IMS-009/ST-1
                           Light Duty Vehicle and
                    Light Duty Truck Emission Performance
                     Warranty; Short Tests and Standards
                               December,  1979
                                   NOTICE

Technical Reports  do  not  necessarily represent final EPA decisions or posi-
tions.  They are intended to present technical analysis of issues using data
which are currently available.   The purpose in the  release  of such reports
is  to  facilitate  the  exchange  of  technical  information and  to inform the
public of  technical  developments which  may form  the basis  for  a  final EPA
decision, position or regulatory action.
                        Inspection/Maintenance Staff
                    Emission Control Technology Division
                Office of Mobile Source Air Pollution Control
                     Office of Air, Noise, and Radiation
                    U.S. Environmental Protection Agency

-------
Abstract

This  report  addresses the relationship  between short test emission  results
and emissions  as measured  on the  "complete"  or certification test  (hydro-
carbons and  carbon  monoxide)  for 1981 and  later model year light  duty vehi-
cles  and  light  duty  trucks.   Several jurisdictions  are  now  using idle  and
loaded  mode  short  tests  as  methods  for evaluating  emissions  from in-use
motor  vehicles.   As  provided  in section  207(b)  of  the  Clean Air Act,  EPA
must  establish  an  Emissions Performance  Warranty  if  a short  test  can  be
developed which  is:   1.   available, 2.  in accordance with good engineering
practice, and  3.   reasonably  capable of  being correlated  with the  Federal
certification  test  (known as  the Federal  Test Procedure or FTP).  The pur-
pose  of  this report is  to evaluate the  correlation  of idle and  loaded short
test  emissions  to  results which will  be obtained  on the certification test
for 1981 model year and  later passenger  cars and light trucks.

In  proceeding  with this work,  a less common  definition of correlation  was
adopted.  While  it is not possible for  the short  tests to predict  on-road
mass  emissions with mathematical certainty,  they  can detect malfunctioning
vehicles with  their attendant high  emissions.   This  functional definition of
correlation  is advanced  in this paper  as the only practicable approach which
satisfies the  statutory  requirement.

This  report  addresses the  issue of  correlation for 1981 and later  model year
vehicles.  For this purpose,  data  from  1975-77 model year federal cars  was
examined.   This  group  represents   the  most   advanced  automotive emission
control  technology  in service for  which adequate data is available.  (Cali-
fornia  cars  would  be  more appropriate for  some purposes but, unfortunately,
sufficient data  does  not exist.)

Three  short  tests are recommended:  idle,  two  speed  idle and two mode loaded.
These  procedures,   necessary  equipment  and  short  test  standards  are   all
described  in Section XI of this report.   These short tests and  associated
standards give approximately  the  same  error  of commission  rate as  the  FTP
(comparing single test results to average  emission levels) while identifying
a substantial  part  of the  excess emissions.  This result is obtained  because
the short  test tends  to  ignore  marginal failures,  i.e.  those vehicles only
slightly above FTP  standards.

Further  information regarding 207(b) short  tests may  be  obtained from United
States  Environmental  Protection  Agency, Inspection/Maintenance Staff, 2565
Plymouth Road, Ann  Arbor,  MI  48105,  (313)668-4367.

-------
                              Table of Contents




     Items                                                  Page




I.   Background, Statutory Requirements                      1




II.  Rulemaking Activities, Direction of Program             2




III. Approach                                                4




IV.  Data                                                    6




V.   Reasonable Correlation                                  8




VI.  Standard Selection                                     11




VII. Two Speed Idle Test                                    12




VIII. Idle Test                                             16




IX.  Two Mode Loaded Test                                   17




X.   Geographical Effects, Light Duty Trucks                20




XI.   Optional Standards                                    22




XII.  Recommendations, Summary                              22

-------
I.   Background, Statutory Requirements

The purpose  of  this paper is to examine  the relationship between  short  test
emissions and actual  emissions  observed  on  the  EPA certification test  pro-
cedure.  (The EPA certification test, or  Federal Test Procedure, is commonly
referred to  as  the FTP).  Many states will be inspecting motor vehicles for
exhaust  emissions  using various  short tests;   repairs  will be required  if
emissions exceed  limits established by the states.  These Inspection/ Main-
tenance  (I/M)  programs  are  mandated by  the Clean  Air  Act  for  areas  with
severe air quality problems.

As an  element  of consumer protection, the Congress  provided  for an emission
control  performance  warranty, Section  207(b)  of  the  Clean Air Act.  Under
certain  conditions,   the  manufacturer will  be  required  to  remedy emission
failures.   However,   before  this  warranty can  become  effective,  EPA  must
promulgate an "approved" short test;

     If  the Administrator determines that

          (i)   there  are  available testing  methods and  procedures  to  as-
          certain whether, when in actual use...each vehicle...complies  with
          emission standards,

          (ii)   such  methods  and  procedures  are  in accordance  with  good
          engineering  practices, and

          (iii)  such  methods and procedures are reasonable  capable of being
          correlated with...(certification tests)...then

     he  shall  establish such methods  and procedures by regulation.   (Clean
     Air Act Section  207(b).)

The first  two  requirements have already  been  satisfied.  As  several juris-
dictions  are  currently  using  idle  and/or  loaded  testing  for   inspection
purposes,  these  tests are  certainly "available".   (The last half  of  the
first  requirement,  ability  to ascertain  compliance  with emission  standards,
is shown by satisfying the  third requirement, correlation with the certifi-
cation  test.)   The  criterion  of good  engineering practice  is  met because
these  tests  can be conducted with  reasonable  demands of test personnel and
equipment,   and  they yield  reasonably  accurate  and  reproducible results.

The  term "reasonably capable of  being  correlated"  is  not  defined  in  the
Clean  Air  Act.   Correlation implies a  mutual  relationship;  two quantitites
which  can  be mathematically related are said  to  be correlated.   Also,  two
methods  of  ranking  objects  are correlated  if  they  put the objects in  the
same order  or  class.  For example,  if eggs are normally graded by weight,  a
method  of measuring the length and  width which achieves the  same  result can
be  said to  be  "correlated."  The  degree of  error which  can be tolerated
before  "correlation"  is  lost is subjective.

