Criteria  for Development of

             Emissions Averaging for Heavy-Duty Engines

                        and  Light-Duty Trucks
                            December  1980
                          Glenn Passavant
                          Chester France
                          John Anderson

Technical Reports do not necessarily  represent  final EPA  decisions
or positions.  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  devel-
opments which may form the basis  for a final EPA decision,  position
or regulatory action.
             Standards Development and Support Branch
               Emission Control Technology Division
           Office of Mobile Source Air Pollution Control
               Office of Air, Noise and Radiation
              U.S. Environmental Protection Agency

                          Table of Contents
I.   Foreword	   2

II.  Background	   2

III. Benefits of Averaging .  .	   3

     A.   Technical Flexibility	   3

     B.   Economic Advantages	   4

     C.   Regulatory Improvement 	 ...   5

IV.  Considerations in Designing an Averaging Program. ....   6

     A.   General Considerations 	 	   6

     B.   Specific Program Considerations	   7

     C.   Equity Considerations	13

     D.   Environmental Impact .	22

     E.   Technological Innovation 	  25

     F.   Long Term Market Effects	25

V.   Design Criteria for a Successful Averaging Program. ...  26

     References	  29

I.   Foreword

     With  the final  implementation of  the  "bubble policy" for
stationary  sources,  EPA is now  investigating  the possible imple-
mentation  of an  averaging concept  for  mobile  source  emissions.
Emissions averaging is meant to provide  motor vehicle manufacturers
with greater flexibility in determining  the mix of vehicles/engines
to produce  and  the  control technology to apply,  while at the same
time permitting no degradation in  the air quality benefits derived
from controlling motor  vehicle emissions.  This paper will examine
the major  benefits of the  averaging concept  and  will identify the
important design considerations which should affect the final form
of any  averaging concept.   The analysis of these important design
considerations will be  used to identify  design criteria which will
be necessary for a successful  averaging  program.

II.  Background

     The  idea  that motor vehicle emissions  should be averaged has
been debated by the government and the motor vehicle  industry  since
prior  to  the  Clean  Air  Act  Amendments in  1970.   However,  in
more recent  times,  the  concept of averaging  was  raised once  again
by  General  Motors and later by  Volkswagen  in  response  to the
light-duty  diesel particulate  emission  standard Notice of Proposed
Rulemaking.   The EPA  staff  thoroughly  analyzed  the two averaging
concepts  proposed  by the  manufacturers,  but  rejected  both  due to
statutory,  equity,   enforcement,   and   localized   impact  problems.
However, at  the same  time,  the Agency took the position of remain-
ing open  to further  consideration of averaging at a later date.jY

     In  response to  the  manufacturers' proposals and  interest
in  averaging,  a  task  force  was established  to investigate the
possibility  of  an averaging  approach  to emissions  compliance in
association  with  the statutory  light-duty  truck  (LDT)  and heavy-
duty engine  (HDE)  NOx emission standards for  the 1985 model  year.

     The  remainder of this paper will be devoted to an  investiga-
tion  of the  averaging  concept  as  it  relates to motor vehicles.
Instead  of  an  analysis of specific  averaging  concepts,  the more
fundamental  concerns  and criteria  will  be addressed.  With this as
the goal,  the remainder of this  paper  is divided into three  sec-
tions.  The  first section will discuss some  possible  benefits  of an
averaging  program.   The second  section  will identify the  important
considerations  in designing an  averaging  program.   The   third
section will  use the  design  considerations  identified  in the
previous  section  to  establish  design  criteria necessary to  a
successful  averaging  program.

     Throughout  the   remainder  of  this  paper the  term  "vehicle"
will  be used  to  represent both  light-duty  trucks and heavy-duty
engines.    Strictly  speaking, however,  manufacturers  certify  the
entire vehicle for  light-duty  trucks  but  only the  engine and
emission control system for heavy-duty engines.

III. Benefits of Averaging

     The major benefit of averaging is the increase in flexibility
which emissions  averaging affords  a  manufacturer.    The  manufac-
turer will have  a much  greater  degree of flexibility  in designing
its  own  internal program to  meet  the emission standard  under  an
averaging approach.   The potential benefits of such  a compliance
approach can be  divided  into  three  broad areas:   technical flexi-
bility,  economic advantages,  and regulatory improvement.

     A.    Technical  Flexibility

     Probably  the  largest  benefit  of an averaging program  for
emissions is the increase in  the technical flexibility available to
the manufacturers.  Instead of every family being  certified to  the
same emission standard,  the manufacturers would now be able to  use
a variety of control strategies  to achieve  compliance  with  an
emission  standard.   For  example, manufacturers can comply with a
standard by achieving  the required emission reductions  in  the
families where  they are  most  readily  available and  not seeking  as
great a reduction  in families  where control  is  more difficult.
Clearly,  the degree of  difficulty  and cost  of achieving emissions
reductions will vary from family to  family.

     In  addition, an  averaging  approach would  allow  the  manufac-
turers to make  greater  use of any emission  trade-offs between  NOx
and the other pollutants.  For example, if HC control was difficult
in  a family,  the manufacturer  could  use the  NOx-HC  trade-off  to
reduce HC and  increase  NOx.   The greater NOx levels could then be
offset  by averaging.  To a  large  degree, the current  approach
limits  the  use of  trade-offs  in the manufacturer's  control stra-

     The  fact  that  an  averaging concept allows  emission offsets
between  lower and higher  emitting families would increase  the time
available to  decrease  the  emissions  from the  higher  emitting
families.   The greater  time  available  would  allow the use of  an
optimized emission  control system  rather than the use  of a quick
fix approach which might be necessary if  time was a prime  factor in
the choice of the emission control  technology applied.

     It  is  possible  that  an averaging  approach  would lead to a
decrease  in  any  fuel  economy  penalty which  might be  related  to
emission  control.   This  is  especially true  in the case of NOx  due
to  the  relationship between   NOx control  and   fuel  economy.   This
potential savings is  further enhanced by the  fact  that the quick
fix technology which  is sometimes  used may  also cause  a  fuel
economy penalty  (e.g.,  retarded  timing).   The  higher  emitting
vehicles  which  would  require the  greatest  total  reductions under
the  current  approach to emissions  compliance could  require  less
total reduction and would thus be able to minimize  the  fuel economy
penalty which  might otherwise be  necessary  with these higher
emitting vehicles.

     Finally,  in  the  area of  emission-related  hardware,  an aver-
aging  approach would allow the  manufacturer  to choose  its  own
control stategy on a  family  by family basis, thus likely reducing
the total  emission  control hardware  costs and allowing the selec-
tive  application of  the most  costly  technology  or hardware.

     B.   Economic Advantages

     An averaging program would  have economic advantages  in  the
areas of marketing, corporate  allocations of  research and develop-
ment  (R&D) funds,  and short-term corporate capital  investment
funds.  Also more desirable  cash  flows could be maintained by the
affected firms.

     The use  of  an  averaging  approach would  minimize  the chance
that  any  family would have  to be dropped  from production  due to
technology or  control cost  concerns.  The  higher  emissions  from
these  families  could be offset  by lower  emissions  from other
families.   This would  have  the added advantage of allowing longer
use of  non-recurring  investments  such as  R&D and  tooling for some

     A  second marketing  benefit of  an  averaging  program would be
the potential change  in  marketing strategies.   For  example, it is
easy  to conceive  a  scenario  in which a manufacturer certifies two
versions of the  same basic engine  at  two different emission  levels.
Version one could have lower emission levels and higher operating
costs  due to emission control.   Version  two  could have  higher
emission levels and  lower operating  costs.   Version  one could be
sold at an artificially lower price to encourage sales and at least
partially  offset   the  higher  operating  costs,  while  version  two
could be sold at  an artificially higher price to discourage sales.
The premium received  on  version  two  could actually be  larger than
the discount offered on  version one,  thus  leaving the manufacturer
with a  larger net profit.  This premium could actually  increase as
the  manufacturer attempted  to  discourage  sales  of  its  higher
emitting   engines,  thus  pushing  the potential  net profit  even

     In relation  to marketing,  an  averaging  program  would allow
the "market  testing"  of  a limited number  of new families  without
the relatively  large  fixed  cost of  R&D associated with  emissions.
The  potentially  higher emission  limits of  these  low sales volume
families could be offset by slightly lower emission limits  from the
higher sales volume  families.   However,  as  the  sales  of this
family  grew, emissions  research and  ultimately  emission reductions
could become necessary to continue production.

