Report No.  EPA 460/3-88-004
Analytical Support for
Emission Factors Development
and Air Quality Assessment

Work Assignment No. 0-01:
Analysis of California
I/M Review Committee Data
Task 4 Report
Analysis of Loaded Mode Testing
on the Potential Effectiveness of
Vehicle Inspection and Maintenance

prepared for
U.S. Environmental Protection Agency
September 30, 1988
prepared by:

Sierra Research, Inc.
1521 I Street
Sacramento, California 95814
(916) 444-6666

-------
          EPA 460/3-88-004
       ANALYTICAL SUPPORT FOR
    EMISSION FACTORS DEVELOPMENT
     AND AIR QUALITY ASSESSMENT

    EPA Contract No. 68-03-3474

     Work Assignment No. 0-01:
       Analysis of California
     I/M Review Committee Data

           Task 4 Report:
  Analysis of Loaded Mode Testing
 on the Potential Effectiveness of
 Vehicle Inspection and Maintenance
           prepared for:

U.S. Environmental Protection Agency
         September 30, 1988
            prepared by:

          Thomas C. Austin
         Thomas R. Carlson
        Kathryn A. Gianolini

       Sierra Research, Inc.
           1521 I Street
        Sacramento, CA 95814
           (916) 444-6666

-------
                       ANALYTICAL SUPPORT FOR
                    EMISSION FACTORS DEVELOPMENT
                     AND AIR QUALITY ASSESSMENT

                           Task 4 Report:
                  Analysis of Loaded Mode Testing
                 on the Potential Effectiveness of
                 Vehicle Inspection and Maintenance
                         Table of Contents



                                                                 page

1.   Summary 	    1



2.   Introduction and Methodology 	    7



3.   Results 	   18

-------
                             1.  SUMMARY



During a recent evaluation of the California vehicle inspection and

maintenance program, an analysis of California surveillance data was

conducted to determine the ability of loaded mode testing to increase

the identification of "excess emissions" in the motor vehicle fleet.

The analysis was possible because steady-state loaded mode tests had

been conducted on most of the surveillance vehicles.   Because NOx

emission control is given high priority in California, NOx emission

cutpoints were included in the evaluation.  In addition, the idle

emission cutpoints used in conjunction with the loaded modes were the

California cutpoints.  The California idle cutpoints are generally

more stringent than the federal "207(b)" cutpoints, especially for

hydrocarbons.



In order to determine potential benefits of loaded mode testing in

other I/M programs, EPA requested further analysis of the California

data using loaded mode cutpoints for HC and CO only,  and the federal

207(b) idle cutpoints.  Since only the 1980 and later model California

models represent vehicles subject to the 207(b) cutpoints, the
                                                 if
analysis was restricted to that model year range.
* However, it should be noted that the federal 207(b) cutpoints apply
only to 1981 and later models.  By expanding the data base with the
1980 model California vehicles, some uncertainty in the results is
introduced by that fact that the performance of the earlier generation
3-way catalyst systems may not represent 1981 and later models.
                                  -1-

-------
The results of the analysis indicate that steady-state loaded mode


testing would increase the excess HC and CO emissions identified.  As


illustrated in Figure 1, using the most stringent loaded mode


outpoints considered, the percent of excess emissions identified


increases from 41.4% HC and 45.8% CO to 60.1% HC and 69.3% CO.
 CD
 CO

 O
 'CO
 .22

 LJJ
 CO
 CO
 CD
 O
 X
 LU
 CD
 O

 CD
 Q_
                          Figure 1

                 Effect of Test Type
           and Loaded  Mode Stringency
          on Excess Emissions Identified
                 (207(b) Idle Outpoints)
    100
•o
 CD   Qrt
H=   90
80


70


60


50


40


30


20


10


 0
                          69.3
             63.8
          54.8
45.8
                            /y//.
                       60.1
                 10.9
                                              16.5
               207(b)
              Idle Only
                      207(b) Idle
                      + Loaded A

                     Test Type
                         207(b) Idle
                        + Loaded B
                                                       HC
                            -2-

-------
However,  as shown in Figure 2, more detailed analysis of the data

indicates that the benefits of loaded mode testing may decrease  in the

future.  Much of the benefit of loaded mode testing for the current

fleet results from big increases in the detection rate for carbureted

vehicles, especially those with closed-loop, 3-way catalyst systems.




                          Figure 2


  Effect of Emission Control System Design

        on Excess Emissions Identification


        (207(b) Idle vs. Idle + Loaded Mode)
TJ
0
H=



I

CO

g
'co
.CO


LJJ

CO
CO

-------
In contrast, the incremental benefits for fuel injected, 3-way

catalyst equipped vehicles are relatively small.  As pointed out by

EPA staff, some uncertainty is introduced by the fact that most of the

carbureted, 3-way catalyst vehicles included in the sample were 1980

model California models.



