ANALYSES  OF TRAFFIC AND AIR QUALITY.
 TRENDS,  TCP EFFECTIVENESS, AND A
VOLUNTARY I/M PROGRAM IN WASHINGTON
           AND OREGON
                       GCA/TECHNOLOGY DIVISION

-------
                    Prepared for

           ENVIRONMENTAL PROTECTION AGENCY
                      REGION X
                  1200 Sixth Avenue
              Seattle, Washington 98101
Project Officer:  Betty A.  Wiese,  Air Programs  Branch
               Contract No. 68-02-1376
                  Task Order No.  13
        ANALYSES OF TRAFFIC AND AIR QUALITY.
         TRENDS, TCP EFFECTIVENESS, AND A

        VOLUNTARY I/M PROGRAM IN WASHINGTON
                    AND OREGON
                 DRAFT FINAL REPORT
                         by

               Rebecca Co  Galkiewicz
                 Maureen To  0°Berg
                Robert Mo  Patterson
                  Frank A« Record
                   GCA CORPORATION
               CCA/TECHNOLOGY DIVISION
               Bedford,   Massachusetts

-------
                            DISCLAIMER
This report was furnished to the Environmental  Protection Agency by the
GCA/Technology Division in fulfillment of  Contract  Number 68-02-1376,
Task Order No. 13.  The contents of this report are reproduced herein
as received from the contractor.  The opinions, findings  and  conclusions
are those of the authors and not necessarily those  of the Environmental
Protection Agency.

-------
                              CONTENTS

                                                                   Page

List of Figures

List of Tables

Acknowledgements

Sections

I      Introduction                                                1-1

II     Recent Trends in Traffic and Air Quality Measurements        II-l
       (In Preparation)

III    Impact of Transportation Control Measures                   III-l

IV     Statistical Analysis of Emissions Data from the  Portland     IV-1
       Inspection/Maintenance Program
                                               GCA/JECHNOlDGY DIVISION

-------
                                 FIGURES

No.                                                                Page

IV-1      Sample Age Distribution                                  IV-7

IV-2      Sample Make Distribution                                 IV-8

IV-3      Sample Weight Class Distribution                         IV-9

IV-4      Sample Engine Displacement Distribution                  IV-9

IV-5      Sample Cylinders Distribution                            IV-10

IV-6      Sample County Distribution                               IV-10

IV-7      Carbon Monoxide Mean Emissions                           IV-18

IV-8      Hydrocarbons Mean Emissions                              IV-19

IV-9      Carbon Dioxide Mean .Emissions                            IV-20

IV-10     Relative Frequency of Hydrocarbons (Idle Mode)t           IV-23

IV-11     Relative Frequency of Carbon Monoxide (Idle Mean)         IV-23

IV-12     Relative Frequency of Log Hydrocarbons (Idle Mode)        IV-24

IV-13     Relative Frequency of Log Carbon Monoxide (Idle  Mode)     IV-25

IV-14     Carbon Monoxide (Idle Mode) by Age                       IV-31

IV-15     Hydrocarbons (Idle Mode) by Age                          IV-32

IV-16     Carbon Monoxide (Idle Mode) by Make                      IV-33

IV-17     Hydrocarbons (Idle Mode.) by Make                         IV-34

IV-18     Carbon Monoxide (Idle Mode) by Control                   IV-35

IV-19     Hydrocarbons (Idle Mode) by Control                      IV-36

IV-20     Carbon Monoxide (Idle Mode) by Mileage                   IV-37

IV-21     Hydrocarbons (Idle Mode) by Mileage                      IV-38

IV-22     Hydrocarbons Mean Emissions - Sample, Before Retest,,     IV-40
          Retest
                                 iii
                                               GCA/TECHNOLOGY DIVISION

-------
                            FIGURES (continued)




No.                                                                Page
IV-23
IV-24
IV-25
IV- 2 6
IV- 2 7
IV- 2 8
IV-29
IV-30
IV-31
IV-32
IV-33
IV- 34
IV-35