     An  averaging approach  allows  manufacturers to  choose where
their  emission  related  R&D  funds  will be spent, and may allow the
emission related  expenditures to be spread more  evenly over  several
years  rather  than lumped into  the  few years preceding the imple-
mentation  of a revised standard.

     One other  potential  savings is related  to  the flexibility a
manufacturer has  in establishing  the emission limits  for each
family.  For  example,  a manufacturer may  choose  to establish the
emission limit  for  a family such that  when  produced  it also con-
forms  to the  emission standards of  one or more  of  its  export
markets.  This would  decrease  both development and  production

     Finally,   and  perhaps  most  importantly,  the increased   flexi-
bility  inherent  in  an  averaging  program could permit  the manufac-
turer  to achieve  its  overall  emissions reductions for the  least
total cost.  Although it is not  possible to  quantify the potential
savings, it  is clear that  the  total cost would  probably be less
than if  averaging were not permitted, and in any case would  proba-
bly not  be more.   This  would  lead to decreased short-term capital
investments in emission control hardware  and  would  be a great
incentive for  emission  control system optimization.  Although the
incentive for  emission control system optimization  is also present
in  a  nonaveraging  approach  to  vehicle emission  control,   little
importance is placed on  optimizing a  system to achieve the required
emission reductions  at  the  lowest cost,  and  it  is not  possible to
optimize product-wide control  strategies  for  total cost.

     In  summary,  potential   economic  disruptions  of  emission
control  regulations  may be minimized with the implementation  of an
averaging approach  for  compliance with  emission  standards.    These
economic benefits  will  be most  pronounced  for  a manufacturer who
markets a  diverse  product line.  A manufacturer with  numerous
families will  have much  more  flexibility  than  one  with  fewer
families when  optimizing  compliance  costs under an  averaging

     C.   Regulatory Improvement

     Under  the current control strategy there has been  little
incentive  to  achieve emission reductions  below  those  required to
meet the emission standards.  Under an averaging approach,  there
will be  a much  greater  incentive to  achieve  the  greatest reduction
possible, because  the extra  reductions  can  then  be used to  assure
compliance of  all  the manufacturer's families/ combinations  within
the class or category of vehicles being averaged.

     In  addition,  it would  be beneficial if the averaging  concept
could  incorporate  additional  incentives  for the manufacturers to
achieve  maximum reductions.    This might be  achieved*if a  program
for  banking and  trading of  excess  emission reductions  could be
implemented.   Such   a program  could  create a market in  emissions,
and could thus  function to reduce compliance  costs  for the industry
as  a whole.   Under  programs such as  these,  manufacturers would be
able to  save  some portion  of  their  excess emission reductions  for
future  use  or  perhaps trade or  sell some portion of their  excess
reductions to other  manufacturers.  However,  these  programs  may not
be  possible under the current Clean Air  Act.

     Finally,  the  implementation  of  an  averaging program,. espe-
cially  in  the long  term,  could  result  in a decrease in waiver
requests which result  under  the current  approach.   Fewer  waivers
would mean  less  total emissions  into the environment.

IV.  Considerations in Designing an Averaging Program

     In light  of  the  considerable  potential benefits which  might  be
realized from an averaging approach to emissions compliance,  it  is
worth  considering  some  of the  factors  involved  in designing
such a program.   To successfully .attain even a significant  share  of
those  benefits  will  require  a  carefully  conceived  program.
Such  a program  would  interact  with essentially  all  areas  of
existing  heavy-duty engine  and   light-duty truck  regulations.
Many  of the  resulting concerns  could  probably  be solved  in  a
straight-forward  manner.   The important ones  are discussed  to give
the reader  a  comprehensive picture of the complicated  impacts  an
otherwise simple  concept  could have  on  EPA's  existing  programs.
Any successfully designed averaging program should minimize any  of
these undesirable impacts.

     A.   General Considerations

     Before dealing  with  specific needs,  there are  some  general
considerations.    First,  an  averaging  approach  would  represent  a
marked change in  the way  the  Agency has to date applied the vehicle
emission requirements of the Clean Air Act.   The existing  program
requires  that each vehicle  in the  class comply with the ...same
emission  standard.   Although  a more  flexible  approach may  be
possible under  the  Clean  Air  Act,  either an  averaging  approach
would  have to be  designed whch  is  harmonious  with the  present
structure of  the Act,  or  recommendations would  have  to be  made  to
Congress for changes in the Act.

     A  second general  consideration  is the overall impact  which a
change  to averaging  could have on the entire motor vehicle control
program.  We  shall see in  subsequent  discussion that a successful
averaging  program  will  impact  nearly all  current  mobile  source
regulatory programs.  The overall motor vehicle control program  as
currently  structured is one  which has  been developed over  many
years with much trial and error.   It  is a mature program which has
achieved  a high rate of  success  in controlling motor  vehicle
emissions.   Manufacturers' warnings notwithstanding,  this  has been
done with no major disruption of the  auto industry.   The Agency has
invested  substantial  resources  in  developing  and  refining  this
program to maximize  its effectiveness.

     Introduction  of a major change  in  this  program (i.e.,  avei—
aging)  can  not be  evaluated  in  a  vacuum,  neglecting  the fact that
the  current  approach  is  in  fact   in  place  and  has  been  for many
years.   A  choice  which might be made  if starting  from  ground
zero with no existing program might not be  the choice starting from
our  current position.   The .current program should not be. reworked
unless  the  changes   show  significant  and  compelling  advantages.


     A third  area for  consideration  concerns the  philosophy  for
standards  setting  embodied  in  the nonconformance  penalty provi-
sions of the Act.   As  shown by  the  legislative history,  these
provisons were  intended  to allow EPA to  set  standards  based upon
capabilities of  what were  identified  as  "technological leaders."
"Technological  laggards,"  which  would  not be able  to  comply with
such standards, would be allowed to continue in production contin-
gent upon payment of a  nonconformance  penalty.   These  concepts of
technological  leaders and  laggards seem  to  assume  a nonaveraging
approach to compliance.  In an  averaging  approach,   a manufac-
turer's  "laggards",  often its bigger vehicle/engine  sizes, could be
balanced out by  its "leaders."   Where only average  levels  are
important,  those  which   are  laggards  or  leaders  would  not  be of
concern   or  require  special  treatment.   Introduction of averaging
would therefore have to  be  reconciled with  the Act's  provisions for
nonconformance  penalties.

     Lastly, it seems most likely  that adoption of  averaging could
lead  to  more complicated regulatory procedures than currently
exist.  For example, accurate and  timely sales forecasts and sales
data  would  be  required  on a  family/control  system configuration
basis.   Heavy-duty engine  manufacturers are not currently required
to  supply  actual sales data.   Light-duty  truck manufacturers
provide  data  for fuel  economy purposes (CAFE),  but it  is  not
likely that  the  families  for fuel economy  would be the  same as
those for  emissions.   Since  sales data  would play  a key  role in
evaluating  compliance  with an emission  standard  under averaging,
sales would have  to  be monitored closely.  Frequent  updating would
be needed to avoid the disruptive  impact of a  significant change in
projected sales that went  undetected for a long period.  The costs
of  enforcing and  monitoring vehicle compliance  may  increase under
averaging.  The  amount  of  assembly line  testing required to reach
enforcement decisions may also increase.