Correlation  is  the most difficult,  and consequently the most crucial,  ele-
ment in establishing an approved short test.   Before they can be sold,  car
and light truck  exhaust  emissions are measured  (using prototype vehicles)  on

-------
the Federal  Test Procedure  (FTP).   It is an expensive,  time consuming and
complicated method designed  to  duplicate urban driving.   This  test takes  a
minimum of  12 hours,  has a specific  preconditioning  sequence and requires
sophisticated gas metering  and  analytical equipment.  If  emission  standards
are met,  a  certificate of conformity  is  issued  by EPA and sales can begin.
While  the FTP  is totally  unsuited  for  inspection purposes,  its intended
functions, prototype  vehicle  certification,  selective enforcement  audit and
recall, are  served well.  The  test  begins with a cold  start after a  simu-
lated overnight  "soak";   many  different driving modes are tested,  including
some highway  operation.   All  of this  is  performed under  controlled labora-
tory conditions  using a  chassis dynamometer  to  duplicate vehicle inertia,
wind resistance and tire  losses.

It should be obvious  that correlating a short, simple inspection  test  to the
FTP is a significant  task.  Compounding this problem is the FTP's cold  start
which  yields  a  substantial  part of  the total hydrocarbon   (HC)  and  carbon
monoxide  (CO)  emissions.  The  cold start also tends  to  be   quite  variable.
During  cold  starts,  a gasoline engine requires a  significantly richer  fuel/
air mixture to insure  that enough fuel is vaporized.  For  most vehicles this
is  provided  by  a "choke" on  the  air supply.  It  is  this rich mixture and
choke  operation  which cause  the increased  cold  start emissions.  (Without
enough  air  for  complete combustion high emissions are inevitable)  For warm
starts and normal operation a choke is not needed.  Practical considerations
prevent  any  I/M  short test from  including  a  cold  start; therefore,   it  is
impossible  to fully  characterize  a vehicle's  true FTP  emissions.  (A car
with a  faulty choke might have very high  FTP emissions and still be "clean"
on  any short  test.)   Since  classical  correlation requires some  sort   of
mathematical  relationship,  [e.g. FTP  emissions =  function(Short  Test)]  no
short  test  can  ever achieve classical correlation.  However,  the short test
can be  used  to correctly identify a substantial number of the high emitting
vehicles; this will be developed below.

The  position taken  in this report  is  that  a short  test which accurately
predicts  FTP failure satisfies the correlation requirement of the  Clean Air
Act; strict  mathematical  correlation  is  not necessary.

A  short  test  should  be  acceptable  if  it  satisfies two  goals.    First,   it
should  identify a substantial  portion of "excess" emissions, i.e. emission
above  the  appropriate standard.   Second, the  number of false  short test
failures,  (called errors of  commission on EC'S),  should  be  as low as pos-
sible.   This is necessary so that  manufacturers  (and  owners whose vehicles
don't  meet warranty  requirements)  will  not  be burdened  by  having to make
unnecessary  repairs.   It is  unfortunate, but  not pertinent,  that a  short
test will not  identify all  non-complying  vehicles.   These  errors  of omission
 (Eo's)  have  no impact on the  manufacturers'  warranty  claims.  In  addition,
the omitted  vehicles  tend to exceed  standards  only by a moderate  amount; the
potential for air quality  improvement  is  not as  large as  with the  correctly
failed  vehicles.

II.  Rulemaking  Activities, Direction  of  Program

On 25  May 1977  the  Environmental  Protection Agency issued a Notice of Pro-
posed  Rulemaking  (NPRM)  on  the  Emission Performance  Warranty,  42 Federal

-------
Register 26742 and 26747.  This rulemaking consisted  of  two  parts:   proposed
alternate short  tests (and associated standard setting  methods) and regula-
tions  to  actually implement the warranty.   For  various reasons, the  imple-
mentation regulations  were withdrawn and were reproposed  on 20 April  1979,
44  Federal  Register  23784.   Before  the Emission  Performance  Warranty  can
become  effective, both  sets of  rules must  be  promulgated  in final  form.

Originally,  five alternative short  tests were  proposed.   These tests  fell
into  three  classes;   unloaded,  steady  state  loaded  (constant speed)  and
transient (varying speed).  Unloaded  testing  includes the  high  and  low speed
idle  modes,  with exhaust  concentrations measured  by simple  "garage"  analy-
zers.   Oxides  of nitrogen (NOx) cannot be determined by the unloaded  tests.
For  steady  state testing, a  chassis dynamometer  is employed  to   impose  a
power  requirement  on  the engine   ("load")  for  specific  vehicle  speeds;
garage  analyzers are  also used for  determining  exhaust concentrations.   As
proposed,  neither  the  unloaded  or  steady  state loaded  procedures  would
determine the  actual  mass of emissions  produced,  only the  concentration  of
exhaust pollutants would be measured.  However, the  transient tests proposed
in the NPRM do determine true mass emissions  at the price  of using  equipment
very  similar  to the  FTP.  The three  types of  tests represent increasing
levels of sophistication with corresponding cost and  complexity.

Along  with  the various short tests,  several  methods were proposed  for sel-
ecting  short  test  standards.   These methods  applied  various statistical
tools  to  determine appropriate standards.    At  the beginning of each  model
year,  the new  vehicle population was  to be divided into  groups  by  make,
engine size, vehicle  size  and technology.  Vehicles  from each group would be
tested  to establish  correlation  between the FTP and  the various  approved
short  tests.    Use  of  many specific  groups was  promoted  to  maximize  the
correlation  obtained.   (If sufficient commonality existed,   groups  of  vehi-
cles  would  be  combined  to minimize  testing.)   One  of the  proposed  metho-
dologies  would  be applied to the data  from each group  to obtain standards.
It would  require testing many cars each year  at considerable expense;   short
test  standards  would  not  be available until  many months  after  the start of
the model year.

