Analysis of the Effectiveness
and Cost-Effectiveness of Remote Sensing Devices
                      prepared for:

            U.S. Environmental Protection Agency
                      May 18, 1994
                    principal authors:

                    Thomas C. Austin
                   Francis J. DiGenova
                   Thomas R. Carlson

                   Sierra Research, Inc.
                      1801 J Street
                Sacramento, California 95814
                     (916)444-6666

-------
Although the information described in this report has been funded
wholly or in part by the United States Environmental Protection
Agency under Contract No. 68-C1-0079, Work Assignment 2-01,
Task 3, to Sierra Research, Inc., it has not been subjected to the
Agency's peer and administrative review and is being released for
information purposes only.  It therefore may not necessarily
reflect the views of the Agency and no official endorsement should
be inferred.

-------
                 Analysis of the Effectiveness
    and Cost-Effectiveness of Remote Sensing Devices
                          Table of Contents

                                                               page

Summary	  1

Introduction	 11

Detecting Excessive Emissions with  RSDs 	 13

Scenario Development	 26

Effectiveness Analysis	 35

Cost and Cost-Effectiveness Analysis  	 43

References 	 SO
Appendix A - Hov Tailpipe Concentrations Are Estimated from
            Concentrations of Diluted Exhaust

Appendix B - Cost Model  Printout from Program Involving RSD
            Measurements at Freeway Ramps
                                -i-

-------
                             Liat of Tables



                                                                    page

1.    Comparison of Alternative I/M Programs 	   1

2.    Emissions Impact Due to Intentional Tampering Known
      to Motorist 	  38

3.    Estimated Effect of Tampering Deterrence 	  39

4.    Comparison of I/M Benefits Achieved Based on the Repair of
      Vehicles Above 3X CO Compared to All 1980 and Later Model
      Vehicles Subjected to Current I/M Program 	  41

5.    Estimated Effect of RSD Screening for Vehicles Subject to
      Enhanced I/M 	  42

6.    Comparison of Alternative I/M Programs 	  47

7.    Cost and Effectiveness of Alternative I/M Programs 	  49
                                   -ii-

-------
                             List of Fizures
                                                                    page

1.   Speed and Emissions Profile for a Typical Driver and
      a Typical Car 	   5

2.   Speed and Emissions Profile for an Aggressive Drive and
      a Typical Car 	   5

3.   Speed and Emissions Profile for a Vehicle with a
      Defective Oxygen Sensor 	   7

4.   Idle CO Distribution for Light-Duty Vehicles 	   8

S.   FTP CO Emissions Distribution for Light-Duty Vehicles
      Sorted by Idle CO 	   9

6.   Composite FTP Emissions Distribution for Light-Duty
      Vehicles Sorted by Idle CO 	  10

7.   FTP NOx Emissions Distribution for Light-Duty Vehicles
      Sorted by Idle CO 	  10

8.   Relationship Between Instantaneous Emissions at a Remote
      Sensor and Average Stop—and-Go Driving Emissions 	  16

9.   Speed and Emissions Profile for a Typical Driver and a
      Typical Car 	  18

10.  Speed and Emissions Profile for an Aggressive Driver and
      a Typical Car 	  18

11.  Speed and Emissions Profile for a Vehicle with a Defective
      Oxygen Sensor 	  19

12.  Average CO Emissionn vs. CO Measured by Remote Sensing	  21

13.  Average HC Emissions vs. CO Measured by Remote Sensing 	 21

14.  Idle CO Distribution for Light-Duty Vehicles 	 23

15.  FTP CO Emissions Dlntribution for Light Duty Vehicles
      Sorted by Idle CO 	 23

16.  Composite FTP Emissions Distribution for Light-Duty Vehicles
      Sorted by Idle CO 	^ .. 24

17.  FTP NOx Emissions Distribution for Light-Duty Vehicles
      Sorted by Idle CO 	 24
                                  -ill-

-------
                      List of Figures continued...
                                                                    ma

18.  Freeway Ramp With Long, Constant Radius Turn and Adequate
      Space for Installation of RSD Enclosures 	 28

19.  Fraction of LA Drives that Include Freeway Driving
      on Maximum Speed	 31

20.  Conceptual Design of Remote Sensing-Based High Volume,
      Centralized Test Station 	 33

21.  Comparison of Underhood Defects Undercover Cars vs.
      Random Roadside Sample 	 37
                                  -iv-

-------
                  Analysis of the Effectiveness

    and Cost-Effectiveness of Remote Sensing Devices



                             Summary

Estimates have been developed for the cost and effectiveness of using
remote sensing devices (RSDs) to facilitate the identification and
repair of vehicles with high emissions.  These estimates have been
compared to cost and effectiveness  projections for conventional vehicle
inspection and maintenance (I/M) programs requiring inspections on a
scheduled basis at fixed locations.  Estimates of decentralized I/M
program effectiveness and all cost  estimates are based on data and
methodologies previously collected  by the California I/M Review
Committee.  (In other states, the costs and effectiveness of
decentralized I/M may be less, but  the costs and effectiveness of the
alternative programs considered  in  the analysis should be similar.)  As
a supplement to a conventional I/M  program, limited use of RSDs in
conjunction with roadside pullovers of high-emission vehicles is
projected to provide additional  emission reductions at a favorable ratio
of cost to effectiveness.  However, widespread use of RSDs to screen
vehicles for more comprehensive  testing or to replace a conventional I/M
program decreases the emission reductions that would otherwise be
achieved.*

As shown in Table 1, California's current decentralized I/M program is
projected to reduce emissions from  vehicles subject to the program by
16.6X for hydrocarbons (HC), 25.31  for carbon monoxide (CO), and 10.4X
for oxides of nitrogen (NOx) in  calendar year 2000."  With the
addition of a remote senoing program, additional deterrence of tampering
is projected to increase the level  of emission reductions to 18.SX for
HC, 27.6X for CO, and 11.OX for  NOx and the overall cost effectiveness
of the program Is projected to improve slightly from $1.75 per pound to
$1.67 per pound***.  The benefit  estimate is based on the expectation
* This  study did not address the combination of widespread use of RSDs
to screen essentially all vehicles for out-of-cycle inspections under a
conventional I/M program.  However,  this  alternative would have
substantially higher coses than the combination of a limited RSD program
for tampering deterrence operating in conjunction with a conventional
I/M program.   While benefits would be marginally higher, the cost-
effectiveness ratio would most certainly  be higher than the alternatives
studied.

~  The difference between this estimate and the 19.6X HC,  IS.31 CO, and
6.7X NOx reductions estimated based on the most recent evaluation of the
California program by the Air Resources Board and the I/M Review
Committee is related to changes In the composition of the  fleet that
will occur by year 2000.

***  Cost effectiveness calculated by dividing annualized costs by the
annualized reductions of HC, NOx, and one-seventh of CO emissions.

-------
Table 1
Comparison of Alternative I/M Programs
Emission Reduction Cost-
(year 2000 forecast) Effectiveness
($/lb. of
HC CO NOx HC+NOx+(CO/7))
Current Decentralized
Program Without RSDs
Decentralized Program
Supplemented With
RSD/Pullover Program
Enhanced I/M
Without RSDs
Enhanced I/M With
RSD/Pullover Program
Enhanced I/M
With RSD Screening
RSDs Only
16. 6X

18. 51


35.81

36.81

12.71

10. OX
25. 3X

27. 6%


34. 6X

35.81

28.71

22.61
10. 4X

11. OX


22. 2X

22. SX

0.3X

0.2X
$1.75/lb

$1.67/lb


$1.09/lb

$1.10/lb

$1.00/lb

$0.97/lb
that about half of all motorists will be deterred from tampering with
their vehicles because they will perceive a substantial risk of being
detected by a remote sensor.  (Complete elimination of intentional
tampering is unlikely because all tampering cannot be detected by RSDs.
The most common forms of deliberate tampering, such as removal of air
injection systems and catalysts",  do not increase emissions  enough to
be detected by RSDs.  More sophisticated motorists who tamper with their
vehicles are expected to learn this.)  Based on the difference in
tampering patterns in vehicles stopped at random by the Highway Patrol
and vehicles volunteered for testing to the California Air Resources
Board (ARB), only about 3X of the vehicles in the fleet appear to have
excessive emissions'" caused by Intentional tampering of which the
currant: vehicle owner is aware.  It should also be noted that roadside
survey data indicate that tampering rates are decreasing over time1"*".
The future benefits of tampering deterrence are likely to be less than
estimated herein.
" The observed frequency of catalyst removal and air pump removal is
dramatically lower in vehicles voluntarily submitted for testing to ARB
compared to vehicles stopped at random by the Highway Patrol.

~ As used  in this report,  "excessive" or "excess  emissions"  are
emissions  above the standards vehicles are certified to meet when  they
are properly" maintained.

*"* Numbers in superscripts denote references listed at end of text.
                                   -2-

-------
The cost of the remote sensing program necessary Co achieve effective
tampering deterrence in the South Coast Air Basin is based on the use of
six teams of RSD van operators, Highway Patrol officers, and inspectors
who move from site to site on a daily basis, with the ability to cover
about 1,000 different sites each year.  The annualized cost is estimated
at $3.3 million.  This cost estimate was developed by using vendor-
supplied estimates for emissions measurement equipment and freeway ramp
modifications, in conjunction with estimates of fully burdened labor
costs for technicians, inspectors, police officers, and their associated
management and support staff, and other miscellaneous costs.   Smaller
metropolitan areas would need fewer teams and would have correspondingly
lower costs.

The same supplemental remote sensing program is projected to increase
the emission reductions from an enhanced I/M program with centralized,
dynamometer testing by about half as much (because an enhanced I/M
program will detect more tampering than the current program).  Without
the use of RSDs, the enhanced program is projected to reduce emissions
by 35.81 for HC, 34.61 for CO, and 22.21 for NOx.  With RSD., the
reductions increase to 36.8X for HC, 35.8X for CO, and 22.5X for NOx.
The additional benefits resulting from the use of RSDs have almost
exactly the same cost per pound as the enhanced program without RSDs,
resulting in a net cost effectiveness of $1.10/pound.

The use of RSDs to screen vehicles for more comprehensive testing was
also evaluated.  This approach is projected to substantially lower the
emission reductions achievable by I/M.  Vehicles with moderately high
exhaust emissions escape detection because of the limitations of RSD
testing.  Vehicles with defects in evaporative emissions control
systems, crankcase control systems, or cold start control systems escape
detection unless they also have very high exhaust emissions.  Most forms
of emissions control system tampering would go undetected on vehicles
that are otherwise well maintained.  As shown in Table 1, the use of
annual prescreening by RSDs, in conjunction with a pullover program,
would reduce the effectiveness of the I/M program to 12.7X for HC, 28.71
for CO,  and 0.3X for NOx.  Based on 100X annual prescreening at a
network of fixed sites, the cost effectiveness of the program
($1.00/pound) would b« only slightly better than that of an enhanced I/M
program and the benefits would be much lower.  The cose estimate for
this scenario was developed by modifying the methodology for estimating
the cost of centralized I/M programs set forth in the California I/M
Review Committee's recent report to the California Legislature1.
Facility size was restricted to three dynamometer test lanes to ensure
that the number of facilities in the South Coast Air Basin (21) would be
at lease as great as the 17 inspection facilities used during the
centralized change-of-ownarship inspection program conducted from 1979
to 1984.  Land size and equipment costs were appropriately modified to
account for a remote sensing lane.

The benefits of a program based on prescreening via remote sensing would
be further reduced to 10.OX for HC, 22.6X for CO, and 0.2X for NOx if
the prescreening were don* at freeway ramps instead of fixed sites.
Under this approach, 29 sets of RSDs and license plate readers would  be
moved, on average once per week, among 1,300 freeway ramps in the South
Coast Basin.  The annualized cost for this element of the program was
                                   -3-

-------
calculated  to be $8.4 million for the South Coast.  This cost estimate
was developed by vising vendor-supplied estimates for emissions
measurement equipment and freeway ramp modifications, in conjunction
with estimates of fully burdened labor costs for technicians and their
associated  management and support staff, and other miscellaneous costs.
The loss in benefits results from the difficulty in getting a valid
reading on  each vehicle when only using freeway ramps.  The cost-
effectiveness ratio of this approach is slightly lower than the other
scenarios considered at $0.97/pound.  However, it should be noted that
there  is a  very optimistic assumption associated with this scenario
regarding the ability to obtain representative drive—by emissions
readings using freeway ramps.  To the extent that accelerations and
decelerations are beyond the limits necessary to ensure representative
emissions,  the number of defective vehicles identified will drop, but
the costs will remain unchanged.  Further field studies are necessary to
determine how much higher than this theoretical minimum the cost
effectiveness ratio is likely to be.

The loss in emissions reductions when RSDs are used as a partial or
complete replacement for I/M occurs because RSDs are not capable of
identifying many commonly occurring emissions-related defects, including
defects in  evaporative emissions controls, crankcase emissions controls,
and cold-start controls.  In addition, currently available RSDs will not
accurately  measure oxides of nitrogen (NOx) and HC emissions.  Although
RSD technology can accurately measure instantaneous CO emission
concentrations, accurate measurements of •drive-by" emissions of CO,
NOx, or HC  are not as effective as dynamometer testing for the detection
of vehicles with emissions-related defects.  The inherent lack of
control over the operating conditions of the vehicles being tested by
RSDs introduces substantial variability in the test results, making it
more difficult to separate defective vehicles from properly maintained
vehicles.   In addition, the inability of RSDs to measure the mass
emissions rate makes the exhaust of all vehicles with the same emissions
concentration appear to be the same even when the mass of emissions may
be substantially different.

The variability introduced by the lack of control over vehicle operating
conditions has been ignored in many previous analyses of the potential
benefits of I/M programs based on the use of RSDs.  Failure to recognize
the inherent variability of vehicle •missions when operating conditions
are not tightly controlled has led to frequent misinterpretation of the
results of  experiments conducted using RSDs to measure emissions from
vehicles in customer service.  Foremost among the incorrect conclusions
drawn  tram  analysis of field data is that 10Z of all vehicles are
responsible for more than 501 of all vehicle emissions.  This conclusion
can only be drawn by assuming that:  1) instantaneous emission
measurements made by RSDs represent the average emissions from each
vehicle measured, and 2) emissions not measured by the RSD  (e.g.,
evaporative hydrocarbons) are perfectly correlated with exhaust  carbon
monoxide (CO) emissions, which are accurately measured by the RSD.  Our
analysis indicates that neither assumption is correct.

Figures 1 and 2 illustrate the fallacy of assuming that instantaneous
remote sensing measurements represent average emissions.  Each  figure  is
a graph of  the second-by-second CO emissions  from a  Chevrolet Lumina

-------
                             Figure  1
                   Speed and Emissions Profile
              for  a Typical  Driver and a Typical Car
    0     100   200    300   400   500    600   700   800
Upper line represents speed.  Filled area represents CO.
Overall CO emissions 2.2 g/m
900
                            Figure 2
                   Speed and Emissions Profile
           for an Aggressive Driver and a Typical Car
    0     100   200    300   400   500    600   700   800
Upper line represents speed.  Filled area represents CO.
Overall CO emissions 39.0 g/m
                                                                10
                                                             -  8
                                                             -  6
                                                             -  4
                                                             -  2
900
                                                                   8
                               -5-

-------
driven from downtown Sacramento  to a residential area in the southwest
portion of the city.  The vehicle was equipped with a portable emissions
measurement system and data were collected to allow concentration
measurements to be converted  to mass emissions*.   The  initial 200
seconds of driving are in the downtown area in mid-afternoon, when the
traffic is relatively light.  The period between about 200 and 500
seconds covers freeway driving.  At about 275 seconds, there is a
freeway-to-freeway interchange.  Beginning at about 500 seconds,  the
route becomes a four-lane surface street on the outskirts of the city
roadway with a speed limit of 40 mph.  For the last 100 seconds,  the
trip is over local streets in a residential area.

As shown in Figure 1, almost all measurements were below IX CO and the
average mass emissions rate for the trip was 2.2 grams per mile (g/mi)
CO.  The average mass emissions rate is very close to the 3.4 g/mi
standard that the vehicle was designed to meet.  Figure 2 shows the
second-by-second CO emissions measured from the same vehicle over the
same road route when driven by a more aggressive driver.  Instantaneous
CO emissions exceeded 3—4X (a level that has been suggested as a
standard for drive-by emissions measured by RSDs) at least thirteen
different times and the average CO emissions for the same trip were
almost twenty times higher at 39 g/mi.  Peak CO concentrations were in
the range of 8-9X, which occurred during hard accelerations.  A thorough
inspection of the vehicle at the end of each trip indicated absolutely
no difference in the condition of the vehicle.  All emission controls
were functioning properly and there were no maintenance-related defects.
The vehicle would pass an idle I/M test with CO emissions of about 0.1X
and an IM240 test (the dynamometer test for I/M programs recommended by
EPA) with CO emissions of about 2 g/mi.  All of the difference in
emissions between the two trips was due to the difference in the way the
vehicle was driven, even though there was no significant difference in
how fast the vehicle was driven.

The second-by-second trace of CO concentrations indicates what would
happen if this vehicle were driven past an RSD.  With the first driver,
the vehicle would appear to be clean.  With the second driver, the
vehicle would appear to be clean if it ware cruising, and to be a gross
emitter if it were being accelerated hard.  In all cases, an inspection
of the vehicle would show no defects.  It can also be seen what would
happen if this saiae vehicle were driven repeatedly past an RSD by both
drivers.  Most of ths> tia* the vehicle would appear to have very low
emissions, but son* of the time  it would appear to have very high
emissions.  Analysis of the data in the manner it is usually reported
would lead to the erroneous conclusion that most of the emissions come
from lass than 10X of the cars.  In fact, it would be th* not car being
seen as a high emitter sometimes and a low emitter at other times.

Careful inspection of the relationship between the emissions traces and
the speed traces shown in Figures 1 and 2 indicates that the .high CO
concentrations caused by the  second driver war* almost always associated
with acceleration.  Theoretically, false failure of the vehicle  by RSDs
* Mass flow through the engine was computed using manifold air pressure
measurements in conjunction with engine  speed.

-------
could be avoided by placing the remote sensors where vehicles would be
unlikely to be accelerating.  However, if eaission measurements are not
made when the vehicle is accelerating, certain emissions-related defects
cannot be detected.  Defects in NOx emissions controls show up well only
during acceleration, which is the mode in which most NOx is generated.
If accurate NOx measurements become available with RSDs, emissions must
be measured during accelerations to determine which vehicles have
excessive NOx emissions.