As Figure 2 shows, the  207(b) idle cutpoints are ineffective in

identifying carbureted, closed-loop, 3-way catalyst equipped vehicles

with excess emissions.  Only 5.3% of the excess HC emissions and none

of the excess CO emissions were identified when these cutpoints were

applied to 24 vehicles  that failed the FTP standards.  Steady-state

loaded mode testing was capable of identifying 76.9% of the excess HC

and 86.4% of the excess CO.



In the case of fuel injected, 3-way catalyst equipped vehicles, the

207(b) idle cutpoints identified 69.6% of the excess HC and 77.5% of

the excess CO from 68 vehicles that failed the FTP.  Steady-state
                                                A
loaded mode testing increased the excess emissions identification rate

to 76.9% for HC and 86.4% for CO.  If the trends shown in Figure 2

hold up, the excess emission detection for 207(b) idle cutpoints will

improve as the fraction of fuel injected vehicles grows.  Steady-state

loaded mode testing will further increase the detection of excess

emissions, but the increase may be smaller than it is when a large

fraction of the fleet is made up of carbureted vehicles.



The conclusions of this study should be considered preliminary because

they are based on the analysis of one particular loaded mode test  (the
                                  -4-

-------
"High Cruise" mode from the Clayton Key-Mode test) that may not be

optimum.  Sierra never determined the reason for the high error of

commission rate that its subcontractor reported when the cutpoints

developed for the High Cruise mode were applied to the 50 mph cruise

data obtained under EPA surveillance testing.  There appears to be

either an unexpectedly sensitive relationship between the loaded test

type and the optimum cutpoints, or there are unexplained differences

between the EPA and California surveillance fleets.   This issue needs

further study.



It would also be useful to consider the incremental effects of loaded

mode testing as a supplement to idle plus 2500 rpm no load testing.

Under Task 3 of our Work Assignment, it was demonstrated that the

addition of 2500 rpm testing increases the failure rate of the I/M

test.  Because that earlier analysis was based on Test Analyzer System

data only (from the I/M test), information was not available on the

percent of excess emissions identified by the addition of 2500 rpm
                                               >
testing.  In addition, under Task 1 of the Work Assignment, it was

demonstrated that the inclusion of 2500 rpm failures had a slightly

positive impact on the emission reductions currently being achieved

under the California I/M program.



As discussed in Section 3, it appears that 2500 rpm testing is much

less effective than the Clayton "high cruise" mode.   However, further

analysis of California I/M evaluation data and the latest available

surveillance data could provide an estimate of how much the
                                  -5-

-------
theoretical benefits of loaded mode testing are reduced by the




addition of a 2500 rpm test mode.
                                 It tt U
                                  -6-

-------
                   2.   INTRODUCTION AND METHODOLOGY



Under a contract with  the U.S. Environmental Protection Agency (EPA)

for "Analytical Support for Emission Factors Development and Air

Quality Assessment," Sierra Research, Inc.  (Sierra) performs a variety

of Work Assignments for the Emission Control Technology Division

(ECTD) of EPA's Motor  Vehicle Emissions Laboratory in Ann Arbor,

Michigan.  Work Assignment 0-01 directed Sierra to perform analysis of

California I/M data for the ECTD Technical  Support Staff (TSS).   Task

number 4 of  that Work  Assignment required an evaluation of how steady-

state loaded mode  testing could affect the  identification of excess

emissions.



The general  direction  provided by TSS was as follows:

        The  discussion and analysis of steady-state loaded mode tests
        on pages 193-210 of the Technical Appendix (to Sierra's
        previous report on the California I/M Evaluation Program)
        assumes that NOx is a target for identification and repair.
        The  contractor shall repeat the analysis with appropriate
        modifications,  but assuming no NOx  inspection cutpoint and no
        interest in NOx identification or reduction.  The contractor
        shall assess the incremental effect of loaded mode HC and CO
        testing over idle-only testing.  The analysis shall be
        performed  using both California and Federal (207(b))
        cutpoints.



Background

The reason that visual and functional inspections are an important

element of I/M programs is that the "no-load" (idle and 2500 rpm)
                                  -7-

-------
testing procedures used to measure emissions are incapable of


detecting certain emissions related defects.  However, Sierra's


evaluation of the California I/M program indicated that


visual/functional checks are performed correctly only about 25 percent


of the time.






The alternative to many of the visual and functional checks is "loaded


mode" testing using chassis dynamometers to simulate actual driving
                                   i

conditions.  Loaded mode testing can reveal defects that are not


apparent at idle or 2500 rpm no-load.  The accuracy of defect


correction depends on the sophistication of the testing procedures


employed.