In  response  to comments received, new data,  a revision  to the  Clean Air Act
which  shortened  the  warranty,  and internal policy  decisions, EPA's position
has been  revised.  Initially,  only  three  tests  are  recommended for promul-
gation  (idle,   two  speed  idle, two  mode  loaded).   This  results   because,
subsequent to  the NPRM, EPA acquired  data revealing  that the idle tests have
higher  than  anticipated  correlation  for  the  latest  technology  vehicles.
Also,  the idle  tests are much  more widely  accepted and "available".   For
those  jurisdictions which wish to use  loaded testing, a  simplified  version
of  the Federal Three Mode has  been  developed.   For  each  test,  a single set
of  standards  will apply   to the  entire  vehicle  population.  This  approach
will  sacrifice  some  potential  accuracy  while significantly  simplifying the
program.  As  will be shown later, a  relatively small number of "gross emit-
ting"  vehicles  contribute  the majority  of  all  excess  emissions.   These
vehicles  are  detected by  the short  tests and standards  in this report.   So,

-------
even if a more sophisticated procedure will detect a larger number of excess
emitters, its additional air quality impact will not be substantially great-
er.

In  this  report,  loaded mode testing of NOx emissions will not be addressed.

III. Approach

By  statute,  the emission  performance warranty is  tied  to state (or local)
vehicle inspection programs.  Only when a car or truck fails  inspection  (and
some  sanction will  be  imposed)  can the  warranty be invoked.   Even  then,
certain mileage  and  maintenance restrictions apply.  It is anticipated  that
many states  will use the emission performance warranty standards as a  lower
limit  for  their  inspection standards; this will achieve the  greatest possi-
ble air  quality  improvement without failing vehicles for  which  the warranty
is  not available.    Actual  I/M standards must be  set  after  considering  the
amount of air quality  improvement needed, alternate approaches,  etc.

It  would  be  possible to develop short test warranty standards by evaluating
vehicles  which  meet all  the warranty  requirements (i.e., age  and mileage
limits, proper maintenance).  Unfortunately, such a data set  does not exist.
Further,  since   these  standards  may have wider application,  a group of  as-
received in-use  light duty vehicles will be the basis for  setting standards.

It  should  be noted  that  a portion of those vehicles  failing the state  I/M
short  test will  not  qualify  for warranty coverage because  of  tampering,  high
mileage and  lack of  required maintenance.

EPA plans to have the  emission performance warranty effective beginning  with
the 1981  model  year.  Promulgation of the warranty regulations,  short  tests
and standards (or procedures to set standards) must occur  prior  to  the  start
of  that model year; this  is  a specific  requirement of  The  Clean Air  Act.
Quite  obviously,  it is  impossible to  test  in-use  vehicles  before  their
introduction.  Instead,  the  standards will be derived by  testing the highest
technology  vehicles for  which  data is  available.  Under   the  NPRM,   this
dilemma  was  to  be   solved by  collecting  data  on  early production  vehicles.
But, with  the enactment of  the  1977  Clean Air Act amendments,  the warranty
period was significantly reduced making  the  proposed  approach  less attrac-
tive;  data gathering  would consume a major part  of  the time available  for
warranty  coverage.

A  comparison reveals  that 1975 through  1977  California  vehicles  have  the
closest  emission standards  to  the forthcoming 1981  and later  Federal limits.

-------
                     Emission Standards  (FTP) - g/miles

                         HC_              CO             NOx

California - car
     1975-6              0.9             9.0            2.0
     1977                0.41            9.0            1.5

Federal - car
     1975-6              1.5             15             3.1
     1977-9              1.5             15             2.0
     1980                0.41            7.0            2.0
     1981 and later      0.41            3.4*           1.0

* Possible waiver to 7.0 for 1981 and 1982.

Unfortunately,  1975-7  California vehicles have not  been tested in  adequate
numbers and  under  appropriate  conditions to establish nationwide short  test
standards.   Instead,  this  paper will use EPA's Portland,  Oregon data base.
Over 2200  (1975 through 1977)  cars have been  tested at an actual I/M lane.
Subsequent  FTP's and  short  tests were  performed by  EPA's  contractor.    An
evaluation  of  these  results  will  produce  short test  standards  for which
correlation with the FTP can be established.

Future Federal  standards  are more stringent than current  levels.   With  the
lower  emission  standards  in  the future,  one  can  reasonably expect lower
short  test  emissions.  Therefore,  short   test  standards  developed using
1975-7 Federal  vehicles can be applied to 1981 and later model  year  vehicles
without  prejudicing the manufacturers rights.  However,  if unforseen tech-
nology is  developed for which the standards are  not appropriate, the manu-
facturer  could  be  allowed  to  apply  for a  separate  short  test standard  for
the effected vehicles.

Short test standards developed in this paper will apply  to light duty trucks
(LOT) as  well  as  passenger cars.   Certification standards  for  light  duty
trucks  for 1979 and  later models were  derived from the 1977 passenger  car
levels.   Adjustments  were  made to account  for the higher weights and larger
frontal  areas;  the same  degree of  emission control  is  required from  1979
light duty trucks as from  1977 light duty vehicles (passenger  cars).

                       Emission Standards (FTP) -  g/mile

                    HC_                  C0_               NOx

     1977 LDV       1.5                  15               2.0

     1979 LDT       1.7                  18               2.3

Light trucks  have  used, and will continue  to use, the same emission control
technology  as  passenger cars.   Most light  trucks  also  use engines  derived
from  passenger cars.   The same certification  test   (FTP)  is  also  used  for

-------
light  trucks.   For these  reasons,  light  trucks  will  respond  to the  short
tests  in  a manner  similar to  light  duty vehicles.   Therefore, short  test
standards derived from 1975-7 cars can be directly applied  to 1980 and  later
light trucks.

For a full discussion of the light truck standards, test procedure, emission
control equipment and related issues, see the rulemaking docket  for the  1979
light truck regulations.