Another example of a problem that is only detected during acceleration
is illustrated in Figure 3.  This figure shows the speed and emissions
trace for the same driver and route as in Figure 1, but with the vehicle
having a commonly found emissions-related defect:  a disconnected or
defective oxygen sensor.  At 21 g/mi, the CO emissions for this drive
were about ten times higher than the level for the drive illustrated in
Figure 1.  Peak CO concentrations increased into the range of 3-4X CO,
but only during accelerations.  Ac other times, CO emissions were
typically below 21.  In order to catch this defect with a remote sensor,
emissions would have to be measured at locations where the vehicle would
be accelerating.  A cutpoint of 3-41 CO would sometimes catch this
defect and always catch the vehicle with no defects being driven hard
(even though the driver could be within the speed limit and not driving
in a reckless manner).
                                Figure 3
                       Speed and Emission* Profile
              for a Vehicle with a Defective Oxygen Sensor
       0    100    200   300   400   500    600   700
   Upper line represents speed.  Filled area represents CO.
   Overall CO emissions 21.0 g/m
900
                                   -7-

-------
   Previous testing programs indicate that the second-to-second variation
   in emissions illustrated in Figures 1-3 is typical of what occurs with
   most vehicles.  Analysis of remote sensing data that is based on the
   assumption that a snapshot of drive-by emissions accurately represents
   the average emissions of each vehicle therefore always overestimates the
   contribution to total emissions represented by the vehicles with the
   highest drive-by measurements.  In addition, previous research2
   indicates that no single mod* of operation correlates well with
   composite emissions of HC or CO.  The following charts illustrate the
   manner in which the reality of emissions variability confounds the
   analysis of data based on snapshot testing.

   Figure 4 shows the distribution of idle emissions for the current light-
   duty vehicle fleet based on analysis of ARB surveillance data in
   conjunction with laboratory test results obtained during the most recent
   evaluation of the California Smog Check program.  The data from the I/M
   Evaluation Program represent those vehicles that "should fail" a
   properly conducted I/M test.  The surveillance data used in the analysis
   were from vehicles that "should pass" I/M.  By using RSDs to look at
   vehicles idling in traffic,  it would appear that 10X of the vehicles are
   responsible for 76Z of the emissions.  However, Figure 5 shows the
   average CO emissions of the same vehicles in stop and go driving, as
   measured by the Federal Test Procedure (FTP).   When ranked by idle CO,
   the highest emitting 10Z of the population is responsible for only 37X
   5.0
   4.5
   4.0
^3.5
£3.0
8  2.5
I  2.0
   1.5
   1.0
   0.5
   0.0
                                   Figure 4
                 Idle CO Distribution for Light-Duty Vehicles
                               4567
                                10-P«rc«ntn« Bin Number
9
10
                             Should Pass I/M D Should Fafl I/M
                                                                   10th Bin * 76%
                                      -8-

-------
                                 Figure  3
          FTP CO Emissions Distribution for Light-Duty Vehicles
                            Sorted by Idle CO
                              4587
                               10-Percsntiie Bin Number
                            Should Pass I/M  D Should Fail I/M
  9
10th Bin * 37%
of the CO eaissions.  Tho significance of the "dirtiest"  10Z  is  further
reduced when their HC end NOx eaissions are considered.   As shown  in
Figure 6, only 31Z of tho composite eaissions of HC, CO,  and  NOx are
represented by the top lOt based on idle CO.  The composite eaissions of
this decile are further reduced to 29Z when the age distribution of the
vehicles is considered.  (There are more older vehicles in the top 10Z
so the average annual VMT for this group is belov average.)   Figure 7
illustrates one reason why the composite eaissions are not accurately
predicted by the idle CO ranking:  only 1SZ of the NOx eaissions are
represented by the top 10Z ranked by idle CO.

By providing an additional deterrent to tampering, reaote sensing  can
serve as a cost-effective suppleaent to a conventional I/M prograa.
However, it is clear that reaote sensing has significant  limitations
that prevent it froa being effective as a partial or coaplete
replaceaent for conventional I/M.  It is also clear that  many previous
analyses of data froa reaote sensing experiments have aisrepresented the
contribution to total eaissions froa those vehicles with  the  highest
instantaneous drive-by eaissions.
                                   ft*
                                   -9-

-------
                                     Figure 6
          Composite FTP Emissions Distribution  for  Light-Duty Vehicles
                                 Sorted by Idle CO
                                  4587
                                   10-Percantfle Bin Number
Comp FTP » HC + NOx + CO/7
Should Pass I/M  D Should Fail t/M
                                             10
                                      10th Bin 3  31%
                                     Figure 7
             FTP NOx Eaiscion* Distribution for Light-Duty Vehicles
                                Sorted by  Idle CO
                                 4      58       7
                                   10-Percentilc Bin Number
                                Should Pass l^l  D Should Fail
                                       -10-

-------
                            Introduction
Since 1990, substantial public interest has been expressed in the use of
remote sensing devices (RSDs) for the measurement of emissions from
vehicles in customer service.  As explained in an earlier Sierra
report3,  RSDs  can accurately determine  the  ratio  of carbon monoxide
(CO) to carbon dioxide (CO,) in the  exhaust plume of a vehicle that
drives through a beaa of infrared radiation projected across a roadway.
The CO/CO, ratio can be  used to estimate  the concentration of CO
emissions in the exhaust because there is a predictable relationship
between the CO/CO} ratio and the CO  concentration.*

Because RSDs can measure exhaust concentrations of vehicles  travelling
along a roadway, some have suggested that the use of RSDs would be a
superior and less costly alternative to conventional vehicle inspection
and maintenance programs under which inspections are performed
periodically at fixed sites on a scheduled basis.  In concept, the use
of RSDs would avoid the cost of land and buildings required at fixed
inspection sites where the number of vehicles tested per hour would be
much less.  However, as pointed out in previous reports by the
California I/M Review Committee and by EPA, there are significant
limitations to the ability of RSD technology to replace conventional
I/M.  Foremost among these limitations is the inherent lack of control
over the operating conditions of the vehicles as they drive by RSDs.
Variability in operating mod* can cause extreme variability in emissions
which frustrates the separation of defective vehicles from properly
maintained vehicles.

Notwithstanding their limitations, the use of RSDs as a supplement to a
conventional I/M program may provide an additional deterrent to
tampering.  EPA's regulations also require an on-road testing element in
areas required to implement "enhanced" I/M programs.  Because on-road
testing will be required in enhanced I/M areas, and because of the
widespread interest that has been expressed in remote sensing in
general, Sierra Research was directed to estimate the cost and
effectiveness of remote sensing under Task 3 of Work Assignment 2-01 of
Sierra's support contract with EPA (Contract No. 68-C1-0079).
*  The CO/CO] ratio can be accurately predicted if the engine is running
richer than the chemically balanced "stoichiometric" air/fuel ratio.
The accuracy of the method is degraded for vehicles using air injection,
operating with leaner air/fuel ratios, or operating with some cylinders
running richer than others.  However, because the range of air/fuel
ratios occurring in gasoline-fueled vehicles is relatively narrow, the
technique is sufficiently accurate for determining the CO concentration
in the exhaust of most vehicles for the purposes of determining whether
the vehicle is emitting CO concentrations substantially in excess of the
standards used in an I/M program.
                                  -11-

-------
The following sections of this report provide background information on
detecting excess emissions with remote sensing, describe how remote
sensing scenarios were developed, explain how the emission control
benefits of each scenario were estimated, and summarize the analyses
conducted to estimate the cost and cost-effectiveness ratio of each
scenario.
                                  ft*
                                  -12-

-------
          Detecting Excessive Emissions with RSDs
Commercially available RSD systems typically include:

     •  an infrared light source and a device for projecting
        the light source across a single lane (up to 40 foot
        width) roadway;

     •  an infrared detector for measuring the amount of
        infrared light transmitted across the roadway (and
        back, if a mirror is used to reflect the beam);

     •  a video system for capturing the license plate number
        of vehicles that drive past the detector with the
        emissions measurements superimposed;  and

     •  a computer control system to- evaluate and record
        apparently valid emissions measurements and transmit
        them to the video recording system.


At certain wavelengths, the infrared light beam that crosses the roadway
is absorbed by different gases including carbon monoxide, carbon
dioxide, and certain hydrocarbons.  Because the wavelengths absorbed by
each gas is somewhat different from the wavelengths absorbed by other
gases, it is possible, by measuring light intensity at several
wavelengths, to infer the concentration of all three of these gases.  As
explained in the following subsection, the relationship between CO and
COj in the  exhaust  of gaaoline-fueled  internal combustion engines is
such that the tailpipe concentration of CO,  and to  a lesser extent HC,
can be predicted from the CO and HC concentration in the path of the
infrared bean, even though a substantial amount of  dilution may have
occurred.
The Theoretical Concept

The ability to determine the carbon monoxide concentration emitted by an
individual motor vehicle after its exhaust has undergone  an unknown  and
variable amount of dilution depend* on the existence of a consistent
relationship between measurable products of combustion.    If,  for
example, it was known that all engines emitted exhaust containing  1SX
carbon dioxide, then th« concentration of other pollutants in the
exhaust could be determined, regardless of the amount of  dilution,
merely by determining the ratio of the measured concentration of each
pollutant to the measured concentration of carbon dioxide.  Although
it's not quite this simple, and not quite this consistent, this is
essentially the way remote measurement of diluted exhaust emissions  can
be used to determine tho concentration of carbon monoxide coming out of
                                  -13-

-------
che tailpipe of che passing vehicle.  Appendix A explains che
relationship between CO and CO} emissions in more  detail.
           of the Concent
Concentration Measurement - Although RSDs can provide reasonably
accurate estimates of CO concentrations, vehicles are not designed or
certified to meet concentration-based emission standards.  The
contribution of exhaust emissions to air pollution is caused by the
product of emissions concentration and exhaust volume, which yields mass
emissions (g/sec or g/mi) .  RSDs can only determine emissions
concentration.  As a result, vehicles with low fuel consumption will
appear to have relatively high emissions compared to vehicles with
relatively high fuel consumption.   Because the range of fuel economy
for vehicles in the existing fleet exceeds 4:1, a fuel efficient
subcompact car emitting 4Z CO can have exactly the same gram/mile
emission rate as a heavier car with a big engine emitting IX CO.
Because the range of light-duty fuel economy is so large, a
concentration-based standard would either be too strict for smaller,
more fuel efficient vehicles, or too lax for larger, less fuel efficient
vehicles .

Examination of tailpipe concentrations rather than mass emissions is a
problem not only for remote sensing based testing, but also for
stationary idle testing, which has a poor correlation with testing of
mass emissions.  However, idle testing, including tests at fast idle,
afford the ability to test under a repeatable well-controlled condition
that is not subject to road or driver variability.  Stationary loaded-
mode testing improves the correlation with mass emissions testing even
more.
Vehicle. BUJaatons Variability and its Effects — The accuracy with which
RSDs can measure emissions from vehicles as they drive by is not the
only factor affecting the ability of RSDs to detect vehicles with
excessive •missions.  An Implicit assumption built into some analyses of
remote sensing is that the instantaneous emissions concentration
recorded by a remote sensor is a good representation of the average
emissions of the vehicle.  This assumption has caused the researchers
from the University of Denver to conclude, "The basic finding ...... is
that a small minority (8.2Z) of the vehicles is responsible for fifty
percent of the carbon monoxide emissions."4  This conclusion would only
be true If there were no variability in the instantaneous emissions of
the individual vehicles driving past the remote sensor.  To the extent
that SOBS; of the highest emission concentrations observed were
associated with infrequently occurring modes of operation, the percent
of the vehicles that actually account for SOX of the emissions
increases.  This fact appears to have been Ignored or unrecognized  In
some previous studies.  As discussed In Sierra's first analysis of
remote sensing3,  studies that have claimed to find consistency in remote
sensing measurements of the same vehicle at different times are not
supported by the raw data used in the studies.
                                  -14-

-------
To a large extent, the claimed benefits of remote sensing as a
supplement or replacement for conventional I/M are due to the assumed
ability of the system to identify 10Z of the motor vehicle fleet that is
responsible for more than 50Z of the motor vehicle emissions.
Proponents of the system have failed to recognize that 10Z of the
operation of anv vehicle is responsible for a disproportionately large
fraction of the total emissions from that vehicle.  It is not just
"off-cycle" operational modes (e.g., very high acceleration rates) that
cause this effect.

The problems with variability can be illustrated with a hypothetical
example.  If each vehicle on the road emitted 1 g/mi 90Z of the time and
9 g/mi 10Z of the time, and if the emissions mode at any instant was
random, then, based on instantaneous measurements of a large sample of
cars driving past a particular point it would appear that 10Z of the
vehicles were emitting 5QZ of the emissions.  In fact, all of the
vehicles would be contributing equally to the total emissions occurring
In tha area.

In reality, of course, there are significant differences between the
average emissions of vehicles in customer service.  The point of the
hypothetical example described above is that the distribution of
emissions in the vehicle fleet cannot be accurately constructed from an
analysis of data collected by remote sensors.  Virtually all cars, even
when perfectly maintained, are capable of generating carbon monoxide
emissions that vary over a wide range.

Automobile manufacturers design vehicles to meet emission standards when
they are tested using the Federal Test Procedure.   That procedure
involves the use of a specific driving pattern, a specific temperature
range, and a specific fraction of operation under "cold start"
conditions."   The average carbon monoxide emissions of properly
maintained, late model vehicles under these conditions is less than
3 g/mi.  However, almost all vehicles will emit substantially in excess
of 100 g/mi when operated under other conditions.  Carbon monoxide
* Correlation with the FTP la critical for any test procedure that might
be used to trigger vehicle maintenance requirements.  The FTP is known
to be a "representative" driving cycle in terms of average speed, stops
per mile, major speed deviations per mile, and minor speed deviation
pattern.  In contrast, it is obvious that certain instantaneous and
infrequently occurring operating conditions wouldn't necessarily be
representative of average emissions in customer service.  More
significantly, vehicle emission control systems are designed to control
emissions during the FTP.  Tha fact that a vehicle might have high
emissions during some condition not represented on the FTP driving cycle
is important, but not as it relates to I/M.   It does no good to fall a
vehicle because of high 'off-cycle" emissions when there is  no
corrective action possiblo for that vehicle.  Identifying vehicles with
high FTP emissions is the surest way to capture vehicles that are likely
to have correctable defects.
                                  -15-

-------
 emissions are substantially higher than average under cold start
 conditions and under relatively high acceleration rates.

 During  a testing program that Sierra conducted at Southwest Research
 Institute, it was common to see vehicles emit less than 10 g/mi of CO
 under some operating conditions (including the official FTP)  and several,
 hundred g/mi under certain acceleration modes.2   Figure 8  illustrates
 the variability problem with respect to the difficulty of using an
 instantaneous measurement to properly identify vehicles with high
 average emissions while not improperly identifying vehicles with low
 average emissions.  The data presented in the figure are actual test
 results obtained during Sierra's recent emissions testing program
 conducted at Southwest Research Institute.

 In Figure 8, the solid black bars show the g/mi CO emissions  for
 vehicles operating on the "IA4", the 7.5 mile stop-and-go driving cycle
 used in the FTP.  The grey bars show the instantaneous CO emissions
 concentration for the same vehicle during a specific operating mode.  In
 "Case 1", a 1984 Oldsmobile, tuned to manufacturers specifications and
 with no emissions-related defects, was measured at just 3.58 g/mi over
 the 7.5 mile trip.  While accelerating at 25 mph, the instantaneous CO
 emissions were 10.35 g/mi for the same car.  Case 2 shows that the same
 vehicle had extremely low CO emissions when cruising at 25 mph.
60
|40
03
CD 30
20
                                Figure  8
                      Relationship  Between
        Instantaneous Emissions at a Remote Sensor
        and Average Stop-and-Go Driving Emissions
                      ••  Average     • Instantaneous
                          Emissions
                                           Emissions
                                       54.86
                                                          54.85
             10.35
      stop&go 25 mph
              accel
        '84 Olds.
        no defects
         Casel
                         stop&QO 25 mph
                                cruise
                            '84 Olds.
                           no defects
                            Case 2
stop&QO 25 mph
        accel
 79 Mustang.
with tampering
   Case 3
stop&go
25 mph
 cruise
 79 Mustang.
with tampering
   Case 4
                               12 O
                                  o
                               10 £
                                   o>
                                   o
                                                                     8
                                                                         
-------
 In  Case  3,  a  1979  Ford Mustang had very high emissions of 54.85 g/mi
 during the  7.5 mile  trip because of a disconnected air pump.  However,
 under the identical  acceleration condition that caused the Oldsmobile to
 emit over 10X CO,  the Mustang emitted only 1.12X CO.  As shown in
 Case 4,  the same Mustang emitted just 0.74X CO during the 25 oph cruise
 mode.

 The implications of  the data shown in the figure should be obvious:  the
 instantaneous emission measurements can't distinguish unambiguously the
 car that is dirty  on the average from the car that is clean on the
 average.

 Figures  9 and 10 provide additional information regarding the
 relationship between instantaneous and average emissions.  Each figure
 is  a graph  of the  second-by-second CO emissions from a Chevrolet Lumina
 driven from downtown Sacramento to a residential area in the southwest
 portion of  the city.  The vehicle was equipped with a portable emissions
 measurement system and manifold air pressure and engine speed data were
 also collected so  that concentration measurements to be converted to
 mass emissions.  The initial 200 seconds of driving are in the downtown
 area in mid-afternoon, when the traffic is relatively light.  The period
 between about 200  and 500 seconds covers freeway driving.  At about 275
 seconds, there is  a  freeway-to-freeway interchange.  Beginning at about
 500 second*, the route becomes a four-lane surface street on the
 outskirts of the city roadway with a speed limit of 40 mph.  For the
 last 100 second*,  the trip is over local streets in a residential area.

 As  shown in Figure 9, almost all measurements were below IX CO and the
 average mass emissions rate for the trip was 2.2 grams per mile (g/mi)
 CO.  The average mass emissions rate is very close to the 3.4 g/mi
 standard that the  vehicle was designed to meet.  Figure 10 shows the
 second-by-second CO  emissions measured from the same vehicle over the
 same road route when driven by a more aggressive driver.  Instantaneous
 CO emissions exceeded 3-4X (a level that has been suggested as a
 standard for drive-by emissions measured by RSDs) at least thirteen
 different time* and  the average CO emission* for the same trip were
 almost twenty time*  higher at 39 g/mi.  Peak CO concentration* were in
 the range of 8-9X, which occurred during hard acceleration*.  A thorough
 inspection  of the  vehicle at the end of each trip indicated absolutely
no difference in the condition of the vehicle.  All emission controls
were functioning properly and there were no maintenance—related defects.
The vehicle would  pa** an idle I/M test with CO emissions of about 0.1X
 and an IM240 teat  with CO emission* of about 2 g/mi.  All of the
 difference  in emission* between the two trip* wa* due to the difference
 in  the way  the vehicle wa* driven, even though there wa* no significant
 difference  in how  fa*t the vehicle wa* driven.