Carbon Monoxide Mean Emissions - Sample, Before
Retest, Retest
Hydrocarbons Mean Emissions - Sample, Before Retest,
Retest and Repair Cost < 0
Carbon Monoxide Mean Emissions - Sample, Before
Retest, Retest and Repair < 0
Cumulative Frequency of Carbon Monoxide (Idle Mode) -
Sample, Before Retest, Retest
Cumulative Frequency of Hycrocarbons (Idle Mode) -
Sample, Before Retest, Retest
Cumulative Frequency of Carbon Monoxide (Idle Mode)
by Control
Cumulative Frequency of Hycrocarbons (Idle Mode by
Control
Cumulative Frequency of Carbon Monoxide (2500 rpm) by
Control
Cumulative Frequency of Hydrocarbons (2500 rpm) by
Control
Cumulative Frequency of Carbon Monoxide (30 mph) by
Control
Cumulative Frequency of Hydrocarbons (30 mph) by
Control
Cumulative Frequency o'f Carbon Monoxide (50 mph) by
Control
Cumulative Frequency of Hydrocarbons (50 mph) by
Control
iv
IV-41
IV-42
IV-43
IV-45
IV-46
IV-47
IV -4 8
IV-49
IV-50
IV-51
IV-52
IV-53
IV-54

                                               GCA/TECHNOLOGY DIVISION

-------
                                  TABLES

No.                                                                Page

III-l     8-Hour VMT Estimates,  Seattle CBD                        III-5

III-2     8-Hour CO Emission Estimates, Seattle CBD                III-ll

III-3     Fraction of Annual Miles Driven by Vehicle Age for        111-12

III-4     1976 CO Emission Estimates for Light-Duty Gasoline        111-14
          Vehicles by Model Year,  Seattle CBD

III-5     8-Hour VMT Estimates,  Spokane CBD                        111-15

III-6     8-Hour CO Emission Estimates, Spokane CBD                111-18

III-7     Fraction of Annual Miles Driven by Vehicle Age for        111-19
          Spokane, Washington

III-8     1976 CO Emission Estimates for Light-Duty Gasoline        111-20
          Vehicles by Model Year,  Spokane CBD

III-9     8-Hour VMT Estimates,  Portland CBD                       111-22

111-10    8-Hour CO Emission Estimates, Portland CBD               111-25

III-ll    Fraction of Annual Miles Driven by Vehicle Age for        111-26
          Portland, Oregon

111-12    1976 CO Emission Estimates for Light-Duty Gasoline        111-27
          Vehicles by Model Year,  Portland CBD

IV-1      Vehicle Data Format                                      IV-3

IV-2      Data Transformation                                      IV-5

IV-3      Sample Age by Make Distribution                          IV-11

IV-4      Age Distribution of Sample and Population                IV-12

IV-5      Make Distribution of Sample and Population               IV-12

IV-6      Control By Make Distribution of Sample and Population    IV-13

IV-7      Age Distribution of Sample and Fail Subsample             IV-15

IV-8      Make Distribution of Sample and Fail Subsample          ,IV-15
                                               GCA/TECHNOLOGY DIViSiON

-------
                            TABLES (continued)

No.                                                                Page

IV-9      Weight Class Distribution of Sample and Fail Subsatnple   IV-16

IV-10     Cylinder Distribution of Sample and Fail Subsample       IV-16

IV-11     Control Distribution of Sample and Fail Subsample        IV-16

IV-12     Fail Criteria                                            IV-17

IV-13     Weighting for Age                                        IV-26

IV-14     Weighting for Make                                       IV-27

IV-15     Weighting for Control by Make                            IV-28

IV-16     Analysis of Variance                                     IV-30

IV-17     T-Test                                                   IV-30

IV-18     Failure Rate and Failure Levels                          IV-55

IV-19     Stepwise Regression                                      IV-57

IV-20     Correlation Coefficient Matrices                         IV-58

IV-21     Engine Tampering                                         1V-59

IV-22     Under Hood Tampering                                     IV-60

IV-23     Correlation of FTP and Short Test Procedures             IV-65

IV-24     Estimation -of Idle Emissions Reductions for a 50         IV-67
          Percent Failure Rate
                                 VI
                                               GCA/TECHNOLOGY DIVISION

-------
                          ACKNOWLEDGEMENTS

2 wish to acknowledge the assistance of Alan M,  Voorhees,  Inc.  who
:ted as subcontractor to GCA/Technology Division arid,  under the direc-
Lon of Mr0  Keith Gilbert, provided traffic  related data and analyzed
is effect of the transportation control measures on VMT and speed.

