EPA-AA-IMS/81-2
                    Derivation of I/M Benefits for Post-1980
                      Light  Duty  Vehicles  for  Low Altitude,
                              Non-California Areas
                               Revised March,  1981


                                  Dave  Hughe s
                                      I

                                     NOTICE
This report  does  not necessarily represent  final EPA decisions  or positions.
It is intended  to present a technical analysis of  the  issue 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.
                        Inspection and Maintenance Staff
                      Emission Control Technology Division
                  Office of Mobile Source Air Pollution Control
                       Office of Air, Noise and Radiation
                      U.S. Environmental Protection Agency

-------
I.  Introduction..

As is  widely recognized,  the post-1980 model  year  fleet  will be predominantly
composed of  vehicles which  employ  what has become known as  Three  Way catalyst
technology.   This  technology incorporates a sophisticated microprocessor-based
engine control  system which holds  the air/fuel  ratio  very close  to stoichi-
ometry,  thereby  allowing   the  Three  Way  catalyst  to simultaneously  convert
Hydrocarbons  (HC),  Carbon  Monoxide   (CO)  and  Oxides of   Nitrogen  (NOx)  to
harmless by-products.   This is  in  contrast to conventional  technology vehicles
(Model Years  1975-1980) which rely  on  mechanical  control  of the air/fuel ratio
and  whose  catalysts  are  only  designed  to convert HC and  CO.   As could  be
expected,  such  a significant shift  in technology will have  an impact  on the
expected in-use  emissions  performance  of these vehicles both with and  without
an Inspection and Maintenance (I/M) program.

This report  discusses  the  derivation of I/M benefits for  the post-1980 Federal
fleet  as  contained  in EPA's  emission  factor model  MOBILE2*.   As  such,  it
relies heavily  on an  earlier report  which presented the derivation  of in-use
emission factors  (i.e.  the  emissions performance  of  the fleet without I/M) for
the post-1980 Federal  fleet [I].**   It is recommended  that  the  major structur-
al points  presented  in that report be  understood first,  so that  the analysis
presented  in this report can be  more  easily  understood.   This analysis will,
however, briefly summarize how the in-use emission factors were  derived.

Both this  analysis  and the emission factor  analysis have relied on a modeling
approach which  structurally  represents  the  trends  observed in  the data,  as
opposed  to  performing  statistical analyses  (e.g.,   linear  regression  with
mileage as the  independent  variable)  on large data bases as  has  traditionally
been done  for earlier model year  fleets.   This  was done due  to  a  lack  of a
sufficiently  large  data  base  on  vehicles  with  representative  Three  Way
catalyst  systems.   There  is a  substantial  data base made up  of  Three  Way
catalyst  systems which  was closely  examined  to discern  patterns  of  in-use
performance  and  to examine  the effectiveness  of the various I/M  short tests.
That data  base  is  not  large enough,  however, nor does  it contain  a sufficient
mileage  spread   to  enable  the  analyses  required under  the more  traditional
approach.
*   MOBILE2  is  a computer  program which models  the emissions performance  of
    the entire vehicle fleet over  time,  both with  and without  I/M.   Figures  of
    percent  benefit  due  to I/M are applied to  the without-I/M case  in MOBILE2
    in order to model the air quality impact of I/M.

**  The numbers  in  brackets  indicate references  which will  be listed at  the
    end of the report.

-------
The following sections  will  first present a brief  description  of the expected
in-use  performance  of  the  post-1980  fleet.    Second,   the  methodology  and
assumptions  used in  deriving I/M  benefits  for  the post-1980  fleet will  be
presented and  discussed.   Finally, issues  related  to the  use  of the benefits
contained in MOBILE2 will be discussed.

II. Expected In-Use Performance of the Post-1980 Fleet.

It is at this point  that  previous knowledge of the  report  which describes the
derivation  of  emission factors  for the  post-1980  fleet would  be  of greatest
advantage.[1]  The  following section  will  briefly  summarize that  derivation.
As such,  it discusses  the  expected modes of failure as  opposed to presenting
the emission factor  equations themselves.  A discussion  of failure modes will
be much more helpful  in understanding  how the  I/M benefits were derived rather
than presenting  the emission factor equations.

The emission factor  model first  breaks  the post-1980 fleet  into two separate
technology  types:  those  vehicles which  employ the  Three  Way  catalyst  tech-
nology  described above,  and those vehicles  which  rely  on more  conventional
technology  (oxidation  catalysts,  air  pumps,  engine  modifications,  EGR).   This
second group is  expected  to  make  up only 7% of the post-1980 fleet.  This is a
small fraction of the  fleet,  and  its emission  factors were derived  differently
than  for  the rest  of  the  fleet.   It  would have  been difficult to  model  I/M
benefits for this  group of  vehicles because  of the  internal methodology used
in deriving their emission  factors.   Given  the  small size  of this  group  of
vehicles, the added  complexity  that would have been  necessary  to include them
in the  calculation of  I/M  benefits,  and the small  increase  in accuracy  which
would result, it was decided to  ignore  them in  this  analysis.   Consequently,
the results derived  for the Three Way technology type have been assumed  to  be
valid for these vehicles as well.