 The second-by-second trace of CO concentration* indicate* what would
happen if this vehicle were driven past an RSD.  With the first driver,
 the vehicle would  appear to be clean.  With the second driver, the
vehicle would appear to be clean if it were cruising, and to be a gross
 emitter if  it were being accelerated hard.  In all ca*es, an inspection
 of  the vehicle would show no defect*.  It can also be seen what would
 happen if this same  vehicle were driven repeatedly past an RSD by both
                                  -17-

-------
                             Figure 9
                   Speed and Emissions Profile
              for a Typical Driver and a  Typical  Car
    0     100   200    300   400   500    600   700   800
Upper line represents speed.  Filled area represents CO.
Overall CO emissions 2.2 g/m
                            Figure 10
                   Speed and Emissions Profile
           for an Aggressive Driver and a Typical Car
    0     100   200    300   400   500    600   700   800
Upper line represents speed.  Filled area represents CO.
Overall CO emissions 39.0 g/m
                                                                10
                                                             -  8
                                                             -  6
                                                             -  4
                                                             -  2
900
                              -18-

-------
drivers.  Most of the time the vehicle would appear to have very low
emissions, but some of the time it would appear to have very high
emissions.  Analysis of the data in the manner it is usually reported
would lead to the erroneous conclusion that most of the emissions come
from less than 10Z of the cars.  In fact, it would be the same car being
seen as a high emitter sometimes and a low emitter at other times.

Careful inspection of the relationship between the emissions traces and
the speed traces shown in Figures 9 and 10 indicate that the high CO
concentrations caused by the second driver were almost always associated
with acceleration.  Theoretically, false failure of the vehicle by RSDs
could be avoided by placing the remote sensors where vehicles would be
unlikely to be accelerating.  However, if emission measurements are not
made when the vehicle is accelerating, certain emissions-related defects
cannot be detected.  Defects in NOx emissions controls show up well only
during acceleration, which is the mode in which most NOx is generated.
If accurate NOx measurements become available with RSDs, emissions oust
be measured during accelerations to determine which vehicles have
excessive NOx emissions.

Another example of a problem that is only detected during acceleration
is illustrated in Figure 11.  This figure shows the speed and emissions
trace for the same driver and route as in Figure 9, but with the vehicle
having a commonly found emissions-related defect, a disconnected or
defective oxygen sensor.  At 21 g/mi, the CO emissions for this drive
              for a
            Figure 11
   Speed and Emissions Profile
Vehicle with a Defective Oxygen Sensor
                                                                  10
      0    100    200   300    400   500   600   700   «nn
  Upper line represents speed.  Filled area represents CO.
  Overall CO emissions 21.0 g/m
                                                               - 8
                                                               -  6
                                                               -  4
                                                               - 2
                                          900
                                                                    8
                                   -19-

-------
were about ten times higher  than the level for the drive illustrated in
Figure 9.  Peak CO concentrations increased into the range of 3-4X CO,
but only during accelerations.  At other times, CO emissions were
typically below 2X.  In order to catch this defect with a remote sensor,
emissions would have to be measured at locations where the vehicle would
be accelerating.  A outpoint of 3-4X CO would sometimea catch this
defect and alwava catch the vehicle with no defects being driven hard
(even though the driver could be within the speed limit and not driving
in a reckless manner).

Previous testing programs indicate that the second-to-second variation
in emissions illustrated in Figures 9-11 is typical of what occurs with
most vehicles.  Analysis of remote sensing data that is based on the
assumption that a snapshot of drive-by emissions accurately represents
the average emissions of each vehicle therefore always overestimates the
contribution to total emissions represented by the vehicles with the
highest drive-by measurements.  In addition, previous research5
indicates that no single mode of operation correlates well with
composite emissions of HC or CO.  Transient operation that causes high
instantaneous emissions may be included in an I/M test, but only if the
test procedure determines averafa emissions during representative
sequence of idles, accelerations, cruises, and decelerations (as in the
IM240 procedure recommended by EPA).  The instantaneous emission
measurements obtained with RSDs is one of the biggest drawbacks of the
technology.

Figure 12 illustrates the effect of the poor correlation between
instantaneous emission measurements and average emissions during a
recent testing prograa conducted by ABB.'  The data presented in the
figure are from an idealized experiment in which vehicles were driven
past a remote sensor at a constant speed.  As the figure shows, 31 of
the vehicles had CO emissions above 3X when driven past the USD.  When
given a laboratory test to determine average emissions in stop and go
driving as measured by the FTP, 30 of the 31 vehicles had emissions that
were at least twice as high as the standards the vehicle was certified
to meet.  One of the vehicles met the emission standards and would be
considered a "false" failure.  Below 3X CO, the number and percentage of
false failures begins to rise.

Based on the results presented In Figure 12, an RSD CO outpoint of 31
would appear to be capable of identifying vehicles with excessive CO
emissions with a relatively low percentage of false failures.  However,
only 32.51 of the excess emissions from the 439 vehicle fleet would be
identified.  59.41 of the excess emissions come from vehicles with
drive-by CO concentrations too low to allow them to be reliably
separated from vehicles that are free from defects.

Figure 13 shows the FTP HC emissions of the same 439 vehicles compared
to the drive-by CO concentrations.  Only 23.91 of the  excess HC
emissions come from the vehicles above 31 CO.

Figures 14-17 provide more detailed information on  the relationship
between instantaneous drive-by emissions and  average emissions  for  a
larger sample of several thousand vehicles.   To minimize  the variability
in drive-by emissions, idle  emission rates are used.
                                   -20-

-------
                              Figure 12
                     Average CO Emissions  (FTP)
                 vs. CO Measured By Remote Sensing
                                439 VitiidM
                              Flgur* 13
                     Avorag* HC Enissions  (FTP)
                 vs. CO Measured By Remote Sensing
80 T	
                                439 ViMdM
                                -21-

-------
 Figure  14 shows  the distribution of idle emissions for the current
 light-duty vehicle fleet based on analysis of ARB surveillance data in
 conjunction with laboratory test results obtained during the most recent
 evaluation of the California Smog Check program.  The data from the I/M
 Evaluation Program represent those vehicles that "should fail" a
 properly conducted I/M test.  The surveillance data used in the analysis
 were  from vehicles that "should pass" I/M.  By using RSDs to look at
 vehicles idling  in traffic, it would appear that 10Z of the vehicles are
 responsible for  76Z of th« emissions.  However, Figure 15 shows the
 average CO emissions of the sane vehicles in stop and go driving,  as
 measured by the  Federal Test Procedure (FTP).  When ranked by idle CO,
 the highest emitting 10Z of the population is responsible for only 37Z
 of the  CO emissions.

 The significance of the "dirtiest" 10X is further reduced when their HC
 and NOx emissions are considered.  As shown in Figure 16, only 31Z of
 the composite emissions of HC, CO, and NOx are represented by the top
 10Z based on idle CO.  The composite emissions of this decile are
 further reduced  to 29Z when the age distribution of the vehicles is
 considered.  (There are more older vehicles in the top 10Z so the
 average annual VMT for this group is below average.)  Figure 17
 illustrates one  reason why the composite emissions are not accurately
 predicted by the idle CO ranking: only 15Z of the NOx emissions are
 represented by the top 10Z ranked by idle CO.

 Avoidance of Detection Bv Motorists — Based on Sierra's review of the
 literature, relatively little analysis has been done on the network of
 RSDs  that would  be necessary to ensure that all, or almost all, of the
 on-road vehicles would be periodically measured.  While relatively few
 RSDs would be theoretically capable of measuring emissions from all of
 the vehicles in  a metropolitan area, use of the roadway network by
 motorists is not random.  No portions of the roadway network are used by
 all motorists.   Depending on the implementation strategy, it could be
 relatively easy  for motorists to avoid driving past the RSDs.

 In cases where avoidance of the RSD is not feasible, avoidance of
 detection may still be relatively easy.  Just as it has become second
 nature  for motorists to slow down when they see a police vehicle at the
 side of the roadway, motorist can be expected to quickly learn that
 coasting past an RSD with the engine off will avoid detection.
 Strategically placed trailer hitches will also eliminate the ability of
 automated, pattern recognition instruments to read license plate
 numbers.

 Under RSD implementation strategies where avoidance of measurement by
 RSDs  is not feasible, it would still be possible for more sophisticated
 motorists to escape from the detection of deliberate tampering.  Many
 forms of deliberate tampering, such as removal of air Injection systems
 and evaporative  control systems, do not increase the emissions measured
by RSDs.  (On late model automobiles, air injection systems are only
used during cold start and warm-up and most evaporative emissions occur
when the vehicle is parked.)  For other forms of tampering, such as
 catalyst removal, emissions may increase by a factor of five or more,
but RSDs are not sufficiently sensitive to detect the increase.   (Most
 late model vehicles equipped with feedback control systems have CO
                                  -22-

-------
   5.0
   4.5
   4.0
   3.5
*  3.0
82.5
I  2-0
   1.5
   1.0
   0.5
   0.0
                                     Figure 14
                 Idle CO Distribution for Light-Duty Vehicles
                                 4567
                                  10-Percantil« Bin Number
                               Should Pass I/M D Should Fan I/M
        10
 10th Bin » 76%
                                    Figure  IS
            FTP  CO Emissions  Distribution for Light-Duty Vehicles
                                Sorted by Idle CO
                                  45       6       7
                                   10-P«rc«ntH« Bin Number
                                Should Pass I/M  D Should FaU I/M
  9
10th Bin  * 37%
                                        -23-

-------
                                   Figure 16
          Composite FTP Emissions Distribution for Light-Duty Vehicles
                               Sorted by  Idle CO
                                4567
                                 10-PtrwntiIe Bin Number
CompFTP * HC •»• NOx + CO/7
                       Should Pass I/M D Should Fail I/M
                                                           10th Bin » 31%
    2.5
    2.0
    1.5
 2 1.0
    0.5


    0.0
                                   Figure 17
            FTP NOx Emissions Distribution for Light-Duty Vehicles
                               Sorted by  Idle CO
i     i     i     i     i     i     i     i     i     in
                               4567
                                10-Percantile Bin Number
                              Should Pass l/M D Should Fail I/M
                                                           9      10


                                                           10th Bin =  15%
                                       -24-

-------
emissions below IX with the catalyst removed.)  This situation will
reinforce the popular belief among "hot rodders* that properly tuned
engines do not need emissions controls to be clean.  Tampering by
performance enthusiasts may not be meaningfully deterred.

The following section of this report outlines the remote sensing
scenarios that were developed to address the limitations of RSDs
summarized above.
                                  -25-

-------
                      Scenario  Development
Based on Che capabilities and limitations  of RSDs  discussed in the
preceding section, it is apparent that remote sensing by itself is not
capable of detecting the majority of excess  emissions.  It is also
apparent that the use of remote sensing under conditions where the
operating characteristics of the vehicle are not controlled would lead
to a substantial number of false failures.   In consideration of these
factors, and in recognition of the fact that unscheduled inspections
would be necessary to deter deliberate tampering,  four alternative
program concepts were examined:

  1. Basic I/M Plus RSD Supplement - The existing, biennial I/M program
     in combination with daily use of quasi-randomly located RSDs with
     roadside pullovers of high emission vehicles  by police;
  2. Enhanced I/M Plug RSD Supplement — An enhanced,  centralized,
     biennial I/M program with chassis dynamometer testing  in
     combination with daily us* of quasi-randomly located RSDs with
     roadside pullovers of high emission vehicles by  police;

  3. Controlled RSD Screening for Enhanced I/M - Installation of remote
     sensors at a network of fixed inspection sites to  provide an
     annual screening test for a more thorough inspection under  an
     enhanced, centralized I/M program, plus quasi-random remote
     sensing with roadside pullovers of high emission vehicles by
     police; and

  4. RSD Only - Intermittent use of remote sensors at a large number  of
     freeway ramps in combination with computerized reading of license
     plate numbers and notification via mail of inspection  requirements
     for high emission vehicles, plus quasi-random remote sensing with
     roadside pullovers of high emission vehicles by  police.
RSD With Roadside Pullovers

Each of these four approaches has one common feature:   daily use of
quasi-randomly located RSDs with roadside pullovers of high emission
vehicles by police.   This feature is necessary to achieve any meaningful
deterrent to tampering.  Without the immediate inspection of a vehicle
determined to have high emissions while driving by a remote sensor,
there will be no clear indication of whether the vehicle has been
tampered with.  Unlike with "photo-radar", where speed limit violations
can be clearly documented without actually stopping the vehicle, there
is no way to determine when high drive-by emissions are caused by
tampering as opposed to a cold engine, lack of maintenance, component
failure, or simply driving habits.  Since notification of the motorist
by mail would provide the opportunity for the temporary repair of
tampered vehicles, immediate inspection of the vehicle is absolutely
                                  -26-

-------
necessary.  Inspection of the vehicle at the roadside eliminates the
ability of motorists to avoid fines by fixing the problem before the
vehicle is inspected.  Only when motorists realize that there will be no
opportunity to make repairs before submitting the vehicle for inspection
can a significant deterrent effect be expected.

While the need for immediate visual inspection of high emission vehicles
is critical to the identification and deterrence of tampering, a survey
conducted by Sierra indicates that it is completely infeasible to use
roadside pullovers under a program designed to test all vehicles on an
unscheduled basis.  Along the most heavily travelled links in the
roadway system, there is no way to flag down and inspect vehicles with
high drive-by emissions without enormous disruption to the flow of
traffic.  Without covering heavily travelled roadways during peak travel
periods, it would not be possible to subject the entire fleet to
inspection.  In order to avoid significant traffic disruption, the RSD-
roadside pullover concept oust be used on a limited basis.  The concept
evaluated by Sierra involves six teams of RSD-equipped vans, police
officers, and inspectors to cover the South Coast Air Basin.  With this
number of teams, it would be possible to perform inspections at over
1,000 different locations each year.  Based on Sierra's survey of
freeway off-ramps (described in more detail below), this would be
sufficient to cover about 70X of the off-raap volume, which is expected
to cover approximately 902 of the vehicle population.  With this level
of coverage, the majority of motorists would be expected to drive
through a "smog trap" at lease once every year.  Although a relatively
small fraction of the vehicle population could actually be detained for
a visual tampering inspection",  this would be expected to provide a
reasonable deterrent to deliberate tampering because the inspections
would be viewed by a large fraction of the motoring public.
Photo-RSD

In order to obtain wide coverage with remote sensing without disrupting
traffic, two different approaches were identified by Sierra.  One
approach is analogous to photo-radar In that vehicles observed to have
high drive-by emissions would be notified by mail of the need for a
comprehensive inspection, including dynamometer teatins; and underhood
inspection, at a fixed site.  This approach Is assumed to be used in the
"USD-Only* scenario.  Exceptions fro* the comprehensive inspection would
be allowed if the vehicle had been recently tested, or was in a grace
period for repair.  For the reasons previously discussed, some of the
apparent high emitters would bo expected to pass the more comprehensive
test and many defective vehicles would escapo detection.
" Allowing 2 hours por day for sot up and tear down and 90X up time,
six sets of RSD systems would provide about 8,000 hours per year of
drive-by testing.  With two visual inspection tesms por RSD van
averaging 4 visual inspections por hour, 64,000 vehicles por year could
be inspected.  This is only IX of inspections performed under a biennial
inspection program.
                                  -27-

-------
Under Che RSD-Only scenario, pairs of RSDs would be located at selected
freeway ramps.  Two units are assumed per site  in order to help reduce
errors introduced by using only one instantaneous measurement.  Freeway
ramps were chosen because, based on an analysis described below, about
85-90X of all of the vehicles that are subject  to I/M are regularly
driven on freeways to some extent.  Other possible sites for RSDs such
as gas stations or single-lane roads were considered, but were rejected
because they did not appear to offer high potential for "seeing" as many
vehicles without an inordinate number of RSDs or without imposing undue
traffic constraints and/or safety hazards.

The ideal type of ramp for remote sensing is probably one, such as the
ramp illustrated in Figure 18, that contains a long constant radius turn
and adequate space for the installation of RSD enclosures (described in
more detail below).  This ramp geometry increases the probability that
vehicles driving past an RSD will be in neither a closed throttle
deceleration nor high power acceleration.  Instantaneous measurement of
vehicle acceleration as it passes the RSD would still be necessary to
screen out the least representative driving modes.  As discussed
earlier, high acceleration rates are known to cause unrepresentatively
high emissions.  Closed-throttle deceleration is also a concern.  During
closed-throttle deceleration, many late modal vehicles are designed to
shut off fuel flow.  In this case there are no emissions to measure.
Vehicles without deceleration fuel shut off may exhibit relatively high
exhaust concentrations because of poor combustion, but they have very
small exhaust volume, so the high concentrations are not a concern.
Ramps which are configured so as to primarily result in closed-throttle
deceleration or near wide open throttle accelerations are unlikely to be
the most suitable for RSDs.
                                Figure 18
              Freeway Ramp With Long, Constant Radius Turn
          and Adequate  Space for Installation of RSD Enclosures
                                   -28-

-------
Further study  is needed  to evaluate  the relative merits of using
specific off-ramps and on-ramps.  Except for ramps with long, constant
radius turns,  on-ramps are likely to prove more acceptable than off-
ramps for emissions monitoring by RSDs.  The number of on-ramps and off-
ramps in Los Angeles  is  about equal, so the type of ramp to be used
would not affect the  current analysis.  In the analysis described below,
the term "ramp pairs* is used.  In most cases, the on-ramp would be more
suitable for monitoring  than the off-ramp in order to maximize the
amount of drive-by operation likely to be within an acceptable
acceleration range (i.e., not a closed-throttle deceleration or a near
wide open throttle acceleration).

