),  all were run twice,  with
4 tests  run  three  times.   Descriptions  of  these  six  vehicles
are included in Table B2 of the Appendix.

     (4)    In addition  to  those  EFP data  included in MOBILE4
analyses   [1,2,3],   EPA  has since  tested  ten  model  year  1987
fuel-injected vehicles  at  ambient  temperature of  50°F  and at
three  fuel  volatility  levels:   14.6,  11.7,  and 9.0  psi  RVP.
Descriptions of  these  ten vehicles are included  in  Table B3 of
the Appendix.
                              -3-

-------
3.0  DATA ANALYSIS


     The following criteria were used in analyzing the data:

     (1)   Each test program was analyzed separately.

     (2)   Within  each test  program,   data  from  each  ambient
temperature level were examined.

     (3)   Since  the  sample sizes  were  relatively  small  (at
each  temperature   level   and   for  each   test  program),   a
statistical significance level of 0.10  was used.

     (4)   One  of  the  test vehicles   from  the  1975  Chevron
program (1973 Pontiac  Catalina) was treated as  if it  were  two
vehicles,  since  it was tested  with  the original carburetor  and
tested  again  with  a  replacement carburetor  (see A.I,  section
2.0) .

     (5)   Between  the two  9.1  psi fuels  used  in  the  Exxon
program (as discussed  in  A.2,  section  2.0),  emissions  from  the
"high octane  Indolene" were used  in this analysis,  since this
is the type of fuel used for certification.

     (6)   The  model  year  1978  vehicle  tested  by  GARB  was
separated from  the other  four  1983+ vehicles  in  the analysis,
since  it  was  built  to  comply with   a  different  evaporative
emission standard.

     (7)   One  of  the  test vehicles   from  the  1988  Chevron
program (1983 Toyota SR-5 pickup) was excluded from  the current
analysis,  as the vehicle was of a different type  (LDGT1).

     Three types  of statistical procedure were  used  to examine
the impact  of fuel volatility  on exhaust emissions:   paired-t
test,  student-t test,  and regression analysis.   Descriptions of
the statistics  and results  are summarized  in  the  following
sections.
                              -4-

-------
3.1  Paired-t Test

     The  paired-t  test  is  a  statistical  procedure  used  to
examine the  effects (exhaust emissions) of  a  certain treatment
(fuel volatility)  which has  been applied  to  the  same  sample
population  (test  vehicle)  under similar  conditions  (i.e.,  at
the same ambient  temperature, FTP speed cycle, etc.).   This  is
a  more  powerful   statistical   tool   than  the  commonly  used
student-t  test   to  be  discussed   later.     In  general,   the
statistic used in paired-t  test  has  the form [10]:

           T = D / s(D)

where:  D  = mean of the differences  Dt/ for i = l,2,...,n,
     s(D) = standard deviation of the differences.

This T statistic  is distributed as  a  t-distribution  with (n-1)
degrees  of  freedom.   The  hypothesis  is  that  the  mean  of  the
differences (D) equals zero.  This hypothesis  is to  be rejected
if  the  absolute  value of   T  is greater  than  a critical value
estimated  from  the t-distribution,   based  on  a  significance
level of 0.10 and sample size n.

     For  each  pollutant, Dt  was calculated as the  difference
between emissions from a high RVP fuel and those from a low RVP
fuel on each vehicle at a given temperature:

           DI = Edigh RVP ~~ ELOW RVP

If  replicate  tests on  a  given  vehicle were  available,  the
average  of   the  emissions   measured    on  each  test   were  used.
Means and  standard deviations  of the differences (Dt)  at  each
temperature  level from  each  test program  were calculated  and
summarized.    Results  from  pre-1980  vehicles  are presented  in
Table 1, and results from 1980+ vehicles are shown in Table 2.

     Also  listed  in  Tables  1  and  2  are  the  probabilities
calculated  from  the  t-distribution.   Probabilities  designate
the numerical  value associated  with  the chances  that  D being
different from  zero are caused  by random  error.  Therefore,  if
the  probability  value  is   less than  or  equal  to   0.10,  this
implies  that D  is  significantly different  from zero,  or  that
the difference between emissions from a high and from a low RVP
fuel is statistically significant.

     The average  percent emission change  is  calculated from the
percent emission change for each vehicle:

           Di / EHigh RVP * 100.0
                              -5-

-------
Therefore,   a  negative  average  percent  emission  change  would
indicate that  on the  average the  emissions  are  increased  by
using a lower RVP fuel.

     In evaluating  paired-t  test  results,   it  should be  noted
that  the  effect  of  fuel  volatility  on exhaust  emissions  is
significant  if  and only  if  there  is  consistency among  all
statistics  being evaluated.

     For example, a determination that  a lower  RVP fuel  (say
6.5 psi  compared   to  8.0 psi)   would  increase   exhaust   HC
emissions at 55°F,  results  from  pairs  of test results at  these
two fuel volatility levels from the API/ATL  program,  where:

     (1)   the   means   (D's)   of  the   exhaust  HC   emission
differences are negative,

     (2)   the probabilities  that  D's  being  equal  to zero  are
less than or equal to the chosen significance level,

     (3)   the  average  percent  emission   change   values   are
negative (consistent with the mean emission  differences), and,

     (4)   there  is a consistent  trend  in  the occurrences  of
the  above   three criteria  at  similar  temperature  levels  and
across  the  test  programs.    For  example,   if  the  exhaust  HC
emission  differences  for  8.4  vs.   6.1 psi  RVP  fuels  were
significant  at  55°F from  the  Chevron  test  program,   it  is
logical  to  expect  that  the  emission  differences  for  similar
volatility  levels  (8.0  vs.  6.5  psi  RVP  fuels)  should  also  be
significant  at  similar  temperatures  (55°F)  from   the  API/ATL
test program.
                              -6-

-------
3.1.1.   Pre-1980 Vehicles

     On examining test results  from  pre-1980  vehicles, the 1978
vehicle  tested  by  GARB  at  75°F  showed  very  large  percent
emission increases from using  a lower RVP fuel.  Since this was
the only pre-1980 vehicle tested by  GARB, and  the only 8.6 vs.
7.2 psi  RVP  emission levels,  no statistical  comparison  can be
made, and thus no conclusion can be drawn.

     For other pre-1980 vehicles,  as can be seen from Table 1,
the majority  of the  paired-t  significance  levels  are greater
than 0.10,  with the following exceptions:
      Test
     Program

     Chevron

      Exxon
      Exxon
      Exxon
op


55
Fuel RVPs
  (psi)

11.8/8.3
75    11.8/6.6
75     8.8/6.6
75     6.8/6.6
Pollutant(s) that showed
 Significant Differences

         HC,CO

           NOx
         HC,NOx
           NOx
This  implies  that,  except  for  the  above  cases,  there  is  no
significant difference  in  emissions  when different  volatility
fuels are used.

     At  each  temperature  level,  the  average  percent  emission
changes, either  increasing or decreasing  by using a  lower RVP
fuels, are also  examined.   The majority of the emission changes
are  small  (say,   within   10  percent),  with   the   following
exceptions:
      Test
     Program

     Chevron

      Exxon
      Exxon
      Exxon
      Exxon
      Exxon
      Exxon
      Exxon
55

75
75
75
75
75
75
75
Fuel RVPs
  (psi)

11.8/8.3

11.8/6.8
11.8/6.6
 9.4/9.1
 9.4/6.8
 9.4/6.6
 9.1/8.8
 8.8/6.6
                  Pollutant(s) that showed
                 10% or more Emission Change
                   Increase      Decrease
HC,CO
 NOx
 HC
CO
CO
CO
CO
CO

NOx
Note that the percent emission changes  from  the Chevron program
at 55°F for  both  HC and CO (11.8 vs. 8.3 psi RVP) and the Exxon
program at 75°F (8.8 vs.  6.6  psi RVP) for HC and NOx were also
statistically significant.
                              -7-

-------
     However,  there is no consistency  in the signs of  the  mean
emission  differences.    For  example,  at  75°F,  mean  emission
differences from  the  Chevron program  at 11.8  vs.  8.3 psi  RVP
for both HC and CO  are  negative (i.e.,  the  HC  and CO emissions
from the lower RVP  fuel  were higher than those from  the  higher
RVP fuel).  But  from  the Exxon program  at the  same temperature
level and  similar fuel  volatility  levels (11.8  vs.   8.8  psi),
the mean HC  and  CO differences  are positive (i.e.,  the  HC and
CO emissions from the lower  RVP fuel were lower  than emissions
from the higher RVP fuel).