Of  chief  concern then  is  the  group  of  vehicles  equipped  with Three Way
catalyst  technology.   This  group  of  vehicles  was  entitled  "Closed   Loop
Vehicles"  in  the   emission factor   analysis,   due  to   the  nature  of  the
microprocessor-based  control system  they employ.   This  system relies  on  a
feedback  signal  from  an  oxygen  sensor  placed in  the  exhaust  manifold  which
tells the microprocessor whether  the air/fuel ratio  is rich or  lean of stoich-
iometry.   The   microprocessor  uses   this   information   in  conjunction   with
information  supplied by  other  sensors  to  adjust  the  air/fuel  ratio at  the
carburetor.  The system is  therefore based  on  a "closed  loop" form  of control,
with  the  signal  from  the  oxygen  sensor  used to  "close  the loop".    Three Way
catalyst vehicles  will therefore  be  referred  to  as Closed Loop vehicles  for
the rest  of this analysis  in order to  parallel the  terminology used in  the
emission factor analysis.

-------
The  emissions  performance of  Closed Loop vehicles  was  modeled by designating
three  distinct categories  of emissions  performance  for  HC and  CO.*  A data
base of  218 in-use, Closed Loop  vehicles was primarily  relied upon to deter-
mine  the incidence  and  the emission  levels of  the various categories.  This
data  base  is  made  up  of  the most  representative  and  advanced  Closed Loop
technology  vehicles for  which data  are  currently available.  (See Section II.
of the emission factor analysis for  a more complete  discussion.[1])

The  first  category of  vehicles  represents  those  vehicles  which  have  lost
microprocessor  control  of the engine  and are  operating in  a  full-rich mode.
This  situation could  result  from  a  number  of  possible  scenarios   including
microprocessor  failure,   tampering,   sensor  failures  and  disconnection  of
electrical  wires.   The  failure  scenarios  possible  will  differ  among  the
various manufacturers, but  in all cases the end  result is  very high  HC and CO
emission  levels.   This  category  of vehicles 'is  called  the Primary category.
This name was  given because  the  vehicles in  the Primary  category,  while  not
initially  very numerous,  contribute  the majority  of  the fleet's  in-use  CO
emissions and  a large  share of the  in-use HC emissions.   The  emission levels
attributed  to  the  Primary category were based  on  data from  ten in-use,  low
mileage  Closed Loop vehicles  from the  data base  which were observed  to have a
Primary  category  failure.   The average emission  levels  of these vehicles were
HC = 3.85 g/mi,  CO =  108.0 g/mi  at an  average  mileage  of  9,163 miles.   The
Primary  category  was given an initial size of 3%  of the  fleet  at zero miles
and  a  growth  rate of 2%  per  10,000 miles.  That is,  as  the  fleet ages,  more
and more  vehicles will experience the  types of failures described above.  (See
Sections  IV.A.2.a.  and  IV.A.3.a  of the  emission factor  analysis for  a  more
complete discussion.[1])

The  second  category of  vehicles  represents  those vehicles which  have experi-
enced misfueling  (that  is, those  vehicles equipped with  catalytic converters
which  have  been  fueled  with  leaded gasoline).   The Misfueling  category  was
modeled  to  make up 8%  of  the  fleet  based on the most  recent  EPA observations
of misfueling  in  the field. [2]  The HC  and  CO emission levels associated with
vehicles  in the  misfueling  category were derived  from recent  EPA misfueling
test programs.   They are  well below  the emission  levels  of Primary category
vehicles  but substantially above  the emission  levels of  the remainder  of  the
fleet.   (See  Sections  IV.A.2.C  and IV.A.3.C  of  the  emission  factor analysis
for a more complete discussion.fi])

The  third  category of  vehicles  basically  represents  the  remainder of  the
fleet.   As  such it  includes  both well-maintained vehicles and vehicles which
have been tampered with  or have  suffered component  wear or component failure,
but  not   of  the  magnitude which  would  lead  to the  Primary  category.   The
emission  levels associated with this category  were  based on data  from over 190
in-use,  low—mileage Closed Loop  vehicles from among  the  data  base  described
above.  This category is referred to as the Secondary category.   (See Sections
IV.A.2.b  and  IV.A.3.b of the  emission  factor analysis   for  a more  complete
discussion.[1])
    The NOx analysis  was  performed  separately and will be briefly discussed in
    the next section.

-------
The fleet  therefore  is broken down into  three  categories of  vehicles for the
HC and CO  analysis,  each  category  with a distinct emissions performance.  Each
category has a unique  zero mile emission  level  and  there  is a common deterior-
ation  rate used  for all the  categories for a given pollutant.   The deteri-
oration rates were  developed for  the  Secondary category  and  then adopted for
the Primary and  Misfueling  categories.  This was done due  to a lack of enough
separate data for the  Primary  and  Misfueling categories to  be able to predict
a unique  deterioration rate for those categories.   As could  be expected from
the preceding discussion, it is chiefly the number  of  vehicles  entering the
Primary category with its  very  high  HC  and CO  emission levels,  and not the
deterioration rates, which  is  of  foremost  concern  from an  I/M perspective.
The following section will present how the I/M benefits were derived.

III. Methodology Used in Determining I/M Benefits.

Basically  two  separate  phenomenon  were accounted  for  in  calculating  I/M
benefits  for  the  post-1980  fleet:    the  identification  of  Primary  category
vehicles  and  the identification of  vehicles experiencing severe  ignition and
misfire problems.