IV.  Data

Information  for  this paper  came from EPA's  Portland  study.   Results  from
over  two  thousand  1975  through 1977 in-use  passenger cars were evaluated.
These  particular  vehicles were  tested  from  September 1977 through  January
1979 and  were  selected  to represent various  types of  emission control  tech-
nology and  not the general vehicle population.  For purposes of  207(b)  this
is rather  fortunate;  any lack of correlation for  a low sales group  will  be
more  evident  than if  the  sample  were  sales weighted.   All  testing was
carried out at an elevation of approximately  50 feet.

No  effort  was  made  to  screen out  maladjusted  or  tampered  vehicles.  The
sample  represents  vehicles  as  they  would  be presented   to  an inspection
station.   As it  is expected that  some  jurisdictions  may  use  the warranty
standards  as  limits  to  their  I/M  cutpoints, this   approach will  give  an
estimate of  the potential benefits of an I/M  program.

Each vehicle received an idle test by the Oregon Department of Environmental
Quality  (DEO).   It was  then  taken to  Hamilton  Test  Systems   (HTS),  EPA's
contractor, where it received a second idle test, a Federal Three Mode  (F3M)
and  a  Federal  Test Procedure (FTP).  DEQ  uses a three-step  idle procedure;
idle  - 2500 rpm  - idle;  the  lowest of  the  two  idle readings  is  used  to
determine  compliance  with Oregon's standards.  The  Federal Three Mode  is  a
loaded  test simulating  steady  state  driving;  it has high  and low  speed
cruises  followed  by an idle.   In order  to determine compliance  with  Federal
Emission  standards,  an  FTP was also run.  A  summary of the data set  appears
in Figure  1.

                                  Figure  1

                                Data Summary

                        Mean                       Mean FTP (g/mile)
Model Year      No.     Mileage                HC_           CO         NOx

1975            636     39174                  1.96         25.30
1976            828     28663                  1.64         20.30
1977            734     17787.                 1.61         22.06

     Overall   2207     28030.                 1.72         22.33      2.3R

Presented  in Figure 2 is  the distribution of FTP emission  values for HC and
CO.   Each  "contour"  line  on  the  graph  indicates the percent   of vehicles
which  would fail the FTP  at any combination  of HC  and CO standards on the

-------
                        Figure 2

                  FTP Failure Rate-(%)
                             75 65    45  35   25  15
   10.0   —I


    6.0   .

    4.0
 s   2.0
 00
o
EC
    1.0


    0.6


    0.4
                                        Standard - 15
                           I
                           6
I   I
  10
 I
20
 I
40
 I
60
f    I
  100
                                  CO - g/mile

-------
line.  For example,  53%  of the sample exceed  the emission standards of  1.5
g/mile HC and 15 g/mile CO.  This is indicated by the 45% and 65% "contours"
which bracket  the  point  representing these standards.  While a large number
of vehicles  exceed new  car  certification levels, many  of  these  are fairly
close to  the standards.   Only 35%  are  over  1.5  times  either standard; at
twice either standard this  falls to  26%.   But  these  26% of  all vehicles
account for  over  85% of  all excess HC and CO emissions, i.e. emissions over
the standards.  Fortunately  these vehicles are easiest  to  detect using  the
short test.

V.  "Reasonable" Correlation

Any emission performance warranty short test must be "reasonably capable of
being correlated"  with  the  Federal  Test  Procedure  as  required  by section
207(b).  But  what  is "reasonable"?  Certainly, the best possible short test
could do  no  better  than  the  FTP.  Since any  test has  some inherent varia-
bility,  even the  FTP  will occasionally  fail  a  vehicle  incorrectly.   Such
errors of commission (Ec's) are unavoidable.  The purpose of this section is
to investigate  the correlation of FTP results  to subsequent FTP tests.  By
this means, we will be able to estimate the number of Ecs which an  "ideally"
correlating  test  will yield.  Intuitively  a "reasonably" correlating short
test will  be permitted a greater number  of Ecs.   Compliance with  the stat-
utory correlation  requirement will be made on  this basis.

FTP variability has  been characterized in "Survey of Sources of Test Varia-
bility in  the 1975 Federal Test  Procedure,"  August,  1978, by Douglas Berg
of EPA's Motor Vehicle Emissions Laboratory.   In  Table B, Mr. Berg  makes  the
following estimates  for  total FTP variability  on  the same vehicle:

                   Percent Coefficients of Variation*

                             HC:  10-25%
                             CO:  15-30%

                *Sample  standard  deviation/sample mean.

These  coefficients of variation  can be used  to  estimate the percentage of
error of commissions if  the FTP were used as a  short test.

In  order to  evaluate  these errors  of commission,  two  assumptions must be
made.   First, it  is assumed  that  test  variability  follows  the normal  (or
bell shaped)  distribution  curve.  Second,  the  Portland data  (FTP  results) is
assumed  to  be  representative  of the  true  emissions  of the general popula-
tion.   It  is acknowledged that these  assumptions are not precisely correct.
However,  their whole  function is to  help  estimate  how well  the  FTP would
function  as  a short test.  Given the  range  of the estimated coefficients of
variation, these assumptions are  adequate.

Cars. from the  Portland  data base were  segregated into groups which  passed
FTP  standards.  All  of these vehicles  were  assumed to have  "true" FTP  valves
equal  to  the  test results observed.   For  any given  car this would not be

-------
d
o
•H
4J
td
•H
V
to
g
 O
o
    o
    n
O
CM
                       Figure 3



              FTP  Errors  of Commission
               .1%
                          T
                         0.10
                                        0.20
0.30
                        CO - Coefficient of Variation

-------
                                            Figure 4

                                Comparison of I/M Idle Standards
                            I/M Standards for late model
                            large displacement cars
    300
                                                                                 o New Jersey
    200  _
0)'
I
ex
i
o
    100 __
                                                          o California
                            Portland
                                      Proposed 207(b) ~ Two sPeed idle
"Standard Locus"
                                                                o Arizona
                                 1.0
                                  I
                                2.0

                               CO - %
 I
3.0

-------
                                    11
correct; however,  for  the entire population  this  assumption should be  rea-
sonable.   It  is expected  that if  a  subsequent FTP  were administered,  the
results observed would  be similar  to  the  first set;  variations  for indivi-
dual vehicles would average out.