     Therefore, it  is concluded  that,  based  on the current data
from pre-1980  vehicles,  the fuel volatility  effect on vehicle's
exhaust  emissions   at   75°F   may  not  be   significant   and
consistent.  The  Chevron test program data  at  55°F  (11.8  vs.
8.3 psi) showed both  significant and consistent fuel volatility
effect on exhaust HC and CO emissions.
3.1.2  1980+ Vehicles

     For  1980  and  later  model year  vehicles, as  can be  seen
from Table  2,  the  majority  of the  paired-t  probabilities  are
also greater than 0.10, with the following exceptions:

      Test             Fuel RVPs    Pollutant(s) that showed
     Program     °F      (psi)       Significant Differences

     API/ATL     42     9.0/8.0              NOx

     EPA         50    14.6/11.7             NOx
     EPA         50    14.6/9.0              NOx
     Chevron     55     8.4/6.1           HC,CO,NOx
     API/ATL     55     8.0/6.5              HC

     API/ATL     80    10.5/9.0              HC
     API/ATL     80    10.5/8.0              HC

     At  each  temperature  level,  the  average percent  emission
changes,  either  increasing or decreasing  by using a  lower  RVP
fuels, are also examined.   The majority of the emission changes
are within 10 percent, with the following exceptions:
                              -8-

-------
                                   Pollutant(s)  that showed
      Test             Fuel RVPs  10% or more Emission Change
     Program     ^F      (psi)      Increase      Decrease

       EPA       50    14.6/11.7       CO
       EPA       50    14.6/9.0         -            NOx
     Chevron     55     8.4/6.1        HC

     API/ATL     80    10.5/9.0         -            CO
     API/ATL     80    10.5/8.0         -          HC,CO
     API/ATL     80    10.5/6.5         -            CO

     At the 42/43°F temperature  level,  only NOx  emissions from
the API/ATL test program  for  9.0 vs. 8.0 psi RVP fuels showed a
statistically significant  mean emission decrease.   The average
percent  emission  change  values  are  all  small  (within  ten
percent increase or decrease).  Therefore,  it  is  concluded that
at  42/43°F  temperature level,  there is no  conclusive evidence
of RVP effect on exhaust emissions.

     At 50/55°F, only the  1988  Chevron program  data  (8.4  vs.
6.1 psi RVP fuels)  showed significant mean emission differences
(increases)  for  all  three  pollutants,  and  close to  an  eleven
percent  HC  emission  increase.    For   similar  fuel  volatility
levels  (at  8.0  vs. 6.5  psi fuels),  the  HC emission increases
from  API/ATL  program was  also significant.  Therefore,  by the
consistency criterion, it is concluded  that  the  fuel RVP effect
on  HC emissions at  50/55°F  may be  significant.   There  is  no
significant  (and  no  consistent)  RVP  effect  on  CO  or  NOx
emissions.

     At   the    75-80°F   temperature   range,  the   probability
statistics  showed  significance  for  HC  emissions  reduction  by
reducing  fuel  volatilities from 10.5  to  9.0,  and  10.5  to 8.0
psi RVP.  However,  the significance  is  lessened  when lower than
9.0  psi  fuel   is  used.   Also,  the  emission  increase   on  CO
emissions when 10.5 psi RVP fuel  is  used in the  API/ATL program
is ten  or more  percent,  although their associated probabilities
are not  significant.   The API/ATL  test results  are consistent
with  the MOBILE4  assumption  —  that there  is  a  small  fuel
volatility  effect   on  exhaust   HC/CO  emissions   when  fuel
volatilities are higher than  the certification level of 9.0 psi
and ambient temperatures are 75°F and higher.
                              -9-

-------
3.2  Student-t Test

     Since  some of  the  test  programs   (such  as  Chevron  and
API/ATL)  used  randomized  test  sequences,  the  paired-t  test
procedure  may  not  be   appropriate.   For   this   reason,   the
student-t  test procedure  was  also  used  for evaluation.   The
student-t test  is  a statistical  procedure used to  compare two
sample   means  and   their   variances.    Therefore,   at   each
temperature level  from each  test  program,  the average exhaust
emissions  from a  high volatility fuel can  be compared  to the
average emissions  from a  low volatility fuel, if  the variances
of the two samples are statistically equal [10]:

           T =  (E1  - E2) / SD

where:  Ei = average emissions from a high volatility fuel,
        E2 = average emissions from a low volatility fuel,
        SD = standard deviation.

The  T   statistic   is   distributed  as  a  t-distribution  with
(nl+n2-2) degrees  of freedom.  The  hypothesis  is that  the two
sample  means   are  equal  (or,  Ei   -  E2 =  0).  This  hypothesis
is to be  rejected  if the  absolute value of T is  greater than a
critical  value estimated  from  the  t-distribution,  based  on a
significance level of 0.10 and the two sample sizes (nl and n2).

     The two sample emission variances can be compared based on:

           F =  SI2 / S22

where:

  SI2 = emission variance from a high volatility fuel, and
  S22 = emission variance from a low volatility fuel.

This F  statistic  is  distributed as a f-distribution with (nl-1)
and  (n2-l)  degrees  of  freedom.   The  hypothesis  is  that  the
ratio of  the  two  variances equals unity.   This hypothesis is to
be  rejected if  the numerical  value  of  F  is greater than  a
critical  value estimated  from  the  f-distribution,  based  on a
significance level of 0.10 and the two sample sizes (nl and n2).

     For  each  pollutant,  means  and standard  deviations  were
calculated  and summarized  at each temperature level  from  each
test program.   Results  from pre-1980 vehicles are  presented in
Table 3, and results from 1980+ vehicles are shown in Table 4.
                             -10-

-------
     Also listed in Tables 3 and 4 are three probabilities:

     (1)  The probability that  the  two average  emissions  being
different from  zero  is  caused by  random  error  — denoted  as
"t-probability."

     (2)  The probability that the  ratio  of  the  two  emission
variances being  different from unity  is  caused  by  random error
— denoted as "f-probability."

     (3)  The  probability  that  the  average  emissions  from  a
high RVP  fuel  are  lower  than the average  emissions  from a low
RVP fuel.  Note  that  this probability has the minimum value  of
0.0  and the maximum  value  of  1.0,  and is not  calculated when
the sample size is less than 5.

     In evaluating  student-t  test results, criteria  similar  to
those  used   for  evaluating  the  paired-t  test  are also  used.
These  criteria  include:   average emissions  being significantly
different (i.e.,  the  t-probability value needs  to  be  less than
or equal to 0.10), and a consistent trend  in  the occurrences  of
significance  at  each temperature  level  and  across  the  test
programs.

     An  additional  criterion  is  used  in  evaluating  student-t
test   results.    The   numerical  values   associated   with  the
probability that  the  mean emissions  from a high  RVP  fuel  being
lower  than  the  mean  emissions from  a  low RVP  fuel  should  be
closer  to the two extremes, either  larger than  0.75  or  smaller
than 0.25.   If'this probability value is 0.50, this would imply
that there  is  a  50-50  chance that the average  emissions from
one  RVP level  fuel will  be lower (or  higher)  than the  average
emissions  from   another   RVP  level   fuel,   hence   the   fuel
volatility effect on  emissions is not significant.  A  note  of
caution is that even if the probability values  are  close to the
two  extremes,   most  of   the  test  programs  examined  here  had
relatively small  sample sizes (ranging from 4 to 13).

3.2.1  Pre-1980 Vehicles

     For pre-1980 vehicles, as can be seen  from  Table  3,  all  t-
and  f-probabilities are greater than  0.10,  with the majority of
the  t-probabilities greater than  0.90.  The  probabilities that
the  high  RVP fuel  emissions  being  lower than  the  low RVP fuel
emissions  are  mostly between  0.30   and  0.70   (with  only  one
exception:   from Exxon program at  75°F,  the  probability that
the  average  NOx emissions  from 8.8  psi  fuel  being lower than
the  emissions  from 6.6  psi  fuel is  0.27).   Therefore,  it  is
concluded that  there  is  no statistically significant fuel  RVP
effect on exhaust emissions at any temperatures.
                             -11-

-------
3.2.2  1980+ Vehicles

     For  1980+ vehicles,  as  can be  seen  from Table  4,  all
t-probabilities  at  each  temperature   level  are  greater  than
0.10.   The  majority  of the  f-probabilities  are  also  greater
than 0.10, with the following exceptions:

                                   Pollutant(s) that showed
      Test             Fuel RVPs    Significant Differences
     Program     ^F      (psi)       in Emission Variances

     EPA         50    14.6/11.7             HC,CO
     EPA         50    11.7/9.0              HC,CO

     API/ATL     80    10.5/8.0              CO
     API/ATL     80    10.5/6.5              CO
     API/ATL     80     9.0/8.0              CO

Note that the significant  differences  in emission  variances are
simply  an  indication that the  degrees  of  dispersion within one
sample are different from those in another sample.