3. Betty Wiese,  Project Officer,  provided continuing guidance and
ssistance throughout the program.   We wish  to thank her and all cooper-
ting agencies for supplying the air quality and  meteorological  data.
inally, we would like to acknowledge the Oregon  Department of
ivironmental Quality, and especially Mr. William Jasper,  for providing
ata and insight  on the voluntary inspection/maintenance program in
srtland.
                              vii
                                            GCA/TECHNOLOGY DIVISION

-------
                              SECTION I
                            INTRODUCTION

This report describes work carried out to assist the States of
Washington and Oregon in developing background information for the
preparation of Transportation Control Plans to achieve air quality
standards in Seattle, Spokane, and Portland.  The report has been
organized into three major sections, each covering one of the princi-
pal work areas of the study.

Section II presents an analysis of vehicle miles of travel (VMT) and
related transportation statistics to determine recent trends and the
degree to which these trends have been reflected in changes in air
quality (carbon monoxide concentrations) as measured by ambient moni-
tors in the three cities.  Apart from the general requirement to docu-
ment these changes, it was hoped that such an analysis would provide
insight into the adequacy of current monitoring procedures for the
evaluation of the effectiveness of transportation control strategies.

Section III provides estimates of the effect of implementing various
transportation control strategies on 1976 carbon monoxide concentra-
tions within the central business districts (CBD's) of the three cities,,
These estimates are made by predicting the changes in VMT and average
speed expected to result from each strategy, and then converting the
resulting VMT-speed data to carbon monoxide emissions.  Ambient concen-
trations of carbon monoxide within the CBD's are assumed to be directly
proportional to emissions.
                                1-1
                                              GCA/TECHNOLOGY DIVISION

-------
Section IV presents the results of a statistical analysis of emission
data collected from a voluntary inspection/maintenance program in
Portland.  The emission testing was performed by the Oregon Department
of Environmental Quality.  Emission data from idle, 2500 rpm, 30 tnph
and 50 tnph loaded dynamometer tests were available for about 6000
vehicles.  The analysis includes development of basic statistics of the
sample, significance tests of various sample parameters, and a compari-
son of the sample distributions of age and make against those of the
entire population of the Portland area.  Cumulative frequency distribu-
tions of emissions broken down by control group are presented as an
aid to setting pass/fail criteria.  Only 100 vehicles were identified
as having had inspection, maintenance, and a second inspection,  and no
definitive conclusions could be drawn regarding post-inspection/
maintenance emission levels.
                                1-2
                                             GCA/TECHNOLOGY DIVISION

-------
         SECTION II

RECENT TRENDS IN TRAFFIC AND
  AIR QUALITY MEASUREMENTS
      (In preparation)
           II-l
                         GCA TECHNOLOGY DIVISION

-------
                              SECTION III
               IMPACT OF TRANSPORTATION CONTROL MEASURES

METHODOLOGY

The transportation control measures evaluated in this section are
grouped into basic packages comprised of those measures currently
being implemented in each city, plus other selected measures  under
active consideration.  The selection of measures for evaluation was
made by EPA, Region X, in consultation with the Department of
Ecology of the State of Washington and the Department of Environ-
mental Quality of Oregon.  The basic package of strategies for each
city includes signal improvements, transit improvements, carpooling,
and parking management.  An inspection and maintenance program is
also considered part of the basic package for Portland, as is the
use of skywalks in Spokane.