To determine  the  FTP's  error of  commission  rate,  coefficients of variation
for HC  and CO were assumed.  For each vehicle  passing the FTP a  probability
of failing was  calculated.  This probability is a function  of how close  the
FTP results were  to be standards.  Various coefficients  of  variation  for HC
and CO  were assumed.   (Probability of failure  was calculated by  determining
how many  standard  deviations  a particular vehicle was  below the standards
and looking up the  probability  in a normal distribution table.  For example;
at an assumed HC coefficient of variation of 0.20, a vehicle whose emissions
are 1.25 g/mile  would  be  1.0  standard  deviation below the  1.5 g/mile  stan-
dard:    1.5 g/mile  - 1.25  g/mile =  0.25 which is 0.20 of  that vehicles  emis-
sions of 1.25 g/mile; or one standardized deviation.  At  1.0 standard  devia-
tion  below the standard  the vehicle  would  have a 16%  chance  of failing  a
subsequent  FTP.   This  process was then  performed for  CO,  and  a  total EC
probability determined).   This  process was repeated for each passing vehicle
to  determine  its  probability  of being  an error  of  commission  (Ec)  on  a
subsequent  FTP.  The  sum of these  probabilities divided  by  the total number
of vehicles is the  estimated FTP EC rate.  Results are indicated  in Figure  3
for a range of HC and CO coefficients of variation.

Assuming coefficients  of  variation of 20%  and 25% for  HC  and  CO,  the  FTP
will  give   approximately  5.3%  errors  of  commission.   (Using the  range of
estimates  for coefficients  of  variation,  EC'S will  be  between  3 and  6%).
This  statement  is  derived from the Portland data sample where  only 47% of
the vehicles  pass  FTP  HC  and CO standards.  If a "cleaner"  population  (more
passing cars)  were tested, errors  of commission would increase because  more
cars  would be  subject  to  the  random  test variability.   (As a  convention,
errors  of  commission are  expressed as a percent  of  the entire  population,
not just those below emission  standards.)

Since  the  FTP, used  as a "short" test, yields approximately 5% errors of
commission,  this  is a criteria against which  the  short  tests can be  evalu-
ated.

VI.   Standard Selection

Figure  4  indicates idle HC and CO  I/M standards for late model, large  dis-
placement  cars in several  jurisdictions.   Also  shown   is  a  general   rela-
tionship between  the  two  constituents,  developed  from  earlier work.    This
relationship,  or  "standard  locus", represents EPA's  best  judgment  (as of
September,  1979)  of the  best  relationship between HC  and CO standards  for
typical  I/M programs.    In developing  this  relationship, test results  from
approximately  300  1975-77 California  cars  were briefly evaluated.    Items
investigated  were   failure  rate,  errors of  commission  and  excess emissions
identified.   Results  of  this  investigation   indicate   that  the "standard
locus"  is   the  center  of a  region where  errors  of commission  and   excess
emissions  identified are constant for a given failure  rate.   Farther  away
from  the   standard  locus,  the  test  loses  some  effectiveness  and errors of

-------
                                   12
commission  increase.   For these  reasons,  short  test  standards will be  de-
veloped from  this  relationship.   (A more thorough study on  the  topic of  I/M
standards  is  now  underway;   results  should  be  available   in  early 1980).

As of  September,  1979  the most stringent I/M program had idle  standards  of
1.0% CO and 225 ppm HC (Portland, Oregon for catalyst vehicles).  The corre-
sponding standard  on the  locus  is 1.0%  CO and  200 ppm  HC.   If errors  of
commission  are  low  enough,   such  a  standard for the  emission  performance
warranty would be desireable.

It  is  expected  that many  future  vehicles will  be designed  for  low  idle
emissions, even after some key parts of the control  system have  failed.   For
these  types of  failures  an  additional  short  test  will  be needed.   To  be
effective,  such  a short  test must exercise  the  fuel  and  emission control
systems  so as to  detect  component failures.   Either  a  high  speed idle  or
loaded mode should accomplish this end.

Standards  will  be developed  for three separate,  but  related,  short tests.
All tests  include  an idle mode.  Different  standards  are necessary because
of the way results are calculated  and  because of potential preconditioning
differences.  The  three  tests are briefly described below;  all  modes except
the 30 mph cruise are performed with transmissions in neutral.

               Idle             - ° Idle

               Two Speed Idle   - ° Idle
                                - ° 2500 rpm  idle
                                - ° Idle

               Two Mode Loaded  - ° 30 mph cruise
                                - ° Idle

The two  speed idle will be addressed first.   This test is used  in  the  Port-
land I/M  program.   Standards of approximately  equal stringency  will then be
developed  for the other procedures.

VII.   Two  Speed Idle Test

An  analysis was  performed  on the  Portland data  using an  idle  standard  of
1.0% CO  and 200 ppm HC.  The two speed idle  test consists  of  two  idle por-
tions  separated by a high speed segment (idle  - 2500 rpm  - idle); the lowest
HC  and CO  levels  of the  two low  speed  idles is used to  compare with  the
standards.  This  is  the procedure used  in Portland.

              Idle Portion only

           Failures             - 37%

           EC'S                 - 4.0%

           Excess HC  identified - 81%

           Excess  CO  identified - 83%

-------
                                   13
This  test satisfies  the  requirement of  reasonable correlation  and is  an
acceptable emission  performance  warranty  standard.  Effects  of other  pos-
sible  standards  are  shown  in Figure  5.   Comparison  of the two speed  idle
test (idle portion only) to the FTP is shown in  Figure  6.