     The  probabilities  that  the  high  RVP  fuel emissions  are
lower than  the low  RVP  fuel emissions  are  also mostly between
0.30 and 0.70, with the following exceptions:

                                                   Probability
      Test             Fuel RVPs                  that 1st Mean
     Program     ^F      (psi)      Pollutant     is < 2nd Mean

     EPA         50    14.6/11.7       NOx          0.21
     EPA         50    14.6/9.0        NOx          0.10
     EPA         50    11.7/9.0        HC           0.29
     EPA         50    11.7/9.0        CO           0.26
     EPA         50    11.7/9.0        NOX          0.29

     API/ATL     80    10.5/8.0        HC           0.23
     API/ATL     80    10.5/8.0        CO           0.16
     API/ATL     80    10.5/6.5        CO           0.19
     API/ATL     80     9.0/8.0        CO           0.29

     Note that  all the above  exceptions  lead  to the conclusion
that the  emissions  from a high RVP fuel  are more likely  to be
higher than the emissions from a low RVP fuel.

     All  these   results  suggest  that   the   average  exhaust
emissions  (HC,  CO,  and  NOx)   from  a  high  RVP fuel  are  not
statistically different than  the  average emissions from a lower
RVP fuel.
                             -12-

-------
4.0  REGRESSION ANALYSIS


     Regression   analysis   is  a   statistical  tool   used  to
determine the  relation between a  set  of  independent  variables
(fuel  volatility  and  ambient  temperature)  and  a  dependent
variable  (exhaust  emissions)  so   that  the  latter   can  be
estimated from the  former.    Since  the  main  purpose  of  this
report  is  to  examine  the  fuel  volatility  effect  on  exhaust
emissions,  the  regression  analysis here serves as a final step
to see  if there  exists any  effect on exhaust  emissions  under a
combination  of   factors   such  as   fuel   volatility,  ambient
temperature,  and the interaction of the two.

     The  results  presented   here  are  in  a  simplified  form.
Basically,  the regression model has the following form:

     E = exp [  a + b*RVP + c*Temp + d*RVP*Temp + e*RVP2
         + f*Temp2 + g*(RVP*Temp)z ]

     where:     E = exhaust emissions in g/mi,
             RVP = fuel volatility in psi,  and,
            Temp = ambient temperature in °F.

A  backward  selection  process is  used  to  determine  the  final
equation  form.    That   is,   a  full  set  of  all  independent
variables,  whenever  appropriate,  is used  in the first  step of
regression  analysis.    The  resulting  coefficients   are  checked
for  significance.   The most  insignificant  variable  (with  the
coefficient having the largest numerical  value in significance
and greater than 0.10) is to be removed  before the execution of
the  next  step of  regression  analysis.   This  selection process
will   continue   until   all   the  variables   remaining   have
coefficients with  significance  levels  less  than  or  equal  to
0.10.  Therefore,  at  the  end  of  the final  step,  the  remaining
independent   variables   are   those   that   would   contribute
significantly in the prediction of the dependent variable.
                             -13-

-------
4.1  Pre-1980 Vehicles

     The  1975  Chevron and  the  1979  Exxon programs  data  were
used  in the  regression  analysis.   Note  that  there  were  two
temperature levels (55 and 75°F) and two  fuel  volatility levels
(8.3  and  11.8  psi  RVP)  in  the  1975  Chevron test  program.
Therefore, only the constant term (a in the above  equation)  and
three  first  degree   independent   variables   (RVP,   Temp,   and
RVP*Temp)  were  used  in  the  initial  step   of  the  backward
selection  process.   The  1979  Exxon program  data were  at  75°F
only (but  with  fuel  volatilities ranging  from 6.6 to  11.8  psi
RVP), thus only the  constant  term and two RVP related variables
(RVP,  RVP2)  were  used  in  the  regression  analysis.   Results
are summarized in the following:
     Test Program
     1975 Chevron
     1979 Exxon
Pollutant   Significant Variables
  HC
  CO
  NOx

  HC
  CO
  NOx
Constant
Temp
Constant

Constant
Constant
Constant
     As can  be seen from the above,  for  pre-1980  vehicles fuel
volatility  is  a  non-significant  independent  variable for  all
pollutants across both test programs.
4.2  1980+ Vehicles

     Data from  two test  programs  were used  in the  regression
analysis:   the  1988  Chevron  and  the  1987  API/ATL  programs.
Note that there  were two temperature  levels  (43 and  55°F)  and
two fuel  volatility levels  (8.1  and 11.4 psi RVP)  in the 1988
Chevron  test  program.   Therefore,   only  the   constant  term
(denoted  as  "a"  in  the  above  equation) and  three  first-degree
independent  variables  (RVP, Temp,  and RVP*Temp)  were used  in
the initial  step of the  backward selection process.   The 1987
API/ATL program  had three  levels  of  temperature  (42, 55,  and
80°F)   and four  levels of  fuel  volatility  (6.5, 8.0,  9.0,  and
10.5  psi RVP),  thus  the  entire  seven  independent  variables
(defined  previously)  were  used in  the  regression  analysis.
Results are summarized in the following:
                             -14-

-------
     Test Program      Pollutant   Significant Variables

     1988 Chevron        HC            RVP,  RVP*Temp
                         CO            Temp
                         NOx           Constant

     1987 API/ATL        HC            RVP,  RVP*Temp
                         CO            Temp2
                         NOx           Temp2

     As can  be  seen, for  1980+ vehicles fuel  volatility  is_ a
significant  independent  variable  for  HC emissions  across both
test programs.
                             -15-

-------
5.0  CONCLUSIONS

     The effect  of  fuel volatility  on exhaust emissions  based
on data  not  included in the MOBILE4 model  was examined.   Three
types of statistical procedure  were  used to examine  the  impact
of  fuel  volatility  on   exhaust  emissions:    paired-t  test,
student-t   test,   and   regression    analysis.    Results   are
summarized in the following:
Statistical
 Procedure
Paired-t
Student-t


Regression




Paired-t
Student-t
Regression
            Results
        Pre-1980 Vehicles

a) The 1975  Chevron  program data at  55°F showed
significant  fuel  volatility  effect  (11.8  vs.
8.3 psi  fuel)  on  HC/CO  emissions,   resulting in
higher emissions when low RVP fuel was used.

b) For  all other  test programs  at  75°F  the RVP
effects were non-significant.

No statistically  significant fuel RVP  effect on
exhaust emissions at any temperatures.

Fuel volatility is not a significant independent
variable for all pollutants.

          1980-t-  Vehicles
a)  At   42/43°F,
non-significant.
the   fuel   RVP   effect
                                                             was
b)  At  50/55°F,  both  Chevron  and API/ATL  data
showed significant  fuel  RVP  effect (8 vs.  6 psi
RVP)  resulting  higher HC  emissions  under  lower
RVP fuel.

c)  At 75/80°F,  API/ATL  data  showed  significant
HC/CO reductions with lower RVP fuel.

Some   showed   significant  differences   in   the
variances.  No  significant  differences  on  the
average  emissions.   Overall, the  emissions  from
one level RVP fuel  are  likely to be  the  same as
the emissions from another level of RVP fuel.

For Chevron and  API/ATL programs,  fuel RVP  is a
significant   independent    variable   for    HC
emissions.
                             -16-

-------
     The algorithm used  in  MOBILE4  was based on the use  of  the
more  powerful  paired-t  test  procedure,  since  in  EPA's  EF
testing program vehicles were tested with  the  higher  in-use RVP
fuel first.   Both Chevron  and  API/ATL program data were based
upon randomized  test  sequence  results.   The  conclusions  from
the student-t  test  would be more applicable for  those studies.
Even if the paired-t test results were  examined  for  Chevron and
API/ATL program  data,  the  majority of   the  data had  shown no
fuel volatility  impact on  exhaust  emissions,  with  significant
fuel RVP  impact  only at 50/55°F  based  on  a relatively small
sample size.

     The  pre-1980   vehicles   were  equipped  with   oxidation
catalysts and  the cold start was controlled mostly  by a choke.
It is  reasonable  to  believe that the cold start portion  of  the
FTP test  might result  in slightly  higher exhaust emissions by
using a lower RVP fuel,  especially  when the  ambient  temperature
is relatively  low and when  vehicles may not be in a "top-notch"
condition.

     For  1980+ vehicles, however,   higher  exhaust HC  emissions
occurred  only  when  8.0 and  6.0 psi fuels  were compared and the
ambient temperature  was  at  50/55°F.  The sample sizes  of these
two test  programs (Chevron  and API/ATL) that showed significant
fuel  RVP  impacts were relatively small  (with  N=13  and  6,
respectively).   Further, data  from  recent  fuel  surveys  [11]
showed  that  only  very  few   brands   of   gasoline   with  fuel
volatility  lower  than  8.0  psi  were commercially  available (in
cities  like Albuquerque, Las Vegas,  Los Angeles,  and Phoenix,
during the  summer months).   The lowest fuel volatility surveyed
was  7.5 psi RVP  from Phoenix  in  the  summer  of  1988,   for  an
unleaded premium gasoline.