     The  purpose  of  discussing  the  procedural  complications  and
potential  for  more  reporting burdens  for  manufacturers  is  to note
this aspect of an averaging  program which  could run  counter  to the
current  thrust  within the Federal  government  to  simplify and  reduce
regulatory  requirements.   The above possible procedural complica-
tions will have to be carefully studied and minimized in the  design
of  an averaging program.   EPA has  adequate experience with current
certification  programs and has made substantial progress in  reduc-
ing  burdens on manufacturers  without  compromising   program  objec-
tives (i.e., abbreviated certification).   Abbreviated certification
may  have to be abandoned  for  at  least the  first  few  years  of an
averaging  program  to  provide  both  EPA and  the  manufacturers
assurance  that  the  averaging provisions  'are being  properly  imple-
mented  and  that undetected  loopholes  and  inconsistencies  are
corrected in a timely manner.

     B.    Specific Program Considerations

     The  various  EPA motor vehicle  emissions  control programs  will
now be examined  to  identify  those  aspects  of  each program  affected


by an  averaging  concept.  The  purpose is  to  identify needs  or
aspects of those programs with  which  a  successful averaging program
will have to deal.   Once these have been  identified, specific
criteria for designing an averaging program can be developed.   The
programs will be reviewed basically in  the  order in which they come
in the life cycle of a typical  certification  family.

     1.   Certification

     The certification program is  designed around the requirements
of Section 206(a) of the Act.   This section requires that EPA issue
a certificate of conformity upon  a showing  by the manufacturer that
any new motor vehicle or engine conforms  to the applicable emission
regulations.  This  certificate is  required before the manufacturer
can  sell or  introduce  into commerce its  vehicles  or engines.
Adoption of an averaging approach  would have a major impact on the
current certification program.

     The  first basic consideration is  that the program chosen must
be able  to be  implemented  in  a straightforward  and uncomplicated
fashion.   With  some  kinds  of  averaging it  could  be  difficult  or
impossible  to  make a  compliance/noncompliance decision  for indi-
vidual engine  families.   If only average  values encompassing
numerous  families  were important,  then individual family emission
rates  would have  little  meaning, and in  fact might fluctuate with
time as  a  manufacturer used  the  flexibility  of averaging to mini-
mize emission constraints  from his viewpoint.  Changes would also
occur  as  sales  projection  updates  were made.  Such an environment
of "moving targets" would make it very  difficult to  operate a
meaningful  certification  program and  at the  same  time avoid bur-
densome  and complicated paperwork.    Some compromises  on  maximum
flexibility might be needed  in  favor  of  a workable program.

     Some  of  these considerations raise a more  basic  question  of
what the  role  of certification  would  be  in  an averaging program.
Assembly line testing  to evaluate actual  production emission levels
could  be used  to provide accurate  emissions  data  for  averaging.
Data  from  preproduction  certification  vehicles  could  quickly
outgrow  its usefulness as it  was  replaced by results  of assembly
line  emission  testing  under  Section  206(b)(l) of  the Act (here-
after, assembly line testing).   After that, the approval  of running
changes,  for  example, might have  to  depend  on  new assembly line
testing  to  evaluate the effect  of the changes.

     Under  some  averaging  concepts,  the entire body of  certifica-
tion regulations  would have  to be  redesigned  away from the current
individualized approach  to vehicle and engine certification.  This
would  entail  a  comprehensive  review of  both the existing regula-
tions  and  the associated  complex of  advisory circulars.   All  of
these  are  now  oriented toward isolated decisions  on a  family
basis.   Under  averaging, the  effect  of  a  given   family's emission
level  could  not  be  determined  without  considering  all  other
families  from  the  same  manufacturer and  associated sales weights

for each.  An emission level that was acceptable at one time might
not be  at another.   Such  a possibility could  play havoc  with
attempts at an orderly certification process  as now understood.  In
addition,  certification  could  conceivably  be  a  single  go/no-go
decision for the manufacturer's  entire  product line, with the
threat being  that  no vehicles could be  produced  if  the projected
average  exceeded  the emission  standard.   The  manufacturer would
then have to reshuffle its  product  line on a  crisis basis to reduce
his average.  This is a high level  of jeopardy which because of its
consequences might be very  difficult to  invoke.  It would certainly
be undesirable from the manufacturer's viewpoint.

     This  paper  has  previously  mentioned the  fact  that  the in-
creased  complexity of certification would probably  result  in the
abandonment  of  abbreviated certification  for  at  least  the  first
several  years  of the  new   program.   There  would  be  an increased
certification workload to:   1)  develop new procedures, 2) implement
and manage  the  new  procedures,  and 3)  handle  the increased data
requirements (e.g., projected and actual sales figures).  In
addition,  it is  reasonable to  expect  that the  learning  process
under an averaging system would  consume resources in refinement of
the program.  This would cover such  areas  as closing loopholes and
dealing with unforeseen complications.

     A  final area  of consideration  concerns  establishment  of
vehicle families and  their  emission  levels.   In a situation
where only  average  emissions count,  the  appropriateness  and need
for  the  current  system of family/control  system determinations
would need  to be reviewed.   An  added complication from the manu-
facturer's point of view is the  fact that  in establishing emission
levels,  it  may  no  longer   have  fixed  standards  against which  to
design.   While  allowing more   flexibility,  averaging  would also
increase the  complexity of  the task  of establishing target levels.

     Many of the possible  disruptions  and changes to the certifi-
cation process could be minimized or eliminated if each family was
certified  to a  manufacturer  specified  level known  as  the  family
emission  limit.   The  family  emission limits would  apply  to each
vehicle  within   the  respective  families   and  when  sales-weighted
could not exceed  the  applicable  emission  standard  for the vehicles
under consideration.

     2v   Assembly Line Testing

     Under the current approach,  some families have emission  levels
well below  the  standard while other  families exhibit more marginal
emission levels.  In  this  scenario,  it is  possible to focus  assem-
bly line testing activities on the marginal families  to  ensure that
they conform to  emission standards.  An  SEA  (Selective  Enforcement
Auditing)  approach  could be  implemented  under some averaging
concepts where  there are  a  small number of  marginal  families.
Under  certain  averaging  concepts,  however,  it  is  expected that
manufacturers  would   have   increased- incentive  to  establish each

engine family  emission  limit  (limits to  which  each  vehicle in  a
given  family  must  adhere)  closer to  the engine  family's  actual
emission  level,  thus  increasing  the  number  of  marginal  families
that EPA may wish to test on the assembly line.   In addition,  some
averaging concepts  could encourage  manufacturers  to establish  a
larger  number  of  families  than those that  currently exist,
so  that  the manufacturer  could  closely  tailor  its  compliance  with
the emission  standard by using  the emission level of a number  of
specific  groups  of vehicles.    This  could conceivably maximize  a
manufacturer's ability  to offset  the high  emission  limit of  one
family against the  low  emission limit of another  family.   Conse-
quently,  the Agency may need  to increase  the number  of vehi-
cles it tests  on the assembly line to maintain the same  confidence
that vehicles  are meeting the applicable emission limits  which  the
Agency has under  the  current  approach.   The  lowest administrative
cost to EPA and the industry for this  increased testing requirement
may occur  if  EPA  required  manufacturers to continually test  a
portion  of their  production  (perhaps  1  to 2  percent)  using  a
sampling  plan  EPA designs to assure  that  a manufacturer tests  a
cross section  of  its production.

     Under  the current approach, EPA  envisions the  continued
use of an assembly  line testing program  that  allows  the  Agency to
make  pass/fail  decisions regarding  whether  vehicles comply  with
emission  limits using  a 10 percent AQL.  In certain  averaging
approaches,  however,  the assembly  line testing  program may  be
required  to  establish  the 90th  percentile emission  level  for
each family which could require more  extensive test data  than  that
required  under a  pass/fail program.