There are approximately  2,300 freeway ramp pairs in the four counties
comprising the Los Angeles Air Basin';  however,  not all  of  these may
provide a suitable site  for RSDs on either the on-ramp or the off-ramp.
In order to estimate  the fraction of suitable ramps, a survey and
analysis was conducted of freeway ramps in Sacramento County, a smaller,
less freeway-dominated metropolitan area.  In the survey, each ramp in
the county was visited and evaluated to determine its suitability as an
RSD site.  It  was assumed that two remote sensing devices would be
placed in protective  enclosures, not less than 20 feet apart, at the
most suitable  ramp (on or off) at each ramp site.  Ramps with multiple
lanes, freeway-to-freeway transitions and with low traffic volumes'
(less than 100 vehicles per day) were classified as unsuitable for RSDs.
For Sacramento County, about 120 ramp pairs out of 225 (S3Z) were found
to be potentially suitable for RSDs.  Based on the fraction of suitable
freeway ramps  in Sacramento and the total number of ramp pairs in L.A.,
it is estimated that  as many as 1,300 freeway ramps may be suitable for
siting RSDs in the L.A. basin.  (At many locations, both the off-ramp
and the on-ramp have  the physical space available.  The 1,300 number is
based on the assumption  that only one of the ramps will be suitable in
terms of typical vehicle operating condition.  As noted above, in most
cases, this is expected  to be the on-ramp.)

Operating RSDs at all suitable ramps continuously would be prohibitively
expensive and  result  in numerous cases of the same vehicle being sampled
repeatedly at  the same site, e.g., a typical freeway commuter might be
"seen" by the  same remote sensor every work day.  It is more efficient
to use a limited deployment of portable RSDs that can be moved, in
pairs, between the 1,300 suitable freeway ramps.  It was therefore
assumed that all suitable freeway ramps would be equipped with RSDs for
an average of  one week per year.

RSDs were assumed to  operate unattended 24 hours per day, 6 days per
week, 50 weeks per year.  It was assumed that one day per week would be
spent on equipment relocating, field calibrations and field servicing.
and that two weeks per year would be spent doing routine servicing  and
other laboratory servicing of equipment.  Assuming, in addition, that
10Z of the RSDs would be needed as spares, 29 RSD pairs  (58 units)'would
be needed for  the Los Angeles Air Basin.

One potential  problem with any roadside monitoring system is driver
avoidance; for example,  if the sampling system  is highly visible at an
off-ramp, drivers of  tampered vehicles might simply bypass that ramp and
exit at a later ramp.  Other avoidance tactics  might include turning off
                                  -29-

-------
 the engine and coasting past the sensor or simply obscuring the license
 plate  to avoid identification.  It does not appear that such avoidance
 activities could be mitigated entirely.  However, to help minimize
 avoidance under the proposed monitoring program, all suitable ramps
 would  be permanently equipped with protective enclosures for RSDs so
 that drivers would not be able to determine if the enclosure contained
 an operating RSD.

 Another problem that exists under any remote sensing program is that not
 all vehicles will be seen by RSDs.  Under this conceptual design,
 drivers who do not regularly use freeway off-ramps, whether due to
 normal driving habits or as an avoidance tactic, would not be seen by
 RSDs.  This group could include drivers who prefer using surface streets
 to avoid freeway congestion, tamperers who wish to avoid detection, and
 drivers making only short trips on surface streets.  In addition,  some
 vehicles, such as older vehicles, may be driven preferentially on
 surface streets to avoid the higher speeds on freeways.

 The fraction of registered vehicles that would be seen annually under
 the RSD deployment scheme described is unknown.  As a rough estimate of
 this number, the freeway ramp volumes for RSD-suitable ramps in
 Sacramento County were compared with the ramp volumes for all ramps in
 the county.  The RSD-suitable ramps were found to represent 70S of the
 total  ramp traffic volume.  Absent a more reliable estimate, this same
 percentage was assumed to apply in other metropolitan areas.  Because
 most vehicles use more than one set of off-ramps, less than 30Z of the
 vehicles are missed by using ramp* that carry 70X of the traffic.

 An alternative estimate of the fraction of vehicles seen has been
 developed based on driving data collected in Los Angeles and in
 Baltimore.  Based on these data, it is estimated that about 85Z of the
 vehicles in a major metropolitan area drive on the freeway at least once
 a week.  This alternative estimate of the fraction of vehicles that
 might  be seen annually by a RSD network was developed, as follows, based
 on information from several field studies.

 To characterize driving patterns of in-use vehicles',  102 drives were
 conducted in Los Angeles In 1992, with 32 of these drives Including some
 travel on the freeway.  Data for all drives were sorted by maximum speed
 using  bin sizes of 5 mph.  For each speed Increment, Figure 19 shows the
 fraction of trips with that maximum speed that were on the freeway.
 Based  on this distribution, a maximum speed of about 55 mph provides a
 relatively well-defined delineation between those drives that included
 freeway travel (those with maximum speeds of 55 mph or less) and those
 that did not.

 A related study10 conducted in several other cities used  instrumented
vehicles to document the driving patterns for a collection of vehicles
 selected to be representative.  Data were collected in Baltimore and
 Spokane.  For Baltimore, data were recorded for one week for each  of
 96 vehicles.  Based on a statistical analysis of the data, 15 of the 96
vehicles (15.6Z) had a maximum speed of 55.5 mph or less.
                                   -30-

-------
                                Figure 19

               Fraction of LA Drives that Include Freeway Driving vs. Maximum Speed
               10  15  20 25 30 35  40  45  50  55  60  65 70 73 80  85
                              Maximum Speed, mph
If the results  in Los Angeles  snd Baltimore  are indicative of driving in
other cities, it suggests  a preliminary estimate that about 1SZ of the
vehicles in a major metropolitan area do not use the freeway in a one-
week period.

It is anticipated that  if  vehicles were instrumented for a full
52 weeks, the fraction  of  vehicles observed  to not use the freeway any
time in a year  may be substantially  lower than the percentage estimated
to not use the  freeway  in  a week.  However,  the hypothetical BSD network
described herein provides  for  sampling for only one week at each
suitable off-ramp.  Based  on this coincidence of sampling periods, it is
believed that the percentage of vehicles not seen by any RSD in a year
(assuming one week of sampling per year by RSD at each freeway off-ramp)
may be approximated by  the fraction  of vehicles that did not use the
freeway In a week.

Many other factors could affect the  estimate of the fraction of vehicles
that would not  be seen  by  remote sensing in  a year.  These include:
vehicles that coast through remote sensing beams or take other steps to
minimize detection, dirty  or otherwise hidden license plates (which
allow RSD readings but  not vehicle identification), vehicles that
accelerate hard between sensors and  thus preclude reliable repeat
measurements, etc.  In  addition, many motorists are expected to use the
same path most  of the time they travel on a  freeway (e.g., such as when
commuting to and from work).   If each freeway ramp is monitored only one
                                   -31-

-------
week each year, ic is likely chaC about 4Z of the motorists who normally
use the ranp may be on vacation during the period that the ramp is
monitored.  Motorists who use freeways only occasionally have a small
probability of using a path that is being scanned by RSD since only 2X
of the available paths are being scanned on a given day.   Most such
factors tend to reduce the fraction of vehicles seen by remote sensing.
Thus, the estimated fraction of 85Z is believed to be closer to an upper
limit of the number of vehicles expected to be seen annually by a remote
sensing network in Los Angeles of the type described.

Vehicles identified by remote sensing as high emitters for a second or
multiple time within a specified time period would not be subject to
further inspection on the premise that the public relations cost of
sending a clean car for several off-cycle inspections due to RSD errors
outweighs the advantage of catching the few persistent tamperers who do
not stop tampering after their first RSD failure.  It is also assumed
that vehicles that had recently passed a centralized inspection would
also be exempted from an RSD-triggered inspection.  Implementation of
these and other administrative functions would require substantial
computer record keeping, as outlined further in the cost analysis
section.
Controlled RSD Screening

The second approach for wide coverage with remote sensing involves a
requirement for annual inspections at a network of fixed inspection
sites at which RSDs would be installed along side of a special test
lane.  This approach was assumed to be used in the "Controlled RSD
Screening for Enhanced I/M" scenario.

Although freeway ranp measurements are a form of screening for more
comprehensive inspections, the approach considered here addresses two
significant limitations of making RSD measurements somewhere on the
roadway network.  As noted earlier, the freeway ramp-based remote
sensing or other field deployments of remote sensing are subject to
variations in vehicle emissions under differing conditions of vehicle
performance.  For example, modern vehicles under wide open throttle
accelerations (or other moderate to hard or uphill accelerations) at on-
ranpe may be expected Co operate with rich air-fuel ratios, resulting in
relatively high exhaust eaissions of CO and HC.  Alternatively, some
vehicles may, upon deceleration at off-ramps or when coasting,
experience fuel shutoff and have no significant eaissions.  While these
two alternative condition* represent opposite extremes, both present
serious challenges to the use of remote sensing equipment to distinguish
between properly and improperly operating equipment.  An equally or more
serious difficulty is presented by the differences between drivers,
which, as described earlier, can dramatically influence Instantaneous
exhaust emission concentrations.  Finally, the roadway based screening
approach does not ensure that all vehicles will actually be observed.
In order to take advantage of some of the benefits of remote sensing as
a quick, relatively-low cost screening tool without the confounding
effects of observing vehicles under widely differing operating
                                  -32-

-------
conditions,  a  scenario  was  investigated for using remote sensing under
controlled conditions,  ac centralized I/M test stations.

Under the Controlled RSD  Screening scenario,  it is assumed that vehicle
owners are required to  drive  their vehicles,  at least once per year
through a high volume,  centralized test station where the vehicles are
subjected to a drive-through  remote sensing test.   The station would
include a test lane equipped  with dual remote sensors to minimize
erroneous readings.   Drivers  would be instructed to proceed through the
test lane at a steady speed,  during which time exhaust concentrations,
speed and acceleration  would  be measured.   Any vehicles that did not
operate within controlled windows of performance would be directed to
drive around the block  and  return for a retest.  Automated license plate
readers would  be used and license plate numbers would be recorded on
computer media along with remote  sensing test results.  Vehicles
identified on  the spot  as apparent high emitters would be subjected
immediately  to a full tailpipe and underhood inspection (electronic
gates and signage would bo  used to direct drivers).   Apparent low
emitters would be recorded  as "passing" and allowed to exit the facility
without intervention, i.e., with  minimal driver delay.  A conceptual
design for a test facility  at which this concept is used is shown in
Figure 20.
                                Figure 20
                   Conceptual Oerign of Remote Sensing-BaMd
                     High Volume, Centralized T««t Station
                                                         r-i
                                   -33-

-------
The major benefit of Controlled RSD Screening is that all vehicles could
be subjected to a screening for excessive exhaust concentrations under
controlled conditions, an important factor for improving the reliability
of remote sensing results.  A relatively small fraction of these,
approximately 72, would be subjected to the expense and inconvenience of
a full tailpipe and underhood test.  There are several disadvantages,
however.  Obviously, the controlled remote sensing-based screening, like
field-deployment of remote sensing, could not identify non-tailpipe
related failures such as failures or tampering of evaporative emission
control equipment.  This approach is also unlikely to work effectively
for detecting NOx emissions problems.  Even if NOx concentration
measurement by RSD can be performed with reasonable accuracy, the
inherent inability of RSDs to measure true mass emission rates is more
of a problem with NOx than with HC or CO because the difference in
emissions between properly functioning vehicles and defective vehicles
are more often smaller than the error introduced by using concentration
measurement instead of true mass measurement.  In addition, NOx
emission* need to be measured under acceleration or under a high load on
a chassis dynamometer.  It may be impractical to achieve the proper
vehicle operating condition when the car is not being tested using a
dynamometer.

Another concern with Controlled RSD Screening is that motorists can
prepare their vehicle for the test.  This preparation could involve the
temporary correction of tampering.  To help deter such behavior and the
excess emission* associated with tampered vehicles, the Controlled RSD
Screening concept could be augmented by the same type of roadside
pullover program described previously.
The following section of the report describes the methodology used to
estimate the emissions reduction* associated with each of the scenarios
evaluated.
                                   ***
                                   -34-

-------
                      Effectiveness Analysis
 I/M programs reduce emissions from motor vehicles In different ways.
 The primary effect of I/M is the identification and correction of
 emissions-related defects resulting from lack of adequate maintenance  or
 unintentional component failures.  A related effect is the incentive an
 I/M program provides for motorists to perform preventive maintenance in
 order to minimize the inconvenience associated with failing an I/M test
 and having to obtain repairs on an expedited basis (before the vehicle
 registration expires).  In addition to this "incentive" for preventive
 maintenance, I/M also provides a "deterrent* to deliberate tampering.
 Many forma of tampering (e.g., catalyst removal) are sufficiently
 expensive and time consuming to correct that the need to pass an I/M
 test deters the vehicle owner from tampering in the first place.

 Sierra's analysis of how programs involving the use of RSDs are expected
 to affect vehicle emissions has considered the effects of both tampering
 deterrence and the ability of the technology to identify unintentional
 emissions-related defects.
Benefi       ^       Deterrence
There is uncertainty regarding the extent to which a remote sensing
program will actually deter tampering.  Even if the probability of being
measured by a remote sensor is assumed to be high, there are two reasons
why tampering may not be effectively deterred:

  1. most owners of tampered vehicles apparently do not know their
     vehicle is defective; and

  2. sophisticated owners will coma Co recognize that most forms of
     tampering cannot be detected by remote sensors.


The tampering deterrence potential of remote sensing cannot be assumed
to be equal to the occurrence of tampering in the fleet.  The latest
roads id* survey results in California Indicate a tampering rate of
15. IX.  However, in compiling the survey results, all defects are
defined as "tampering" that could have been due to the intentional
removal, modification, disconnect ion, or disablement of an emissions"
related part.  There is considerable evidence that many motorists are
unaware that their vehicles contain such defects.  Many vehicles -
recruited from the general public for emissions testing by ARB contain a
full range of emissions-related defects.  In fact, the pattern of
defects that exists in the vehicles recruited by ARB is similar to the
pattern of defects observed during random roadside inspections conducted
with the assistance of the California Highway Patrol (CHP).  The pattern
of underhood failures in the undercover cars recruited by ARB for the
last evaluation of the California I/M program was evaluated to determine
                                  -35-

-------
 whether  they were  representative of those vehicles stopped by the CHP
 during the  1992  random roadside inspection program.  Figure 21 shows how
 the  pattern of defects in  the cars volunteered for testing by the
 general  public compared to the pattern of defects in random roadside
 sample.  The defect pattern for the random roadside sample was
 calculated  for those vehicles that failed the roadside inspection (for
 either exhaust emissions or underhood problems) in order for the
 comparison  to be between "should fail" vehicles (which the undercover
 car  sample  represents).  In addition, the sample of vehicles volunteered
 for  testing was  normalized so that the percentage of cars in each model
 year range  exactly matched the random roadside sample.

 As the figure shows, the pattern of defects between the two populations
 is similar.  In  some categories (e.g., PCV, thermostatic air cleaner,
 evaporative controls, and EGR functional test), the volunteered vehicle
 sample exhibits  more tampering than the random roadside sample.   In
 other categories (e.g., catalyst and air injection system components),
 the volunteered  vehicles exhibit somewhat less tampering.  In other
 categories  (carburetor/fuel injection system defects, oxygen sensor
 defects, EGR visual defects), the results are almost identical.   The
 largest  difference in defect rates exists in the case of catalyst
 removal where the owner would be expected to know that the catalyst is
 an emissions-related device and that it had been removed intentionally.

 Translating the  results shown in Figure 21 to the entire fleet
 (including vehicles that would not fail a properly performed I/H test),
 catalyst tampering is a factor in about IX of the vehicles on the road,
 and  the  rate of  catalyst removal is about 90S lower for vehicles
 volunteered for  testing.  In the only other category where a greater
 incidence of defects was observed in cars stopped at random, the air
 injection system, about 21 of the vehicles stopped at random exhibited
 air  injection related defects, while the defect rate for vehicles
 volunteered for  testing was about 2/3 less.

 Based on the comparison of the defects observed in vehicles stopped at
 random compared  to vehicles volunteered for testing by motorists who
 presumably would not knowingly offer tampered vehicles to the state, it
 appears  that the intentional tampering rate is currently between 2-3X.
 It is not reasonable to expect that the apparent intentional tampering
 rate under the current I/M program would continue under an enhanced
 program.  Under  the current program, the quality of tampering
 inspections is poor and only about SOX of all defects are identified by
 the first Smog Check station to which a vehicle is taken for testing.
 The currant California program also gives motorists the opportunity to
 shop around for  a station that will pass their tampered vehicle or to
 take their vehicle to a garage where they have established a business
 relationship and where the technician may be willing to overlook certain
 defects  that the motorist does not wish to correct.  This further
 reduces  the effectiveness of the current California program in
 eliminating tampering.  Under an enhanced program with centralized
 testing, it is expected that over 90S of the tampering will be detected.
With this level  of detection, it is unlikely that most motorists will
 re-tamper with their vehicles after repairs have been necessary to get
 through  the inspection process.  In the unlikely event that half of  the
motorists would  go to the expense and effort to tamper with their
                                  -36-

-------
                                Figure 21
                     Comparison of Underhood Defects
               Undercover Cars vs. Random Roadside Sample
                           ("Should Fail" Vehicles)
          PCV  TAC  Ev«p  CAT  EQR  EGR Cvfa/Fl SpwK WOT**  O2   Mr   Air   A»   A*
                                Fune      Timing  Light Smor Pump  Baft  V*w Ptunb
                                 Type of Defect
vehicles after they have been repaired, the net benefits of a program
that deterred tanpering would be cut in half compared to what might be
expected under current I/M programs.