     As suggested by the data,  using fuels with volatilities at
6.0 or  6.5  psi RVP  could result in a  small  exhaust  HC emission
increase  at  50/55°F  ambient.   But  the benefit of using a higher
psi fuel to offset this  small exhaust  HC emissions increase for
the  early  morning   trip even   in  the  high   temperature  ozone
season is probably cancelled out by the higher  evaporative hot
soak HC  emissions generated in the midday when  the  majority of
daily trips  occur and the  ambient  temperatures  can be  over 90
degrees Fahrenheit.

     To summarize,   in  analyzing the  currently  available data,
there  were  inconsistencies  in results.   Under  the  unlikely
combination of relatively low temperature and less than 8.0 psi
RVP fuel volatilities, a small  increase  in exhaust HC emissions
is  negligible when  compared with  the benefit of a  relatively
larger decrease  in  evaporative  hot  soak emissions.    For  these
reasons,  the MOBILE4  model  assumed  no fuel volatility effect on
exhaust emissions when the  fuel RVP is below  the  certification
level of 9.0 psi.
                             -17-

-------
     This  assumption  is adequate  for most  of  the areas  with
moderate  ambient  temperature  profiles   in  high  ozone  season
(with  daily  minimum  of  60  to  the  maximum  of  84°F) .    The
combination  of  lower   (say,  50/55°F) ambient  temperature  and
lower (6.0) psi RVP fuel is expected to be rare.   Even in  areas
with  more  extreme ambient  temperature  profiles  (say,  70  to
96°F),  the  suggested  fuel  volatility  required  by  1992  is
7.8 psi  RVP in  EPA's  Final Rule  for Phase II  Fuel  Volatility
Control.

     It  is  concluded  that the  assumptions  used in MOBILE4  are
adequate  for  all  situations  likely  to  be encountered  in  real
world  modeling.    There  may  be uncertainties  when  estimating
exhaust  emissions  in areas  with  low  RVP fuels (less  than  9.0
psi)  at  temperatures  greater than 40°F.   More  data  should  be
collected   and   analyzed  before   these   uncertainties  can  be
addressed.
                             -18-

-------
                            Table 1
                     Paired-t Test Results
                       Pre-1980  Vehicles
Amb
Temp.
(°F)
55
75
75
75
75
75
75
Fuel
RVP
(psi)
11.8
vs.
8.3
8.6
vs.
7.2
11.8
vs.
8.3
11.8
vs.
9.4
11.8
vs.
9.1
11.8
vs.
8.8
11.8
vs.
6.8
               N
Mean
Emission
Difference
r Poll (g/mi)
Probabilities
that the
Differences
Standard are from
Deviation Random Error
Chevron Data (1973-76 Vehicles)
8
8
7
HC
CO
NOX
-0.1561
-1.6387
-0.0329
GARB Data (
1
1
1
HC
CO
NOx
-0.2220
-2.9700
0.2100
0.16214
2.39650
0.08321
1978 Vehicle)
0.0296*
0.0943*
0.3364
-
Chevron Data (1973-76 Vehicles)
8
8
7
8
8
8
8
8
8
8
8
8
8
8
8
HC
CO
NOx
Exxon
HC
CO
NOx
HC
CO
NOx
HC
CO
NOx
HC
CO
NOx
-0.0359
-0.7213
-0.0114
0.15777
2.81200
0.09651
0.5406
0.4917
0.7647
Data (1974-77 Vehicles)
-0.0186
-0.3258
0.0164
0.0185
0.3456
0.0729
0.0189
0.5236
-0.1018
-0.0205
0.8361
0.0195
0.16978
1.04230
0.27891
0.07007
1.23500
0.28561
0.18292
2.85160
0.26899
0.23953
1.59310
0.14529
0.7654
0.4060
0.8728
0.4795
0.4546
0.4939
0.7789
0.6195
0.3202
0.8157
0.1813
0.7155
 Average
 Percent
Emission
  Change
                                                            -18.87
                                                            -12.70
                                                             -2.00
                                                            -25.00
                                                            -29.38
                                                             17.65
                                                             -5.60
                                                              1. 15
                                                             -0.38
                                                              4.95
                                                             -7.69
                                                             -0.91

                                                              1.25
                                                              4.64
                                                              1.89

                                                              8.46
                                                              4.09
                                                             -7.56

                                                              5.98
                                                             10.73
                                                             -1.62
*  Probability  value   is  less  than  0.10,   an  indication  that  the
emission differences are significant.
                             -19-

-------
                        Table 1 (Continued)
                       Paired-t Test Results
                         Pre-1980 Vehicles
Amb
Temp.
(°F)
75


75


75


75


75


75


75


75


75


Fuel
RVP
(psi)
11.8
vs.
6.6
9.4
VS.
9.1
9.4
vs.
8.8
9.4
vs.
6.8
9.4
vs.
6.6
9.1
vs.
8.8
9.1
vs.
6.8
9.1
vs.
6.6
8.8
vs.
6.8
               N


Mean
Emission
Difference
r
Poll
(g/mi)
Exxon Data
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
HC
CO
NOx
HC
CO
NOx
HC
CO
NOx
HC
CO
NOx
HC
CO
NOx
HC
CO
NOx
HC
CO
NOx
HC
CO
NOx
HC
CO
NOx
-0.0488
0.4936
0.1274
0.0371
0.6714
0.0565
0.0375
0.8494
-0.1181
-0.0019
1.1619
0.0031
-0.0301
0.8194
0.1110
0.0004
0.1780
-0.1746
-0.0390
0.4905
-0.0534
-0.0673
0.1480
0.0545
-0.0394
0.3125
0.1213

Standard
Deviation
(Continued)
0.21001
2.40380
0.07821
0.19777
1.46260
0.29223
0.13328
3.63790
0.35424
0.14134
2.25150
0.30659
0.15198
3.19480
0.30274
0.20357
2.85570
0.37707
0.26185
1.21800
0.24638
0.23439
2.26850
0.27382
0.07394
1.98830
0.24735
Probabilities
  that the
 Differences
  are from
 Random Error
                                                 0.5324
                                                 0.5796
                                                 0.0025*

                                                 0.6119
                                                 0.2353
                                                 0.6015

                                                 0.4523
                                                 0.5301
                                                 0.3770

                                                 0.9711
                                                 0.1878
                                                 0.9778

                                                 0.5925
                                                 0.4917
                                                 0.3342

                                                 0.9960
                                                 0.8650
                                                 0.2316

                                                 0.6862
                                                 0.2922
                                                 0.5594

                                                 0.4438
                                                 0.8588
                                                 0.5910

                                                 0.1757
                                                 0.6701
                                                 0.2082
 Average
 Percent
Emission
  Change
                -1.02
                10.73
                 6.95

                -5.28
                10.14
                 1.57

                 2.90
                 8.41
                -7.86

                 0.64
                14.53
                -2.42

                -7.69
                14.25
                 5.90

                 6.14
                -2.57
                -11.15
                 3.35
                 5.24
                   80
 -4
                -3.66
                 6.23
                 3.59

                -2.51
                 6.66
                 4.42
*  Probability value  is  less  than  0.10,  an  indication  that  the
emission differences are significant.
                             -20-

-------
                        Table  1  (Continued)
Amb
Temp.
75
75
Fuel
 RVP
(psi)  N
8.8
vs.
6.6

6.8
vs.
6.6
8
8
8

8
8
8
                       Paired-t Test Results
                         Pre-1980 Vehicles
          Mean
        Emission
       Difference
   Poll  (g/mi)
               Standard
               Deviation
                   Probabilities
                     that the
                    Differences
                     are from
                    Random Error
                       Exxon Data (Continued)
HC
CO
NOx

HC
CO
NOx
-0.0676
-0.0300
 0.2291

-0.0283
-0.3425
 0.1079
0.06025
1.19070
0.27782

0.09717
1.76110
0.10130
0.0156*
0.9452
0.0524*

0.4380
0.5994
0.0196*
                         Average
                         Percent
                        Emission
                          Change
-10.62
  6.87
 11 . 53

 -8.83
 -1.25
  7. 58
*  Probability  value  is   less  than  0.10,
emission differences are significant.
                                     an  indication that  the
                             -21-

-------
                              Table 2
                       Paired-t Test Results
                           1980+ Vehicles
Amb
Temp.
Fuel
RVP
(psi)
N
Probabilities
Mean that the
Emission Differences
Difference Standard are from
Poll (g/mi) Deviation Random Error
Chevron Data (
43