     The sanctions applicable  to a manufacturer when assembly
line  test  data  indicate that  a  family fails  to meet  a  family
emission  limit,  or when  the manufacturer  in  effect exceeds  the
applicable emission  standard  (i.e.,  when the  average, sales-
weighted  emissions from the manufacturer's fleet exceeds  the
applicable  emission standard), must be articulated in any averaging
program.   In  certain  circumstances a  family  failure  could cause  a
manufacturer to exceed the standard.   In  other cases,  it  may  not be
practicable for  EPA  to make any decisions  regarding a  manufac-
turer's  "average"  compliance  status  until  the  end  of  the model
year.   Averaging  schemes  must somehow  address  this  problem.

     3.   Nonconformance Penalties (NCPs)

     The  payment  of nonconformance penalties  is  intended  to allow
heavy-duty engines and light-duty trucks  with GVWR above  6,000  Ibs.
to  be  produced under certain circumstances even though they  cannot
meet established emission standards but can meet established upper
limits.  (The  Act  does  not  provide for NCPs for light-duty  trucks
under  6,000 Ibs.  GVWR.)   In averaging a manufacturer balances  its
higher emitting .vehicles against  its  lower emitting vehicles  to
meet an  emission  standard.  Therefore, nonconformance penalties as
currently perceived by  EPA  may not be  applicable  to an  averaging

situation.   However,  an averaging approach with  a  stringent  emis-
sion standard  and  stringent assembly  line  compliance requirements
may still  leave  some  manufacturers with a need  for  NCPs.   Indeed,
some manufacturers  could  receive  little or  no benefit from  some
averaging  concepts  and  thus,  could  in  effect still  be  operating
under the current approach.

     NCPs  may  have to  be  modified  for use  with averaging.    For
example,  when  an  averaging approach  uses  family emission  limits
(limits  each vehicle  in the  family  must meet),  the  NCPs  might  be
used for those families which exceeded their limits.   Or NCPs  might
be applied in a broader way to a manufacturer's whole product  line.
This would happen  at  the end of the  model  year in  the case where,
based  upon revised sales  volume  data or  emissions  data,   a  manu-
facturer's sales—weighted  average  of family emission limits exceed
the applicable emission  standard.   In any case,  payment of an NCP
should  remove  a  manufacturer's competitive advantage  over another
manufacturer that has  achieved compliance.

     NCPs  could be a  very  important  part in  any averaging  approach
to  emissions  control.   NCPs  could  provide much  relief  from  the
jeopardy manufacturers  might  encounter  as  sales  change throughout
the model year.

     4.   Recall
     The present  recall  program functions by  identifying  a  poten-
tially nonconforming  group  of vehicles;  testing vehicles  from that
group and,  if a determination is made that a substantial number are
not meeting an  applicable  standard,  requiring  that  all  vehicles in
that group be modified to meet the standard.  It may be possible to
continue to  operate  the  recall  program in this  manner under  an
averaging approach  if family emission  limits are established.   If,
however, under  averaging,  family  emission limits  are set  nearer
actual  emission  levels  or  the  number  of  families  increases,  an
equally effective recall  program may be more complicated  and
necessitate more  in-use testing.   The  need  for more tests  may make
it  infeasible to  conduct  a heavy-duty engine  recall  program.   EPA
anticipates that  it will  be  extremely difficult  to procure  in-use
heavy-duty engines  for recall testing.   Therefore,  if  the testing
burden  increases,  the heavy-duty  engine  recall  program may  become

     An  averaging program  could  also complicate a manufacturer's
remedy  for  the  nonconforming in-use  vehicles  the  recall  program
identifies.  For example, a manufacturer may not wish to recall the
nonconforming family  but, instead, recall a family more amenable to
emission reduction  at lower cost.   In  that  case, EPA would have to
approve the  manufacturer's  remedial  plan  and may require  the
manufacturer  to conduct additional emissions  tests  on the  vehicles
the manufacturer  elects  to  recall  to  determine  the emission level
of  the  fix.   If  a recall  of new vehicles or  engines  is  involved
when assembly line  testing  reveals that  a family is not meeting its

family  emission  limit an averaging program may complicate the
recall of  these vehicles.   This  is because for recalls of current
model year  vehicles  it may be impossible  to  know what a manufac-
turer's final average emission level  will be during the model year.
It may be necessary  to base recalls  of current model year vehicles
or engines  solely on family emission limits based on projections.
If, however, these projections were  not accurate, it may result in
EPA ordering recalls, that were not necessary,  or having to order a
second recall on  a  family because the original fix did not reduce
the emissions to a level  which would  allow  the manufacturer to meet
the  applicable emission standard under  an  averaging  approach.

     Another issue under an averaging concept  is how to credit the
manufacturer for the vehicles  or  engines  in the recalled class when
determining overall compliance with the emission  standard.  In most
recalls  not every  nonconforming  vehicle  or  engine  is  repaired.
Sales  volume  is a  factor  in determining  compliance with  the
emission standard.   Therefore, a  manufacturer  should  be account-
able for the emissions and number of  vehicles  that are not actually
repaired in a recall.

     5.   Inspection and  Maintenance  Programs  (I/M)

     In the present  case, where EPA  is looking only at NOx averag-
ing,  there  would  be very little  impact  on I/M programs.   This is
because I/M programs presently measure only HC and CO emissions.  A
physical inspection  might be  used  to look  for damaged EGR valves,
but  this would not be  affected by  use of averaging concepts.   I/M
is also not currently  applied  to heavy-duty vehicles.  However, it
would include light-duty trucks.

     On a broader perspective, where  potential future applicability
to  pollutants  other  than  NOx  is considered,  the impact  of an
averaging  concept on I/M  programs could be  radical.   This is
because I/M enforcement is  inherently on a vehicle by  vehicle
basis.   Each vehicle must go through a pass/fail decision process
to determine  if  corrections  are  needed.    Therefore, the operators
of an I/M program require clear limits which they may apply to each
vehicle passing  through an  inspection station.  Averaging may
result in  the  multiplication of limits  to  a  myriad  of  values
applicable  to different  families.  This could then create the need
for  determining  separate idle test  cut  points  for  each  of these
limits.  To evaluate  the  status  of a vehicle  could  involve identi-
fication of the family and perhaps the control system configuration
as  well as  the  make  and  model  year.   Such  identification would
probably have to  be  added  to  the  engine  label  by the manufacturer.
The  engine  family identity would  then lead  to an applicable  family
emission limit.

     Prospects such as  these  just  outlined could threaten the
viability  of  workable  I/M  programs.   I/M programs now form key
elements  of the  State  Implementation  Plans  of many  states, and
their  number  is  increasing  rapidly.   Their  role  in  the control

of  in-use emissions must  not  be  jeopardized.   Any  emissions
averaging program must be  chosen  to be compatible with  workable I/M

     6.   Allowable  Maintenance/Parameter Adjustment  Provisions

     Both of  these  areas  apply by nature  to  individual  families.
Changing standards compliance to an  averaging  approach should have
no effect on  either  one.

     7.   Proposed Durability Procedure

     This program is also carried out  on a family specific basis
to determine deterioration factors  (dfs).^/  Therefore, aver-
aging  would  not  directly impact its  operation.   However, there
are  ways in  which  averaging  would  affect  the application of
the  durability  program  results.   For  example, because  of the
ability  to  trade off high  emissions  for  one  family against low
emissions  for another,  averaging would reduce  the  jeopardy to
individual families  which could result  from adverse performance
of  an  in-use fleet.   On  the other hand,  as  preliminary dfs are
replaced by  in-use  dfs,  averaging would complicate  the  conse-
quences  of an erroneous  preliminary  df.   If  the in-use dfs in-
dicated that  the true average of  a prior year's production exceeded
the  standard, then  it  is not clear  what the options  would be.
Under  non-averaging regulations, EPA might verify  the revised
emissions for the erroneous  family df  by selective in-use  testing
from that  family.    However, with averaging  that  might  be  insuf-
ficient.  This is because the actual  existence  of a non-compliance
situation could  in some averaging concepts  also  depend on the  true
in-use  emissions  of all   the other  families  produced  by the  same
manufacturer.   Determining  those  values  could  be  prohibitively
costly in terms  of testing resources.