However, as with any  sampling process, predicting  the number of  Ecs  (like  an
election result) is not an exact  process.  For idle  standards of 1.0% CO and
200  ppm HC an  estimate taking into  account  the impact of  test variability
was  made.  The  sample of 2207 vehicles was randomly divided into 22 subsam-
pies of  100  cars each.  For  each subsample  the number  of  Ecs was computed.
Considering  this to  be  a  sample  of 22  programs,  nonparametric tolerance
intervals  based  upon  the  distribution of order statistics were derived  to
estimate  expected programmatic   effects.  The  following  statements can  be
made with 90% confidence:

          - 89% of the  I/M programs will have  EC'S  less than 5%

          - 92% of the  I/M program will have EC'S  less  than 6%

Of  course,  these error of  commission rates were  derived  by an  analysis  of
the  idle  standard applied to groups  of  1975  to 1977 model cars.  The  emis-
sion performance  warranty will apply  to  1981  and later vehicles.   For  pas-
senger  cars,  idle emissions  and errors of commission  should  be much  lower
since  the newer  vehicles  will   be  certified to  lower emission  standards.
Light  trucks will  experience  about   the  same  EC  rate, they use  the  same
emission control  systems as the earlier cars.

The  emission performance  warranty can  also  be applied to the high  speed
(2500  rpm) segment.   For 1975 through 1977 model  cars  the  low speed idle  is
an  effective  test.   But,  with changing  technology the high  speed  segment
(2500  rpm) may  be needed to  identify high emitters.  (EPA  has obtained  test
results verifying this  occurance.  For some future  technology cars,  the  2500
rpm  segment   identifies certain  failure  modes  where  the  normal idle  does
not.)   Results   of  the  idle  test are  summarized below,  standards  of  1.0%
CO/200 ppm HC were used.  A summary of these results is as  follows;  they all
meet the statutory requirement of reasonable correlation:

                    Two Speed Idle Test Summary
                     Failure
                     Rate
              Excess Emissions Identified
              Ecs          HC          CO
Idle  (minimum of      37%
       two  idles)
 2500 rpm

 Combined
22%

42%
4.0%


2.8%

6.0%
81%


60%

87%
83%


59%

90%
     Note:  Test  sequence  is:  idle  - 2500  rpm - idle.   Idle results are
     the  lowest of  the  two HC  and two CO values observed.   Standards of
     1.0% C0/200ppm HC  were used  for both  idle and 2500 rpm.

-------
     100 _
                             Figure 5


                    Impacts of Possible Standards

                (Two Speed Idle Test — Idle portion only)
U-l
•H
4J


01
CO

O
•H
CO
CO
•H

s

CO
co
0)
0
X
w
      80 -
      60 ..
40 -
      20 >
                  HC
                    I


                   10
                       20
 I


30
I

40
  I

50
                                                            .  10
                                                                   % Ecs
                                                           -   5
                               Failure Rate %

-------
                              15
                          Figure 6
          Comparison of Two  Speed Idle Test to FTP
                    (Idle portion only)
cu
•H
4-1
c
0)
CO
CO

-------
                                   16
While  the  two speed  idle test  is  a specified sequence  (idle - 2500 rpm  -
idle),  three  ways exist  to  take emission  results.   (See Section XI for  a
full test description).  These methods are as follows:

     1.  Idle mode only.  Emissions  from the two idle  sequences  are measured.
     The lowest  HC  and CO readings  observed  are compared against  the stan-
     dards.   Emissions from the  2500 rpm  mode are  either  not measured  or
     their  results  are disregarded.   (This mode  is preconditioning  for  the
     second  idle).    Of  course,  if  the  vehicle  passes   on  the first  idle
     sequence, the  remainder of the  test need  not  be  run.  (The standard  is
     1.0% CO/200 ppm  HC).

     2.  2500 rpm mode only.  The vehicle is briefly  idled prior to  the  2500
     rpm mode.   Emission measurements need only be  taken at 2500 rpm,  the
     second idle is deleted.  (The  standard is  1.0% CO/200 ppm HC).

     3.   Combination of idle and 2500 rpm.   The  full  test  sequence,  idle-
     2500  rpm-idle,  is  run  with  emissions  measured under  all modes.   To
     pass,  a  vehicle  must be under  the standards at both  idle and 2500  rpm.
     As with  the "Idle mode only"  procedure  described above, the  lowest  HC
     and  CO readings  from  the two  idle  sequences  are compared against  the
     standard.   If  the  vehicle passes the first  idle,  the second idle  se-
     quence may  be  deleted.   (Standards  for both  modes are 1.0%  CO/200 ppm
     in HC).

VIII.   Idle Test

This  is the  simplest short test;  exhaust  concentrations are measured  with
the  vehicle idling  in neutral.  (This procedure  is  followed in  New Jersey
and  other locations.)   To  achieve   the same  level  of stringency as  the  two
speed  idle (idle mode only),   slightly higher standards  will be  necessary.
(The  two  speed idle  uses the  lowest of two observed  idle measurements,  any
single  measurement  procedure will  require  higher  numerical  standards at the
same  level  of  stringency).

For  standards of 1.2%  CO and  220 ppm HC  (hexane)  the idle test  yields  the
following  results:

                          Idle Test  Summary

      Failure                                       Excess  Emissions Identified
      Rate                 Ecs                       HC	CO	

      39.4%                5.6%                       82%       84%

These  compare quite  closely with those for  the  idle mode of the  two speed
idle.   This test meets  the  statutory requirement  of  reasonable  correlation,
i.e.  it results in  low Ecs  and a high fraction  of excess emissions  identi-
fied.

-------
                                    17
IX.  Two Mode Loaded Test

This paper  presents a simplified  loaded  test for use with  the emission per-
formance warranty.  Derived from  the Federal 3 Mode,  the  simplified two mode
test has  identical test conditions for  all vehicles:  30 mph cruise at 9.0
hp  load  followed  by  an idle  in neutral.   Following  is a  history  of  its
derivation and an  estimate of  its effectiveness.