11.
vs
8.
4
t
I
11
11
11
HC
CO
NOx
-0
-0
-0
.0245
.2940
.0154
API/ATL Data (
42


42


42


42


42


42


10.
vs
9.
10.
vs
8.
10.
vs
6.
9.
vs
8.
9.
vs
6.
8.
vs
6.
5
,
0
5
,
0
5
,
5
0
,
0
0
.
5
0
.
5
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
HC
CO
NOx
HC
CO
NOx
HC
CO
NOx
HC
CO
NOx
HC
CO
NOx
HC
CO
NOx
-0
-0
-0
0
-0
-0
0
-0
-0
0
-0
0
0
0
0
0
0
0
.0158
.1868
.0242
.0058
.6477
.0042
.0212
.0910
.0003
.0217
.4608
.0200
.0370
.0958
.0238
.0153
.5567
.0038
1981-83 Vehicles)
0
0
0
.05334
.70524
.05239
0
0
0
.1593
.1969
.3537
1983-86 Vehicles)
0
1
0
0
0
0
0
1
0
0
1
0
0
1
0
0
1
0
.05272
.41350
.08108
.06246
.85635
.09735
.13222
.40310
.09061
.05820
.25630
.02258
.13564
.32390
.06949
.13396
.30050
.06911
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
.4950
.7592
.4981
.8281
.1231
.9206
.7111
.8800
.9932
.4036
.4101
.0822*
.5336
.8662
.4391
.7904
.3424
.8972
                                                             Average
                                                             Percent
                                                            Emission
                                                              Change
                                                             -4.29
                                                             -3.27
                                                             -1 . 10
                                                             -2.32
                                                             -0.49
                                                             -4.20

                                                             -0.34
                                                             -7.91
                                                             -1.42

                                                             -3.51
                                                             -2.40
                                                             -0.49
                                                              1.85
                                                             -8.53
                                                              2.79

                                                             -1.21
                                                             -3.02
                                                              3.23

                                                             -3.15
                                                              4.48
                                                              0.42
*  Probability value  is  less  than  0.10,  an  indication  that  the
emission differences are significant.
                             -22-

-------
                        Table 2 (Continued)
                       Paired-t Test Results
                           1980+ Vehicles
Amb
Temp.
Fuel
RVP
(psi)
N
Probabilities
Mean that the
Emission Differences
Difference Standard are from
Poll (q/mi) Deviation Random Error
EPA Data (1987 Vehicles)
50


50


50


14.
vs
11.
14.
vs
9.
11.
vs
9.
6
,
7
6
t
0
7
t
0
10
10
10
10
10
10
10
10
10
HC
CO
NOx
HC
CO
NOx
HC
CO
NOx
-0
-3
0
0
1
0
0
4
0
.1197
.2380
.0760
.0901
.0070
.1260
.2098
.2450
.0500
Chevron Data (
55


8.
vs
6.
4
(
1
11
11
11
HC
CO
NOx
-0
-0
-0
.0453
.6707
.0192
API/ATL Data (
55


55


55


10.
vs
9.
10.
vs
8.
10.
vs
6.
5
,
0
5
,
0
5
.
5
6
6
6
6
6
6
6
6
6
HC
CO
NOx
HC
CO
NOx
HC
CO
NOx
0
-0
0
0
0
-0
-0
0
-0
.0008
.0033
.0100
.0200
.1317
.0250
.0200
.0217
.0100
0.
10.
0.
0.
3.
0.
0.
14.
0.
55136
63600
10501
19988
79760
13377
71109
12000
11106
0
0
0
0
0
0
0
0
0
.5097
.3609
.0479*
.1878
.4234
.0155*
.3752
.3666
.1883
1981-83 Vehicles)
0.
0.
0.
03536
90075
03536
0
0
0
.0017*
.0331*
.1022*
1983-86 Vehicles)
0.
0.
0.
0.
1.
0.
0.
1.
0.
04873
95143
09050
05916
04910
07113
08155
32410
09077
0
0
0
0
0
0
0
0
0
.9682
.9935
.7975
.4453
.7709
.4286
.5742
.9696
.7981
                                                            Average
                                                            Percent
                                                           Emission
                                                              Change
                                                             -3.02
                                                            -10.08
                                                              7.48

                                                              5.77
                                                              5.39
                                                             13.42

                                                              1.66
                                                              5.89
                                                              5.97
                                                            -10.56
                                                             -7.91
                                                             -3.61
                                                              0.24
                                                             -2.13
                                                              1.81

                                                              2.51
                                                              0.13
                                                             -2.59

                                                             -4.33
                                                             -3.39
                                                             -0.95
*  Probability value  is  less  than  0.10,  an  indication  that  the
emission differences are significant.
                             -23-

-------
                           Table 2 (Continued)
                          Paired-t Test Results
                              1980+ Vehicles
Amb
Temp.
iin
55


55


55


75


80


80


80


80


Fuel
RVP
(psi)
9.0
vs.
8.0
9.0
vs.
6.5
8.0
vs.
6.5
8.6
vs.
7.2
10.5
vs.
9.0
10.5
vs.
8.0
10.5
vs.
6.5
9.0
vs.
8.0
               N
Probabilities
Mean that the
Emission Differences
Difference Standard are from
r Poll (g/mi) Deviation Random Error
API/ATL Data (Continued)
6
6
6
6
6
6
6
6
6

4
4
4
HC
CO
NOx
HC
CO
NOx
HC
CO
NOx
GARB
HC
CO
NOx
0
0
-0
-0
0
-0
-0
-0
0
.0192
.1350
.0350
.0208
.0250
.0200
.0400
.1100
.0150
0.
0.
0.
0.
0.
0.
0.
0.
0.
Data (1983-88
-0
-0
0
.0095
.2375
.0025
API/ATL Data (
6
6
6
6
6
6
6
6
6
6
6
6
HC
CO
NOx
HC
CO
NOx
HC
CO
NOx
HC
CO
NOx
0
0
-0
0
1
-0
0
1
-0
0
0
0
.0300
.6050
.0167
.0558
.1400
.0108
.0208
.0013
.0258
.0258
.5350
.0058
0.
0.
0.
05181
96958
09165
04488
78920
06388
04940
64012
11962
Vehicles)
01555
43485
03304
0
0
0
0
0
0
0
0
0

0
0
0
.4064
.7469
.3925
.3070
.9412
.4777
.1041*
.6913
.7711

.3089
.3546
.8893
1983-86 Vehicles)
0.
1.
0.
0.
1.
0.
0.
1.
0.
0.
1.
0.
03271
10700
08920
05748
59360
10500
03917
59720
07883
04364
26230
09157
0
0
0
0
0
0
0
0
0
0
0
0
.0746*
.2383
.6664
.0632*
.1401
.8105
.2494
.1852
.4586
.2067
.3468
.8821
 Average
 Percent
Emission
  Change
                                                              1.97
                                                              1.85
                                                                65
 -5
                                                             -4.30
                                                             -0.76
                                                             -3.54
                                                             -6.75
                                                             -3.77
                                                              1.15
                                                             -4.14
                                                             -2.46
                                                             -3.03
                                                              7.82
                                                             13.88
                                                             -2.92
                                                             13.19
                                                             20.55
                                                             -2.93
                                                              7.51
                                                             16.86
                                                             -3.78

                                                              5.61
                                                              5.55
                                                             -0.35
*  Probability  value  is  less  than  0.10,  an  indication  that  the
emission differences are significant.
                             -24-

-------
                        Table 2 (Continued)
Amb
Temp.
80
80
               N
                       Paired-t Test Results
                           1980+ Vehicles
                  Mean
                Emission
               Difference
           Poll  (g/mi)
                   Standard
                   Deviation
                    Probabilities
                      that the
                     Differences
                      are from
                     Random Error
                      API/ATL Data (Continued)
9.0
vs.
6.5

8.0
VS.
6.5
6   HC   -0.0092
6   CO    0.3963
6   NOx  -0.0092
6   HC
6   CO
6   NOX
-0.0350
-0.1387
-0.0150
0.05572
1.09570
0.05435

0.08283
0.86434
0.06964
0.7036
0.4162
0.6966

0.3481
0.7105
0.6203
                         Average
                         Percent
                        Emission
                          Change
-0.90
 2. 69
-1.02

-8.50
-7.51
-1.21
                             -25-

-------
                               Table  3
                       Student-t Test Results
                          Pre-1980  Vehicles
 Amb
Temp.
(°F) Poll
55
75
75
75
75
75
HC

CO

NOx

HC

CO

NOx

HC

CO

NOx

HC

CO

NOx

HC

CO

NOx

HC

CO

NOx

11.8
8.3
11.8
8.3
11.8
8.3
11.8
8.3
11.8
8.3
11.8
8.3
11.8
9.4
11.8
9.4
11.8
9.4
11.8
9.1
11.8
9.1
11.8
9.1
11.8
8.8
11.8
8.8
11.8
8.8
11.8
6.8
11.8
6.8
11.8
6.8
    Average
   Emissions
N   (g/mi)
t-prob.   Variance  f-prob.
                                  0.4108