The impact on VMT and speed of the measures included in each  strategy
package has been estimated in a cumulative manner.  Then, the addi-
tional measures have been added one at a time, and the combined effect
of the package plus each individual strategy has been estimated.  The
analysis of the control measures is described separately for  each
city, although the analysis techniques and some of the factors applied
are common to more than one area.  These analyses result in estimates
of VMT by speed and vehicle classification and thus provide requisite
inputs to emission calculations.  Initially, the VMT for each city
was estimated by various speed levels, as shown previously in Table
II-   .  To simplify the emissions calculations needed in this sec-
tion, weighted average speeds have been calculated for the lower
                              III-l
                                              GCA TECHNOLOGY DIVISION

-------
speed categories shown in Table II-  .  Also VMT have been split into
vehicle classifications used by EPA in its current emission calculation
procedures.  VMT estimates are for the peak 8-hour traffic period of
the day.

It must be kept in mind that this evaluation is undertaken from a
"sketch planning" approach rather than through detailed analysis.
Each central business district is treated as a unit and is not sub-
divided by activity groupings such as land use.  Estimates of the
effects of strategies are made using general factors, guideline values,
and rules of thumb.  In some instances, limited confidence could be
placed in some of the basic counts and other input data.  Because of
these limitations, it is, in our judgment, appropriate to  use the re-
sults of these analyses for general planning purposes only.   The actual
effectiveness of strategy packages can only be ascertained by care-
fully designed monitoring programs which provide data for   before -
and- after  comparisons.

The emission calculations were carried out using procedures  described
by EPA in the following two draft documents, dated October 1974:
(1) Supplement No. 5 for Compilation of Air Pollutant Emission Fac-
tors, Second Edition, and (2) Attachment No. 1 for Compilation of
Air Pollutant Emission Factors, Second Edition, Projected  Mobile
Source Emission Factors,  prepared by Kircher and Masser.  These draft
documents were made available for use prior to publication in AP-42
so that the latest emission factors and recommended calculation pro-
cedures could be incorporated into the strategy evaluation methodology
employed in this study.  The overall methodology used in calculating
carbon monoxide emissions is nearly identical with that outlined by
Kircher and Armstrong as  presented in AP-42 and so is not  described
in detail here.  The principal procedural changes incorporated in the
revised methodology are:
                              III-2
                                              GCA/TECHNOLOGY DIVISION

-------
         A redefinition of vehicle class.  The revised classifi-
         cation is as follows :

         « Light-duty gasoline  vehicles (passenger cars)

         o Light-duty gasoline  trucks (gross weight of 8,500
           pounds or less)

         o Light-duty diesel vehicles (passenger cars)

         « Heavy-duty gasoline  vehicles (gross weight over
           8,500 pounds)

         e Heavy-duty diesel vehicles (trucks and buses)

         e Motorcycles

         The use of separate sets of emission factors for each
         calendar year instead  of the use of deterioration
         factors.  Emission factors are tabulated by model year
         for various calendar years for light- and heavy-duty
         gasoline-powered vehicles.

         The use of separate speed correction factors for
         different model years.  Speed correction factors are
         presented mathematically by model year for light-duty
         gasoline vehicles operating between 15 and 45 miles
         per hour.  It is recommended that, until other data
         are available, these factors be used for light-duty
         trucks.  It is also recommended that pre-1968 light-
         duty vehicle correction factors be used for pre-1970
         heavy-duty vehicles, and that 1968 light-duty ve-
         hicle factors be used  for post-1969 model years of
         heavy duty vehicles.  These recommendations have been
         followed in calculating the emissions presented in
         this section.
The fraction of total vehicles in use by age was calculated using
passenger car and truck registration data compiled by R.L.  Polk and
Company for the three urban areas.   King County data were used for
Seattle; Spokane County data were used for Spokane; and the data for
the counties of Clackamus, Multnomah, and Washington were combined
for Portland.  Passenger car registration data were used directly in
the calculations of light-duty gasoline vehicles for the years 1971
and 1972.  Average value's for 1971, 1972, and 1973 were used for the
1976 calculations.  The Polk statistics do not separate light-duty

and heavy-duty trucks.  The procedure adopted for trucks was to use
                              III-3
                                              GCA/TECHNOLOGY DIVISION

-------
the Polk truck data for calculations of light-duty gasoline truck
emissions, and the nationwide statistics for heavy-duty vehicles
given in the draft copy of Supplement No. 5 for Compilation of Air
Pollution Emission Factors.  Nationwide statistics of average annual
miles driven by vehicle age have been used in all cases.