     Durability  may  also  impact  the  flexibility  of averaging via
its  impact on dfs.   Each  year  the dfs  would be  changed  to  reflect
the  results  of  more recent in-use  data.   These  changing dfs  will
affect  the   manufacturer's  flexibility  in establishing emission
targets for  the  new year's production, and thus,  may serve to
reduce the benefits  of averaging.

     C.   Equity Considerations

     It  is clear that  ideally  any  averaging  approach which is
implemented  for  NOx or  any other pollutant should  not put any
manufacturer  at  a disadvantage  nor give  any other manufacturer an
advantage.   All manufacturers  of light-duty  trucks or  heavy-duty
engines  should  receive  fair treatment  if  an averaging  program is
implemented.    This   section will  investigate  possible  inequitable
situations which must be  avoided if  at all possible when designing
an  emissions averaging  program.   As  an introduction, a brief
industry overview will  provide  the background information necessary
to support this  discussion.

     1.   Industry Overview

     Thirteen foreign and  ten  domestic manufacturers produced and
sold approximately 3,530,000  light-duty  trucks and heavy-duty
vehicles  in 1979.   Table 1 contains a general overview of the
corporations competing in this market ^3_/  Table  2 contains approx-
imate  sales  in  each of the vehicle  and  engine classes under con-
sideration, for  the year 1979._4/

     As can be  seen  in the tables, the light-duty truck (LDT)
market  is spread among twelve manufacturers, with the vast majority
of  the sales still held  by  the  large  domestic  manufacturers (85
percent of  the  1979 market).   The captive imports of the domestic
manufacturers and  the other  remaining imports were  15  percent of
the 1979 market.   This  fraction has been gradually increasing and
may continue to  increase in the years ahead.

     Only  domestic  manufacturers  sell heavy-duty gasoline-powered
(HDG)  engines in  the U.S.  This  market  is  clearly led by General
Motors and  Ford,  but  no manufacturer has less than ten percent of
the market.   Three  of  the four  manufacturers in this group also
sell gasoline-powered buses.

     The  heavy-duty  diesel  (HDD) engine market is  dominated  by
domestic  manufacturers.   Domestic  manufacturers  produce  about 95
percent of the heavy-duty diesels  sold  in the  U.S. each year.  This
market is  led by  Cummins  Engine Co. and General  Motors (DDA), but
Caterpillar,  International  Harvester  (IHC)  and   Mack  Trucks also
hold substantial  market  shares.   General Motors, Cummins, and IHC
are  the  primary producers of heavy-duty diesel  engines for  buses.

     The  remainder of  this  section will discuss  several  equity
conflicts which may arise and  should be avoided or minimized  in the
formulation and implementation  of any averaging program.

     2.   Averaging by Engine and Vehicle Type

     An  ideal  averaging  program for  NOx or  any other  pollutant
should deal  fairly with all manufacturers  affected.   In terms of
equity,  a fundamental question arises.   Should  averaging be per-
mitted across vehicle type, engine type,  both, or none?

     Averaging  across  vehicle  type  implies   that  a manufacturer
would  be  permitted  to offset higher  emissions in one vehicle
class  by  lower emissions within  another vehicle class.   This
would  ultimately mean that the sales-weighted emission  levels
of  one  class  with higher  emission  levels  would  be offset by
the lower  sales-weighted emission levels from  another  class.   Thus,
for  example,  higher heavy-duty  engine  emissions  could be  offset by
lower  light-duty  truck  emissions,  such that   the  sales-weighted
average  emission   level  would  meet  the emission  standard.   This
concept  could be  exclusive  by engine type (i.e.,  gasoline-fueled
only)  or  include both engine types.   Averaging by engine  type means

                                                  table 1

                                      1979 Truck and Bus Producers  I/

Scania Vabis
Toyo Kogyo







Produces Produces HDG
LDT LDDT Engines


.X X X








Produces Produces
Engines Engines





X •

I/   Data gathered from EPA Certification Division.

                                                   Table  2

                                         1979  Estimated U.S.  Sales I/

Mitsubish i
Scania Vabis
Toyo Kogyo

LDT 2/
- •

HDG Engines 3/

- '
- >
• -

HDD Engines 3/
• •
. -
. —
Bus Engines Bus Engines
_ . —
— —
— —
6,700 2,900
- -
13,000 1,100
- -
- 160
• .•
- -
- —
- -
- -
- ' • -
— —
— • —
- ' -

\_l   Calendar Year 1979.

2J   0-10,000 Ibs. GVWR:  Data was obtained  from MVMA,  Automotive
News, March 24, 1980, and discussions with several manufacturers.

3/   MO.OOO .Ibs. GVWR - estimated from MVMA data.

that a manufacturer  can offset  higher emissions in one  engine
subclass   (e.g.,  heavy-duty  diesel)  with  lower  emissions  in  the
other subclass (e.g.,  heavy-duty gasoline-fueled engine).  The same
comparison could hold  for light-duty  trucks.

     Thus, there  are  four  combinations  to  consider:   1) averaging
limited by vehicle class and engine  type (within heavy-duty diesel
families  only),  2) averaging limited by vehicle class  (within
heavy-duty engines only), 3) averaging limited by engine type only
(diesel only  -  but  both light-duty trucks  and heavy-duty engines)
and finally,  4)  unrestricted averaging (all four vehicle and engine
types could be averaged).

     As is probably obvious, there are certain equity problems tied
to any one of the four  approaches.   Averaging  limited by vehicle
class and  engine  type would  provide  the most increased flexibility
to the manufacturer  with  a diverse product line,  and  for some
manufacturers would provide  virtually no increase in flexibility.
Table 3 shows the number of families  each manufacturer certified in
each class and thus, demonstrates  the problems with this approach.

     Averaging restricted by vehicle  class  would create some of the
same problems but to a greater  degree  in the  case  of heavy-duty
engines.    Table  3 shows that of  the 17 companies which certified
heavy-duty  engines  in   1979,  2  certified gasoline-fueled  engine
only, 13 certified diesel engines  only,  and 2 certified both engine
types.   In the case of heavy-duty  engines allowing  averaging within
a vehicle  class  would provide  increased flexibility  for  only 2 of
the 17 companies involved.

     The  case is much  the  same  for light-duty trucks.   Of the
twelve companies  which  certified  light-duty  trucks in 1979 only 3
certified  light-duty diesel  trucks.   Thus, only  3 of the 12 manu-
facturers would have any increased flexibility.  However,  for those
3 the benefits could be quite great.   The high ratio of gasoline to
diesel light-duty truck sales for  these manufacturers would greatly
aid  in the minimization of any problems related to diesel gaseous

     Averaging  limited   by  engine type  would  have  some positive
implications.   For  the gasoline-powered vehicles, all  of the
manufacturers of heavy-duty gasoline-fueled engines  also produce
gasoline-powered light-duty trucks.  Thus,  4  of  the 12 producers of
light-duty  trucks would  also have heavy-duty gasoline-fueled
engines  to  include  in averaging.   No increase in flexibility
would accrue to the other eight manufacturers of  light-duty trucks.

     For  diesel  engines even  less  of  an  increase in flexibility
results.   Of the 3 light-duty diesel truck manufacturers only two
also produce heavy-duty diesel engines.   Thus, only  2  of the 12
light-duty truck  and  2  of  the  15  heavy-duty diesel engine manufac-
turers would have increased flexibility.

     Unrestricted averaging would  provide a great increase  in
flexibility, but most  certainly  some companies  would benefit
substantially more than others.  Large, diverse companies would  be
able to average all of their  product  lines and the more specialized
would have far more limited benefits.

     It may not  be  possible  to develop an entirely equitable
averaging program.    Indeed,  at  least  5  of the manufacturers  in
Table 3 would not benefit from  any averaging approach, because they
certify only one family.   Under any  of  the 4 combinations described
above,  the manufacturers with  diverse  product  lines and many
families will  benefit more  than  the  specialized or  more  limited
manufacturers.   In the long  run it  may be  that  no approach  would
benefit all manufacturers or benefit  them all  to  the same  degree.