Loaded mode testing has been considered particularly important in


detecting excessive NOx emissions because high NOx emissions do not


occur until the vehicle is under load and because certain NOx control


devices (such as EGR) are non-functional at idle.  However, certain


types of HC and CO related defects are also difficult to detect during


no-load testing.  For example, partially poisoned catalysts are more


effective at reducing emissions under no-load conditions than when the


exhaust gas volume increases during actual driving.





Loaded mode tests are of two types -- steady-state tests, where


vehicles are operated at a fixed speed/load condition and transient


tests where vehicles are driven at varying speeds to include


acceleration,  deceleration and cruise modes.  The type of dynamometer


instrumentation required to measure emissions and the time required to
                                 -8-

-------
perform the two types of tests are different, and the transient test




requires considerably more expensive equipment than the steady-state




test.  During Sierra's earlier evaluation of potential improvements to




the California I/M program, both steady-state and transient loaded




mode testing were evaluated.









The conclusions of the earlier analysis are summarized in Figure 3.




As the figure shows, loaded mode testing was shown to be capable of




substantially increasing the detection of excess emissions.  Using




California's idle and 2500 rpm emission cutpoints, only 54% of the




excess HC and CO emissions were detected.  Vehicles which failed the




test were responsible for 19.8% of the excess NOx emissions.  The




earlier analysis showed that steady-state loaded mode testing could




increase the detection of excess emissions into the range of 68-78%




for all three pollutants.  Using the more complicated transient test,




90-95% of the excess emission were detected.









Modifications to the Earlier Analysis Requested by EPA




The analysis conducted for the California I/M Review Committee cannot




be directly applied to I/M programs where NOx emission reductions are




of no interest.  There was no attempt in that analysis to isolate the




benefits of loaded mode testing on HC and CO only since some of the




excess HC and CO emissions identified could have resulted from




vehicles that only failed the NOx cutpoints.  Exclusion of the NOx




testing would also be expected to affect the overall failure rate and




the error of commission rate.  In addition, the idle cutpoints used  in




the analysis are unique to California.
                                  -9-

-------
         by
                   Figure 3
     Excess Emissions Identified
Various Emission Measurement Options


CO
o
'at
CO
1
CO
CO
CD
o
X
LU
**•«
O
3*



100
90
80

70
60


50


40
30
20
10
0
               Ide Only
                                             92.5
                                                 94.8
                Steady State
                  Loaded
Transient
 Loaded
To evaluate  the potential benefits of loaded mode testing in other

areas of the country,  the Technical Support Staff requested that the

analysis be  repeated using the federal 207(b) cutpoints  for idle-only

and without  the consideration of NOx testing.  In addition, TSS

requested that Sierra  focus on the less expensive, steady-state

testing mode.




Methodology

Test Procedure Selection - Sierra's analysis of loaded mode testing

for the California I/M Evaluation Committee was based primarily on the
                               10-

-------
surveillance data collected by ARB (Series 1-8).  The data were

supplemented with EPA Emission Factor test data for a 50 mph steady-

state loaded mode and the corresponding FTP-  However, differences in

the test methods limited the extent to which the data sets could be

combined.  The loaded-mode steady-state test conducted by the ARB was

the "Clayton Key-Mode" test (where the test speed and horsepower

setting of the dynamometer is a function of the weight of the

vehicle).  1980 and later model vehicles also received the "MVIP

Loaded Mode" test (where the test speed is 40 mph and the horsepower

setting of the dynamometer is a function of the number of cylinders

and weight of the vehicle).  The test used during the EPA surveillance

programs was performed at 50 mph cruise using the certification

horsepower setting for each vehicle.



Table 1 summarizes the loaded mode test procedures that have been used

in the California Air Resources Board's surveillance testing programs.

The Series 1-3 programs cover model years through 1978.  1980 and
                                                ;>
later models began showing up in the Series 5 program.  As the table

shows, there are several loaded modes available for 1980 and later

models.  (It should be noted that all tests specified in the table

were not consistently performed on all vehicles.)