Based on the assumption that intentional tampering under an enhanced I/M
program can be represented by 0.5Z of the vehicles operating without
catalysts and IX of the vehicles operating without functional air
injection systems, the net effect on fleet average emission levels is
estimated to be 0.018 g/mi HC, 0.18 g/mi CO, and 0.005 g/mi NOx.  This
estimate is based on assuming a cold start catalyst efficiency of about
60X for HC and CO and SOX for NOx (an estimate for the average aged
catalyst in customer service), and non-catalyst emission rates of
2.S g/mi HC, 25 g/mi CO, and 2 g/mi NOx.  The increase in emissions for
0.5X of the vehicles running without catalysts would cause average
emissions to increase by 0.008 g/mi HC, 0.08 g/mi CO, and 0.005 g/mi
NOx.  If air injection systems are assumed to reduce emissions by SOX
for HC and CO, the increase in emissions associated with IX of the
vehicles having tampered air injection systems would be 0.01 g/mi HC and
0.1 g/mi CO, based on the assumption that aged vehicles with catalysts
and properly functioning air injection systems emit about 1 g/mi HC and
10 g/mi CO.  Since air injection is no longer used on most vehicles,
this estimate of the Increase due to air injection tampering serves as a
surrogate for potential tampering with other cold start controls (e.g.,
electrically heated catalysts) that may become a factor in the future.
                                  -37-

-------
Using  the emission  reductions described above, the benefits associated
with the elimination of deliberate tampering under an enhanced I/M
program have been computed by dividing the estimated decrease in
emissions by the baseline exhaust emissions forecast for calendar year
2000.  The baseline forecast was produced by ARB's CALIMFAC model
combined with evaporative and crankcase emissions estimates from the I/M
Review Committee's  latest analysis1.   These baseline  rates  are 0.95  g/mi
HC, 7.78 g/mi CO, and 0.80 g/mi NOx.  Under an enhanced program,  the
emission benefit associated with the complete elimination of tampering
is estimated at 1.9Z for HC,  2.3Z for CO, and 0.6Z for NOx, as shown in
Table  2.  Under the current program, with less effective identification
of tampering, the theoretical benefits would be twice these levels.
However, it is unlikely that a remote sensing program would be
completely effective in eliminating tampering.  If the use of remote
sensing becomes widespread, most motorists will quickly learn that the
system can be beaten.  In locations where immediate pull-overs are used,
it will be obvious  to approaching motorists what is happening.   Turning
off the roadway to  avoid an inspection will be an alternative.   Turning
off the ignition and coasting through the sensor beam will also defeat
the system.  In addition, the sensitivity of remote sensing is such that
catalyst removal and air injection system tampering cannot be detected
on a vehicle that is still running closed-loop.  Sophisticated motorists
will learn this.
Table 2
Emissions Impact Due to
Intentional Tampering Known to Motorist
HC CO NOx
Baseline
Catalyst Tampering
Impact
Air Injection/Other
Tampering Impact
Net Tampering Impact
Change in Baseline
0.9S g/mi
0.008 g/mi
0.01 g/mi
0.018 g/mi
1.9X
7.78 g/mi
0.08 g/mi
0.10 g/mi
0.18 g/mi
2.3Z
0.80 g/mi
O.OOS g/mi
0.0 g/mi
0.005 g/mi
0.6Z
Assuming only half of the motorists recognize that there  Is no
significant risk of having tampering detected by remote sensing,  the
potential emissions benefits associated with the deterrence of  tampering
through a remote sensing program in combination with enhanced I/M are
reductions of IX for HC, 1.2Z for CO. and 0.3Z for NOx.   Under  a
continuation of the current decentralized program, the reductions are
estimated at 1.9Z for HC, 2.3Z for CO, and 0.6Z for NOx.
                                  -38-

-------
Table  3  shows how  the benefits of  tampering deterrence presented in
Table  2  are projected to affect  the emission reductions achieved under
an  I/M program.  Based on ARB's  CALIMFAC model, the calendar year 2000
emission reductions  from vehicles  subject to the current decentralized
I/M program will be  16.61 for hydrocarbons (HC), 25.3X for carbon
monoxide (CO), and 10.4X for oxides of nitrogen (NOx).  With the
addition of a remote sensing program, additional deterrence of tampering
is  projected to increase the level of emission reductions to IB.51 for
HC, 27.61 for CO,  and 11.OX for  NOx.
Table 3
• Estimated Effect of Tampering Deterrence
Emission Reduction
(year 2000 forecast)
HC CO NOx
Current Smog Check Program
Without RSDs
Current Smog Check Supplemented
With RSD/Pullover Program
Enhanced I/M Without RSDa
Enhanced I/M Supplemented With
RSD/Pullover Program
16.61
18. SX
35. 8X
36. 8X
25.31
27.61
34. 6X
35. 8X
10. 4X
11. OX
22. 2X
22. SX
Without the use of RSDs, ARB's model predicts that an enhanced program
with dynamometer testing will reduce exhaust emissions by 28.2X for HC,
34.6X for CO, and 22.2X for NOx."  Accounting for the efficient repair
of evaporative and crankcase emissions defects, the  total HC emissions
reduction for an enhanced program increases to 35.8X, aa explained in
the I/M Review Comnitteen report to the California Legislature1 and aa
shown in the third row of Table 3.  With greater tampering deterrence,
the reductions increase to 36.8X for HC, 35.8X for CO, and 22.SX for
NOx.
' The CALIMFAC modal does not specifically address IM240 testing, but
rather a program that uses steady-state loaded mod*  testing  in
conjunction with comprehensive visual and  functional inspections
performed with the same  level of  thoroughness aa used by ARB
technicians.  The benefits of such  a program are assumed to  represent
"enhanced" I/M for the purposes of  this analysis.
                                  -39-

-------
Benefits of Comprehensive Testing of Vehicles Screened hy

As described in the previously mentioned report to the California
Legislature1,  the ARB's CALIMFAC computer model can be  used to  estimate
the emission benefits of an I/M program based on the reductions achieved
from the repair of vehicles that occurs during one program cycle.  This
approach is based on the assumption that the ratio of the benefits
achieved during one cycle to the compound benefits that occur after
several program cycles will be similar across a range of alternative I/M
programs.  This approach recognizes the fact that calculation of the
benefits of any I/M program requires estimation of both the "residual*
and current benefits of identifying and correcting vehicles with
emissions-related defects.

In the most recent Review Committee analysis of the benefits of the
current California I/M program, data from the 1,100 undercover vehicles
were analyzed using the CALIMFAC model.  This model contains information
regarding the distribution of vehicles in the fleet by 25 model years,
16 technology groups, and five "emissions categories."   Based on
thousands of tests of vehicles recruited from customer service by ARB
over several years, each model year-technology combination is broken
into "normal", "moderate", "high*, "very high" and "super* emitter
categories.  Because the model is pollutant specific, there may be
different distributions for HC, CO, and NOx.  Based on data from
laboratory tests conducted by ARB and experience gained during the 1986
evaluation of the I/M program, a probability of I/M failure is
associated with each emitter category-technology group combination as a
function of the test procedure (e.g., idle vs. loaded mode), emissions
standards stringency, extent of visual and functional inspection, and
quality of inspections.  Repair actions cause vehicles to migrate
between categories.  The baseline migration pattern is a function of the
actual migration pattern observed during the 1986 evaluation of the
program.  Alternative migration patterns for higher repair cost ceilings
and better repairs have been developed based on data from vehicles
repaired by ARB technicians.                      *

In the absence of I/M, there is also migration occurring between emitter
categories as vehicles age.  This non-I/M migration pattern is also
based on the actual performance of vehicles in customer service without
I/M tests.  By running the model with and without the I/M module
activated, the difference in average emissions can be estimated.
Depending on the assumptions used regarding inspection accuracy and
repair effectiveness, several cycles through an I/M program are
necessary before the benefits are seen to stabilize.  This is  the result
of the fact that the residual benefits of a previous inspection cycle
last for more than two years.

To estimate the benefits of an I/M program under which vehicles are
recruited or screened for comprehensive testing by remote  sensing, an
analysis was conducted using data from the 1,100 undercover vehicles
recruited during the latest evaluation of the current program.  These
vehicles represent the portion of the fleet that should  fail a properly
performed I/M test.  Based on the steady state, idle and 2500  rpm CO
concentrations for these vehicles, a subset of  the sample  with CO
                                  -40-

-------
emissions above 32 was selected.  Based on the earlier-referenced ARB
test program in which the correlation between FTP and drive-by emissions
was determined, this CO concentration was selected as the most stringent
drive-by standard that could be used without an excessive number of
false failures.  The FTP emissions reductions actually achieved when
these vehicles were taken to Smog Check stations were then compared to
the FTP emissions reductions achieved for the entire sample of 1,100
undercover vehicles.  Table 4 presents the results of the analysis for
the 1979 and later model-year vehicles.  (The values in the table
account for the cumulative effect of multiple I/M cycles using the same
methodology described in Appendix A of the I/M Review Committee's report
to the Legislature.1)
Table 4
Comparison of I/M Benefits Achieved Based on the Repair of
Vehicles Above 3Z CO Compared to All 1980 and Later Model Vehicles
Subjected to Current I/M Program
HC CO NOx
Vehicles 23Z CO
All Vehicles
Percent of Maximal
Benefits
12.61
23.81
52.91
16.61
20.91
79. 4Z
-0.2Z
6.2Z
O.OZ
The values shown in Table 4 do not represent the maxtmtm achievable I/M
benefits since the analysis is based on the current performance of
repair facilities involved in the Smog Check program.  The analysis was
conducted only to establish the ratio of I/M benefits associated with a
screening approach that separates vehicles above 3Z CO from the rest of
the fleet.  The ratios established were then applied to the emission
benefits that the CALIMFAC model predicts for an enhanced I/M program.

The benefits for an enhanced I/M program, with and without RSD
screening, are shown in Table 5.  As shown in the first row of the
table, the CALIMFAC model predicts exhaust emission benefits for a
program using high quality inspections and dynamometer testing to be
28.2Z for HC.  As shown in the second row of the table, the HC emissions
benefits of enhanced I/M are predicted to drop to 22.1Z if, as would be
the case with RSD screening, there is no inspection for evaporative and
crankcase emissions defects on vehicles that pass RSD screening.  (The
22.1Z HC benefit, as explained in the I/M Review Committee's report to
the Legislature, represents the total HC benefits when only 30Z of
evaporative and crankcase defects are identified.  Since only vehicles
with relatively high exhaust emissions would fail RSD screening, many
evaporative and crankcase emissions-related defects would go undetected.
Assuming only 30Z of evaporative and crankcase defects are identified is
consistent with the fact that less than one-third of the vehicles that
would fail a conventional enhanced I/M test are expected to fail RSD
screening.)  As shown  in the third row of Table 5, screening all
                                   -41-

-------
Table 5
Estimated Effect of RSD Screening
for Vehicles Subject to Enhanced I/M
Emission Reduction
(year 2000 forecast)
HC CO NOx
Enhanced I/M Without RSDs
Based on CALIMFAC Modal
Results for Exhaust Emissions
Enhanced I/M Without RSDs
Based on Total Emissions
Total Emission Reductions
Projected for RSD Screening for
Enhanced I/M
Emission Reductions for
RSD Screening for Enhanced I/M
Plus RSD/Pullover
28.21
22. IX
11.71
12. 7X
34. 6X
34. 6Z
27. 5X
28.71
22.21
22. 2X
O.OX
0.3X
vehicles for enhanced I/M testing using remote sensing produces
estimated benefits of 11.7X HC, 27.SX CO, and OX NOx.  In the case of
HC, the 11.7X benefit vas computed by applying the ratio of HC benefits
for vehicles above 3X CO to the total HC benefit for exhaust,
evaporative, and crankcase emissions (0.529 x 22.IX - 11.7X).

RSD screening in conjunction with a pullover program is expected to
provide additional tampering deterrence.  With routine inspections
performed at centralized facilities, the additional tampering deterrence
adds IX to the HC reductions, 1.2X to the CO reductions, and 0.3X to the
NOx reductions.
                                   ***
                                   -42-

-------
             Cost and Cost-Effectiveness  Analysis


 The  cost analysis  involved independent estimates  of cost for four
 different  inspection processes:  1) remote sensing in conjunction  with
 immediate  visual inspection (RSD/Pullover);  2) remote sensing at a
 network of freeway ramps with notification to owners of high-emission
 vehicles provided by mail (Photo RSD); 3) remote  sensing at a network of
 fixed sites where vehicles would be driven down a dedicated lane while
 speed measurements are made by laser rangefinder  or radar (RSD
 Screening); and 4) comprehensive Inspections of vehicles observed  to
 have high  drive-by emissions at a network of centralized facilities
 equipped with chassis dynamometers (Dyno Testing).  Each of the
 scenarios  under evaluation incorporates at least  two of these processes.
 In addition, a baseline calculation was made based on adding the
 RSD/Pullover concept to the current decentralized program.

 The majority of the cost analysis was performed using a modified version
 of the centralized I/M cost calculation model used to develop the  cost
 estimates  contained in the I/M Review Committee's latest report  to the
 California Legislature.1   The  methodology used was patterned after
 manual I/M cost worksheets developed by EPA in the late 1970s.11-12  The
 model produces a detailed "ground-up" cost estimation for alternative
 enhanced I/M programs in the L.A. basin.  The model accounts for  all of
 the individual cost elements of an I/M program (e.g., direct labor,
 land, buildings, equipment, overhead, etc.).  Capital costs (e.g.,
 equipment  purchases) are amortized and converted  to annual costs  after
 accounting for the money needed to make the capital investments.
Annualized capital costs are added to recurring costs (e.g., wages) to
 determine  the total annual cost.  After accounting for an assumed profit
margin of  10Z, the annual cost is then divided by the average annual
number of  I/M tests projected for the contract period to produce  a
 cost/test  estimate.

Assumptions are incorporated for both cost/unit values (e.g., building
 construction costs for dyrao testing facilities, in $/ft2) and the  number
of units required for the program (e.g., the square footage of buildings
needed at  each inspection facility).  Most of the key assumptions are
 provided in Table 16 of reference 1.  Many of the assumptions regarding
 centralized testing costs were obtained from one  of the major I/M
 testing contractors and all of that information was considered
 •reasonable* by other contractors with whoa representatives of the
Review Committee consulted.  Miscellaneous estimates were obtained from
building contractors.  Supplemental estimates for RSD-related costs were
 obtained from RSD equipment suppliers and contractors to the California
 Department of Transportation  (Caltrans).

Using the  computer model, costs were projected for each of the four
 scenarios  described in th« section on Scenario Development.  In the case
 of the "Basic I/M Plus RSD Supplement" scenario,   the model was used only
 to calculate the total annualized cost of the supplemental RSD program,
                                  -43-

-------
which was then divided by the annual testing volume to obtain an
additional cost per test to add to the $28 cost per test for the current
program used in the Review Committee's last report to the Legislature.
In other cases, the model was used to calculate the cost per dvno test
for the entire program (including both the RSD and dvno testing
components).  In each case, the costs were based on data applicable to
the South Coast Air Basin.  Overall costs would be similar in other
areas of the state, with some adjustments resulting from differences in
land costs and prevailing labor rates.

Many key assumptions and parameters affecting the cost of the remote
sensing program, such as the number of RSDs to be used and the number of
ramps to be equipped with permanent enclosures, have already been
discussed.  Other key factors determining program cost are discussed
below.

The fraction of vehicles seen by remote sensing that are identified as
high emitters determines the number of vehicles required to be brought
in for centralized inspection and, if appropriate, repair and retesting.
Based on the previously described USD/FTP correlation analysis performed
by ARB, 6.8Z of the fleet would be flagged as excessive emitters using a
3Z CO cutpoint.  (This assumes measurements are restricted to periods of
steady speed operation; under less controlled conditions, more vehicles
would be expected to fail.)  For the sake of simplicity, it was assumed
that all vehicles identified as high emitters by remote sensing using a
31 CO cutpoint would also fail a conventional I/M test.  It was also
assumed, based on current I/M data for all modal years and for
California as a whole, that the percentage of all vehicles failing a
conventional I/M test, such as IM240, for other than tailpipe CO would
be about SXU (overall, about 201 fail, with 16Z failing due to excess
tailpipe emissions).
Centralized Dvno Tearing

As noted above, the cost of the network of inspection facilities for
comprehensive vehicle testing involving chassis dynamometers was
computed using the same basic model used for the centralized I/M cost
estimates included in the I/M Review Committee's most recent report to
the California Legislature.  An error in an earlier version of the model
was corrected and this lowered the cost per test of a centralized I/M
program by about $0.50 per test.  More significantly, all of state
administration and enforcement costs were computed using a subroutine
built into the modal that was not used previously.  In previous model
runs, tha $7 per certificate fee currently used to cover state
administration and enforcement costs in California prograa was assumed
to be carried forward.  It was recognized that tha administration and
enforcement costs of a centralized program would ba lower; however,
maintenance of the same $7 per test charge was intended to provide
additional resources for mechanic training and service information
distribution.  Under this analysis, tha actual cost savings due to the
reduced administrative and enforcement burdens with a centralized
program were computed to ensure that substantially different I/M
concepts would be evaluated in the most equitable manner.

-------
Controlled RSD Screening

The RSD Screening element of che "Controlled RSD Screening for Enhanced
I/M" scenario involves modification to the basic centralized I/M model
to incorporate  the additional land and paving required for the drive-by
testing in addition to the RSD equipment.  The site plan for the
centralized, dyno testing facility with the RSD-equipped lane presented
earlier was used to estimate the additional land and paving needs.   The
costs used for  RSD equipment are consistent with those for the "Photo
RSD" approach,  which are described below.
Photo RSD

The par-unit cost to equip ramps with protective enclosures was
estimated based on the cost of enclosures, concrete pads,  conduit for
electrical service, and installation.  Two key assumptions were that
Caltrans owned and would cede the right-of-way for remote  sensing setups
at no cost and that no pavement would be required for remote sensing
crews that serviced the enclosures.  Both of these assumptions would
require further investigation and confirmation by Caltrans before they
could be used In developing detailed program plans and costs.

The cost of fabricated enclosures was based on an estimate provided by
M£M Electric, on* of several electrical concractors identified by
Caltrans.  M&M Electric provided cose information and drawings for four
enclosures (two transmitter cabinets and two receiver cabinets) per
site, to be manufactured by Tesco Controls Inc.  The cost estimates are
based the specifications provide by Tesco.  Other per-raap cost
estimates were based on personal communications with Caltrans14 and M&M
Electric15.  Total cost per ramp site for the modifications was
estimated at $15,700.

The capital cost of remote sensing devices was assumed to be $150,000
per ramp.  This cost assumes two infra-red light sources and receivers,
a control unit, vehicle speed measuring equipment (for screening remote
sensing data to minimize errors caused by undue accelerations between
the two instantaneous measurements), a video camera and video "frame
grabber* to record license plates for subsequent analysis in a
centralized laboratory, and optical storage equipment to record and
transfer data between th« field and the laboratory.

The capital cost of a centralized laboratory for RSD servicing and for
data processing was based primarily on rough estimates and Information
adapted from the I/M cose spreadsheet.  One unusual item required is an
optical license plate reader, which is used to determine license plate
numbers in an automated fashion.  As far as we know, this equipment is
not yet available commercially.  According to one recent draft
report16, it may be available soon at an approximate cost of  $35,000
per unit; throughput is not known.  It was assumed that four optical
license plate readers would be used at the centralized facility to read
video data recorded at the off-ramps.  Overall, capital costs for the
central laboratory were estimated at only about 10Z of the cost of
                                  -45-

-------
equipping ramps.  An estimate of $1.83 per vehicle exceeding the drive-
by emissions threshold was used for the cost of notifying owners of
vehicles.  This estimate is expected to cover all owner identification
and notification costs, including postage costs for the initial and
follow-up notifications to motorists.

Three two-person field crews were assumed to provide field service and
calibration and to transport remote sensors between off-ramps.   It was
assumed that each team would drive 600 miles per week and that  total
travel cost, including vehicle amortization, would be $0.7S/mile.