     3.   Production Volumes  and Characteristics
     As shown in Table 2 the sales of light-duty trucks and heavy-
duty engines are not  spread  evenly over all manufacturers involved.
For  example,  the four  major  domestic manufacturers sold  over  85
percent  of  the light-duty trucks  and  one manufacturer alone
accounted for over 40 percent of the market.

     For  heavy-duty  engines,   the scenario  is  similar.   General
Motors  and Ford each  sell about fifty percent  more heavy-duty
gasoline-fueled engines  than  International  Harvester  or Chrysler.
In  the  heavy-duty diesel market,  Cummins  and General Motors  -
Detroit Diesel  Allison  (DDA)  each sell  at  least  70  percent  more
than  their  nearest  competitors,  Caterpillar,  Mack,  and  IHC.

     Sales  volume  is  important  primarily  because  of  the  flexi-
bility  it gives manufacturers.   A  manufacturer  such  as  General
Motors which has a large share  of each  of the three markets under
consideration potentially has  a  great  advantage  in that  its  re-
quired  reductions could be spread  over many  sales and  diverse
product lines.   Large emission  reductions required by some families
could  be offset  by obtaining a small  increment  of additional
reduction from another family which is easier to control, and has a
large sales volume.   In contrast, this  same flexibility is avail-
able to  International  Harvester,  but  its sales are  so  much lower
that  its  flexibility  is greatly reduced  as compared  to  General
Motors.   This  potential inequity exists  for light-duty trucks  and
both types of heavy-duty engines.

     In addition to  the  broader equity areas discussed previously,
the  characteristics  of  the  different  manufacturers'  product lines
must  also  be  considered.   Probably the single  most  important
consideration   is  the  number  of LOT, HDG,  and HDD certification
families  of each manufacturer.   Table  3  contains  the  number  of
families  certified by  each  manufacturer  for 1979.   The  number  of
families  is  especially  important  because  it  is  a  strong indicator
of  the  flexibility  available  to the manufacturers  for averaging
emissions.   Engines  within  a  family  are  expected  to  have  similar


Scania Vabis
Toyo Kogyo
        Table 3

1979 Engine Families _!/

 LPT          LDDT






 I/    Based  on  data gathered from EPA Certification Division,

eraission  characteristics,  while  engine  families are  expected to
have  different  emission characteristics.   A manufacturer  with a
variety of  families will  probably have some  families  from which
reductions  are easily obtained, and some  other families whose
emissions are  much more  difficult to reduce.   This manufacturer
will  have more  flexibility than a manufacturer whose product  line
is much  smaller.   Examples of  this  situation  are demonstrated by
Caterpillar and Mack in the HDD market and  Ford  and Chrysler  in the
HDG market.

     A second area of  product  line characteristics which should be
considered is the  type of  drivetrain (i.e. 2 wheel drive, 4 wheel
drive) used  in  light-duty  trucks.  Since the LDT emission  testing
procedure is a vehicle test, the type of  drivetrain is an important
factor  in the ability  of  the vehicle to  meet  the emission stan-
dards.   Of  the  five  major domestic  manufacturers,  all  sell  four
wheel drive  light-duty trucks but three  also sell two-wheel drive
LDTs.3/   Thus  the manufacturers  which sell  only four wheel drive
LDTs TAMC,  IHC) may  have  less  flexibility  in  complying  with an
averaging program  than manufacturers  which sell both two and  four
wheel drive  LDTs.   Table  4  contains  a  breakdown of the manufac-
turers by drivetrain type.

     A  third  area  of  product line characteristics which should be
considered is the engine size mix (See Table  4).  This is especial-
ly  true  for LDTs which certify  to a grams  per mile standard.  A
manufacturer which has a limited product  line,  with primarily large
engines, will be at a  disadvantage when  compared  to  a manufacturer
which has a much more diverse  engine  size  mix.   For  example,
Chrysler  has  six engine families  ranging  from  225  CID to  360 CID
while Ford  has five  engine families  ranging  from 300  CID to
460 CID.

      The  same  argument might  be  made for  heavy-duty engines,  but
the  fact  that  the heavy-duty  standards are on  a  g/bhp-hr basis
somewhat  decreases  the inequity.   A  gram/brake-horsepower emission
standard allows  an engine which does  more work (produces more
power)  to have a  higher  engine-out   emission  rate.   As a  general
rule, as  CID increases horsepower increases,  but  not  at the  same
rate, so  there is still the possibility that  inequities  could exist
due  to  the  need to average heavy-duty emissions on a mass  basis.

    'A  final product line characteristic  which  should be  considered
is  the  variety of vehicle  inertia weights available for the  pro-
ducers  of LDTs  and the impact of these  inertia weights  on  the
emission  levels  (See  Table 4).   LDTs with higher inertia  weights
usually also  have  larger  CID engines.   These  two  factors together
make  it more  difficult for  the larger vehicles  to meet  a grams per
mile emission  standard.   A  manufacturer  with a wide  range of
available  inertia  weights  within  each engine  family and within a
manufacturer's  product line would have  increased  flexibility  when
compared  with  a  manufacturer  with  a more  limited  product  line.


Toyo Kogyo
                               Table 4
                 1979 LOT Fleet  Characteristics  I/




CID Range
Inertia Weight
Classes Range


I/Based on data gather  from EPA Certification  Division.

2J   IHC certified two  2WD configurations  in 1979,  but  sales  were

     4.    Foreign and Domestic Manufacturers

     The final potential inequity which needs consideration is the
possibility that  an averaging  program  might discriminate against
either foreign or  domestic manufacturers,  by changing their rela-
tive market positions.   Based on the number of families  involved,
it may appear  that  the  relative position of the domestic manufac-
turers would improve.  However, what must also be considered is the
relative ease with  which most of the imported LDTs meet  the emis-
sion  standards  due to  their  smaller  engine  sizes and lighter
vehicle inertia  weights.   In  terms  of  diesels,  the foreign manu-
facturers  would  not receive  as much flexibility  as  the domestic
manufacturers because of the  smaller variety  of engine  families.

     D.    Environmental  Impact

     The  implementation  of  an  averaging  program  for   light-duty
trucks and  heavy-duty engines may have  both nationwide and local-
ized air quality impacts  which must  be studied.

     1.    Nationwide Air  Quality Impacts

     Assessment  of the  impact of  an  averaging  program  on the
national ambient air quality  is  directly tied to the change in the
per vehicle  emission rate  which occurs  as  a result of going from
the current  approach to an  averaging  approach  to compliance with
the emission standard.   The per  vehicle emission rate, in turn,  is
tied to the low mileage production targets.

     Several different factors  may  affect the low mileage targets.
Obviously  the most  important  factor  which affects the value of the
low mileage  target  is the actual value of  the emission  standard.
Under the  current  approach,  every  vehicle  must meet the emission
standard.   Consequently,  the vehicle   should be  at or  below the
emission standard  for its  full useful  life. Under  some forms  of
averaging,  only   the  average  emissions  of  a  manufacturer's  pro-
duction must  be below  the  standard,  so  the  overall  fleetwide
emission  levels  may  be  higher  than under  the current  approach.
This  increase  in  overall  emission  rates would have unacceptable
environmental  consequences,  so any   averaging program which allows
an increase in the  average per vehicle  lifetime  emissions  would not
be acceptable.

     The second  factor which has a strong impact on the  low mileage
emission target  is the  deterioration factor.   In the past engine
out NOx  emissions  have  shown little deterioration, but the use  of
add  on  devices  (such  as catalysts) to  control  NOx emissions may
lead  to  an  increase  in  this  deterioration.    This has  generally
been  the case with EGR and  3-way  catalyst systems.   Under the
current  approach the compliance strategies  for different families
have  been  similar,  so  the   deterioration  factors  have   also been
approximately  the  same  value.   However,   under  an averaging ap-
proach,  the increased  flexibility may  allow a marked increase

in the NOx control strategies  used.   This  occurrence could lead to
a change  in  the  characteristically uniform values of most deteri-
oration factors.   This  in  turn would increase the potential that
families  with  substantial  deterioration will  go undetected.