                                  0.4496

                                  0.4659




                                  0.4708

                                  0.3298

                                  0.4630
Chevron Data (1973-76
8
8
8
8
8
8
8
8
8
8
8
8
Exxon
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
8
1.07
1.23
16.51
18.15
1.41
1.44
0.97
1.01
11.29
12.01
1.42
1.43
Data
1.13
1.15
11.42
11.74
1.79
1.77
1.13
1.11
11.42
11.07
1.79
1.72
1.13
1.11
11.42
10.89
1.79
1.89
1.13
1.15
11.42
10.58
1.79
1.77
0.7036

0.7292

0.9290

0.9322

0.8763

0.9766

(1974-77
0.9734

0.9517

0.9597

0.9713

0.9480

0.8183

0.9731

0.9181

0.7723

0.9716

0.8715

0.9511

Vehicles)
0.5894
0.7034
81.8400
90.3830
0.3878
0.4147
0.6657
0.7050
68.5760
96.9190
0.4328
0.4654
Vehicles)
1.0317
1.3688
110.7300
112.9500
0.4538
0.3554
1.0317
1.0045
110.7300
105.6700
0.4538
0.3217
1.0317
1.3846
110.7300
89.5090
0.4538
0.4981
1.0317
1.5271
110.7300
95.4060
0.4538
0.3275
                                  0.3593

                                  0.4899

                                  0.3777



                                  0.4864

                                  0.4762

                                  0.3306



                                  0.3538

                                  0.3930

                                  0.4527



                                  0.3088

                                  0.4246

                                  0.3389
  Prob.
1st Mean
is < 2nd
                               0.64

                               0.63

                               0.53




                               0.53

                               0.56

                               0.51
                               0.51

                               0.52


                               0.48



                               0.49


                               0.48

                               0.41



                               0.49


                               0.46

                               0.61



                               0.51

                               0.44


                               0.48
                             -26-

-------
                         Table 3 (Continued)
                       Student-t Test Results
                          Pre-1980  Vehicles
 Arab        Fuel        Average
Temp.       RVP        Emissions
(°F) Poll  (psi)    N   (g/mi)
75    HC    11.8    8
             6.6    8
      CO    11.8    8
             6.6    8
      NOx   11.8    8
             6.6    8

75    HC     9.4    8
             9.1    8
      CO     9.4    8
             9.1    8
      NOx    9.4    8
             9.1    8

75    HC     9.4    8
             8.8    8
      CO     9.4    8
             8.8    8
      NOx    9.4    8
             8.8    8

75    HC     9.4    8
             6.8    8
      CO     9.4    8
             6.8    8
      NOx    9.4    8
             6.8    8

75    HC     9.4    8
             6.6    8
      CO     9.4    8
             6.6    8
      NOx    9.4    8
             6.6    8

75    HC     9.1    8
             8.8    8
      CO     9.1    8
             8.8    8
      NOx    9.1    8
             8.8    8
                              Prob.
                            1st Mean
t-prob.   Variance  f-prob.  is < 2nd
Exxon Data (Continued)
1.13
1.18
11.42
10.92
1.79
1.66
1.15
1.11
11.74
11.07
1.77
1.72
1.15
1.11
11.74
10.89
1.77
1.89
1.15
1.15
11.74
10.58
1.77
1.77
1.15
1.18
11.74
10.92
1.77
1.66
1.11
1.11
11.07
10.89
1.72
1.89
0.9311

0.9260

0.7011

0.9466

0.8996

0.8488

0.9499

0.8683

0.7230

0.9976

0.8232

0.9916

0.9600

0.8782

0.7219

0.9995

0.9718

0.5940

1.0317
1.4230
110.7300
107.5400
0.4538
0.3919
1.3688
1.0045
112.9500
105.6700
0.3554
0.3217
1.3688
1.3846
112.9500
89.5090
0.3554
0.4981
1.3688
1.5271
112.9500
95.4060
0.3554
0.3275
1.3688
1.4230
112.9500
107.5400
0.3554
0.3919
1.0045
1.3846
105.6700
89.5090
0.3217
0.4981
                    0.3410     0.53

                    0.4851     0.47

                    0.4258     0.36


                    0.3467     0.47

                    0.4661     0.45

                    0.4494     0.43


                    0.4941     0.48

                    0.3834     0.44

                    0.3337     0.63


                    0.4445     0.50

                    0.4148     0.42

                    0.4584     0.50


                    0.4802     0.52

                    0.4750     0.44

                    0.4504     0.37


                    0.3413     0.50

                    0.4161     0.49

                    0.2891     0.69
                             -27-

-------
                         Table 3  (Continued)
                       Student-t Test Results
                          Pre-1980  Vehicles
 Amb
Temp.
(°F) Poll
75
75
75
75
75
HC

CO

NOx


HC

CO

NOx


HC

CO

NOx


HC

CO

NOx


HC

CO

NOx
9.1
6.8
9.1
6.8
9.1
6.8

9.1
6.6
9.1
6.6
9.1
6.6

8.8
6.8
8.8
6.8
8.8
6.8

8.8
6.6
8.8
6.6
8.8
6.6

6.8
6.6
6.8
6.6
6.8
6.6
                  Average
                 Emissions
              N   (g/mi)
8
8
8
8
8
8

8
8
8
8
8
8

8
8
8
8
8
8

8
8
8
8
8
8

8
8
8
8
8
8
                                                   Prob.
                                                 1st Mean
                     t-prob.   Variance  f-prob.   is < 2nd
Exxon Data (Continued)
1.11
.1.15
11.07
10.58
1.72
1.77
1.11
1.18
11.07
10.92
1.72
1.66
1.11
1.15
10.89
10.58
1.89
1.77
1.11
1.18
10.89
10.92
1.89
1.66
1.15
1.18
10.58
10.92
1.77
1.66
0.9457

0.9234

0.8541

0.9046

0.9775

0.8578

0.9489

0.9491

0.7115

0.9170

0.9953

0.5033

0.9636

0.9467

0.7244

1.0045
1.5271
105.6700
95.4060
0.3217
0.3275
1.0045
1.4230
105.6700
107.5400
0.3217
0.3919
1.3846
1.5271
89.5090
95.4060
0.4981
0.3275
1.3846
1.4230
89.5090
107.5400
0.4981
0.3919
1.5271
1.4230
95.4060
107.5400
0.3272
0.3919
0.2971     0.53

0.4481     0.46

0.4909     0.57


0.3287     0.54

0.4911     0.49

0.4006     0.43


0.4503     0.52

0.4675     0.48

0.2969     0.36


0.4861     0.54

0.4075     0.50

0.3799     0.27


0.4641     0.52

0.4393     0.53

0.4095     0.37
                             -28-

-------
                              Table 4
                       Student-t Test Results
                           1980+ Vehicles
 Amb
Temp.
(°F) Poll
43
42
42
42
42
HC

CO

NOx

HC

CO

NOx

HC

CO

NOx

HC

CO

NOx

HC

CO

NOx
11.4
8.1
11.4
8.1
11.4
8.1
10.5
9.0
10.5
9.0
10.5
9.0
10.5
8.0
10.5
8.0
10.5
8.0
10.5
6.5
10.5
6.5
10.5
6.5
9.0
8.0
9.0
8.0
9.0
    Average
   Emissions
N   (q/mi)
t-prob.   Variance  f-prob.
                                  0.4644

                                  0.4688

                                  0.4379
             8.0
Chevron Data
13
13
13
13
13
13
0.54
0.56
10.22
10.47
0.55
0.56
API/ATL Data
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
0.70
0.71
10.47
10.65
0.76
0.79
0.70
0.69
10.47
11.11
0.76
0.77
0.70
0.68
10.47
10.56
0.76
0.76
0.71
0.69
10.65
11.11
0.79
0.77
(1981-83
0.8817

0.9315

0.9323

(1983-86
0.9424

0.9532

0.8378

0.9784

0.8277

0.9721

0.9132

0.9753

0.9978

0.9199

0.8815

0.8650

Vehicles)
0.1198
0.1263
52.1010
54.5590
0.1421
0. 1558
Vehicles)
0.1375
0.1366
26.8440
30.9760
0.0408
0.0386
0.1375
0.1281
26.8440
23.5510
0.0408
0.0403
0.1375
0.0775
26.8440
22.3920
0.0408
0.0425
0.1366
0.1281
30.9760
23.5510
0.0386
0.0403
                                  0.4971