One issue  that  needs  to  be addressed before proceeding  has  to do with which
model  year case  of  the  emission  factor analysis  was used  to determine I/M
benefits.[1]  Those  readers  familiar with  the  emission  factor  analysis  will
note that  three  separate  cases were  modeled  for the post-1980 fleet:  the  1981
model year fleet, the  1982  model year fleet  and the 1983  and beyond model year
fleet.  Separate analyses were required  for  1981 and 1982 due to  the presence
of the Clean  Air Act Section 202(b)(5) waiver  fleet:  those  cars that received
a waiver of the  CO  standard from  3.4 g/tni to 7.0 g/mi in 1981 and 1982.   1981
and 1982  are also   the model  years  when the  benefits which accrue  from the
Parameter  Adjustment regulations  (44 F.R.   2960)  are phased  into  the  model.
The differences  between  the various model year  cases are not  great.  The  1983
and beyond case  was  used as the base  case in determining I/M benefits.   This
was because  the  1983  and beyond case is the most representative  case for the
overall timeframe for which I/M benefits  need to be  calculated.   It would  have
added  a   significant measure  of  complexity without  an  appreciable  gain  in
accuracy to account  for the small differences in the L981  and 1982 model years.

A.  Identification of Primary Category Vehicles.

The  only  significant  issue  involved here is  what  percentage  of  Primary
category  vehicles  can be expected to be identified  by  the various  I/M  short
tests.  Since Primary category vehicles emit such high levels  of  HC and CO and
indeed contribute the  majority  of  the overall  fleet's in-use  CO emissions and
a large share of the fleet's HC emissions, it is reasonable to expect that  an
I/M short  test  be   capable  of  identifying   most of these  vehicles.  This  is
especially  true  if  excess  emissions   (i.e.   emissions  above   the   Federal
standards)  are  considered  instead  of  total  emissions.    Primary  category
vehicles contribute  an overwhelming share of  the excess  emissions, which are
of chief  concern from an I/M perspective.   The ability  of the  various  short

-------
tests to  identify  Primary category vehicles was  quantified  by examining short
test  data from  the  ten  in-use  Primary  category vehicles  mentioned earlier.
Table III.A.I.  presents  the  results  of  applying inspection cutpoints  to  the
short  test  data.    The  individual  data  from  each  of  the  10  vehicles  is
presented in Appendix A.


                                  Table  III.A.I
    Ability of the Various Short Tests to Identify Primary Category Vehicles


                            Applicable           Percent of Primary Category
Short Test                  Cutpoints                Vehicles Identified*

Idle(N)                   1.2% CO, 220 ppm HC               50%

2500 rpm/Idle(N)          1.2% CO, 220 ppm HC               60%

Loaded Two Mode           1.2% CO, 220 ppm HC               70%
(30 mph/Idle(N))
At  first  inspection,  Table  III.A.I  suggests definite incremental  benefits  in
going from  the Idle(N)  test  to the  2500 rpm/Idle(N) test  to the  Loaded  Two
Mode  test.   A closer  inspection,  however,  reveals  that making  a  distinction
between the  2500  rpm/Idle(N)  and  Loaded Two Mode  tests on  the  basis of  the
data  currently on hand  is  unwarranted.   As can  be  seen in Attachment A,  one
vehicle,  the second  1979  Mercury Marquis,  "passed"  the  2500 rpm/Idle(N)  by
only  0.02%  CO.  Given the fact that  this one vehicle was only a very margin-
ally  "passing" vehicle  and  given the   fact  that   previous  examinations  and
technical judgement do  not  support a significant distinction  between  the  2500
rpm/Idle(N)  test and  the Loaded Two Mode test,  EPA has decided to  associate  a
70% Identification Rate of Primary category  vehicles  with  the  2500  rpm/Idle(N)
test.  That  is,  the  Loaded  Two Mode  and 2500 rpm/Idle(N)  tests  are judged  to
be  equally  capable  of identifying  Primary  category  vehicles  - at  a  70%
Identification Rate.  The selection of an Identification Rate  to  be associated
with a given short test  is  significant  in terms of  the emission benefits which
result from  the model.  The  significance of  the Identification Rate of Primary
category  vehicles will  become more  apparent  as the  report  progresses.    In
summary,  given  the   limited  data  available,  it is  not  warranted  to  make
distinctions  of  such  consequence where previous  experience  and  technical
judgement do not support the distinction.
*  This  quantity,  the  percent  of Primary  category vehicles  identified by  a
given short  test,  is  referred  to as  the Identification Rate  in the  "User's
Guide to  the Mobile Source  Emissions  Model:  MOBILE2".  Section  2.2.4  of  that
report discusses the Identification Rate.

-------
50% was used  as the default value  within MOBILE2 for  the  Identification Rate
of Primary  category vehicles  in the calculation  of I/M benefits.   As  can be
seen  in  Table  III.A.I.  this is  the Identification  Rate associated  with the
basic Idle(N)  test.   Section 4  of  this report will deal with  the inner work-
ings  of  MOBILE2  and  the  use  of  alternate  Identification  Rates  in greater
detail.

Identification  Rates  of  60%,  70%,  80%  and  90%  were  also  evaluated  and
incorporated  into  MOBILE2  to   provide   flexibility   for   those  states  and
localities who  choose  to use the Loaded  Two  Mode  test  or the Idle(N)/2500 rpm
test, and  also to  provide flexibility  for modifications if the  addition of
more  data  at a later  point changes  the  percent of Primary  category vehicles
identified by the various short tests.

Those Primary  category vehicles which  are  modeled as  being  identified  in the
I/M  process  are  all  assumed  to  be  repaired.   The  emission  levels of  the
repaired vehicles  were modeled by  assuming 8% of the  repaired  vehicles enter
the Misfueling  category and adopt  the  Misfueling category's emission levels.
The rest  are  assumed  to  adopt  the  emission  levels  of  the  Secondary category.
Eight percent are  modeled to enter the Misfueling category  since  the overall
misfueling  rate of  8%  is  assumed  to  occur  uniformly throughout  the  fleet.
Therefore  8%  of the vehicles  in the Primary category have also  experienced
misfueling.  When  these vehicles are repaired,  the  engine  control  system is
returned to normal  operation but the catalyst (and  oxygen sensor)  will  still
be  damaged  due  to misfueling.    Thus,  these  vehicles  should  be  assigned
emission levels which reflect a history of misfueling.