     A third  factor  influencing the level of the low mileage target
is the value  of  the acceptable quality level   (AQL)  which must be
achieved  during production.   Under  the  current approach,  a 10
percent AQL  is required  for  heavy-duty  engines and  light-duty
trucks.   This  10  percent  AQL  requires that virtually all engines
meet the emission standards.   Variability  during  engine production
will cause some uncertainty as to the ability  of  a manufacturer's
families  to  pass assembly line  testing.  To account  for  this
production variability  manufacturers  generally  decrease  low
mileage targets by a small  increment.  In addition, it is  often the
case that  manufacturers  adjust  their low mileage targets to provide
themselves statistical confidence in their  ability to pass assembly
line testing.  Under an averaging approach, if the 10 percent AQL
is dropped or altered,  the  potential  exists for a loss  in air
quality benefits  due to the  slight  increase   in  the  per vehicle
emission  rate.   However,  if   the  10  percent  AQL concept  were re-
tained under  an  averaging  approach  then the air  quality benefits
could  be  retained.   This could  be done  if,  for example,  a 10
percent AQL were  applied to each family emission limits.

     In summary,  three potential losses of emission reductions must
be avoided.  First,  average per vehicle emissions over the useful
life must remain below the  applicable standard.  Secondly, to
retain the level  of  air  quality benefits achieved under  the current
approach  emissions  deterioration  must  be closely  monitored  and
controlled for all  of the families  included.   And,  finally,  the
"cushion"  the manufacturers use because of the 10 percent AQL and
assembly  line  testing must be retained.   All   of  these components
could  be  retained  virtually intact  if  every family.cert ified
to some predetermined emission limit.

     2.   Localized  Impacts

     Local or  perhaps  regional impacts would  occur  if a dispro-
portionate number of vehicles  emitting above the  emission standard
were  operated within a  limited  geographical area.  There are
several heavy-duty vehicle types which  may have  a tendency to be
operated primarily within  an  urban  area.   The  most logical example
is  transit buses used in urban/ metropolitan areas.   In 1977
transit  buses accounted  for  over  70 percent of all  non-school
buses,5j  These buses accumulate over 60 percent  of  their miles in
urban  areas.6/   Clearly,  if  diesel bus  engines emit above any
standard under averaging,  the  potential for air quality  degradation
exists.  This  problem is compounded by the fact that  the  number of
transit  buses in  use  should increase   in  the coming decade.

     A  second area  of  potential  problems  is medium-duty diesel
trucks.    These   trucks are  primarily Class VI (19,500-26^000

lb  GVWR) and  are primarily competing in the  gasoline-powered
heavy-duty truck and  bus market.   This  class  of  vehicles  is
especially  important  for  several  reasons.   Diesel  engines  used
in  this  type of  vehicle are high speed-low horsepower,  with a
tendency toward higher emissions,  and  a  higher cost to reduce these
emissions.   It is quite  likely that  the engines  in  this  type  of
vehicle would  be  certified above  the  standard.   Another important
aspect of  the  medium-duty  diesel  group is the  percentage  of  the
time which this type  of vehicle  spends in urban areas.  Whereas the
bulk  of  the heavy-duty  diesel trucks accumulate  less than  25
percent  of their mileage in  urban  areas, medium-duty diesels
accumulate  about  35  percent  of  their  mileage   in  urban areas,T_I
Finally,  both EPA and the manufacturers  expect a marked increase in
the use  of diesel  engines in  Class VI trucks  and buses.   The
percentage  is  expected  to  grow  from  8 percent  in 1978 to about 41
percent in 1988,8/  These three  factors,  inherently high emissions,
greater than average  urban vehicle  miles, and increasing sales, may
lead to localized negative air quality impacts.

     A third group of vehicles  which  may impact  local air quality
is  the  heavier  light-duty  trucks  (6,001-8,500  Ib GVWR).   Higher
inertia weight trucks  such  as standard  pick-ups  and  vans are used
more in urban areas  than their lighter inertia weight counterparts.
This  is  primarily because these  vehicles  are  used  for business
purposes such as  hauling and  delivery and not as much for personal
transportation.   The  emissions  from  these heavier vehicles (espe-
cially those with larger CID  engines) are more difficult to reduce
than those  from compact trucks  and mini-vans.  With the concentra-
tion  of  the heavier  light-duty trucks  in  urban  areas  and their
inherently  higher emission  rates, the  potential  exists  for  air
quality degradation.

     Finally,  negative  regional impacts associated with a concen-
tration of  light-duty trucks within  a  given area  may occur.   For
example,  cities  such as  Buffalo,  Pittsburgh,  and Cleveland which
receive heavy  snow falls  and sometimes have  quite hilly terrain,
will have  a larger  concentration  of  four  wheel  drive (4WD) light
trucks.   The  concentration of  these  vehicles in urban  areas,  and
the slightly higher  emissions of  4WD vehicles over  their  2 wheel
drive counterparts  may lead  to increased emissions in these areas
over  what .would  occur  if every vehicle had to  meet the emission
standard.   If  averaging was being considered for  light-duty vehi-
cles,  the effect of  increased  use  of diesel  engines  in  taxicab
fleets would also have to be investigated.

     3.   Heavy-Duty Emissions

     The  fact  that heavy-duty emission  standards are on a g/bhp-hr
basis  introduces one  additional complication under  emissions
averaging.   Because   the  standard  is  based on a work amount, the
engine which produces more horsepower and  thus  does  more work, is
allowed  to have  a  larger  amount  of  emissions   in  terms  of mass.
Thus, this  engine could produce more  emissions  and do more useful

work, but  still  have the same  g/bhp-hr  emission  level  as  a dif-
ferent engine  which  produces less emissions  and  does  less work.

     The g/bhp-hr type  of  emission  standard with  its  inherent
trade-off between useful  work  and emission's mass already incor-
porates a form of averaging  into the  heavy-duty class.  The larger
more powerful heavy-duty  engines  which normally produce a greater
mass of  emissions  than their smaller, less  powerful counterparts
are  not  penalized in  demonstrating  compliance with  the emission
standards,  beacuse their greater emission's mass is offset by more
horsepower.  Conversely,however,  this  is  not  the case for vehicles
such as light-duty trucks which must  demonstrate compliance with a
grams per mile emission  standard.  The larger  vehicles/engines must
comply with  the  same emission standards  as their  smaller counter-
parts.  Each vehicle in the  class is  limited  to the same amount of
emissions per distance traveled.   The larger vehicles/engines would
have a more difficult time  complying with the emission standards
than their  smaller  counterparts.   For vehicles classes complying
with the grams  per  mile type of  emission standards,  averaging is
especially  advantageous  because  the  smaller  and   larger vehicles/
engines  within  the  class  can  be  offset against one  another.

     To protect air  quality we  would want  to be sure  that the
total mass  of emissions emitted  did  not increase under  an aver-
aging  approach.  Direct trade-offs  by  g/bhp-hr between  two or
more, families  would not provide  this   protection because the
absolute mass  of emissions  per engine would  vary from family to

     To allow emissions averaging  with heavy-duty engines un-
der  a  g/bhp-hr  emission standard,  the  averaging would  have to
be  done using  a mass-based  system, for example,  total grams
of  NOx  per test  or  grams per mile  of NOx  based  on some  average
emission factor  or  some average  distance covered  per test cycle.
However,  this could  have the  effect of decreasing  the  benefits some
manufacturers could  gain from an  averaging program.