                                  0.4395

                                  0.4770



                                  0.4699

                                  0.4446

                                  0.4955



                                  0.2720

                                  0.4236

                                  0.4821



                                  0.4728

                                  0.3855

                                  0.4815
  Prob.
1st Mean
is < 2nd
                               0.56

                               0.53

                               0.53
                               0.53

                               0.52

                               0.58



                               0.49

                               0.58

                               0.51


                               0.46

                               0.51

                               0.50



                               0.46

                               0.56

                               0.44
                             -29-

-------
                         Table 4  (Continued)
                       Student-t Test Results
                           1980+ Vehicles
 Arab        Fuel
Temp.        RVP
(°F) Poll  (psi)
42





42





50





50





50





HC

CO

NOx

HC

CO

NOx

HC

CO

NOx

, HC

CO

NOx

HC

CO

NOx

9.0
6.5
9.0
6.5
9.0
6.5
8.0
6.5
8.0
6.5
8.0
6.5
14.6
11.7
14.6
11.7
14.6
11.7
14.6
9.0
14.6
9.0
14.6
9.0
11.7
9.0
11.7
9.0
11.7
9.0
Average
Emissions
N (q/mi)
t-prob.
Variance
API/ATL Data (Continued)
6 0.71
6 0.68
6 10.65
6 10.56
6 0.79
6 0.76
6 0.69
6 0.68
6 11.11
6 10.56
6 0.77
6 0.76
EPA Data
10 0.81
10 0.93
10 12.98
10 16.22
10 0.90
10 0.83
10 0.81
10 0.72
10 12.98
10 11.97
10 0.90
10 0.78
10 0.93
10 0.72
10 16.22
10 11.97
10 0.83
10 0.78
0.8486

0.9750

0.8417

0.9356

0.8446

0.9746

0.1366
0.0775
30.9760
22.3920
0.0386
0.0425
0.1281
0.0775
23.5510
22.3920
0.0403
0.0425
(1987 Vehicles)
0.7459 0.3468

0.6204

0.4124

0.7187

0.7876

0.1837

0.5582

0.5060

0.5610

0.9768
78.5890
334.3400
0.0469
0.0351
0.3468
0.2596
78.5890
57.0110
0.0469
0.0361
0.9768
0.2596
334.3400
57.0110
0.0351
0.0361
           Prob.
         1st Mean
f-prob.   is < 2nd
                                                     0.2744

                                                     0.3652

                                                     0.4592



                                                     0.2972

                                                     0.4786

                                                     0.4776
                                                      0.3365

                                                      0.3201

                                                      0.3519



                                                      0.0307*

                                                      0.0073*

                                                      0.4833
            0.43

            0.49

            0 .43



            0.47


            0.43

            0.49
                                                      0.0695*     0.62

                                                      0.0211*     0.68

                                                      0.3365      0.21
            0.37

            0.40

            0. 10



            0.29

            0.26

            0.29
*  Probability  value  is   less  than  0.10,  an  indication  that   the
variance differences are significant.
                             -30-

-------
                         Table  4  (Continued)
                       Student-t Test Results
                           1980+ Vehicles
 Amb
Temp.
 °F) Poll
55
55
55
55
55
HC

CO

NOx

HC

CO

NOX

HC

CO

NOx

HC

CO

NOx

HC

CO

NOx
8.4
6.1
8.4
6.1
8.4
6.1
10.5
9.0
10.5
9.0
10.5
9.0
10.5
8.0
10.5
8.0
10.5
8.0
10.5
6.5
10.5
6.5
10.5
6.5
9.0
8.0
9.0
8.0
9.0
             8.0
Average
Emissions
N ( q/mi )
Chevron
13
13
13
13
13
13
API/ATL
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
Data
0.36
0.40
6.18
6.75
0.63
0.65
Data
0.54
0.54
7.95
7.95
0.79
0.78
0.54
0.52
7.95
7.82
0.79
0.80
0.54
0.56
7.95
7.93
0.79
0.80
0.54
0.52
7.95
7.82
0.78
0.82
t-prob
(1981-83
0.6719

0.7344

0.9278

(1983-86
0.9965

0.9990

0.9386

0.9125

0.9602

0.8453

0.9177

0.9933

0.9378

0.9180

0.9585

0.8039

Variance
Vehicles)
0.0465
0.0573
15.2730
20.2720
0.2009
0.2071
Vehicles)
0.0996
0.1085
20.6770
19.3310
0.0383
0.0578
0.0996
0.0892
20.6770
19.0470
0.0383
0.0552
0.0996
0.1142
20.6770
17.1600
0.0383
0.0554
0.1085
0.0892
19.3310
19.0470
0.0578
0.0552
f-prob.



 0.3617

 0.3158

 0.4795
 0.4636

 0.4715

 0.3315



 0.4537

 0.4652

 0.3494



 0.4420

 0.4214

 0.3480



 0.4177

 0.4937

 0.4803
  Prob.
1st Mean
is < 2nd
   0.66

   0.63

   0.53
   0.50

   0.50

   0.47



   0.46

   0.48

   0.57



   0.54

   0.50

   0.53



   0.46

   0.48

   0.59
                             -31-

-------
                         Table 4 (Continued)
                       Student-t Test Results
                            1980-1- Vehicles
Amb
Temp.
(°F)
55





55





75





80





80







Poll
HC

CO

NOx

HC

CO

NOx

HC

CO

NOx

HC

CO

NOx

HC

CO

NOx

Fuel
RVP
(psi)
9.0
6.5
9.0
6.5
9.0
6.5
8.0
6.5
8.0
6.5
8.0
6.5
8.6
7.2
8.6
7.2
8.6
7.2
10.5
9.0
10.5
9.0
10.5
9.0
10.5
8.0
10.5
8.0
10.5
8.0
Average
Emissions
N (q/mi) t-prob.
Variance
API/ATL Data (Continued)
6
6
6
6
6
6
6
6
6
6
6
6
CARB
4
4
4
4
4
4
0.54
0.56
7.95
7.93
0.78
0.80
0.52
0.56
7.82
7.93
0.82
0.80
Data (
0.33
0.34
4.04
4.27
0.45
0.45
API/ATL Data
6
6
6
6
6
6
6
6
6
6
6
6
0.35
0.32
4.35
3.75
0.89
0.90
0.35
0.30
4.35
3.21
0.89
0.90
0.9160

0.9921

0.8871

0.8324

0.9652

0.9142

0.1085
0.1142
19.3310
17.1600
0.0578
0.0554
0.0892
0.1142
19.0470
17.1600
0.0552
0.0554
1983-88 Vehicles)
0.9595 0.0640

0.9282

0.9897

(1983-86
0.6989

0.6275

0.9125

0.4307

0.2858

0.9401

0.0647
11.2500
14.3310
0.0733
0.0647
Vehicles)
0.0192
0.0149
4.9908
3.7726
0.0719
0.0593
0.0192
0.0086
4.9908
1.1420
0.0719
0.0467
                                                                Prob.
                                                              1st Mean
                                                     £-prob.  is < 2nd
                                                      0.4783

                                                      0.4496

                                                      0.4819



                                                      0.3967

                                                      0.4558

                                                      0.4984
                                                      0.4967

                                                      0.4235

                                                      0.4605
                                                      0.3944

                                                      0.3831

                                                      0.4193



                                                      0.1989

                                                      0.0657*

                                                      0.3237
                    0.54

                    0.50

                    0.55



                    0.58

                    0.52

                    0.46
                    0.36

                    0.33

                    0.54



                    0.23

                    0. 16

                    0.53
*  Probability  value  is   less  than  0.10,
variance differences are significant.
an  indication  that  the
                              -32-

-------
 Amb
Temp.
(°F) Poll
80
80
80
80
                         Table 4 (Continued)

                        Student-t Test Results
                            1980+ Vehicles
N
 Average
Emissions
 (q/mi)
t-prob.  Variance
           Prob.
         1st Mean
f-prob.   is < 2nd
                      API/ATL Data  (Continued)
HC

CO

NOx

HC

CO

NOx

HC

CO

NOx

HC

CO

NOx

10.5
6.5
10.5
6.5
10.5
6.5
9.0
8.0
9.0
8.0
9.0
8.0
9.0
6.5
9.0
6.5
9.0
6.5
8.0
6.5
8.0
6.5
8.0
6.5
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
0
0
4
3
0
0
0
0
3
3
0
0
0
0
3
3
0
0
0
0
3
3
0
0
.35
.33
.35
.35
.89
.91
.32
.30
.75
.21
.90
.90
.32
.33
.75
.35
.90
.91
.30
.33
.21
.35
.90
.91
0

0

0

0

0

0

0

0

0

0

0

0

.8137

.3521

.8680

.6883

.5676

.9659

.9131

.6763

.9507

.6514

.8331

.9150

0
0
4
1
0
0
0
0
3
1
0
0
0
0
3
1
0
0
0
0
1
1
0
0
.0192
.0253
.9908
.3246
.0719
.0658
.0149
.0086
.7726
.1420
.0593
.0467
.0149
.0253
.7726
.3246
.0593
.0658
.0086
.0253
.1420
.3246
.0467
.0658
0

0

0

0

0

0

0

0

0

0

0

0

.3838

.0859*

.4628

.2794

. 1078*

.3994

.2874

.1377

.4560

.1299

.4373

.3576

0.41

0.19

0.56

0.35

0.29

0.48

0.54

0.35

0.52

0.67

0.58

0.54

*  Probability  value  is  less  than  0.10,   an  indication  that  the
variance differences are significant.
                             -33-

-------
                          References

1.    Shih,  Celia,  "Analysis of  Impact of  Fuel RVP on  Exhaust
     Emissions      at      75°F      Ambient      Temperature,"
     EPA-AA-TEB-EF-88-01,  May 12,  1988.