The 40 mph cruise mode added during the "Series 4" surveillance

testing program does not appear to be ideal because the test mode is

influenced more by the number of cylinders of the engine than the

weight of the vehicle.  It would be counter-intuitive if this were

proven to be the optimum approach.  The weight  of the vehicle would
                                  -11-

-------
                               Table 1
                      GARB Surveillance Program
                            Test Routines
                                          Speed             Load
Test Series         Test Modes            (mph)          (horsepower')

   1-3          Idle                        0                0

                Low Cruise
                     2000-2800 Ibs        22-25             4-6
                     2801-3800 Ibs        29-32             8-10
                     >3801 Ibs            32-35            10-12

                High Cruise
                     2000-2800 Ibs        36-38            13-15
                     2801-3800 Ibs        44-46            21-24
                     >3801 Ibs            48-50            27-30

   4-6          As in 1-3, plus:

                40 MPH Cruise Mode
                      <4 cylinders        39-41           9.0-11.0
                     5-6 cylinders        39-41          13.5-16.5
                >7 cyl., <3250 Ibs        39-41          16.0-19.0
                >7 cyl., >3250 Ibs        39-41          18.5-22.5
   7-8          As  in 4-6, plus:

                2500 rpm
appear to be a more effective parameter for determining the

dynamometer load which best  simulates a typical vehicle load during

actual driving.  The 40 mph  cruise mode was ignored and the analysis

of steady-state loaded mode  testing was based exclusively on the "High

Cruise" mode of the Clayton  Key-Mode.
                                   12-

-------
Cutpoints - The loaded mode emission cutpoints used in the earlier



analysis were based on an error of commission rate analysis.  CARS



staff had specified that two sets of cutpoints be utilized, one that



results in an error of commission rate of 2 percent and a second that



results in a rate of 5 percent.  Error of commission (E ) rates were



calculated from the formula:





      E  =  Number of vehicles failing short test + passing FTP

                          Total sample of vehicles





Analysis of the surveillance data to derive cutpoints conforming to



the 2 and 5 percent E  rates was performed by calculating a mean and
                     C


variance of loaded mode emissions for vehicles passing FTP standards



in each technology type/model year group.  Assuming a normal



distribution, the one-sided "tail" representing 2 percent and 5



percent of the population respectively was translated to a cutpoint.



The method provided cutpoints that were highly variable between



technology and model year groups and, in most cases, were numerically



very low.  To account for instrumentation accuracy and variability



between technology groups, somewhat higher (i.e., less stringent)



cutpoints than indicated by the error of commission analysis were



developed.  As shown in Table 2, two different levels of HC and CO



stringency were used.  Values labeled "A" are the less stringent



cutpoints and values labeled "B" are the more stringent cutpoints.







Using these cutpoints, the actual EC values for the current California



I/M cutpoints for idle (or idle + 2500 RPM for 1980+ vehicles), as



well as the values of E  resulting from the addition of the loaded
                                  -13-

-------
                               Table 2

                        Loaded Mode Cutpoints
            Used in the California I/M Evaluation Program
Model Year
1975-1979

1980

1981+


(A)
(B)
(A)
(B)
(A)
(B)
HC (ppm)
100
50
100
50
100
50
CO (%)
1.0
0.5
1.0
0.5
1.0
0.5
NOx (ppm)
1600
1600
1000
1000
750
750
mode test were computed.  For 1980 and later models, the error of

commission rate  for  the  idle and 2500 rpm cutpoints was 0.0%. i.e.,

all of the vehicles  that failed the no-load cutpoints also failed the

FTP standards.



The error of  commission  rate for the less stringent loaded mode

cutpoints ranged from  0-4.6%, depending on the technology category.

For the more  stringent cutpoints, the error of commission rate ranged

from 0-8.8%.  Although NOx  testing affected the error of commission

rate, Sierra  estimated (and subsequent analysis confirmed) that the

same HC and CO cutpoints could be used in the analysis requested by

EPA.  Maximum error  of commission rates were still in the target

ranges when NOx  cutpoints were excluded.  (Note that these are not the

207(b) cutpoints.)



Using the same HC and  CO cutpoints and less stringent NOx cutpoints

(1000 ppm for all model  years), Sierra's previous analysis indicated
                                   14-

-------
much higher error of commission rates for the EPA data base.  Since

there were no significant differences associated with the no-load

testing, it appeared that the 50 mph cruise mode used on the EPA

surveillance fleet produced significantly different results than the

Clayton Key-Mode test used on the California surveillance fleet.  As a

result, the two fleets could not be combined and the analysis was

restricted to the use of Clayton Key-Mode tests on the California

fleet.



Computation of Excess Emissions Identified and Other Statistics - The

effectiveness of any short test has always been gauged by several

measures.  The most important measures are:
     O  Error of Commission Rate  - or the percentage of the vehicles
        that fail the short test when they should not, and

     O  Excess Emissions Identification Rate - or the percent of
        emissions in excess of standards identified.
In addition to these two measures, the overall failure rate is often a

factor related to the public acceptability of an I/M program.  Errors

of Omission (vehicles that pass the I/M test but fail the FTP) are

also of some interest.  The "high emitter" identification rate is of

special interest as Sierra's prior analysis found that most of the

emission benefits of repair are from high emitters (defined as

vehicles emitting in excess of 5 times California FTP standards).