Primary category vehicles  which  are not  identified  by  the I/M  short test at a
given inspection  are  assumed  to  also  not be  identified  at  all  subsequent
inspections.  For example, the 50%  of Primary category  vehicles  not identified
at the first  round of  inspections  in  the default  case will  thereafter  always
remain in  the fleet as  Primary  category vehicles.  This reflects  the assump-
tion  that  some  Primary category failure modes will  have  low emissions at  the
operating  points  tested by the  given short test and  are not capable of ever
being identified by  that  short test.   This  assumption  is  conservative (results
in  lower  benefits  from the I/M program)  relative  to  the possible competing
assumption that some Primary vehicles  (50% in the default  case)  are able  to
pass  a given  inspection due to a random  process which  occurs independently at
each  inspection.    No   evidence  exists  to  support  or contradict  this  less
conservative assumption.

Those Primary  category vehicles which  are  identified  and repaired  in the  I/M
process  are  subsequently  assumed   to  be  eligible  to  reenter  the  Primary
category at the same rate as other  vehicles  in  the  fleet.   These  vehicles are
not exempt from ever entering  the Primary category again.  This is  because of
the  number  and variety  of  scenarios  which can  lead  to  Primary  category
operation as well as due  to  the  effects of change of ownership  in conjunction
with  tampering by -a new owner.

It  should also be  noted  at  this  point  that  the  likelihood  of a vehicle
entering  the  Primary category is modeled  as  slowly increasing  as a vehicle
accumulates  mileage.   That  is,  as  a  vehicle  ages,  it  is more  likely  to

-------
experience  a Primary  category failure,  due to  increased component  wear and
decreased incentive  for the owner to  properly  maintain his vehicle.  This was
modeled by  relying on  the  Primary category  growth  rate used in  the emission
factor  analysis.   A  "constant" 2%/10,000  mile growth  rate  was used  in that
analysis, but since the  2%  is  coming  from an ever smaller group of non-Primary
category  vehicles,  the  assumption of  a constant  rate implies  an increasing
likelihood  of  Primary  category failure as  a vehicle  ages.   For  example,  at
zero miles,  97% of the fleet  is  not  in the Primary category and  out of this
group 2% of  the total fleet enters the Primary category over the first 10,000
miles.   At  100,000  miles, only  77%  of  the  fleet   is  not  in  the  Primary
category, yet  the  same number of  vehicles,  2%  of the  total  fleet,  enters the
Primary  category  over  the next  10,000  miles.   Thus,  the emission  factor
analysis models an increasing  likelihood  for a  vehicle to  enter  the Primary
category as  it  ages.   The  algorithm which  describes this  tendency  was used to
determine the growth  of the Primary  category between inspection  points for the
calculation of I/M benefits.

The  failure rate  of  Primary  category  vehicles (not to be  confused  with the
Identification Rate)  will  be low,  especially after  the first years of  an I/M
program's implementation.   This is due to the fact that the growth rate of the
Primary category is modeled to be only 2%/10,000 miles.   There  are therefore
simply not  that many  Primary category  vehicles  to identify, although identify-
ing  the  ones present  in the field  results  in substantial emission benefits.
Table  IV.A.I  includes  an  example  of the  expected failure  rate   of  Primary
category vehicles.

The  identification of  Primary  category vehicles  in the  I/M  process  accounts
for  all  of  the CO  benefits contained  in  MOBILE2 and  roughly half of  the  HC
benefits.   The  HC  benefits  resulting  from  the   identification  of  Primary
category vehicles  cannot be presented exactly  since  they were  calculated  in
conjunction  with   the  HC  benefits  which result  from  the  identification  of
vehicles with ignition/misfire problems.   A summary of  total  benefits  will  be
presented  later.   To  give  the reader  an indication of the  magnitude  of the
benefits resulting  from the identification  of  Primary   category  vehicles,  for
the  default case   (50%  Identification Rate  corresponding  to  use of  the basic
Idle(N) test)  the  CO  benefits  are approximately 30% after the  first year  of
program  implementation  and the   HC  benefits  are  approximately   15%.   These
figures  of  percent  benefit  represent  the percent  of  emissions  reduction
attributed to I/M as compared to the base case (without I/M).

NOx Penalty Issue.

The  reader  may wonder  if  there  should not be a NOx "penalty" associated  with
the  HC/CO  "benefits"  which  accrue  from  the  repair  of  Primary  category
vehicles.   This  would  be  because  of  the very  low  NOx emission levels  which
result when a  vehicle  is   in  the  Primary category's rich mode  of  operation.
Low  NOx  levels  result  in  this situation due to  two parallel phenomenon:   the
low  NOx  levels  which  are  always   produced  by an engine during  rich modes  of
operation  and  the  increased   NOx  conversion  efficiency of  the  Three  Way
catalyst during  rich  operation.   The  emission  factor  analysis  accounted  for
the  low NOx levels  of  Primary category  vehicles by  creating a separate  NOx
category, the Low category,  which  parallels  the  growth  of  the Primary category