2.    "Reference  Material  for  Phase  II  Volatility Control  FRM
     Docket,"  EPA Memorandum  from  Celia  Shih,  TEB,   to  Tad
     Wysor,  SDSB, March 26,  1990,  Attachment C.

3.    "Reference  Material  for  Phase  II  Volatility Control  FRM
     Docket,"  EPA Memorandum  from  Celia  Shih,  TEB,   to  Tad
     Wysor,  SDSB, March 26,  1990,  Attachment D.

4.    Welstand, J.  S.,  "Effects  of Reduced  Volatility  Gasoline
     on  Exhaust   Emissions,  Driveability,  and  Fuel  Economy,"
     June  10,   1983,  Chevron   Research  Company,   Richmond,
     California.

5.    "The Effect  of  Fuel  Volatility  Variations on  Evaporative
     and  Exhaust  Emissions,"   API   Publication  4310,   Exxon
     Research and Engineering Company, Linden,  New Jersey,  May
     1979,  prepared for the American Petroleum Institute.

6.    "Evaluation of Test  Results  from the 8 RVP Fuel  Evaluation
     Project,"   State  of   California   Air   Resources   Board
     Memorandum from Dave  Coel  to Steve Albu, August  18,  1988.

7.    Gething, J.  A.,  and  J.S.  Welstand,  "Summary  Report  on the
     Effects  of  Reduced  RVP Gasoline on  Early  1980's  Model
     Vehicle  Exhaust   Emissions   and  Their   Interaction  with
     Temperature," February  l,  1988,   Chevron Research  Company,
     Richmond, California.

8.    Welstand,   J.   S.,  Chevron  Research   Company,   Personal
     Communication with Celia Shih, EPA, April 5, 1990.

9.    "Emissions  and  Vehicle Performance  with  Lower  RVP  Fuels,
     Final Report," Automotive Testing Laboratories,  Inc.,  East
     Liberty, Ohio,  January 27,  1988,  sponsored  by the  American
     Petroleum Institute.

10.  Neter,    John   and   William  Wasserman   "Applied   Linear
     Statistical  Models,"  Richard D.  Irwin,  Inc., 1974.

11.  "National   Fuel   Surveys,"   Motor   Vehicle   Manufacturers
     Association of  the  United  States,  Inc.,  both  summer  and
     winter of 1986,  1987, and  1988.
                             -34-

-------
APPENDIX
  -35-

-------
                           Table Al

                     List of  Test Vehicles
              1975 Chevron Research Test  program
                   (All California Vehicles)
Veh.tt

 l
 2A*
 2B*
 3
 4
 5
 6
 7
Model
Year   Make/Model
Engine Garb
  CID  Bbl
1973   Plymouth Fury            360   2
1973   Pontiac Catalina         400   2
1973   Pontiac Catalina         400   2
1974   Volkswagon Super Beetle   98   l
1975   Chevrolet Impala         350   4
1975   Ford LTD                 400   2
1975   Ford Pinto               140   2
1976   Oldsmobile Cutlass       350   4
Transmission
Auto/Manna1

    A
    A
    A
    M-4
    A
    A
    A
    A
    Vehicles #2A and  tt2B  were  the same vehicle, tt2A  was  tested
    with  the  original  carburetor  and tt2B  was tested with  a
    replacement carburetor.
                             -36-

-------
                                Table  A2

                         List of Test Vehicles
                    1979  Exxon  Research Test  Program
                 (All Automatic Transmission Vehicles)
    Veh.tt

     A
     B
     C
     D
     E
     F
     G
     H
Model
Year

1976
1975
1974
1974
1977
1976
1975
1974
Make/Model

Ford LTD
Olds
Valiant
Vega
Granada
Buick
Dodge
Grand Prix
Engine  Carb
CID     Bbl
400
455
225
140
302
350
318
400
2
4
1
2
2
2
2
4
Catalyst
Equipped

   yes
   yes
   no
   no
   yes
   yes
   yes
   no
—   *    Vehicles  B,  E,  and  G were  California  vehicles.   The
         engines  of  vehicles  B   and   G  had  been  overhauled
         previously,  at  3,000  and 6,000  miles,  respectively,
         prior to the testing.
                                 -37-

-------
                           Table A3

                     List of Test Vehicles
             1988 CARB Five-Car Evaluation Project
                   (All California Vehicles)
         Model                                   ft of    Fuel
Veh.tt    Year     Make/Model            Engine   Cyl     System

 1       1987     Olds Delta 88         3.8L     6       PFI
 2       1987     Toyota Camry          2.0L     4       PFI
 3       1988     Ford Tempo            2.5L     4       TBI
 4       1983     Toyota Tercel         1.5L     4       Carb
 5*      1978     Chevrolet Caprice     5.OL     8       Carb
 *   Vehicle  was  analyzed  separately,  since  this vehicle  was
     built for compliance with  a  different  evaporative emission
     standard.
                             -38-

-------
                             Table Bi

                      List of Test Vehicles
                1988 Chevron Research Test Program
                     (All California Vehicles)
      Model
Veh.tt Year
1
2*
3
4
5
6
7
8
9
10
11
12
13
14**
1981
1982
1982
1983
1982
1983
1981
1983
1983
1983
1983
1982
1983
1983
Make/Model
Phase  Engine
             Dodge Aries
             Chev Cavalier
             Olds Ciera
             Honda Accord
             Ford Escort
             Chev Cavalier
             Datsun 210
             Datsun Maxima
             Olds Cutlass
             Mazda GLC
             Toyota Corolla
             Toyota Corolla
             Lincoln Town Car
             Toyota SR-5 Pickup
I, II
I, II
I, II
I, II
I, II
I
I, II
I, II
I, II
I, II
I
II
II
I, II
L-4
L-4
L-4
L-4
L-4
L-4
L-4
L-6
V-6
L-4
L-4
L-4
V-8
L-4
 Displ
(Liter)

 2.6
 1.8
 2.5
 1.8
 1.6
 2.0
 1.2
 2.4
 3.8
 1.5
 1.6
 1.8
 5.0
 2.4
        Trans
 Fuel   Auto/
System  Manual
Garb
Garb
TBI
Garb
Garb
TBI
Garb
PFI
Garb
Garb
Garb
Garb
TBI
Garb
A
A
A
M-5
A
A
M-5
A
A
M-4
M-5
M-5
A
M-5
*  Vehicle was  tested with the  original  carburetor on  Phase  I,  and
   with the replaced carburetor on Phase II.

** Vehicle was  excluded from  the  analysis,  since  this vehicle  was
   categorized as a truck.
                             -39-

-------
                       Table B2

                 List of Test Vehicles
               1987 API/ATL Test Program
   (All Federal Vehicles with Automatic  Transmission)
       Model                                   Fuel
Veh.tt  Year    Make/Model        Engine       System

CD-I   1985    Olds Cutlass         3.8L      Garb 2V
CD-2   1986    Pontiac Grand AM     3.OL      PFI
CD-3   1983    Mercury Cougar       3.8L      Garb 2V
CD-4   1985    Ford T-Bird          5.OL      TBI
CD-5   1985    Plymouth Horizon     2.2L      Garb 2V
CD-6   1985    Chev Cavalier        2.OL      TBI
                       -40-

-------
                           Table B3

                     List of Test Vehicles
                     EPA/E&D Test Program
                     (All Federal Vehicles)
        Model                                      Fuel
Veh tt   Year     Make/Model             CID        System

 1      1987     Pontiac 6000           151        TBI
 3      1987     Ford Taurus Wagon      182        PFI
 6      1987     Dodge Shadow           135        TBI
 7      1987     Ford Taurus Wagon      182        PFI
10      1987     Olds Cutlass           173        PFI
19      1987     Chrysler Lebaron       135        PFI
20      1987     Mercury Cougar         231        TBI
35      1987     Chev Cavalier          121        TBI
43      1987     Pontiac 6000           151        TBI
56      1987     Chev Cavalier          121        TBI
                             -41-

-------