This analysis has not been prepared in the form of a Transportation
Control Plan.  Its purpose is only to estimate the emission reductions
that can be expected from strategy packages currently being imple-
mented plus probable reductions from other selected strategies.   It
does not provide a total cumulative estimate of emission reduction or
discussion of social, economic, and other impacts.

SEATTLE

VMT Estimates
Table III-l lists the estimated 8-hour VMT for Seattle's CBD after
implementation of the various transportation control measures.   The
1971 and 1976 totals are based upon daily estimates made by the city
in 1974 for the area bounded by Virginia Street, 8th Avenue, 1-5,
Yesler Way, and the Alaskan Way Viaduct.  The growth trend follows
that of the 2nd Avenue permanent counting station.   Speeds reflect
1974 survey data.  The breakdown by class is based  on registration
data, Origin-Destination survey data (including trip length), and  ve-
hicle classification counts.  In this class breakdown, lowest confi-
dence  must be attached to the motorcycle estimates.  The 1976 base is
an approximation of data along pre-energy crisis trends.  It does  not
represent an expected condition, but rather is a starting point for
estimating the changes expected by 1976 if the strategies take effect
by then.
                               Ill-4
                                              GCA/TECHNOLOGY DIVISION

-------
Tab le III-l.  8-HOUR VMT ESTIMATES,  SEATTLE CBD
1971 Base
i
!
1976 Base
t
SIGOP
Transit program
t
Car pool (max.)
*
Parking Management
— JB— Reduce Parking 20% (a)
(b)
»
— 3»- Replace 20% curb
with off-street
	 2»- Flex-time
— x»«- Road pricing
	 3°— Delivery Restrictions
— S»- I and M
— **- Exclude nonretrofit
' 	 *»— Predictive exclusion
VMT
Avg.
speed
(mph)
12.9
12.9
14.9
14.9
14.9
14.9
13.5
14.9
14.9
15.1
15.2
15.0

See

Light-
duty
autos
44,021
46,616
46,616
43,268
41,620
41,620
58,299
40,620
41,532
41,620
36,938
41,620

text

Light-
duty
trucks
3,988
4,303
4,303
4,303
4,303
4,303
4,303
4,303
4,303
4. ,3 03
4,303
2,797



Heavy-
duty
trucks
1,023
1,103
1,103
1,103
1,103
1,103
1,103
1,103
1,103
1,103
1,103
717



Diesel
1,074
1,159
1,159
1,159
1,159
1,159
1,159
1,159
1,159
1,159
1,159
753



Motor-
cycles
1,023
1,986
1,986
1,986
1,986
1,486
1,986
1,986
1,986
1,986
1,986
1,986



Total
51,128
55,167
55,167
51,820
50,171
50,171
66,850
49,171
50,083
50,171
45,489
47,873



                  III-5
                                 GCA/TECHNOLOGY DIVISION

-------
The following paragraphs discuss the rationale used in preparing the
VMT estimates presented in Table III-l.

SIGOP - The city's before and after studies did not detect average
speed changes but found sharp reductions in delay and numbers  of stops.
The survey method ("floating car") may not be sensitive to small speed
changes.  Therefore, the effect of this signal timing improvement is
taken as a 2 mph step in average speed.

Transit - METRO has embarked upon a transit improvement plan including
more buses, better levels of service, park-ride and express facilities,
downtown flow improvements, free downtown bus rides,  and other steps.
Success in terms of increased patronage is already seen.  A delay from
initial development schedules.is expected, however, largely due to
delivery of new buses at a time later than originally expected.