As an alternative  to the idle  test, 1/M  jurisdictions may employ loaded mode
testing.  Currently, Arizona uses a loaded mode  preconditioning for its idle
test,  other  states and  localities  are considering loaded  tests.   To  the
extent  possible,   the  emission performance warranty  should accomodate these
procedures.

Unfortunately,  previously  there  was  no  standard  loaded  mode  test  in which
the test parameters are  the same  for all cars.

In  the  NPRM,  EPA  proposed the "Federal  3 Mode" (F3M), consisting of 50 and
30 mph  cruises  followed by an idle  in neutral.  Vehicles were divided into
four weight categories  for  the  purpose of setting  dynamometer  load.  This
test was  subsequently modified before the Portland program  to  provide many
more categories of dynamometer adjustment  as well  as  an idle  in  Drive for
automatic  transmission cars.    EPA's experience  with this test  revealed the
virtues of  a  procedure with fixed test  conditions.  In  performing a proce-
dure with  multiple test  conditions,  much time  is spent  determining  the
proper  classifications  for  a particular  vehicle;   errors  can  occur.   Any
possible  increase in correlation from the multiple  classification  will not
be great enough to offset  the  extra complexity.

For  this  reason a simplified  loaded procedure  with  one  test condition will
be evaluated.

As  specified  in the NPRM, the Federal 3 Mode low  speed  cruise for vehicles
up  to  2500 pounds was 30  mph  with a  9.0  hp load.   This  will be adopted for
all  vehicles.   The high speed cruise  will be deleted to  reduce test length.
(An  optional  high speed cruise  will  be  permitted,  but  warranty  standards
will not  apply).   For safety  reasons,  the  idle mode will be run in neutral
instead of  Drive  for automatic transmission vehicles.

For  heavier vehicles,  this test  condition is less severe than the original
F3M.   This power  level  is  equivalent  to a modest  acceleration  for small
vehicles  and  a  very  low acceleration for a  larger  car,   see Figure  7.
(During the FTP,   a vehicle will  see acceleration  rates of up to 3 mph/sec.)

In  order  to  achieve  the  same general  level  of stringency  it  is  necessary
that  slightly  higher  standards  be employed for  the  idle  portion  of  the
simplified  loaded  test than for the idle test.   During  the Portland program,
idle exhaust  emissions averaged  approximately  0.2% CO  and 25  ppm  higher on
the  F3M than  the  idle test (idle portion of two speed  idle).  These results
were obtained from manual transmission  cars, thus  avoiding any increase due

-------
              18
             Figure 7

    Loaded Mode Power Settings
   10 -
ex
I
I-
o
CO
rt
n

-------
                                   19
to automatic  transmissions  being idled in  "Drive"  (F3M)  as opposed  to  neu-
tral  (idle test).  This offset is probably  caused in part by the  loaded-mode
"preconditioning"  under  the F3M procedure  and, in part, by  the use of  the
lower  of  the  two  idles  in  the two  speed idle test.   Consequently,   idle
standards for the  loaded test will be set at 1.2% CO and 220 ppm  HC.

To set  standards  for the loaded mode it is necessary  to calculate a  theore-
tical exhaust concentration.  One must assume both an  exhaust emission level
(grams/mile) and fuel economy (miles/gallon).   (Several additional facts  and
assumptions  enter   in  the  calculation;  fuel  density,  fuel   composition,
stoichiometric combustion,  standard  temperatures and  pressures,  etc.  Minor
discrepancies in  these  ancillary facts will not upset the  approximate  out-
come).  From  this  point  one can  apply  the  standard combustion  equations to
predict  an  exhaust  concentration.   For example,  a  car traveling  30  mph,
emitting 7.0  g/mile  CO  and 0.41  g/mile  HC  (1980 standards), and getting 50
miles/gallon  will  have  exhaust  concentrations  of approximately  1.0% CO  and
180  ppm HC.  Vehicles  with fuel  economies less than 50 miles/gallon  will
consume more fuel  and have corresponding higher  exhaust flows.   For  the  same
mass  emissions  (g/mile),  exhaust  concentrations (%,  ppm)  will  have to  be
lower.  Air pumps,  employed  on some  vehicles,  will  increase  exhaust  flow
with the same effect on concentration.

A  very conservative example was used  in  the paragraph  above; a  vehicle
getting 50  miles/gallon  at  30 mph while its engine produces over 9.0 horse-
power.   (Tire and drive  train  losses  are not considered  in  setting  the
dynamometer to 9.0 hp.)   Such a vehicle would have to have a very efficient
engine, less than  0.4 Ib fuel/horsepower -  hr.   It is  doubtful  that  any  real
vehicle would consume less fuel  and have a  lower exhaust flow.  Also, it  was
assumed  that  HC and CO were being  emitted at  the  level  of the standards.
Again,  this is  very conservative; a vehicle which  meets emission standards
will  be very clean  at  a  steady cruise under   low power  levels when fully
warmed  up.   (These conditions describe the loaded mode test condition.)   To
pass  the  FTP,  such clean operation  is  essential to balance inherently  high
cold  start  emissions.  The  same standards used for  the  idle portion,  1.2%
CO/220 ppm HC, will  pass the hypothetical example in the previous paragraph;
this  is a  conservative  standard and will have very few errors  of commission
when applied to 1981 and later passenger cars.

These  same  standards can  also   be used  for light  trucks  even though their
1980  standards  are  significantly higher,  (1.7 g/mile  HC;  18  g/mile  CO).
However,  light  trucks  consume significantly more fuel with resulting larger
exhaust flows.   The appropriateness  of  these standards for light trucks  can
be verified by  examining how 1975-7 cars in the Portland data  base  respond.
As explained  previously,  these  vehicles are technologically similar  to  1981
and  later model  light trucks and should respond similarly  to the same short
tests.   This  comparison will  also  serve  as  an "upper  limit" for  1981  and
later  passenger cars;  their cleaner  emissions will give lower  Ecs   with
expected decreases in failure rate and excess emissions identified.   Results
are  stated  below  (Test  conditions were the same or more severe  than  the  two
mode  test  recommended.   Results  here are estimates only for  the loaded
mode).