-------
                                        8

(See  Section  V.A.  of  the  emission  factor  analysis   for  a  complete  dis-
cussionfl]).   It  stands to reason that  when  Primary/Low category vehicles are
repaired, the  low NOx emission levels will  rise as the  HC/CO  emission levels
fall.   Thus,  a  theoretical NOx  "penalty" results  from I/M.  This  issue was
examined in a  similar fashion as  was done in developing  the  HC/CO credits for
identifying  Primary  category  vehicles.   A model  was constructed  wherein Low
category vehicles were  identified  and  repaired.   A comparison was  then made
between  the  with and  without  I/M cases.  The  results of that  analysis showed
only a very small NOx penalty.  In  a "worst-case"  scenario which set the upper
bound for a  possible NOx penalty, the  "percent penalty"  was  only 2% at 50,000
miles and 2.7% at 100,000 miles.  The  reason  why the NOx "penalty" is so small
whereas  the  HC/CO  benefits  are  fairly  large   (15%/30%)  has  to  do with the
difference between  the category emission levels for HC/CO  as compared  to NOx,
and  the difference  between  the  contribution   of  the  category  deterioration
rates for HC/CO  and NOx.   For  HC and especially  for  CO,  the  gap  between the
emission  levels   of  the  Primary  category and  the other  categories  is  much
larger  than between the NOx Low  category  and the other  NOx  categories.  Thus
there  is less difference  for  NOx  before  and  after repair when  the repaired
vehicles enter a  new category  and  therefore  a  small  "penalty"  results.  Also
for  NOx the  deterioration  rates  of  the various categories play  a  much larger
role  in  determining  the   fleetwide   composite.    Therefore   shifts  between
categories are much less significant.   Given  the small size of  the NOx penalty
and realizing that  identifying  so small  a  percentage would be overstepping the
inherent accuracy of  the model, the NOx penalty  issue was ignored.

B.  Identification  of Vehicles With Ignition/Misfire Problems.

Ignition and  misfire problems  have  historically been the  cause  of a signifi-
cant portion  of  excess HC  emissions.   While a  modeling  approach such  as  has
been used in  this analysis catlnot accurately account  for the emissions impact
of every possible  vehicle  malfunction,  ignition/misfire problems  were judged
to have a significant enough  impact  on in-use HC emissions that they should be
accounted for.  Post-1980 vehicles can  be  expected to be  free  of  many  histor-
ical maladjustments  (e.g.  idle mixture)  due to  technology  changes but  will
still be  prone to  ignition/misfire  problems.   Post-1980 vehicles will still
experience  shorting  and cracking   of  ignition  wires,   spark  plug  failures,
cracking of distributor  elements  etc.,  all of which will  lead  to  excessive HC
emissions.   These  types of failures are  readily  identifiable  by  the  various
I/M short tests since  they  usually result  in  greatly increased  HC emissions at
all modes of operation.

A  final introductory  point which  needs to  be made is  that  ignition/misfire
problems are  assumed to occur  in proportion to vehicle age.   That  is, as  a
vehicle ages,  it  has a greater likelihood of experiencing  an ignition/misfire
problem.  This point is significant in  that  it largely  determined which  data
could best be used  to examine  the  I/M  benefit  resulting  from  the  identifica-
tion of vehicles  with ignition/misfire  problems.  In particular,  data confined
to small mileage intervals are not suitable.

-------
Me thodology.

There  is  an important  difference  in the way  benefits  were  calculated for the
identification  of vehicles with  ignition/misfire problems  as  opposed  to the
identification  of Primary  category  vehicles.   The presence of Primary category
vehicles  (i.e.  vehicles with a failure of  the computer control  system  with a
resultant  rich  mode  of operation)  among  the  fleet was clearly  delineated in
the  emission factor  analysis  through  the  establishment  of a category.   The
category  grows  with  time  (in  the absence  of  I/M)  and its  size  and therefore
its  effect can  be calculated  at  any  point  by knowing  the category  growth
rate.  The calculation  of  I/M  benefits  for these vehicles  was therefore fairly
straightforward,  since   the  presence of  these  vehicles  in  the  fleet  is  so
clearly delineated.

The  emission factor analysis did not account for the  impact of  vehicles with
ignition/misfire   problems  by  creating   a  corresponding  "ignition/misfire"
category,  however.   Rather,  it  accounted  for  the  impact  of  these  vehicles
implicitly through the  HC  zero-mile emission  levels assigned to  the Secondary
and  Misfueling   categories  and  the  HC  deterioration   rate   used  for  all
categories.*  Repair  of ignition/misfire  problems in an I/M  program is modeled
by adjusting deterioration rates, rather  than  by adjusting the  distribution of
vehicles among categories.  The number  of vehicles failing the I/M test  due to
misfires  and their HC  emission  levels have  not  been modeled directly;  only
their  effect on  the  average  emissions of all  vehicles  in  a  given  category
(Primary, Secondary, or Misfueled) appears in the model.

The  Volvo/Saab  fleet  described  in Section  II.D.  of  the  emission  factor
analysis played  the major  role in determining the I/M  benefits  resulting from
the  identification of vehicles  with ignition/misfire problems[1],   This  fleet
contains 104 1978/1979  Volvos and Saabs which  all met  the  criteria established
to screen out those vehicles with maladjustments which can  be expected to be
* The HC zero-mile emission  level assigned  to  the  Secondary  category was based
on  data  from  191  representative,  in-use vehicles.   One  of  these,  a  1980
General  Motors  six  cylinder Citation  with  12,879  miles,  had  its number  1
cylinder shorted  in the distributor  cap.   FTP HC emissions  from this  vehicle
were 4.8  g/mi.   This  one  vehicle contributed  8%  of  the  average HC  emission
level of the  191  vehicles  in the Secondary category data  base.   The impact  of
this vehicle  was also carried  over  into  the  Misfueling category  since  the
zero-mile emission level of  the  Misfueling  category was  arrived at by applying
a multiplicative factor to that of the Secondary category..