Appendix B contains a printout of one of the model runs for a scenario
involving monitoring at freeway ramps.  As shown in the upper right-hand
corner of the first page of the printout,  the annualized cost for this
element of the program is $8.4 million.
RSD/Pnllover

Six supplemental pull-over teams were assumed.  This estimate was based
on the number of teams necessary to have the capability to cover at
least 1,000 different locations each year.  That degree of coverage,
involving approximately the same number of sites as the ramp-based Photo
RSD approach, is expected to result in about 85-90X of all drivers
seeing a pullover operation at least once each year.  The number of
vehicles identified as high emitters by these teams and pulled over for
testing is expected to be snail (about 64,000 per year) relative to the
number identified by remote sensors operating continuously at the off-
ramps.  The ability to perform immediate visual inspections is the
bottleneck.  Assuming a 6.8Z RSD failure rate, just under 1 million
vehicles could theoretically be inspected and detained for underhood
inspection, which is about 12X of the fleet in the Greater Metropolitan
Los Angeles area.  Because of the visual inspection bottleneck, only
about IX of the fleet is projected to require a dyno retest based on
failing the RSD/Pullover program.  Repeat tests on specific vehicles
will reduce the actual fraction of the fleet subject to inspection using
this approach.
As discussed previously, these tesas are intended to function primarily
as a highly visible deterrent to tampering and off-ramp avoidance.
Costs for this element of the program are summarized in the upper
right-hand portion on the second page of the example printout in
Appendix B.  The major cost elements include $280,000 each for vans
equipped with RSD systems, personnel costs (including police officers).
and equipment maintenance.  The annualized cost of the program is $3.3
million.
C(ffi Summary

Table 6 summarizes the results of the analysis performed using Sierra's
cost model.  The annual inspection rate, inspection facility
requirements, and cost per dyno-tested vehicle is presented.  The size
of the dynamometer testing facilities (lanes per facility) was
                                  -46-

-------
Table 6
Comparison of Alternative I/M Programs
Annual Inspection Inspection Cost/Car
Inspection Elate Lanes Stations Tested
(USD/Full) Required Required Fully
Current Program
Current Program
Supplemented
with
RSD/Pul lover
Program
Enhanced I/M
Without RSD»
Enhanced I/M
Supplemented
With
RSD/Pullover
Program
Enhanced I/M
With Annual RSD
Screening
Photo RSD
OX/58. 5X
11.8X/59.3X
OX/58. 5X
11.8X/59.3X
100X/6 . 8X
78.8X/5.4X
2,160
2,160
240
240
54
21
720
720
60
60
18
21
$28.18
$28.82
$18.50
$19 . 14
$36.10
$45 . 52
restricted as necessary to ensure that the number of facilities in the
South Coast Air Basin would be at least as great as the 17 inspection
facilities used during tha centralized change-of-ownership Inspection
program conducted from 1979 to 1984.  (In the case of the decentralized
program scenarios, the number of lanes (bays) and facilities is the
minimum necessary to adequately serve the public without facilities
designed for high-volume testing.)

As shown in the table, tha addition of the RSD/Pullover program adds
$0.64 per test to the current program or an enhanced centralized
program.  The greater efficiency of centralized testing results in a
much lower cost of $19.14 per test (compared to $28.82 for decentralized
testing) even though dynamometer testing is included in the centralized
scenario, but not the decentralized scenario.

Cost per dyno test for the scenarios that involve RSD screening are
substantially higher because fewer defective vehicles are identified and
the economies of scale are not as favorable.
                                  -47-

-------
 Cost Effectiveness

 Using  the costs per test described above, in conjunction with the
 emission reduction estimates described in the previous section,  cost
 effectiveness ratios in dollars per pound emissions reduced were
 calculated.  Annualized costs for each scenario were divided by the
 annual emissions reduction using ARB's standard approach of combining HC
 and NOx emissions with one-seventh of CO emissions.  (The discounting of
 the CO accounts for the relatively greater air quality degradation
 caused by each pound of HC and NOx.)

 In computing annualized costs, the testing cost per inspected vehicle is
 multiplied by the fraction of the vehicles that are tested each year.
 The testing cost per inspected vehicle is determined by the cost per
 test and the retest rate.  Repair costs per tested vehicle are then
 computed based on the assumption that the total repair cost under the
 current program (estimated at $60 per failed vehicle and $12 per tested
 vehicle) will increase in proportion to the emission reductions achieved
 under more effective I/M programs.  The repair cost per vehicle tested
 under an alternative program is therefore $12 multiplied by the ratio of
 emission reductions estimated for the program and then multiplied by the
 ratio of vehicles tested each year.  Average cost per on road vehicle is
 then computed by multiplying the cost per inspected vehicle by the
 fraction of vehicles inspected.

 The denominator of the cost effectiveness ratio was computed by applying
 the percent reduction in emissions calculated in the previous section to
 the baseline emission rates for calendar year 2000 produced by ARB's
 CALIMFAC model, adjusted to reflect evaporative and crankcase emissions.
 These baseline rates are 0.95 g/mi HC, 7.78 g/mi CO, and 0.80 g/mi NOx.
Annual emission reductions per vehicle were calculated based on an
 assumed 10,000 miles per year of operation for the average vehicle.  The
 effectiveness of the RSD-Only scenario was adjusted downward from the
RSD Screening for Enhanced I/M scenario to account for the fact that
 only 85-90X of the vehicles would be seen.

 The results of the analysis are summarized in Table 7.
Caveats

Because no I/M program based on remote sensing currently exists, there
are significant uncertainties related to the costs, capabilities and
limitations of programs such as the one described here.  Uncertainties
include how many of the registered vehicles would be seen by remote
sensing, what fraction would be apparent high emitters and what fraction
would be true high emitters.  As discussed previously, these relatively
uncertain elements can have an important effect on program cost.

Other major uncertainties also exist.  Some of these have been mentioned
already, including Caltrans' ability and willingness to cede right of
way and the availability of optical license plate readers.  In addition,
efficient program operation may require substantial analysis of optimal
site selection for remote sensors, screening criteria for analyzing

-------
Table 7
Cost and Effectiveness of Alternative I/M Programs
Failure Repair Cost-
Cost Per Rate on Cost Per Effectiveness
BAR90/Dyno BAR90/Dyno Inspected ($/lb. of
Test Test Vehicle HC+NOx+CO/7
Current Prograa
Current Prograa
Supplemented
with
RSD/Pul lover
Prograa
Enhanced I/M
Without RSDs
Enhanced I/M
Supplemented
With
RSD/Pullover
Prograa
Enhanced I/M
With Annual RSD
Screening
Photo RSD
$28.18
$28.82
$18.50
$19.14
$36.10
$45.52
20X
21X
20X
21X
100X
100X
$12
$13
$21
$22
$66
$66
$1.75
$1.67
$1.09
$1.10
$1.02
$1.00
disparate measurements for the same vehicle at the same sit* and/or at
different sites, "grace" periods for repair, suitable contingencies, and
approaches for incorporating the latest remote sensing technology (such
as HC and NOx measuring equipment, if that should become available).

Given the limitations and uncertainties identified in this analysis, the
cost estimates presented here should be considered for strategic
planning only.  A more dttailed fiscal analysis, Including a detailed
examination of the uncertainties identified above, should be conducted
prio.r to proceeding with an I/M program based on remote sensing.
                                   ***
                                  -49-

-------
                               References
1.   "Evaluation of the California Smog Check Program and
     Recommendations for Program Improvements - Fourth Report to the
     Legislature," California I/M Review Committee,  February, 16,  1993.

2.   T.C. Austin, et al., "An Evaluation of Loaded Mode I/M Testing at
     Service Stations," Report No. SR88-12-02, Sierra Research,  Inc.,
     December 7, 1988.

3.   T.C. Austin, et al., "An Evaluation of 'Remote  Sensing' for the
     Measurement of Vehicle Emissions," Report No. SR90-08-02,  Sierra
     Research, Inc., August 28, 1990.

4.   D.H. Stedman and G.A. Bishop, "An Analysis of On-Road Remote
     Sensing as a Tool for Automobile Emissions Control," University of
     Denver, March 1990.

5.   T.C. Austin, et al., "An Evaluation of Loaded Mode I/M Testing at
     Service Stations," Report No. SR88-12-02, Sierra Research,  Inc.,
     December 7, 1988.

6.   Summary of data on 439 vehicles supplied by Mark Carlock,  ARB
     Mobile Source Division.

7.   Personal communication, Caltrans, January 1993.

8.   "1986 Ramp Volumes on the California State Freevay System."
     District 3, State of California,  Business, Transportation and
     Housing Agency, Department of Transportation, June 1987.

9.   "Characterization of Driving Patterns and Emissions from Light-
     Duty Vehicles In California," (DRAFT) prepared for the California
     Air Resources Board by Sierra Research, March 5, 1993.

10.  "Light-Ducy Vehicle Driving Behavior: Private Vehicle
     Instrumentation," Volume 1: Technical Report, (Draft final report)
     prepared by Radian Corporation for the U.S. Environmental
     Protection Agency, August 24, 1992.

11.  "Centralized I/M Program Cost Calculation Worksheet," U.S.
     Environmental Protection Agency, August 1979.

12.  "Decentralized Private Garage I/M Progrsm Cost Calculation
     Worksheet," U.S. Environmental Protection Agency, August 1979.

13.  CVS Neva. Sierra Research, November 1992.

14.  Personal communication with CALTRANS, January 1993.
                                  -50-

-------
15.   Personal communication, M6M Electric, January 1993.

16.   "Feasibility Study for the Use of Remote Sensors to  Detect High
     Emitting Vehicles," Hughes Environmental Systems,  prepared for the
     South Coast Air Quality Management District,  November 6,  1992.
                                  -51-

-------
                              Appendix A
               How Tailpipe Concentrations Are Estimated
                 From Concentrations of Diluted Exhaust
The Theoretical Concept and its Limitations

The  ability to determine the carbon monoxide concentration emitted by  an
individual motor vehicle after its exhaust has undergone an unknown  and
variable amount of dilution depends on the existence of a consistent
relationship between measurable products of combustion.   If,  for
example, it was known that all engine's emitted exhaust containing 15Z
carbon dioxide, then the concentration of other pollutants in the
exhaust could be determined,  regardless of the amount of dilution,
merely by determining the ratio of the measured concentration of  each
pollutant to the measured concentration of carbon dioxide.  Although
it's not quite this simple, and not quite this consistent, this is
essentially the way remote measurement of diluted exhaust emissions  can
be used to determine the concentration of carbon monoxide coming  out of
the  tailpipe of the passing vehicle.
Combustion Chemistry

Although gasoline is a complex mixture of hydrocarbon molecules,  octane
(C|Hlt) is typical of the average molecule.   Perfect combustion of octane
in air is represented by the following equation:

     C,HU + 12.5 0,  + 471*2 - 8C02 + 9H3° + 47M3                     CD


As shown in the above equation, one molecule of octane reacts with 12.5
molecules of diatomic oxygen to form eight molecules of carbon dioxide
and nine molecules of water vapor.  Assuming air consists of 21Z oxygen
and 79X nitrogen, forty—seven molecules of diatomic nitrogen are
present, but do not participate in the reaction.  After condensing the
water out of the exhaust products, the concentration of carbon dioxide
in the exhaust is 14.5X (8+(8+47)).  For theoretically perfect and
complete; combustion, there are no carbon monoxide emissions.  The
air/fuel ratio that provides just enough air for complete combustion is
called the •stoichiometric" air/fuel ratio.  For the combustion of
octane, the stoichiometric air/fuel ratio on a volumetric basis is
59.5:1 ((12.5447)*!).  However, the air/fuel ratio is usually reported
on a mass basis:  the pounds of air per pound of fuel.  Accounting for
the molecular weights, the stoichiometric air/fuel ratio for octane is
about 15:1.

Carbon monoxide emissions occur when there is insufficient air to
completely oxidize  the hydrocarbon molecules to carbon dioxide.  An
                                   A-l

-------
 oxygen deficiency exists  when the  air/fuel  ratio  is  "richer"  than
 stoichiometric  (i.e,  less than 15:1  in the  case of octane).   When there
 is  a deficiency of air, all of the available hydrogen will be
 preferentially  oxidized to water vapor.  The remaining carbon atoms will
 be  oxidized first to  carbon monoxide,  and then to carbon dioxide  to the
 extent oxygen is available.   When  all  of the oxygen  is exhausted,  some
 carbon monoxide remains.   Based on these idealized assumptions, the
 relationship between  carbon monoxide and carbon dioxide emitted can be
 computed from the air/fuel ratio.  The following equation shows how the
 idealized combustion  equation must be  modified to account for varying
 amounts of air  and fuel and to account for  the possibility that carbon
 monoxide is in  the combustion products:

      CjH,, + m 0, + 3.76(m)N, -» aCO  + bCO, + 9H,0 + 3.76(m)Nj        (2)

                              a
 Using the  coefficients  in the above equation, the relationship between
 CO  and CO} in the exhaust of an engine burning an eight-carbon fuel
 (octane) will be such that:

      a + b  - 8;  and (3)

      a/2 +  b -  m - 4.5  (4)
The reason why the sum of  the number of moles of CO and CO} must be
eight is that there were eight carbon atoms in the fuel molecule.  The
equality shown in equation 4 is a little more complicated.  The  amount
of oxygen available for the formation of CO and CO} is 4.5 moles less
than the total amount of oxygen initially available because 9 moles  of
water vapor will be formed from the oxidation of the hydrogen (and 9
moles of HjO contain 4.S mole* of 0}).  The remaining  oxygen  (m - 4.5)
is divided between the carbon dioxide and the carbon monoxide.   Because
there is only one oxygen atom in each CO molecule, each mole of  CO uses
only half as much oxygen as each mole of CO},  hence the coefficient of
the CO term is divided by  2.  Because the relationships between  the
coefficients of the CO and CO} terms shown in equations 3 and 4 must be
simultaneously satisfied,  there is only on* possible solution to the
equation for a particular  air/fuel ratio.  However, it Bust be noted
that equation 2 is a sisq>lified fora of ths* combustion equation  for
octane that applies only when the air/fuel ratio is equal  to or  less
than stoichiometric.  With leaner mixtures (A/T>15:1), some oxygen is
left over.

Using ths idealized combustion equations, Figure A-l  shows the
relationship between carbon monoxide and carbon dioxida over a wide
range of air/fuel ratios.  The relationships shown in the  figure show
why tailpipe emission concentrations can be predicted from diluted
exhaust in the wake of a passing vehicle.  There la « unique CO/CO,
ratio for every CO concentration at the tailpipe.  As an example, for  an
                                   A-2

-------
                                Figure A-l
                        Theoretical Relationship
                         Between Air/Fuel Ratio,
               Carbon Monoxide, and Carbon Dioxide
    20
                   Stoichiometric A/F for Octane
o
o
 »*                        •-•-..
**                              ••».
o
o
  . 10  -   —-1
                                 Carbon Monoxide
    5	
                                       « C\J
                                       3 O
                                         O
                                       2O
                                         O

                                       1
      11:1      12:1      13:1      14:1      15:1     16:1     17:1      18:1
                               Air/Fuel Ratio

Note: Dry Basis (corroded tor wctervepor)
  engine emitting about  7Z CO, the CO/002 ratio i* about 0.75.  At  the
  tailpipe, the CO] concentration would be about 9.SZ.  A* the  vehicle
  passes a remote sensor, the concentration of CO  and COj behind  the
  vehicle vill be measured to be ouch lover because of the dilution that
  is occurring as the  exhaust is mixed with the ambient air.  However,  the
  ratio of CO to CO] should remain unaffected.  Regardless of the amount
  of dilution that has occurred, a CO/CO, ratio of 0.75 indicates that  the
  engine was emitting  7X CO.
  Errors Tntmdnced bv Non-Ideal
  It must be noted that the relationships between CO, CO,, and air/fuel
  ratio illustrated in Figure A-l are idealized.   In real engines, there
  are numerous factors that affect the relationships illustrated in the
  figure.  One factor is that some of the fuel exits the  engine without
  being oxidized as far as carbon monoxide.  When this  occurs, there are
  hydrocarbon emissions to account for.  Another factor is that fuel and
  air are not perfectly mixed in the engine.  Some cylinders operate at
                                    A-3

-------
different air/fuel ratios than others.  Even within the same cylinder,
there are regions of the combustion chamber that are leaner than others.
Due to air/fuel ratio inconsistencies, real engines emit carbon monoxide
emissions even if the overall air/fuel ratio is stoichiometric or
leaner.  Differences in the hydrocarbon composition of commercially
available gasolines also affect the relationships shown in the figure,
especially when oxygen-containing additives, such as alcohols, are used.
Emission control systems also affect the relationships between emissions
and air/fuel ratio.  Engines running richer than stoichiometric can
almost completely eliminate CO emissions through the use of air
injection systems, thermal reactors, and catalytic converters.

Sierra has performed an analysis of the range of effects that the non-
idealized nature of combustion in real vehicles can have on the ability
to detect tailpipe CO concentrations, based on measurement of C0/C07
ratio.  Based on this analysis, C0/C02 ratio is  a poor  predictor  of
actual air/fuel ratio, but a fairly good predictor of "excess air
adjusted* carbon monoxide concentrations in vehicle exhaust.  The term
"excess air adjusted" is used because the remote sensor is blind to the
extra oxygen and nitrogen consumed by an engine or injected into the
exhaust system by an air pump.

If, for example, an engine is running at 12:1 air/fuel ratio, Figure 2
would predict that the CO concentration emitted by the engine would be
11.12X and the CO/CO] ratio would be 1.72:1.  If the engine is equipped
with an air pump that injects exactly the amount of air theoretically
required to complete combustion, and if that caused the oxidation of
half of the CO, then actual CO concentration would drop to 4.SX and the
CO/CO] ratio  would decrease to 0.46.   However, based on Figure 2,  the CO
concentration would be predicted to be S.OX, over 10X higher.  Although
the apparent error is over 10X, it is due entirely to the fact that the
unreacted oxygen and its associated nitrogen was diluting the exhaust of
the vehicle.   Generally speaking, it is desirable to correct for the
effects of dilution anyway because dilution does not reduce that actual
mass emission rate of a vehicle, just the concentration.  Since it is
the mass emission race of vehicles that affects air quality, a
measurement device that is insensitive to the effects of dilution  is
actually preferable.  (The predicted air/fuel ratio associated with a
CO/CO, ratio  of 0.46 is  13.52,  an error of 12.7X.)