Using two stringency levels of loaded mode HC and CO cutpoints, a

total of six different sets of emission cutpoints were needed to

perform the analysis requested by EPA:
                                  -15-

-------
     1.  207(b) idle outpoints only;

     2.  California idle outpoints only;

     3.  207(b) idle outpoints plus less stringent loaded mode
         outpoints;

     4.  207(b) idle outpoints plus more stringent loaded mode
         outpoints;

     5.  California idle outpoints plus less stringent loaded mode
         outpoints;

     6.  California idle outpoints plus more stringent loaded mode
         outpoints.
As discussed above, the loaded mode outpoints used in the analysis

were the same for all 1980 and later models (see Table 2).  However,

the California idle outpoints are a function of technology.  The

California outpoints are compared to the federal 207(b) outpoints in

Table 3.



                               Table 3

               Idle Emission Cutpoints for I/M Programs
                    California vs. Federal 207(b)

                                   California          Federal 207(b)
     Technology                    HC       CO          HC       CO

No Catalyst                      150 ppm   2.5%       220 ppm   1.2%

Oxidation Catalyst/no AIR        150 ppm   2.5%       220 ppm   1.2%

Oxidation Catalyst w/AIR         150 ppm   1.2%       220 ppm   1.2%

3-Way Catalyst                   100 ppm   1.0%       220 ppm   1.2%
For 2500 rpm testing (not included in the analysis), the federal

207(b) outpoints are the same as the outpoints used at idle.  In
                                  -16-

-------
addition, California also uses the federal 207(b) outpoints for 2500

rpm testing.



For each of the six sets of outpoints, Sierra determined which 1980

and later vehicles in the surveillance data set would have failed an

I/M test using those cutpoints.  Sierra then calculated the fraction

of emissions in excess of the FTP standards for the surveillance

fleet  that were captured by each set of cutpoints.  The error of

commission rate, the error of omission rate, the failure rate, and the

fraction of "high emitters" identified were also computed.  (High

emitters were defined as vehicles 5 or more times the FTP standards

for one or more pollutants.)  All of these calculations were done for

the 1980 and later fleet as a whole and for each of five technology

categories.  Those categories were:


      1.  Carbureted vehicles equipped with oxidation catalysts and
          air injection (Carb/OC/Air);

      2.  Carbureted vehicles equipped with open-loop, 3-way
          catalysts (Carb/30L);

      3.  Carbureted vehicles equipped with closed-loop, 3-way
          catalysts (Carb/3CL);

      4.  Carbureted vehicles equipped with closed-loop, 3-way
          catalysts and clean up oxidation catalysts  (Carb/3CL+OC);
          and

      5.  Fuel injected vehicles equipped with closed-loop, 3-way
          catalysts (FI/3CL).
                                 41 II j£
* excluding vehicles for which  loaded mode test results were not
available
                                  -17-

-------
                             3.  RESULTS



The analysis conducted indicates that a substantially higher

percentage of excess emissions can be identified when a steady-state

loaded mode is added to the federal 207(b) idle test.  Figure 4

summarizes how the addition of steady-state loaded mode testing to the

California idle cutpoints increases the excess emissions identified in

the California surveillance fleet for less- and more stringent levels

of loaded mode testing (loaded A and B, respectively).  For the more

stringent loaded mode test cutpoints, excess emissions identified

increase from 45.5% HC, 48.2% CO, and 10.7% NOx to 62.5% HC, 70.7% CO,

and 18.8% NOx.



Figure 5 shows a very similar trend when the loaded mode tests are
                                               r>
added to the federal 207(b) cutpoints.  For the more stringent loaded

mode test cutpoints, excess emissions identified increase from 41.4%

HC, 45.8% CO, and 8.4% NOx to 60.1% HC, 69.3% CO, and 16.5% NOx.



Figure 6 shows the effect of loaded mode addition when both the

California idle and federal 207(b) idle baselines are plotted on the

same chart.  As the figure shows, the incremental benefits of the more

stringent idle cutpoints are maintained when loaded mode testing is

added.
                                  -18-

-------
                   Figure 4
           Effect of Test Type
      and Loaded Mode Stringency
     on Excess Emissions Identified
         (California Idle Outpoints)
100
        California
        Idle Only
California Idle
 + Loaded A

Test Type
California Idle
 + Loaded B
                                             HC


                                             CO


                                             NOx
                    -19-

-------
0)
CO

o
"co
.CO


LJJ

CO
CO
o
o
X
LU
(D
O

0)
Q_
                      Figure 5

               Effect of Test Type

         and Loaded Mode Stringency

         on Excess Emissions Identified

              (207(b) Idle Outpoints)
100


 90


 80


 70


 60


 50


 40


 30


 20


 10


 0
              69.3
   63.8
54.8
            207(b)