Previous calculations of transit impact on CBD VMT used estimates of
service improvements and mode use modeling techniques.   If the present
trends of increasing patronage hold until 1976, the previously esti-
mated VMT reduction will be achieved.  However, in view of probable
delays in bus deliveries, the present trends were truncated to reflect
a delay in 1975 program completion to about 1977.  This made a dif-
ference of about 7 percent in the VMT reduction due to transit as com-
pared to previous estimates.  The transit impact is estimated as an
overall reduction of about 6 percent from the 1976 base.

Car Pools - The Federal Department of Transportation has sponsored sig-
nificant car pool programs in nearly all urban areas.  These programs
have typically concentrated on contacting potential employee groups
and providing matching and publicity services.  Follow-up surveys of
effectiveness have not yet become available in sufficient quantity for
use in this study.  However, if an active program such as Seattle's is
maintained, the share of employees in car pools might be doubled.  If
                               III-6
                                              GCA/TECHNOLOGY DIVISION

-------
so, VMT's would be reduced by 3 or 4 percent.  Since influences such
as energy prices and availability can strongly affect car-pooling,
this estimate is probable close to a maximum reduction.

Parking Management - This strategy embodies two measures,  preconstruc-
tion review and a prohibition on increasing the CBD parking supply
above November 1973 levels.  Preconstruction review of parking pro-
posals should not affect VMT.  It is viewed as a means  of  monitoring
and enforcing a "lid" on parking supply.  For Seattle,  if no strate-
gies were implemented, the parking demand would meet the "lid" in
1975 or 1976.  However, effective transit and car pool programs should
reduce VMT  and parking demand in parallel, so that the closest
approach of demand to the lid may have already been reached.  There-
fore, no impact has been estimated for this measure.

Reduction in Parking Supply - If the "lid" on parking supply were
dropped, say to 80 percent of its present value, two effects would be
expected.  First, since demand for parking could not be expected to
drop in the short run and if only short-term parkers are affected,
additional search-for-parking VMT would be created along with idling
while waiting for a parking space.  Analysis of parking arrival and
accumulation patterns, as related to available supply,  shows nearly
17,000 VMT created at an estimated 10 mph.  (This is reflected in
Table III-l, estimate (a).)  On the other hand, scarcity of supply
could lead to higher costs.  It is estimated that an increase in park-
ing fees of 78 percent would counter-balance the new VMT through di-
version to transit or to'new destinations.

A reduction in employee spaces could be effected through deletion of
reduced rates for all-day parking.  A rough doubling of long-term
costs for employees could drop light-duty auto VMT from 41,621, to
40,621 increasing the average speed to 14.9 mph  (estimate (b)).
                              Ill-7
                                              GCA/TECHNOLOGY DIVISION

-------
Replace Curb Parking with Off-Street - Replacement of curb parking with
off-street could increase street capacity (and possibly speed), might
reduce the search-for-parking VMT if the new spaces are easy to locate,
and would be costly.  This measure was analyzed by assuming that 20
percent of the curb spaces were displaced in the most congested areas.
A 1-mph speed increase was assumed (the difference between peak and
off-peak) for those travelling at 15 mph for the peak 2 hours only.

Staggered Hours and Days - Staggered hours is referred to in this study
as "flex-time" denoting that employees would choose their own travel
hours xtfithin limits set by employers (as opposed to a rigid staggered
hours concept).  This approach is desirable from a labor relations
point of view and has been demonstrated to reduce peak vehicle and
transit loadings.  A 1-mph increase in peak hour speeds for employees
was assumed as a maximum impact of such a program.  VMT probably would
not be shifted significantly out of the critical 8-hour period.

No additional effect is estimated for "staggered days."  Present
"4-40" plans would impact Mondays and Fridays only unless major dis-
ruptions are imposed on employee work patterns.