The HC  deterioration rate  used  for  the various categories is  also  assumed  to
include the effects of  vehicles  with  ignition/misfire  problems.   That  is,  part
of the deterioration rate  is attributed  to the  increasing  presence of vehicles
with ignition/misfire  problems  and  correspondingly  high  HC emission  levels.
These problems are assumed to occur more  frequently at higher mileage,  thereby
adding a "misfire" component to the deterioration rate.

-------
                                       10

prevented by  the  advent of  EPA's  "Parameter Adjustment" regulations  (44 F.R.
2960).  The Volvo/Saab  fleet was used  due  to its having a wider mileage spread
than  the  other Closed  Loop  vehicle fleets.   The presence  of  a wide  mileage
spread was  considered  necessary  in order  to  better  see  the  effects of  the
increasing likelihood  of  vehicles  to  experience ignition/misfire  problems  as
they  accumulate  mileage.   The calculation  of  I/M  benefits was  done in  the
following manner.  First,  the  data from the  Volvo/Saab fleet was  put  through
an I/M simulation  where vehicles  were  identified by only using  an HC cutpoint
of 200  ppm.    The  I/M  short  test  used  as  the  basis  of  evaluation was  the
Idle(N)/2500  rpm  test  where  emissions  are  measured  and   evaluated  at  both
Idle(N) and 2500  rpm.   A cutpoint of  200 ppm HC was used.   Putting  the  fleet
of 104  Volvos and Saabs  through  the  I/M  simulation  resulted  in 2  vehicles
being identified as failures.  The mean HC emissions of these two  vehicles  was
4.18  g/mi  at  an  average  mileage  of  15,365  miles.  One  vehicle seemed  to  be
experiencing intermittent misfire and  the other  had  a failed spark plug.   Both
of these cars would have only  been identified using  the HC  cutpoint.   That  is,
they would not have been failed by  a corresponding CO cutpoint of  1%  CO.   They
were  classic HC-only failures.

The   second  methodological  step   in  determining   ignition/misfire   benefits
involved developing  a deterioration rate  for  the  various   categories  without
the  presence  of vehicles  with ignition/misfire  problems.   That  is,  assuming
that   an  I/M  process  would identify these vehicles  as  requiring  maintenance,
what  would the HC  deterioration rate be without their presence.  This was done
by  calculating  the  percent  reduction   due  to  removing   the  two   vehicles
described above.   This  percent reduction was  then  applied   to the  HC  emission
factor equation developed  for  the  Secondary  category of Closed  Loop  vehicles:
HC  =  0.23 g/mi  +  0.12  g/mi/10,000 miles  x  (Miles/10,000)   (See  Section
IV.A.2.b. of  the emission  factor  analysis   [1]).*   The percent  reduction was
applied  at  the Volvo/Saab fleetwide  average  mileage   of  13,786 miles.  This
determined one point  on the  new line.   A second point  on  the line was chosen
to be the zero mile  level  of the  HC emission factor  equation for the  Secondary
category:  0.23  g/mi.   This was  done even  though  this   point  reflects  the
presence of a  vehicle with  ignition/misfire  problems as discussed above  (the
Citation with  a cylinder shorted in the distributor cap), due to the  fact that
the  I/M  process  would not begin  to identify these vehicles until  their  first
inspection at  one year of age (approximately  14,000 miles).

Table III.B.I. presents the pertinent  values  used in the above development and
Figure III.B.I. presents a graphical illustration of  the concept.
* Since  the  Volvo fleet most  closely resembles a fleet of  Secondary  category
vehicles,  the  new HC  deterioration  rate  was calculated  using the  Secondary
category's emission  performance  as a base.   The  new rate was  then applied  to
the  Primary  and  Misfueling  categories  as  well.    Primary  and  Misfueling
category vehicles  are  expected to experience ignition/misfire  problems at  the
same rate and with the same proportional effect as Secondary  category  vehicles.

-------
 u
1.0 J-
0,54-
Figure 3EC.&..1.
                        13,784
                  /o,ooo
                                      HO, 000

-------
                                       12

                                 Table III.B.I.

                    Values Used  in the Development of an HC

                    Deterioration Rate Without the Presence

                   of Vehicles with Ignition/Misfire  Problems

Average HC emissions of total
Volvo/Saab fleet                       = 0.462 g/mi (n=104)

Average fleetwide mileage              = 13,786 miles (n=104)

Average HC emissions without
vehicles with ignition/misfire
problems                               = 0.389 g/mi (n=102)

Percent reduction                      = (0.462 - 0.389/0.462)xlOO =  15.8%

Emission factor equation of            HC = 0.23  g/mi +  (0.12  g/mi/10,OOP miles)*
Secondary category                                   (miles/10,000)
Closed Loop vehicles[1]


HC emission level from the above
equation at 13,786 miles:

      HC = 0.23 g/mi + (0.12) * (13,786/10,000) = 0.395  g/mi

HC emission at 13,786 miles
with 15.8% reduction:

      HC = 0.395 - (0.158X0.395) = 0.333 g/mi

New Emission Factor Equation accounting for the identification
      of vehicles with ignition/misfire problems:

    a. Zero-Mile = 0.23 g/mi

    b. Deterioration Rate = 0.33 g/mi - 0.23 g/mi = 0.07 g/mi/10,OOP miles
                                  13,786 miles

-------
                                        13

The reason why a  regression  analysis  was not performed on the Volvo/Saab fleet
after removing  the vehicles with ignition/misfire  problems  can best be under-
stood by  considering the development of deterioration  rates for the Secondary
category  as  presented  in  Section  IV.A.2.b.  of  the emission  factor  ana-
lysis [1].   The  reader will  note that  although a  regression analysis  of  the
complete Volvo/Saab  fleet  was  performed in that development (after deletion of
some cars for parameter  maladjustment) , the result was not  applied  for the HC
case.   Instead,  the  result  for CO  (the ratio  between in-use  performance  and
Certification performance)  was  used  to replace  the  HC  results because  the
latter was judged  to  be  unreliable.