     E.   Technological Innovation

     Although an averaging program will increase  the manufac-
turer's compliance options,  this  flexibility  will  probably  lead to
a  reduction  in innovation  in the  emission control  technology
field.   Any  reduction  in technology forcing is   probably  counter
to  Congress's  intent  in the provisions  of the Clean  Air Act.
Averaging may discourage innovation, because of the  emissions
offset  strategy available  to  the manufacturers.   Instead of a
manufacturer  investing  emission  related  R&D  funds  to  attempt to
bring  every  family  into compliance,  it will be able  to offset
higher emissions from some families, with  lower emissions  from  other
families.   This  emissions offset option removes much  of  the  incen-
tive to  reduce  emissions   from the higher emitting  families.
Implementing  an  averaging approach  to compliance  will also  allow
the  manufacturers  the option of  manipulating the  sales volume in

each  family to assure compliance with  the  standard  and  at the
same  time  continue  to permit  the production of  higher emitting
families.  The  implementation  of  an averaging program may lead to
stagnation in  the drive to  develop cleaner engines.    Conversely,
emissions averaging  would  allow  the  manufacturers  to innovate on
just a few engines  in perhaps one  family with little consequences
from any failure.

     As  mentioned  previously,   technological  innovation  could
probably be encouraged, if a system of banking and perhaps trading
of  emissions reductions could  be included.  This  system could
provide  additional   incentive  to  spur  technological  innovation.

     F.   Long Term Market  Effects

     As  was  shown in  the  equity discussion  presented  previously,
the  potential  exists  for  market  disruption  due  to  the inherent
characteristics of   an  averaging  program.    Under  some  averaging
approaches,  large and  diverse manufacturers  would   have  a  clear
advantage over  small or more  specialized manufacturers.   If  this
discrepancy were  to continue, the  position of the large manufac-
turers  in  the marketplace  would be  enhanced as  they  used  their
greater  resources and  sales  to manipulate the market.   In the long
term the sales of the more  specialized companies  could be  decreased
as  a  result  of  competition with  a  competing  product  from a  larger

     In  the  area  of foreign versus domestic  producers,  it appears
that  an  averaging program  would  favor domestic manufacturers.  In
the area of  heavy-duty diesel engines,  there is no doubt that  the
greater  sales  and diversification  of the domestic producers  would
allow them to improve their market positions.

     The  domestic producers of  light-duty  trucks  may  be slightly
favored  under  an averaging  program  due  to their greater  product
diversification.  This product diversification would  allow them to
use  the  lower emissions from their  smaller  trucks  to offset  the
higher  emissions  from their larger trucks.   Foreign  manufacturers
sell  more  limited product  lines, and the characteristics of  their
vehicles and engines would probably allow all foreign  manufacturers
to  meet  the standards under  the  current approach or an  averaging
approach.   Any averaging  program should strive to minimize  these
equity  concerns   and  the  problems  they  create.    However,  a  com-
pletely  equitable emissions  averaging program may  not be possible.

V.    Design Criteria  for a Successful Averaging Program

      Having identified the  benefits and  design considerations
 in  an averaging  program  it  is   now  possible to  isolate  specific
design  criteria which  should  ideally  be met for  a  successful
averaging  program.   These criteria have  been determined primarily
from  the design  considerations   which were discussed  in the pre-
ceding  section, and  are  aimed at  minimizing if not eliminating the

potential problems  of  an  emission control strategy based on  averag-

     The design  criteria identified  are outlined below, together
with an explanatory paragraph.

     1.   Any averaging  approach  implemented  for  mobile  source
emissions must have a  valid  legal base.    The  averaging approach
must  be consistent with the  statutory  requirements  for  setting
standards,  including  the requirement  in section 202(a)(l)  to  set
standards applicable to emissions from "any class or  classes of  new
motor  vehicles  or new  motor  vehicle  engines."   The averaging
approach must adhere  to  statutory  limits  for  the  affected  vehicle
and engine classes.

     2.   Any averaging  program ultimately  implemented must  be
administratively  practical and compatible with existing EPA pro-
grams.   The  program should  be  designed  such  that:   compliance  can
be  clearly determined,   there  is  a  point  (or points) at  which a
manufacturer   is  held  accountable for  the  emissions  performance of
its  production,  the sanctions   (including NCPs)  for noncompliance
are  clearly  defined and  the in-use  emission  program is not ren-
dered  ineffective.    The concepts  of  family-based  certification,
assembly line testing  (SEA),  NCPs,  Recall  and  I/M must  remain
relatively unchanged due  to  their critical role in the total mobile
source control strategy.

     3.   Each family should  have a fixed certification emission
level  which   it  must  maintain  throughout the model  year at a high
pass  rate.   This  design criterion  is  necessary  to ensure  the
continued  effectiveness  of  certification,  SEA,  NCP,  Recall,  and
I/M,   as well as  to  assure consistency with  the  structure of  the
act.   Without  a fixed  emission  level,  certification would be
almost  worthless,  and enforcement  and  I/M  would  become virtually

     4.   Any averaging program must give equivalent air quality
benefit to a nonaveraging approach and must  not  allow any sub-
stantial localized impacts.   Air  quality  degradation  can be
minimized, if not eliminated, by establishing  as strict an emission
standard as  feasible,  and setting a maximum level on the emissions
allowed from any  family.   This  maximum level  will especially
address  possible local impacts.

     5.   Any averaging program should be true "regulatory reform,"
it  should make  compliance less difficult  for the industry and
reduce  compliance costs for  consumers and industry alike.  Changes,
just for the sake of changing,  do little more  than cause confusion.
Any averaging program  should   provide  a  substantial  increase  in
flexibility  as  compared to the current approach.  The  program
should  be  designed and  implemented in as simple  a  manner  as pos-
sible  so as  to,  at  the very least, not  increase the administrative
burden  on  the regulated  industry or EPA.   Above  all, the  average

per vehicle cost of  compliance,  including  fixed  costs,  should not
be higher than  that for the  current approach.    An  averaging
approach should be at least as  cost  effective  as compliance  under
the current approach.

     6.   An averaging  program  should not increase any manufac-
turer's economic  jeopardy which might be caused by emission control
regulations.  EPA expects that manufacturers will easily  be able  to
predict the emission  levels  of the families they certify, but  sales
projections eighteen  or  more months  into  the future  are  difficult.
Some mechanism  must  be  available  to  allow the manufacturers  to
accommodate errors in their sales projections.   Also, the failure
of an  individual family in  an assembly line  testing  (SEA) program
should not  jeopardize the production  of  other families.   NCPs  or
perhaps  other  mechanisms  could  aid in  minimizing  jeopardy.

     7.   Any averaging program should  benefit members of the
regulated industry  without causing  a disproportionate level of
advantage or disadvantage.  Ideally,  no competitor  in the  market-
place should have its relative position in the market significantly
affected by  emissions  averaging.  However, due  to  the  character-
istics  of the manufacturers in the  marketplace,  it may not  be
possible  to  develop  a completely equitable approach  to emissions
averaging.   It  is  desirable, however,  to  provide increased flexi-
bility to  as  many manufacturers as possible  provided that  signi-
ficant marketplace disruptions do not  occur in  the short  or  long



y   See 45 FR 14496, (March 5, 1980).

27   See 44 FR 40784, (July 12, 1979) and 44 FR 9464, (February 13,
     1979)  for more information on  in-use durability testing.

3/   Data gathered  from EPA's Certification  Division,  and  Summar-
~    ized in 44 FR 42444, (July 19, 1979).

4/   Sales data was  taken  from MVMA  data,  R.L.  Polk data published
     in  Automotive  News,  and supplemented  by  conversations  with
     manufacturers when necessary.

57   Motor Vehicle Facts and Figures, 1978, MVMA data.

6j   EPA memo, Urban/Rural Vehicle Miles Travelled by Mobile Source
     Category, Marcia Williams, December 4, 1975.

Tj   The Development of an Emission and Fuel Economy Computer Model
     for Heavy-Duty  Trucks  and Buses, John H.  Johnson  and  Anil B.
     Jambekar, SAE paper 780630.

&J   Regulatory Analysis and Environmental Impact of Final Emission
     Regulations  for 1984  and  Later  Model Year Heavy-Duty Engines,
     U.S. EPA, OMSAPC, December 1979.