-------
                                         20
                          Loaded Test Summary

                       Failure          Excess Emissions Identified
                       Rate             Ecs          H£          C0_

Loaded (Estimates)     24%              4.3%         59%         54%

Idle                   40%              5.2%         83%         85%

Combined               46%              7.0%         93%         93%

     Note:   Test  sequence is 30 mph  at  9.0 hp load followed  by an idle  in
     neutral; standards are 1.2% CO/220 ppm HC for each mode.

The two mode loaded test is a specified sequence.  However, as  with the  two
speed idle test, three ways exist to take emission measurements.

     1.  Loaded Mode only.  Standards of 1.2% CO/220 ppm HC are  used for  the
     loaded mode, the concluding idle may be disregarded.

     2.   Idle Mode  only.   The  vehicle  undergoes the  30  mph  cruise  for  a
     brief  period,  10-30  seconds.    An  emission  measurement  need  only  be
     taken for the  idle mode.  Standards are 1.2% CO/220 ppm HC.

     3.  Combination of Loaded and Idle.  The full test sequence is run with
     standards of 1.2%  CO/220 ppm HC applied to both modes.   If these stan-
     dards are exceeded on either mode, the vehicle fails.

As indicated  above, the loaded test and associated standards comply with  the
requirements  of  Section 207(b).   Errors of commission are within acceptable
limits and a  substantial part of the excess emissions are identified.  Light
trucks  (1981 model year)  should respond  in  a similar  manner to the above
table.  Actual EC'S for 1981 passenger cars will be lower due  to more strin-
gent Federal  Standards.

X.  Geographical Effects, Light Duty Trucks

To gain  additional  confidence in the application of the short  test standard
to other geographic areas, a similar analysis using EPA emission factor data
was performed.  All currently available EPA emission factors and restorative
maintenance  test  data on 1975 through  1979 models  was used.   (Analysis  was
done in October 1979, test results on 3706 cars and light trucks were avail-
able).  This data has been collected over  a  number of years  in 8 different
cities; it  represents the best cross section available.  For  this analysis,
results from  the  idle test (using the recommended standards, 1.2% CO/220  ppm
C,)  were  compared  with  FTP emissions; failure rates  and  percent errors  of
commission were calculated.  As can be seen in Figure 8, the results compare
closely to those  observed in Portland.  It is also evident that  light trucks
are treated  slightly  less stringently than automobiles.

-------
                                        22
    XI.  Optional Standards

    The  short  test  standards  developed above  are  appropriate  for anticipated
    future vehicles.  It is possible,  but unlikely,  that control technology will
    be  developed for which  these  standards are  not  appropriate.   Under these
    circumstances  specific  standards  or modified  procedures  should be esta-
    blished for  the effected vehicles.
    XII.  Recommendations  Summary

    Three short tests can be  used  for the Emission  Performance Warranty; idle,
    two-speed  idle and  two-mode loaded.   Initial standards for  1981  and later
    model light trucks and passenger cars should be as set forth in Figure 9.

                                       Figure  9
                                  Short Test Summary
    Name

    Description
    (test modes)
    Standards
    Equipment
1.2% CO
220 ppm HC

Garage
Analyzer
               Two Speed Idle
               1.
               2.
               3.
    Idle
    2500 RPM
    Idle
lr.0% CO*
200 ppm HC*

Garage Analyzer
Tachometer
Two Mode Loaded

1.  30 mph/9.0 hp
2.  Idle
1.2% CO**
220 ppm HC**

Garage Analyzer
Dynamometer
          *  Standards apply  to the  lowest HC  and lowest  CO  emissions observed
          during  the two idle  portions  of the test.   Standards also apply to the
          2500  rpm  mode either  independently  or in  conjunction with  the idle
          portions.

          **  Standards apply to the  loaded mode and  idle,  either In combination
          or  independently.

          Note:   All idles  are  in  neutral

    As  experience Is gained with  both the warranty  program and the later tech-
    nology  vehicles,  data can  be  collected and more stringent  standards,  if
    appropriate,  proposed  in  future  rulemakings.

    Standards  have  been  set   so  that the  error of commission  rate  is approxi-
    mately the  same as  for the FTP.   As a result,  the short test concentrates on
    vehicles with  very  high   emissions.   Failure  rates are much  lower than for
    the FTP.  Because of  the  heavy  contributions from these "gross emitters", a
    substantial  portion of the total excess emissions is identified.
U£. GOVERNMENT PRINTING OFFICE: 1979- 651-112/0131

-------
                                       21
                                 Figure 8

                  Idle Test  Comparison For Various  Cities

City
Chicago


Denver


Houston


Los Angeles


St. Louis


Washington


Detroit*
Phoenix


Composite



Type
LDV
LOT
Total
LDV
LOT
Total
LDV
LOT
Total
LDV
LDT
Total
LDV
LDT
Total
LDV
LDT
Total
LDV
LDV
LDT
Total
LDV
LDT
Total

Vehicles
792
27
819
416
27
443
321
27
348
371
27
398
625
23
648
339
27
366
100
574
10
584
3538
168
3706
Number
Failures
332
3
335
207
13
220
168
13
181
115
8
123
199
7
206
136
5
141
35
202
3
205
1394
52
1446
Percent
EC
42
1
43
8
1
9
23
0
23
23
2
25
27
1
28
22
1
23
5
42
0
42
192
6
198
Fail
42
11
41
50
48
50
52
48
52
31
30
31
32
30
32
40
19
34
35
35
30
35
39
31
39
EC
5.3
3.7
5.3
2.0
3.7
2.0
7.2
0.0
6.6
6.2
7.4
6.3
4.3
4.3
4.3
6.5
3.7
6.3
5.0
7.3
0.0
7.2
5.4
3.6
5.4
Portland *  LDV                                                   39       5.6

     LDV = Passenger car (Light duty vehicle)
     LDT = Light duty truck

  * No LDT's tested
  Source:  EPA test programs through October 1979 for 1975-79 model vehicles.

-------