The  end product  of  this  portion  of the  development is a  deterioration  rate
which represents   the average  HC deterioration  of post-1980 vehicles  without
the  presence of  vehicles  with  ignition/misfire  problems.   This deterioration
rate  is  applied   to  all  three categories  of  vehicles  (Primary,  Secondary,
Misfueling)  since  ignition/misfire   problems  are  assumed   to  have  a  similar
impact on all three.

The  third and  final methodological  step  is  fairly  straight-forward.   Simply
put, at the  first inspection point  after the I/M  process is initiated,  the HC
emission performance  of  the various  categories  is  assumed  to drop down to  the
line representing  the category without  the  presence of vehicles with ignition/
misfire problems.   That  is, I/M is  assumed to  identify those  vehicles  and to
cause them  to be  repaired.    The emissions  performance of  each of the  three
categories of vehicles  in the fleet  drop down to  a new line for the category.
The three categories  can of  course be simply weighted together  to give  a fleet
composite.

Following  the  inspection,  each category  readopts its previous deterioration
rate  (0.12  g/mi/10,000  miles).  This  is  because after  the  I/M  inspection,
vehicles can be  expected to develop  new ignition/misfire problems  at  the  same
rate  of  occurrence  as   before  the  inspection.   At  subsequent  inspections
however, each category will once again drop  down  to the  line which represents
the  category's   HC  emissions  without  ignition/misfire   problems.    Figure
III.B.2.  graphically illustrates  this  concept  for the  first three  inspec-
tions.   It  examines what  is  happening   for  the   Secondary  category only.
Exactly  parallel   scenarios  are  modeled   for  the  Primary  and  Misfueling
categories.

The I/M benefits  which accrue  from  identifying vehicles with ignition/  misfire
problems account for  roughly half of  the HC benefits  contained  in MOBILE2.   CO
benefits  for this phenomenon  were  not  addressed since it  is  overwhelmingly  a
HC phenomenon.

-------
                                        14
IV. Post-1980 Model Year I/M Benefits as Contained in MOBILE2.

A.  Introduction.

The  preceding sections  presented  the  assumptions  and  methodologies  used  to
model the impact of an Inspection/Maintenance program on the post-1980 fleet's
emissions.  These  assumptions and methodological  steps  were incorporated into
a  computer  model  which tracked  the fleet  through  an  Inspection/Maintenance
program over  time.   This was done in order  to calculate actual figures of the
percent  benefit  due  to I/M (i.e.  the percent  reduction  in HC/CO  emission
levels due  to I/M).   A  detailed  discussion  of the computer model  will not  be
given here, although  a copy  of the computer program is  attached.   (Appendix
B).   The  computer  program  simply incorporates  the  methodology developed  in
this report.

The output of the  program  is a  series  of large matrices.  Each matrix contains
the benefits  which are modeled to accrue for  a  given  Identification  Rate  of
Primary category vehicles.   Each  matrix has  nineteen  rows  of  benefits.  These
rows  contain  the  benefits which  are   modeled  to  accrue  for  an I/M  program
starting-up in  any  of  the  nineteen calendar years  following a given  model
year.  The computer program is based on an assumption that  the maximum life  of
a  vehicle is 20  years.   Therefore  as  the  start-up  of an I/M  program  comes
later and later  than  the  introduction  of a given model  year, the number  of
years for which  I/M benefits are  calculated  steadily decreases.   For  example,
for the  1981  model year fleet,  if  the  I/M  program begins in  January  of  1983,
I/M benefits  need to  be calculated for  the following  18  years.  If  the  I/M
program doesn't begin  until  January of  1990,  however,  1981  model  year  vehicles
would only  be active  participants of   the program for 11 years.   There  would
therefore only be a need to calculate benefits for 11 years in that case.

The output of the  computer program presented  in  Appendix B.  then,  is  a series
of matrices  which  provide the HC and   CO  benefits  due  to I/M  under a  wide
variety  of  scenarios.   These  scenarios are  determined  by  the Identification
Rate  of  Primary category  vehicles  and the  year of  I/M  program start-up  in
relation  to the model year.   Table IV.A.I presents an example of the  benefits
stored in one  of the matrices.  This example  gives the benefits resulting from
using  the  default  Identification  Rate of  50% and  with  the  I/M  program
starting-up in the first year of a model year.  The total array of  benefits  as
contained in MOBILE2 will  not be  presented due to its size.  Table IV.A.I also
presents  the  failure  rate  of  Primary   category  vehicles  resulting  from  the
above scenario.   It should  be  noted that this failure  rate does not  include
the identification of  vehicles with  ignition/misfire  vehicles.  These  vehicles
can be  expected  to contribute  between  2% and  6% to  an I/M program's  failure
rate, especially as given model year fleets  age.

-------
                                        15


                                 Table IV.A.I.