            Idle Only
           60.1
      10.9
 207(b) Idle
 + Loaded A


Test Type
                 16.5
                              207(b) Idle

                              + Loaded B
                        -20-

-------
                  Figure 6
.a>


"c



CO

o
"co
.22


LU

co
co

-------
Table 4 contains the data plotted in Figures 4, 5, and 6.  In

addition, Table 4 shows the differences in excess emissions identified

for each of the five technology categories that were evaluated.  As

the table shows, there are large differences in how the different

categories are affected by loaded mode testing.  These differences are

also illustrated in Figures 7 and 8.  The number of vehicles in each
                               Table 4

              Excess Emissions Identified by Short Tests
           GARB Surveillance Data  - 1980 and Later Vehicles
Idle Stringency:  California
  Sample
  Group
All Vehicles

CARB/OC/AIR
CARB/30L
CARB/3CL
CARB/3CL+OC
FI/3CL
      Idle
       CO    NOx
45.5  48.2  10.7
45.4  40.8   8.8
39.9  56.1   4.9
 5.3   0.0  27.8
42.8  41.7   9.8
70.8  77.5  15.4
             Excess Emissions (%)
                Idle + Loaded A
                HC    CO    NOx

               59.0  66.1  13.2
                              Idle + Loaded B
                              HC    CO     NOx

                             62.5  70.7   18.8
59
62
76
51
79
.9
.4
.9
.2
.8
64
74
86
54
86
.2
.5
.4
.5
.6
15
6
27
10
..15
.0
.3
.8
.8
.4
65
77
76
52
79
.9
.0
.9
.4
.9
78
84
86
54
86
.4
.6
.4
.7
.6
21
17
35
16
18
.4
.1
.1
.3
.4
Idle Stringency:  Federal  207(b)
  Sample
  Group
 HC
Idle
 CO
NOx
All Vehicles   41.4  45.8    8.4
CARB/OC/AIR
CARB/30L
CARB/3CL
CARB/3CL+OC
FI/3CL
45.0
34.1
5.3
35.0
69.6
40.0
56.1
0.0
36.4
77.5
8.8
4.9
24.4
4.3
14.4
59.5
56.6
76.9
43.4
78.6
Excess Emissions  (%)
   Idle + Loaded A
   HC    CO    NOx

  54.8  63.8  10.9
                                          63.4
                                          74.5
                                          86.4
                                          49.2
                                          86.6
 Idle + Loaded B
 HC    CO    NOx

60.1  69.3  16.5
15.0
6.3
24.4
5.3
14.4
65.5
77.0
76.9
47.6
78.7
77.6
84.6
86.4
51.8
86.6
21.4
17.1
31.7
10.8
17.4
Legend:  OC = oxidation catalyst,  30L =  open-loop  3-way catalyst,
         3CL = closed-loop  3-way  catalyst,  GARB =  carburetor,
         FI = fuel injection
                                  -22-

-------
of the technology categories  is  given  later in Table 5  (page 26).  The




dramatic increase in excess emission for the CARB/3CL category may be




partially due to the relatively  small  size of that category (29




vehicles).









With California idle cutpoints as  the  baseline, Figure  7 shows that




there are large differences in the effect of adding loaded mode




testing between technology categories.  The incremental effect of




adding loaded mode testing on excess emissions identified for fuel




injected, closed-loop vehicles is  relatively minor (see the right-most




group of bars in each chart).  However, the percentage  of excess




emissions identified is  greatly  increased by the addition of loaded




mode testing for carbureted,  closed-loop vehicles (see  the center




grouping of bars).









Figure 8 shows that the  effect of  adding loaded mode testing to the




federal 207(b) idle test is very similar.  There is no  significant




difference in the percent of  excess emissions identified for fuel




injected, closed-loop vehicles and there is a substantial increase in




excess emissions identified with loaded mode testing for carbureted,




closed-loop vehicles.  None of the excess CO emissions  from




carbureted, closed-loop  vehicles is identified with the idle test but




86.4% is identified with loaded  mode testing.  The effect on the




identification of excess HC emissions  is similar.  Only 5.3% of  the




excess HC is identified  in carbureted, closed-loop vehicles with idle-




only testing, but 76.9%  of the excess  HC is identified  with loaded




mode testing.
                                  -23-

-------
                     Rgure 7

  Effect of Emission Control System Design

      on Excess Emissions Identification
            (California Idle Outpoints)
  100
   80
CO
8  70
(0
m
60