As noted above, hydrocarbon emissions are another source of error.
When misfire occurs, a portion of the air/fuel mixture passes through
the engine unreacted.  This dilutes the exhaust gas and tends to lower
the tailpipe CO concentration, but has no effect on the CO/CO, ratio.
For a misfiring vehicle running at 13:1 air/fuel ratio, the actual CO
concentration is 10.7X lover than predicted by the relationship between
CO/CO, ratio  and CO shown in Figure 2.  Again,  the reason for the -
discrepancy is primarily dilution and the error is acceptable, if  not
desirable.  (In this case, the air/fuel ratio is still accurately
predicted.)
                                   A-4

-------
     Effects of Fuel Economy Differences

The  developers of the University of Denver remote sensing system refer
to the system by the acronym  "FEAT", for "Fuel Efficiency Automobile
Test.*  This  acronym was selected by the developers based on their
belief that the system could  detect when a vehicle is operating at a
stoichiometric air/fuel ratio and that such operation is the most
efficient.  As explained above, the system is not capable of determining
the  actual air/fuel ratio at  which an engine is operating.  In addition,
stoichiometric operation is not the most efficient air/fuel ratio.  (Due
to the "pumping loss" associated with throttled, part-load operation of
an Otto cycle engine, peak efficiency generally occurs at relatively
lean air/fuel ratios where less throttle and less pumping loss occurs.)

Another limitation of tha system is related to fuel efficiency.  The
system is capable of predicting excess air adjusted CO concentration,
but  not mass emissions (grams per mile).  As a result, vehicles with
very low fuel consumption will appear to have relatively high emissions
compared to vehicles with relatively high fuel consumption.  Because the
range of fuel economy for vehicles in the existing fleet exceeds 4:1, a
fuel-efficient subcompact car emitting 4X CO can have exactly the same
gram/mile emission rate as a  heavier car with a big engine emitting IX
CO.  Because the range of light-duty fuel economy is so large, a
concentration-based standard  would either be too strict for smaller,
more fuel efficient vehicles, or too lax for larger, less fuel-efficient
vehicles.
                                   ***
                                   A-5

-------
             Appendix B
Cost Model Printout for Program Involving
  RSD Measurements at Freeway Ramps

-------
                                           ZONES PER FACILITY

                                               COST     % Of TOTAL
AUGUST 18.1993      REMIM5.WK3

       TOTAL ANNUAU2EO COST OF PROGRAM:

COST OF CENTRALIZED I/M PORTION OF PROGRAM       t27.194.478
COST OF RS PORTION OF PROGRAM                  98.381.400
OWNER NOTIFICATION AND FOLLOWUP ADMIN. SUPPO    $1.040.015
SUPPLEMENTAL PULLOVER PROGRAM(DETERTAMPER)    $3.341.733
TOTAL ANNUALIZED COST.                          (39.9S7.61S
NUMBER OF CENTRALIZED TESTS/YEAR                 877.878
TOTAL COST PER CENTRALIZED TEST                    (45.52
                                                              68,1%
                                                              210%
                                                               2.6%
                                                            .   8.4*
                                                             100.0%
GENERAL ASSUMPTIONS:
RSDS ARE MOVED AROUND BETWEEN OFF-RAMPS ONCE PER WEEK
TWO REMOTE SENSING DEVICES AT EACH INSTRUMENTED RAMP
RAMPS ARE EQUIPPED WITH 4 PERMANENT ENCLOSURES WITH POWER TO EACH
NO CHARGE FOR LAND (ASSUME CALTRANS OWNS * CEDES ALL RIGHTS OF WAY)
RSDS OPERATE 24 MRS/DAY. 0 DAYS/WK. UNATTENDED
APPARENT HIGH EMITTERS SUBMIT ANNUALLY TO CENTRALIZED INSPECTION
CENTRALIZED. TRANSIENT LOADED MODE TESTS
OPTICAL LICENSE PLATE READERS USED AT CENTRALIZED DATA ANALYSIS SITE
SPARE REMOTE SENSING DEVICE SETUPS:                             10.0%
ANNUAL WEEKS OF OPERATION PER R8O SETUP:                          50
COST FOR VEHICLE OWNER NOTIFICATION, PER INITIAL TEST PER YEA       91.00
AREA-SPECIFIC ASSUMPTIONS:                                      SCAB

NUMBER OF RAMPS TO BE EQUIPPED WITH BOXES:                       1300
FRACTION OF VEHS ASSUMED TO PASS BY RSDS AT LEAST ONCE/YR         87.5%
FRACTION OF VEHS PASSING BY REMOTE SENSING WHICH RESULT
       IN VALID RS READINGS AND VALID LICENSE PLATE READINGS:        80.0%
FRACTION OF VEHS READ BY RS (PRODUCT OF ABOVE TWO VALUES):        78.6%
FRACTION OF VEHS READ BY RS THAT ARE SEEN AS HIGH CO EMITTE        8.6%
VEHICLE FRACTION ASSUMED TO FAIL CENTRAL INSPECTION
       FOR REASONS OTHER THAN TAILPIPE CO:                       6.8%
NO. OF RSDS MOVED PER WEEK PER 2 PERSON TEAM                      10
MILES DRIVEN PER RS SERVICE TEAM PER WEEK                         600
FRACTION OF APPAR. HIGH CO EMITTERS THAT ARE HIGH CO EMITTER        50.0%
COMPUTED PARAMETERS:

NUMBER OF RSD SETUPS NEEDED:                                     29
NUMBER OF 2-PERSON TEAMS FOR MOVING/FIELD SERVICE OF RSO:             3
FRACTION OF TOTAL VEHICLES THAT ARE SEEN BY RS AND CORRECTLY
       IDENTIFIED AS HIGH EMITTERS:                               2.7%
FRACTION OF VEHICLES COMING IN ANNUALLY FOR INSPECTION:            12.4%
PROGRAM COSTS:

COST OF REMOTE SENSING PART OF PROGRAM (EXCLUDES PULLOVER):

CAPITAL COST OF RSDS                                  94350000
CAPITAL COST OF ENCLOSURES AT RAMPS (DETAIL 1 BELOW)   920 410 000
CAPITAL COST OF CENTRAL LAB FOR RS ft DP (DETAIL 2 BELOW)  9l'91o'ooO
ANNUALIZED CAPITAL COST                               95 334 000
ANNL COST OF CENTRAL LAB/OFFICE SUPPORT (DETAIL 3 BELO   91 'eftS^OO
ANNUAL EQPT MAINTENANCE COST (IN LAB) Q10%               9942 000
ANNUAL COST OF FIELD SERVICE/MOVING RSDS (DETAIL 4 BELO    9420 200
TOTAL ANNUALIZED COST FOR RS PART OF PROGRAM           96 381 400
                                                                                       COST DETAIL. REMOTE SENSING PART:

                                                                           1 CAPITAL COST/RAMP:
                                                                             CONCRETE PADS (91500 X 4)                                 96 000
                                                                             ENCLOSURES (2(19600+2091000)                              93.200
                                                                             CONDUIT. INSTALLED (500 6910/FT) AND ELECTRIC SERVICE         96.SOO
                                                                             COST FOR PREPARING ONE RAMP. INSTALLED                   915.700
                                                                             TOTAL CAPITAL COST. ALL RAMP SETUPS                    920.410.000

                                                                           2 CAPITAL COST OF CENTRAL LAB FOR RSD SERVICE AND DATA PROCESSING:
                                                                             LAND AND BUILDING (ESTIMATED)                           91 000.000
                                                                             ELECTRONICS A OPTICAL EQUIPMENT LABORATORY             9250.000
                                                                             AUTOMATED LICENSE PLATE READERS (4^935.000)              9140 000
                                                                             OFFICE EQUIPMENT                                       920000
                                                                             COMPUTER EQUIPMENT                                   9500 000
                                                                             TOTAL CAPITAL COST FOR CENTRAL OFFICE FACILITIES          91.910.000

                                                                           3 RS. CENTRAL OFFICE AND LAB SUPPORT STAFF                ANNUAL
                                                                             (EXCLUDES ADMIN SUPPORT FOR OWNER NOTIF&FOLLOWUP)     SALARY
                                                                             POSITION/AREA                              NUMBER® 40 HRS/WK
                                                                             PROGRAM ADMINISTRATOR                           1     967500
                                                                             TECHNICAL OFFICERS                               3     952500
                                                                             ENGINEERS                                       3     952.500
                                                                             DATA ANALYSIS/STATISTICAL STAFF                    3     937.500
                                                                             PROGRAMMERS AND DP SUPPORT STAFF                 3     952.500
                                                                             CLERICAL AND SECRETARIAL STAFF                    4     922500
                                                                             SUBTOTAL (SALARY X PERSON-YEARS)                        (742 500
                                                                             OVERHEAD AND FRINGE                         1000%    9742500
                                                                             OFFICE AND ELECTRONIC EQUIPMENT MAINTENANCE (Q20%/YR)    9182 000
                                                                             SUPPLIES (610% OF MAINTENANCE)                          916*200
                                                                             TOTAL. FULLY BURDENED                                91 665 200
                                                                           4 RS. FIELD SERVICE CREWS

                                                                             SUPERVISOR 1
                                                                             INSTRUMENT SPECIALIST 1
                                                                             SUBTOTAL (SALARY X PERSON-YEARS)
                                                                             OVERHEAD AND FRINGE
                                                                             LABOR. FULLY BURDENED
                                                                             VEHICLE OPERATING COSTS
                                                                             TOTAL
                                                    ANNUAL
                                                    SALARY
                                           NUMBER @ 40 HRS/WK
                                                                                                                             1
                                                                                                                             5

                                                                                                                         1000%
                                                      937.500
                                                      927.500
                                                     9175.000
                                                     9175.000
                                                     9350.000
                                                      970.200
                                                     9420.200

-------
          COST ASSUMPTIONS:
          CAPITAL COST OF INSTRUMENTS FOR 1 REMOTE SENSING SETUP:         »150.000
                TWO IR SOURCES AND RECEIVERS. CONTROL UNIT.
                SPEED MEASURING EQPT(2). VIDEO CAMERA. FRAME GRABBER.
                OPTICAL DATA STORAGE EQPT (ALL PORTABLE EQPT). SOFTWARE
          CAPITAL COST/RAMP(DETAIL AT RIGHT):                              915.700
                DETAILED SITE SURVEY. ENGINEERING DESIGN. PERMITS
              '  TWO SMALL ENCLSURES FOR SOURCE UNITS. ONE MEDIUM ENC. FOR SINGLE
                RECEIVER. ONE LARGE ENC. FOR SECOND RECEIVER AND
                CONTROL UNIT. FOUNDATIONS AND POWER FOR EACH ENC..
                INSTALLATION AND CHECKOUT
          CAPITAL RECOVERY FACTOR                                        0.20
          SERVICE VEHICLE. ASSUMED TOTAL COST/MILE (INCLUDES VEH AMOR        90.75
SUPPLEMENTAL PULLOVER PROGRAM TO DETER TAMPERING

COST PER RS-BASED PULLOVER TEAM (EXCEPT AS NOTED AT BOTTOM):
CAPITAL COST OF RS VAN WITH 2 RSDS
CAPITAL COST OF OTHER FIELD-BASED TEST EQUIPMENT
ANNUAUZED COST OF EQUIPMENT
          $200.000
           925.000
           961.000
RS AND PULLOVER CREW
CHP OFFICER
SUPERVISOR 1
INSTRUMENT SPECIALIST 1
INSTRUMENT SPECIALIST 2
SUBTOTAL (SALARY X PERSON-YEARS)
OVERHEAD AND FRINGE
FIELD LABOR. FULLY BURDENED '
NUMBERQ 40 HRS/WK
      1
      1
      2
      1

 100.0%
T
N)
SUPPLEMENTAL CENTRAL OFFICE SUPPORT/PULLOVER TEAM
CONTACTING OWNERS FOR REPAIR

OTHER RECURRING COSTS:
          SUPPLIES Q5«
          TRAVEL (60 MILES/MEMBER/DAY Q9.30)
          CHP VEHICLES
          EQUIPMENT MAINTENANCE QlOtt
SUM OF OTHER RECURRING COSTS:

TOTAL ANNUAL COST PER TEAM

NUMBER OF TEAMS

COST FOR PULLOVER SUPPLEMENT-TOTAL FOR ALL TEAMS
 943.000
 937.500
 927.500
 932.000
9167.500
9167.500
9335.160

 971.203
 912.000
                                                                                                                                             915.250
                                                                                                                                             922.500
                                                                                                                                              99.261
                                                                                                                                             930.500
                                                                                                                                             977.511

                                                                                                                                            9556.054
                                                                                                                                          93.341.723

-------
CENTRALIZED PORTION OF THE PROGRAM:

I. PROGRAM PARAMETERS

A. Aaaumptiona
2.'
3.
4.
5.
6.
7.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19
20.

23.
9t
26.
97
26.
29.
rraaenmne rupuauon
Annual Auto Population Growth Roto (%)
Inspaction Frequency (Annual • 1. Biennial • 2)
Annual Inflation Roto (%)
Annual Interact Rate (%)
Program Length (yoara)
Annual Operating Houra
Lane EHdency Factor
Hourly Lano Throughput
Stringency Factor (%)
Annual Percent Additional Taata (%)
Porcant of Initial Toata at Cantrateod Factty (%)
Porcont of Rotoata at Controlled FodNry (%)
Land AcquWtton Coat par aquore tool
Corwtruction Coata par aojuara toot
Monthly OMco Spaco RortoJ Coat par aquaro toot
Paving Coata par oquoro toot
Training Coat par mochonlo

Initial Daalgn. Eng. A Eval.(% of MUott or 9/Vehtdo)

Property Tax Rate par Ful Vate (contractor program
Tasting Lanoa par Aero of Land
CVS Tort Col Throughput Aoata/yHoat col)
rur
GRT
FREQ
INF
INT
PRL
HR8
EFF
Can/hr
8TR
XTR
INlT
RT8T
9LANOM2
9CONSm2
9RENTM2
9PAVO«2
Moch/1 000
9MCH
S1PI
mpi
IDE
ROE
9CPT
only) PRT-
PTYPE •
LPA
TPCV8
r.Mi.ooj
2.7%
1
4.0%
8.0%
5
2.600
50%
65%
20
20.0%
5.0%
100.0%
100.0%
914.00
960.00
91.20
91.60
2
9360.00

10.0%
01%
SO 10
1.00%
c
2
1000
                                                                           90.43
           B. Parameter Calculation*

    1.  AVERAGE ANNUAL POPULATION

           PRL-1       I
    AAP - POP K | SUM (1 * CRT) | / (PHL x FREQ) •            6.361.372
           Ml

    2.  Avorago Annual Cartalizod T«tt» Pafformad

A. INITIAL TESTS DUE TO RS - AAP x (%HI EMITTERS FROM RS)      568.574

B. RETE8T8 DUE TO TAILPIPE CO FAILURES FROM (A).
           (A) K(%APP HIGH EMITTERS THAT ARE HIGH)-        264.2S7

C. RETE8T8 DUE TOOTHER FAILURES FROM (A>-
           | (A) • (B)) x (%FAILING FOR OTHER THAN CO)           25.017

O. ANNUAL NO. OF CENTRALIZED TESTS FROM ALL SOURCES-      677.670

    3.  ANNUAL LANE CAPACITY

       CAP • HR8 x Cara/hr        CAP -                  52.000


    4.  Tola! Taat Lanoa Roqukod

       LAN -  TST / (CAPxLLPxEF   LAN-                     40

-------
II.  INITIAL COSTS TO PROGRAM OPERATOR

A.  Estimate* of Fectty Requirement*
   1.
inspection FadHry Sahara Footage Requirement* (enter data In aq ft)
            LAND
           Bute per
            Faculty
                                 LANES
                 Additional
                  per Lane
                      21.780
     Lane
                                             1.000
OFFICE. ETC.
  Basic par
   Facttty
          750
             Additional
             per Une
                           PAVING
                           Basic per
                            Facaty
                                 375
   2.    AaocatlonofLaneeendSe
        LANE ALLOCATION
           Lane* per     0
            FacHtty       ThtaSka
                    2            20
                                 Total
            TOTAL
                           (FAQ
                          20
                                 (UN)
LAND
 Each
Faddy
    43.560

  LND •
                                            BUILDING
                                             LANES
                                 Total      Each Facility
                                   871.200       2.000
                                                                      Additional
                                                                       per Lane
                                                                           3.200
                                                OFFICE. ETC.
                                        Total     Each Facility
                                          40.000         750
                                                                                                                          PAVING
                                                                                                                Total     Each Facility     Tola!
                                                                                                                  15.000        6.775     135.500
                                   671.200
                             LNS  -
                                          40.000
                                                                                                            OFC •
                                                                  15.000
                                                                                                                                    PVG  -
                                                                                                                                                135.500
                                                                  Annual Salary    Overhead  Salary/Benefit*
            Portion
        Sta. Manager
        Asst. Manager
        InapecttonTacta
        Service Rap
Q40hnVweek
           1
           1
           4
           1
                                      1.2S  perhdtty
                                      1.25  parhcHy
                                         5  per teal lane
                                      1.25  parlKttly
                              per Employee
                              O40hrafwk
                                   •27.750
                                   $10.890
                                   I10.MO
                                   (23.200
                                                                                                             Direct Coat* Total Training    Hiring    Total Training
                          A Fringe   par employee   Duration of   of Instruction    Coat per
                               30%             Instruction (hr»  par Trainee   Employee
                              $6.325      $45.004         320        $400       $5.050
                              $5.807      $32.464         320        $400       $4.306
                              $5.007      $32.484         160        $200       $2.190
                              $6.060      $37.700          80        $200       $1.360
                                                                                       Cost per   6. Hiring Cost
                                                                                       Employee  per Employee
                                                                                           $200      $6.150
                                                                                           $100      $4.496
                                                                                            $50      $2.249
                                                                                            $50      $1.410
   1.    ConatiucUon and Land AcquJiMnn Coete
        a.  LandAcquMtM
        b.  Paving
        c.  Conatructton    <
        d. TOTAL BuMng
                  •  LND x $LANDtt2 -
                  • PVO • $PAVG«2 •
                   (OFC « LNS) M $CONS«2
                    $PAV)NQ«$CONSTR-
                $LANOAQ -
                  $PAVING
                $CON8TR  •
                $BUtLDtNO •
             $12.106.800
               $203.250
              $3.300.000
              $3.503^90
                                     2.  Inapactlon FadUty Peraonnal Hiring and Training Co*U

                                        a.  Station Manager*:    $6150 x  125/facil  x FAC              $153.750
                                        b.  AM! Stat. Manager*: $4406 x  1.25 /facH  x FAC              $112.450
                                        c.  InapectkmTech*.:    $2240 x  5.00/lane  x LAN              $449.600
                                        d.  Cuat.Serv.Repa:    $1410 x  125/tacit  x FAC               $35.250
                                        a.  TOTAL ln*pectton Factory Penonnel
                                           Hiring A Training                        •  FIELDPERS      $751.250
        Fadty Preparation and Equtpmam Coat»
           (Category)
            Faddy
           Tail Lane