Road-Pricing - This concept calls for some manner of charging for use
of roadways.  The charge could be higher during periods when demand is
high, and low when demand is low.  Methods of charging include bridge
tolls,' special vehicle licenses or stickers, increases in parking
costs, and various means of automatically detecting and billing.
Problems with the concept range from technology (in detection) to lack
of applicability for bridge tolls (impact too few CBD trips and too
many others).  Parking cost increases may not be effective unless of
a very large order.

To estimate a probable reduction from this strategy, it has been
assumed that the cost of auto use to and from downtown could be
                               III-8
                                              GCA/TECHNOLOGY DIVISION

-------
increased four times.   This assumption,  when used  with  a  mode use
model, results in an approximate 11-percent reduction in  automobile
VMT.

Daylight Delivery Restrictions - Prohibition of truck curb  use during
the day would reduce VMT, increase speeds, and cause disruptions  in
present business schedules.  About 35 percent of truck  travel in  down-
town may be associated with delivery and service.   This share of  the
three truck categories is deleted for this strategy and a slight
speed gain is estimated for the peak travel of remaining  vehicles.

Inspection and Maintenance - This measure does not affect VMT or  speed,

Exclusion of Nonretrofit Vehicles - Such a strategy assumes some  sort
of retrofit program in operation, a method for identification of  modi-
fied vehicles, and enforcement.  The effects of an exclusion would  de-
pend on the devices and model years specified.

Predictive Exclusion - This strategy would call for the exclusion of
classes or all vehicles from the CBD on days when an episode is pre-
dicted.  Exclusion could be based on vehicle type (such as  trucks),
vehicle age and equipment  (retrofit), CBD employment (with a license
or pass), or time of day.  The amount of exclusion required would de-
pend upon the demonstrated effectiveness of other measures being  used
and the predicted severity of the episode.  It could be achieved  by
adjusting the time period of exclusion and/or the type and number of
vehicles excluded.  Prohibitions on entering downtown from 10 a.m.  or
11 a.m. onward would allow employees in and might reduce VMT by 36
percent.  Prohibition starting earlier, say 6 a.m., would exclude em-
ployees and effectively shut down the CBD for the day.

Problems, other than economic impact, include the cost of the exclu-
sion operation  (barricades, detour, personnel, publicity, etc.),
                               III-9
                                              OCA/TECHNOLOGY DIVISION

-------
predictive ability, and potential traffic confusion and congestion
just outside of the CBD.

CO Emission Estimates

Table III-2 presents estimates of carbon monoxide emissions within the
Seattle CBD over the 8-hour period of maximum travel during the day by
vehicle class.  These estimates were made using the procedures des-
cribed earlier in this section, and the VMT-speed data given in Table
III-l.  Table III-3 gives the fraction of annual miles driven by
vehicle age data used in making these calculations.

It can be seen from these estimates that the strategy package currently
being implemented will fail by a wide margin to achieve allowable emis-
sions by 1976.  Specifically, if a 55 percent reduction from 1971 base-
line levels is required to achieve the allowable emissions level an
additional reduction of 1097 kg per 8 hours will be required beyond
those attained with the strategy package (i.e. 3482 kg - 2385 kg).
Table III-2 also shows that none of the additional strategies when
added singly  to the basic package, is sufficient to achieve the allow-
able emissions level.

The probable  effect of an inspection and maintenance program on carbon
monoxide emissions from light-duty automobiles has been estimated by
EPA to be a 10 percent reduction, assuming a 50 percent initial failure
rate and annual inspection using an idle emissions test.  The use of
loaded emissions tests has been estimated to achieve a 12 percent re-
duction in carbon monoxide emissions.  Current EPA policy, as expressed
in the draft  Attachment No. 1 for Compilation of Air Pollution Emission
Factors, October 1974, also allows a 10 percent reduction per pre-1975
light-duty trucks.  Since over 90 percent of emissions from light-duty
trucks will come from pre-1975 vehicles in 1976  (see Table III-4), an
overall credit of 10 percent  is a reasonable value to apply to light-
duty vehicle  emissions for an inspection and maintenace program.  This