50
o  40
m

•5  30

"1  20
           -7-
                                          [Z]HC


                                              CO


                                              NOx
         Carb/OC/Air      Carb/3CL       F1/3CL
                 Carb/3OL     Carb/3CL+OC


  (California Idle + Steady State Loaded Mode)
   100
co
to
111

CO
CO
CD
o
X
LLJ
1
                                               HC



                                               CO


                                               NOx
        Carb/OC/Air      Carb/3CL       F1/3CL
                Carb/3OL     Carb/3CL+OC

                Emission Control System


      Note: OC = oxidation catalyst
          SOL = 3-way catalyst, open loop
          3CL - 3-way catalyst closed loop
                       -24-

-------
                    Figure 8
Effect of Emissions Control System Design
     on Excess Emissions Identification
            (207(b) Idle Outpoints)
IUU
TJ
£ so
1 80
1 70
CO
•2 60
« 50
CO
8 40
UJ
•5 so
"i 20
o
1 10
0
.
-
-
-

-
L
_
_
-
-

/
/
/
/
/
/
I
t
1,
//



>ol
/
/
/
/
/
^
yi

|
^
X





^3 (71









|
8
V


r — ~T\









/
/
/
/
/
f
/
1
^
^
^



v*l
/
/
/

/
/
/
/
/
/
2
'/
^
'l
/.

t
^
',







s
A
                                                NOx
      Carb/OC/Air     Carb/3CL       FI/3CL
              Carb/3OL    Carb/3CL+OC
  (207(b) Idle + Steady State Loaded Mode)
                                             J HC


                                             1 CO


                                             I NOx
      Carb/OC/Air      Carb/3CL        FI/3CL
             Carb/3OL     Carb/3CL+OC
             Emission Control System

     Note: OC - oxidation catalyst
        SOL = 3-way catalyst open loop
        3CL = 3-way catalyst, closed loop
                    -25-

-------
For other carbureted vehicles, the benefits of loaded mode testing are

significant, but not nearly as dramatic.



Table 5 presents the error of commission rates and sample sizes for

the 1980 and later models in the surveillance fleet.  (Note that

although the excess emissions identification rate for the idle testing

of carbureted, closed-loop vehicles was extremely low, there were 24

of these vehicles in the sample that failed the FTP standards.)  Error
                               Table 5

                   Error of Commission Rates (%)
                Computed From GARB Surveillance Data
                      1980 and Later Vehicles
Idle Test Stringency:  California
                                        Error of Commission Rates (%)
Sample
Group
All Vehicles
CARB/OC/AIR
CARB/30L
CARB/3CL
CARB/3CL+OC
FI/3CL
Sample
Size
659
155
85
29
254
106
Passing
FTP
194
36
31
5
77
38
Idle
Cutpoints
0.0
0.0
0.0
0.0
0.0
0.0
Idle+Loaded
Cutpoint A
0.8
0.0
1.2
0.0
1.2
0.9
Idle+Loaded
Cutpoint B
2.4
1.3
5.9
0.0
2.4
2.8
Idle Test Stringency:  Federal  207(b)
                                        Error of Commission Rates (%)
Sample
Group
All Vehicles
CARB/OC/AIR
CARB/30L
CARB/3CL
CARB/3CL+OC
FI/3CL
Sample
Size
659
155
85
29
254
106
Passing
FTP
194
36
31
5
77
38
Idle
Cutpoints
0.0
0.0
0.0
0.0
0.0
0.0
Idle+Loaded
Cutpoint A
0.8
0.0
1.2
0.0
1.2
0.9
Idle+Loaded
Cutpoint B
2.4
1.3
5.9
0.0
2.4
2.8
                                  -26-

-------
of commission rates were extremely low for all outpoints and




technology categories and there is no difference between the error of




commission rates for the California idle cutpoints plus loaded mode




and the federal idle cutpoints plus loaded mode.  Only the carbureted,




3-way, open-loop vehicles exceeded an error of commission rate of more




than 3%, but even this group was only just over 5% using the most




stringent loaded mode test.