        Post-1980 I/M Benefits for.a  50%  Identification Rate and Program
                   Start up in the  First  Year  Of a Model Year
                  Primary Category
Benefit Year      Failure Rate*        HC Benefit*         CO Benefit*

      13                   89
      2                 1                   23                  28
      31                   28                  30
      41                   31                  31
      5                 1                   33                  31
      6                 1                   35                  32
      7.1                   36                  32
      8                 1                   36                  32
      9                 1                   37                  32
     10                 1                   37                  32
     11                 1                   38                  32
     12                 1                   38                  32
     13                 1                   38                  32
     14                 1                   38                  32
     15                 1                   39                  32
     16                 1                   39                  32
     17                 1                   39                  32
     18                 1                   39                  32
     19                 1                   39                  32
* Percent.

-------
                                       16
B.  MOBILE2.-
The  complete matrices  of  benefits  described  above  are  stored  in  MOBILE2.
MOBILE2 has  the  capability  to  internally  combine  the  I/M emission  benefits
resulting  from  all model  years  of vehicles  (before  and after the  1981  model
year) and to arrive at a fleetwide figure of  the percent  benefit due  to I/M in
any given  calendar year.   A detailed description of  how  that  is done  will not
be  given  in  this report.   Those  readers  interested in  exactly  how  MOBILE2
applies I/M  benefits  to calculate a  fleetwide figure  can find a discussion of
the  basic   concepts  in the  report  which  describes the  derivation  of  I/M
benefits for earlier  model year  fleets.   (That  report had not  been released in
final form at the  time when  this report was released,  so  an EPA identification
number could not be referenced.)

C.  The Default Case and Optional Cases.

The MOBILE2  default case  for post-1980 vehicles includes  a 50% identification
rate of Primary  category  vehicles and full benefits  from identifying  vehicles
with ignition/misfire problems.   It  is  not  possible  for users of MOBILE2  to
modify  the  ignition/misfire  benefits.   All  I/M programs  are assumed to  be
capable of  identifying  these vehicles.   The  rate of  identification of Primary
category vehicles  can be modified, however.   MOBILE2  contains  benefits  result-
ing  from  60%,  70%,  80%, and  90%  Identification Rates in  addition to  the
default value of  50%.   The 50%  default  value  is currently  associated  with the
use of  the Idle(N) test.  Users of  MOBILE2 who wish  to  use other  Identifica-
tion Rates  should be prepared  to demonstrate that  the  rate  they  use can  be
achieved  with   a  specific  short test  procedure   and   specific   inspection
standards.    The  Identification  Rates are therefore  related to specific  short
tests.   This is  in contrast to earlier model year  benefits as calculated  by
MOBILE2 which vary  based on  stringency factors.   Stringency factors are  a
function of HC and CO cutpoints.

D.  Mechanic Training.

There are  not assumed  to be any  incremental benefits from mechanic  training
for post-1980 vehicles.   This  is not because mechanic training is assumed  to
be of no importance  for the  post-1980 fleet.  Rather  it  is due to  the  expected
nature of repairs  on  these vehicles  and  their inability  to pass an I/M retest
if  they have not been correctly repaired.  Since post-1980 vehicles rely  on a
complex network of  interactive electrical/mechanical  components, the repair  of
Primary  category  vehicles   is  expected  to  rely  chiefly  on  correct  system
diagnosis usually followed by component  replacement  or  repair.   Repairs  are
therefore  expected to  be largely  of a  component  nature  as  opposed to  the
largely adjustment nature of repairs  for more conventional  technology  vehicles
(e.g. idle mixture  adjustment, timing adjustment).  One effect  of  this  will  be
to  make  emission-related  repairs much  more  of  an all—or—none issue.   It  is
assumed that  a  Primary  category  vehicle will  either  be  successfully diagnosed
and  repaired  with  correspondingly  low  emissions   (equal to Secondary  or
Misfueled  vehicles)  or  it  will  be  incorrectly  diagnosed  and   incorrectly
repaired with correspondingly very  high emissions.   An incorrectly  repaired
vehicle  is  unlikely  to  be  able  to  pass  an  I/M  retest.  A similar  logic
prevails for repairs  on vehicles with  ignition/misfire  problems.   Repairs  on

-------
                                       17
these vehicles are also largely of  a  component  nature  and  the emission results
of repair are also very much  an all-or-none issue.   Thus,  there is not assumed
to be a benefit  resulting from mechanic training for  post-1980 vehicles  since
a  vehicle will  not  be  able  to  pass  the  I/M  test  until the owner finds  a
mechanic who can correctly repair his vehicle.*

It may  be the case that as more  data on post-1980  vehicles becomes available,
benefits  for  the  identification  and  repair  of   Misfueling  and  Secondary
category  vehicles  emitting above  the standard  may  be practical  and  justifi-
able.   At this point  the available  data  is insufficient  to support such  an
analysis  and  the effect is  assumed  to be small.  Should this situation change,
and  Secondary and Misfueling  category vehicles are also  shown to  contribute
significant I/M benefits, mechanic  training  might also play more of a role for
the post-1980 fleet.
* EPA recommends that  local I/M planners  provide for and encourage  mechanic
training  for post-1980  vehicles.   This  is  to  ensure  an  adequate  number  of
trained mechanics to be  able to  handle  the  repairs generated  by an I/M program
as well  as to help  ensure  competitively-based  pricing in  the repair  market-
place.

-------
                                       18
References

1.  "Derivation of  1981  and Later Light Duty Vehicle Emission  Factors  for  Low
    Altitude, Non-California Areas", EPA-AA-IMS/80-8.

2.  EPA Internal Memorandum;  August 2,  1979; from Benjamin R.  Jackson,  Deputy
    Assistant  Administrator,  Mobile Source  and  Noise  Enforcement,  to  all
    Regional Administrators.

-------