         Central Oflk*
            TOTAL
                          Number
                          20
                          40
                           3
                           1
8Ke Preparation
    Each

      $11.000
Total

 $440.000
                                                 $440.000
                                              ($PREP)
                                                                 Teat Equipment
                                                                      Each
                                                                           Total
                                                                                   O«ce A Other Equlpmt.     Data Proc. Equlpmt.
                $145.000   $5.600.000
                                             $5.600.000
                                                                          $3.000

                                                                         $14.000
                                                                         $40.000
                                                        Total
                                                         $60.000

                                                         $42.000
                                                         $40.000
                                                        $142.000
Each
$50.000
$5.000
$25.000
$100.000


Total
$1.000.000
$200.000
$75.000
$100.000
$1.375.000

TOTAL
$1.060.000
$6.440.000
$117.000
$140.000
$7.757.000
(JEQUIP)

-------
           C.
              1.   SUit-Up Poraonnal
                    Portion ATM
                  Program Administrator
                  AfaaAdminlilnlora
                  Technical OAcm
                  Data Ana»y»t»/Stattatteal SMI
                  Ctorical and Secretarial SMT
                      TOTAL
'; Avaraga No. of
Annual Satan/
Numborpar Numbarpar Total Number Startup Hour* Total Ptraon a 40 hra/waak
Central 0*ce SataHtta Ofice ofEmployeM per Employe* YMI» per Employee
1
0
4
2
0


0
1
2
1
3


1
3
10
S
IS
34

2.080
400
280
280
140


1.0
0.6
1.3
0.7
1.0


167.900
162.500
$52.500
$37.500
t22.SOO


OvwhMd
A Fring*
stm
I33.7SO
$31.250
$26.250
$18.750
$11.250



Total Salaiy
•ndB«n«ftU
$101.250
$54.087
$106.010
$37.861
$34.075
$333.281
($CSBP)
              2.   Stait-UpTfaMngandHMn«Cott»
                    Position Am
                  Ataa AdnrinMnton
                  TachnlcalOAcara
                  Clarical and SacrataiW SMI
                      TOTAL
                                  ToMNumbar

                                            I
                                            3
                                            10
                                            S
                                            15
                                            34
                                           Annual Satory
                                           •ndBanate
                                           par EfflBtoyaa
                                               9101.290
                                                983.750
                                                978.750
                                                956.290
                                                933.750
  Duration of
Instruction (hn)
(P« Employ*.)
          40
          40
          60
          40
          40
OractCottof     Total
 botruction  Training Cott*
 (PofEmpI)   (AiEmph)
      $400
      $400
      $400
      $100
      $100
 $2.347
 $8.608
$34.286
 95.808
911.236
  Hiring
  Coito
(ParEmpI)
   $10.000
    $5.000
    $1.000
      $200
       $50
Total Training
 PkM Hiring
 (AIIEmpto)
    $12.347
    $21.609
    $44.288
     $6.808
    $11.866
    $87.138
 ($CHIRE)
T
Ul
2.
CaleuMlon of 84art-Up AdmlnMrabV* Cart*

a.  Start-Up Panonnal          • 9C8BP
                                9IPI or 9IPVMNcta « POP                •

                  c. lnHMPnaramOadBn.Ena.tEML
                                HIDE M (91ANOAQ«9BUILOINO«9EQUIP)
                                or $IOEAwh. i POP                      •

                  d. MachantoTratabw Mala opanUd praanm only)

                                Machfl.000 « (POP/(FREQx 1^)00)) x $MCH -

                  a. AAnlnWnalMl>*i»oni^HM>aandTralnlna -     9CHIRE  -

                  *. TOTAL Start-Up AdmlnlaliathroCort*                    •
                                                                  9333.281


                                                                       $0



                                                                92.345.705



                                                                      N/A

                                                                   987.138

                                                                92.776.125

-------
D. ToUl Initial CoaU to Program Opmior

   1.   Land Acquisition • 9LANQAQ                             •   t12.196.800
   2.   Total Building - 9BUILCXNG                              -    $3.503.250
   3.   FWd PanMimal Training A Hiring « $FIELDPER8              -      9751.256
   4.   Fadtty Preparation ft Equlpmanl • 9EQUIP                  -    97.757.000
   5.   Land Purchase tRfrSate Fee -           6.0%
                                  x 9LANDAQ K 2- 9LNDFEE  •     91.463.616
   6.   TOTAL Start-Up AdmMstrattw Costa - 9AOUIN               •    92.776.125
   7.   SUBTOTAL  -  9FIELOPER8 « 9EOUIP + 9ADMIN - 98TARTUP-   912.747.W1

   6.   TOTAL Initial Costs • ILANOM) • IBUILMNO * 9STAftTUP    •    920.446.041
III.  RECURRING COSTS TO PROGRAM OPERATOR

A. PWMIWMI Cods

   1.   CMbalO«c«Opintt« Staff

         PMlUonAna
       TachnlcalOMMra
       Data Ana»ilaff»a»itkal Staff
       Ctorical and Swralarial Staff
       SUBTOTAL (salary K paraon yaan)
       Otrartwad « Frin0*
           TOTAL
          1
          4
          2
          a

      50.0%
9C8TAFF-
               Annual Salaiy
              O40lmnwaak
                                                         2.
                                                              SataUla CflcM Opamtlng Staff
                    937.500
                    922.500
                   9467.500
                   9243.750
                   9731.250
Number
Position Area PorCttce ,
Area Administrator 1
Technical OAcera 2
Data Anatysh/Stetlttleal Staff 1
Clerical and Secretarial Staff 3
SUBTOTAL (salary x person years)
Overhead * Fringe 50.0%
Total per SataWe Offfce
TOTAL 9SSTAFF •
Annual Salary
Q40hr*/wael
J62.5OO
952.500
937.SOO
922.500
9272.500
9136.250
9406.750
91.226.250
       a. Station Managara;
       b. AatMant Station Uanaganv
       c. InapacttonTachnlctam:
       d. CtMtomarSantoaRapK

       a. SUBTOTAL

       f. OwaihaadandFi«n0aBanaffl>

       g TOTAL Inapacttan Fadtty
          Oparaling Staff
9 salary x Mad x  FAC
9 salary x Mad x  FAC
9 salary x Mane x  LAN
       x Mad x  FAC
        30.0%


      9FSTAFF
                               9693.750
                               9499.750
                             93.996.000
                               9560.000

                             95.771.500

                             91.731.450


                             97.502.960

-------
                  •Hanaou* Total Rooming Coata
                   OmcoRanlal
r


2.
a
•).
4.
S.
0.
7.
a.
8.


in
IV*
11.
.
13.
14.
•CSTAFF x $RENT«2 x 12 - 128.000
•SSTAFF x $RENTA2 x 12 • $49.360
TOTAL RENTAL COSTS •
Support SMVfcM to FacNUoB
Baalcpwladllty - (12.000
Additional par tana •
TOTAL SUPPORT SERVICES •
op«ff«*m'Bnniaai
N2Gaa • $30/250 Totto • $105.945
Cal.Oaao*- $240/2 moJIana • $57,800
Mac. SmpkM - $250/hwa - $10.000
I8UPP •
TravJ $1.000 /hoi x FAC •
PubSc &ksms*SK m $RPI or $RPVMli.HAAP -
Equipment UaMananca
10% M ($EQUIP - $PREP) -
Anmial Program Oaalgn. Eng. A EvaL
%ROE M ($LANOAO » $BUILOU4G * $EQUIP)
Of $RO£AMh. x AAP •

limmnoa Coite (whom apptobla)
•BuNdingt'fate • 0.3%
•OaooConlonM'rala - 4.5%
Taal/Oate Pros. Equip.* rate - 0.4%
•BuMtag^eoat • $2.571
*OMc« Cont^rtt* cult • Ht.^ftft
ToftfOala Prac. Eojulp.* co* • $27!a61
TOTAL INSURANCE COSTS -
$C8TAFF * $88TAFF * $F8TAFF •
Additional Uachanfc Training (ttala program only)
MKM1.000K(AAfM>OP/FREQyi.OOOx$MCH •
Pvoparty TaMM (44Abadftf prooram onry)
PRT x ($BUILO1NG » $LANQAQ * $EQUIP) •
Hiring and TiaMng Coats Qua to Employaa Tumovw
20% x ($FIELOPERS * $CHIRE)
TOTAL RacuolnQCoata - $RECUR -

$73.440

$240.000
$172.045
$20.000
$3.585.380
$731.700
$23.457
$87.788
(% of Initial
Capital Coat)


$38.822
$8.480.450
N/A
$234.571
$188.678
$14.848.041

-------
           IV.  ANNUALIZEO COSTS TO PROGRAM OPERATOR
           A. AvangaRacwrlng&ateAccoui^ (or Inflation
1.
2.
3.
4. "
5.
7.
8.
g.
10.
11.
12.
13.
14.
OAcaRantal
Support Sarvfcaa
Oparattng SuppMaa
Traval
Public Information .
Annual Program Daalgn. EnghmrtnQ • Evaluation
Compute Procaaalng of Taste
Panama! Coat*
Proparty Taxaa (contractor pwgnm)
Hiring and Training Coats
TOTAL Racurring Coats
9RENTI
9SUPTI
9SUPH
9TRAVI
9APUBI
9AOESOI
9CMPRI
9INSI
9STAFFI
9AMTI
9TAXI
9CHIREI
9RECURI
979.SSS
9259.983
9187.348
921.665
93.694.758
92M10
995.097
13BB72
9 JV.VPV •
910.246.170
9254.102
9183.806
916.082.189
                                                                                                                For aach coat.        PRL        I

                                                                                                                           Coat x | SUM  (1 + INF) ] / PRL
                                                                                                                                     1-0
                                                                                                               (doaa not Includa amortisation of Initial costs)
           B. Amortization of InMW Coats
T
oo
              1.
2.
     Vakia of Item Ramab*«
     at and of Program
Oanaral Formulas

  VAE  • 9TTEM x (OEPR • PRL) / OEPR
(tMtwra D£PR to dtfmdfation pMtod)

                      PRL
Praaant Otecountad Vatua of (Urn
RamaMng at and of Program      POV •  VAE/(I

VakiaofPrtndplatobaPald       PRIN  • 9ITEM - PtjV
Of ovar Langth of Program

Annual Paymant of InUal                             PRL       PRL
LoanPtualntaroat       9PMT • PRIN x INT (1 * INT) / (1 * INT) -1

(Note: Ir^aat Rate Ak«aoV Account* for Matkm)
                                Catenation fef WUal SpacHo Coat Etomanto

              1.    LandAcquialllon:
                        Bu^SalOtocounlRala   -         204%
                        (RaMkmVAE • (1 • Mac Rata)« 9LANDAO)
                        Oapractatton parted 
-------
V. DISTINGUISHING STATE FROM CONTRACTOR OPERATED PROGRAM

A. State Operated Program

                     btepecttort Fee          -  IANNUAL/TST  -    N/A
VI.  PROGRAM COSTS TO STATE (Ca

A. MB

   1.   MM Teat Cat I
               1
       ^SUBTOTAL

           TOTAL

   3.   TaatCalStaiMlp
                                     OS
                                     0.1
                                     at
                                     at
                             (MTSTM
CeA
  997.800
  •39.000
  132.000
  »27.500
  (20.700
  tIMM
         tatViMcliiTaated • «MT8T    -
    fMTSTS • IMfTST « T8T            -
           CCCLtS • MTSTS / TPCVS
«.  Coat af Teat Cell
    StagteTeat Cat Caat -  (MBtDfl
    IMBinflf  • IMtlDO H  0CEU8
                                                        am
                                                         •71
                                                     OOOjOOO
d.  Coat af Teat <
    8ln01a Teat Cat Caat •  (MEOUtP       •
    IMTOIlim •  IMEOUa?  M «CELL8    •

a.  TatalCaoMCaat
    (MCAP • (MBLOOS * (MEOUP8
       f.  MUalTaatCal
                      Coat • IMT8TW -
       0.  HMng and TnWng Coat tor Teat Cel Paraonnal
           •MPERS i        (2.000 SEmptoyee   -
                                                     (400.000
                                                     (400,000
(000.000

 (40.369


  (d.OOO
       h.  TOTAL Slart-Up Teal Cal Coata ((CEUJN)
                                                                                          B.  Contractor Operated Program

                                                                                  1.  Total AnmialMd Coata to Contractor • (ANNUAL      -
                                                                                  2.  Contractors Net Rat      10 O*           .
                                                                                  3.  TOTAL AnnuaHzad Contractor Program Coals • (CONTR -
                                                                                  4.  ArmuaazedContractorCotlparVehicle- (CONTR/TST-
                                                                                                                                              (21.096.102
                                                                                                                                               »2.108.6IO
                                                                                                                                              923.205.712
                                                                                                                                                   12643
                                                                                                 Recurring TM| Cal Paraonnal
                                                                                                 PcsSSonAfes
                                                                                                 Supervtaor
                                                                                                 Bnirlalrt 3
                       Person Years    Cost
                                1    $37.500
                                I    $35.000
                                1    (32.000
       11                        1    $27.500
 SUBTOTAL   (MIPERS-)          4   (132.000
Overhead A Fringe               86%   $125.400
   TOTAL               IMT3TAN "   (257.400

Teal Cal Recurring Coata

a.  Recurring Teat Cal Para. Coat - (MTS   (257.400
                                                                                                b. Recurring Vehicle Recruamenl CoeU
                                                                                                    Racn*nantCoat/VeNclo  - (RCTM
                                                                                                    Annual Coal - (RCTMT X M4TST
                                                                                                c.  Ra
                                                «aflnUalTeatCelCoat •  %MMAI
                                                Annual Coat - (MMAINT x (MCAP

                                            d.  TOTAL Recurring Tail Cel Coat*
                                                                                                                                  (0
                                                                                                                                    (120.000

                                                                                                                                    (377.400

-------
                   Amortization «l InkM Tact Cal Costs to 8taU

                                            PHL         PRL
                       tCELUN « INT (1* INT)  /  (1 * INT)   -1  - $CIPT -        $182.967
                                                               tCLANN  •       $406.834
                                                                                             Taat Call Portion Of Annual Faa to Motorist*

                                                                                             •. Taal Cal •Irapacllon Fae" - $TOTL / TST
                                                                                         toes
Rooming Taat CM Coate to State Accounting hr
                PftL-1         I
    8CELLAN i I  SUM (1»INF) ) / PW.
                   TOTAt AiMMlMd TMl

                       $8TPT » MANN
              1.
                                            9STATE
                                                             Amud Safety
r
»-•
o
0.4
OJ
0.1
0.1
0.1
0.0
0.1
0.1
0.1
0.1
                       TOTAL
    $87.500
    $82.500
    $62.800
    $27.500
    $37.500
    $36/MO
    $30.000
    $37.500
    $22.500
    $25.000
    $20.000
   $123,500
($CAP1N)
              2.
                                     iccomr

                   a.  MMalTaalCa8Caato

                   I. TOTALMMCoatitoSMa
                             • 8CAPM * SBEMN-

                              • HMftJ  •  $FTRM   •


                                « 1.000) K $MCH


                                   $0.000 M8JTAFF

                                   •  $CEIUN

                                       •  $8TIN
                 $571.211
                                                                 86%
                                                          O»art»ad
                                                           A Cf^u^k
                                                              $84.125
                                                              $23.750
                                                               $8.875
                                                               $2.613
                                                               $3.563
                                                               $3.325

                                                               $3.563
                                                               $2.138
                                                               $2.375
                                                               $1.800
                                                             $117.325
                                                          ($B£NIN)
                 $240.825

                  $37.250


               $5.709.758


                  $85.000

                 $848.385

               $8.885.188
                                      Duration of  OtradCoilof     Total
                                     iMtiuctton (hr»  Iraliuction  TraMng Coala
40
40
60
80
40
80
80
80
40
40
40
                                                                                                                      $800
                                                                                                                      $400
                                                                                                                      $400
                                                                                                                      $200
$400
$400
$100
 $50
 $50
 $1.6
 $1.602
 $2.418
 $2.815
   $821
 $1.748
 $3.108
 $1.842
   $533
   $531
   $435
$16.150
HWng
Co»J»
$10.000
$5.000
$3.000
$100
$100
$200
$100
$200
$100
$50
$50
$16.800

Total Training
Plus titling
$11.606
$6.602
$5.4 18
$3.115
$1.021
$1.846
$3.308
$2.042
$633
$561
$485
$37.250
($FTRN)

-------
C. Recurring CocU

   1.   CtnlnlAdmln
        PMttkMiAra*
       Program AdmMrintor
       Deputy AdmMMrator
       AUo EmlMloM Oyim
       EmlMloM TM*«
                                              AftnudS^wy
                                            $07.900
                                            $02.900
                                            $52.900
                                            $27.900
                                            $37.900
                                            $39.000
       8*cnUiy
       CMtfTypM
         SUBTOTAL^     OMTAFF)*          13

          TOTAL                  $CAPAN

  2.   ToMftocunlngCMtotol

       a  C«MM/

       b.
                                            $29.000
                                            $20.000
                                           $479.000
                                           $451.250
                                           $•20.250
   1.
               MCH « (AAP-POFYFMeQyi.000 x Mwttl.000




       •.   Tiwri -                        $300   IfSTAFF -

       _   CjM^k f^^^H^M^M^^Oi •              f AK M ftf^EAIlIO •


       k.  ConfcMlMiativteM                              -
                                                $STRC
                     PRL        ML
    $STW • INT(1*INT)  / (1*MT)  -1   - $STPT
                                          $SANN
J.  TOTAL lUeuntaBCMto to
                    PW.-1         I
           $8TMC •  | SUM (1*IMF) ) / PW.
                     M
       AiMMNnd PngtMi COM* to OM* fCwMtad Part Only)

          Mm* land 2 (rt«M)                   » $STATE
                                                        $310.432

                                                        $3n.400

                                                          $3.000

                                                          $0.900
   $3.729

$2.134.207
                                                             $1.070.059
$2.311.010
                                                      $3.000.705

-------
VII.  OVERALL PROGRAM COSTS FOR CONTRACTOR OPERATED PROGRAM
A. AnMMHMd Program Co«te
    tSTATE + SCONTM + RS +NoMfc* Admin «P«*MMf  -«TOTL-    t30.fi67.aiC
B.  AnnuriFMtoMolMM(AM«n«»AMMMlPiegmmCMlpirC«nlralndTMl)
                                         fTOTUTST  •        $45.52

-------