Table 6 presents the failure rate, error of omission rate and "high




emitter identification rate" for the surveillance vehicles based on




different test types.  Errors of omission are vehicles that fail the




FTP standards but pass the I/M test.  "High emitters" are defined as




vehicles with either HC or CO emissions equal to or greater than five




times the California standard.  As the table shows, loaded mode




testing significantly increases the I/M failure rate, reduces the




error of omission rate, and increases the percent of "high emitters"




identified.  However, the sample sizes for the various technology




categories was low.  The number of "High Emitters" identified was:









             All Vehicles  	 42




             CARB/OC/AIR 	 13




             CARB/30L  	  3




             CARB/3CL  	  1




             CARB/3CL+OC 	 18




             FI/3CL  	  7
                                  -27-

-------
                               Table 6

            Failure Rate (F),  Errors of Omission (0) and
                High Emitter Identification Rate (H)
          GARB Surveillance Data - 1980 and Later Vehicles
I/M Stringency:  California

 Sample             Idle
  Group	
                                    Idle + Loaded A
All Vehicles
6.8  63.7  52.4
                                    12.1  59.2  71.4
I/M Stringency:  Federal 207(b)

                     Idle
  Sample
  Group
All Vehicles
5.1  65.4  47.6
                     Idle + Loaded A
                                    10.5  60.8  66.7
                                         Idle + Loaded B
19.3  53.7  73.8
CARB/OC/AIR
CARB/30L
CARB/3CL
CARB/3CL+OC
FI/3CL
9.0
4.7
6.9
6.3
8.5
67.7
58.8
75.9
63.4
57.6
46
66
0
50
71
.2
.7
.0
.0
.4
18.7
14.1
13.8
9.4
10.4
58.1
50.6
69.0
61.4
56.6
69
100
100
61
85
.2
.0
.0
.1
.7
25.2
30.6
17.2
15.4
16.0
52.9
38.8
65.5
56.7
52.8
76.9
100.0
100.0
61.1
85.7
17.9  55.1  71.4
CARB/OC/AIR
CARB/30L
CARB/3CL
CARB/3CL+OC
FI/3CL
8.4
3.5
3.4
3.9
6.6
68.4
60.0
79.3
65.8
55.7
46
66
0
38
71
.2
.7
.0
.9
.4
18.1
12.9
10.3
7.1
8.5
58.7
51.8
72.4
63.8
54.7
69
100
100
50
85
.2
.0
.0
.0
.7
24.5
30.6
13.8
13.4
14.2
53.6
38.8
69.0
58.7
50.9
76.9
100.0
100.0
55.6
85.7
The effect on I/M failure rates is illustrated in Figure 9.  As the

figure shows, the idle-only failure rate increases by about a factor

of two when the less stringent loaded mode test cutpoints are added,

and by a factor of three when the more stringent loaded mode test

cutpoints are used.
                                  -28-

-------
    o
   •«
   DC
    O
   "CO
   LL
         50
         40
30
20
         10
                                   Figure 9

                       Effect of Test Type

                          on Failure Rate
                                         California Idle


                                         207(b) Idle
19.3
                                             17.9
                  6.8
                   Idle Only     Idle + Loaded A  Idle + Loaded B

                                Test Type
Although not addressed in the task description,  it seems that it would


be useful to consider the incremental effects  of loaded mode  testing


as a supplement to idle plus 2500 rpm no load  testing.  Under Task 3


of our Work Assignment,  it was demonstrated that the addition of 2500


rpm testing increases the failure rate of the  I/M test.   Because that


earlier analysis was based on Test Analyzer System data obtained from


I/M stations, information was not available on the percent of  excess
                               -29-

-------
emissions identified by  the  addition  of  2500 rpm  test.  In addition,




under Task 1 of  the Work Assignment,  it  was demonstrated  that the




inclusion of 2500  rpm  failures  had a  slightly positive  impact on the




emission reductions currently being achieved under  the  California I/M




program.









Figure 10 provides a sense of how 2500 rpm testing  would  affect the




incremental benefits of  loaded  mode testing.  Based on  our Task 3




report ("Incremental Effects of 2500  RPM Testing  and Alternative Idle




Outpoints on I/M Results," December 1, 1987), the I/M failure rate for




1980 and later models  increased by 56% when 2500  rpm testing was added




to the federal 207(b)  idle test.   In  contrast,  the  failure rate




increases by 251%  when loaded mode testing is added.  The available




data seem to indicate  that loaded mode testing  is much  more effective




in identifying defective vehicles than the 2500 rpm test.  (It should




be noted that the  difference in idle  failure rates  for  surveillance




data and Test Analyzer System data reflect the  higher average mileage




for the 1980 and later models recently tested in  the I/M program




compared to the  average  mileage for vehicles tested in  the




surveillance program.)








Further analysis of California  I/M evaluation data  and  the latest




available surveillance data  could provide a more  refined estimate of




how much the theoretical benefits of  loaded mode  testing are reduced




by the addition  of a 2500 rpm test mode.
                                  -30-

-------
                  Figure 10
         Effect on Failure Rate
of Adding Loaded and 2500 RPM Tests
         to the 207(b)  Idle Test
  Idle-Only   Idle + Loaded
 	Surveillance Data	
                 Data Source
Idle-Only   Idle + 2500
	TAS Data	
                  -31-

-------