                             111-10
                                              GCA/TECHNOLOGY DIVISION

-------
                                 Table  III-2.   8-HOUR CO EMISSION ESTIMATES, SEATTLE CBD
0
O
O
z
Q
b
g
O
     M
     M

1971 Base
1976 Base
SIGOP
Transit Program
Car Pool (max.)
Parking Management
Reduce parking 207» (a)
(b)
Replace 20% curb with
off-street
Flex -time
Road pricing
Delivery restrictions
I and M
Exclude nonretrofit
Predictive exclusion
Emissions (kg/8 hours)
Light -
duty
autos
4,407
3,454
3,100
2,877
2,768
2,768
4,180
2,701
2,762
2,739
2,419
2,754

See tex

Light -
duty
trucks
522
395
355
355
355
355
383
355
355
351
350
228

t

Heavy-
duty
trucks
309
290
264
264
264
264
281
264
264
260
260
171



Diesel
31
33
33
33
33
33
33
33
33
33
33
22



Motor-
cycles
32
62
62
62
62
62
62
62
62
62
62
62



Total
5,301
4,234
3,814
3,591
3,482
3,482
4,939
3,415
3,476
3,446
3,124
3,240



Additional
reduction
needed

1,849
1,429
1,206
1,097
1,097
2,554
1,030
1,091
1,061
739
855



NOTE:  Allowable emissions, assuming a 55 percent  reduction from 1971 base  levels,
       are 2,385 kg/8 hours.

-------
figure assumes that all such vehicles entering the CBD are covered by
the program, however, and must be reduced appropriately if' this is not
the case.  Calculations based on the estimates provided in Table III-2
and a 10 percent reduction in emissions from the first two vehicle
classes show that an inspection and maintenance program will provide,
very roughly, one-third of the reduction required beyond that achieved
by the combinations of strategies analyzed.  It should be noted that in
the case of the parking reduction strategy (a), emissions are estimated
to increase substantially as a result of increased search time.  In
this case, an inspection and maintenance program would provide only 18
percent of the additional reduction required.

Table III-3.  FRACTION OF ANNUAL MILES DRIVEN BY VEHICLE AGE FOR SEATTLE
Vehicle age
0
1
2
3
4
5
6
7
8
9
10
11
12
>13
Light -duty autos *
1971
0
0.063
0.108
0.128
0.119
0.108
0.104
0.089
0.067
0.057
0.043
0.025
0.021
0.068
1976a
0
0.089
0.116
0.107
0.106
0.105
0.097
0.086
0.073
0.058
•0.044
0.030
0.023
0.066
Light -duty trucks
1971
0
0.056
0.114
0.147
0.121
0.108
0.089
0.065'
0.047
0.039
0.030
0.019
0.020
0.145
1976a
0
0.097
0.122
0.107
0.106
0.103
0.087
0.071
0.054
0.041
0.030
0.022
0.019
0.141
Heavy-duty vehicles
1972b
0
0.130
0.133
0.123
0.124
0.085
0.068
0.058
0.050
0.040
0.031
0.024
0.017
0.117
 Average for calendar years 1971,  1972,  and 1973
 Nationwide
                               111-12
                                             GCA/TECHNOLOGY DIVISION

-------
Table III-4 provides a breakdown of emissions by model year for the se-
quence of strategies shown in Tables III-l and III-2.  Estimates of the
reduction likely to be achieved by selected retrofit and exclusion mea-
sures can be worked out by means of this table.

SPOKANE

VMT Estimates
Table III-5 presents the estimated 8-hour VMT for the Spokane CBD,  de-
fined for this study as the area bounded by Spokane Falls  Boulevard,  3rd
Avenue, Monroe Street, and Division Street,  This area includes  con-
siderable through traffic, perhaps half of the total.  The analysis for
Spokane was approached in the same manner as for Seattle,  and so many
of the same general comments apply.  The rationale used in preparing  the
VMT estimates is discussed in the following paragraphs.

Signal Improvements - By the end of 1975, it is likely that all  CBD in-
tersections will be controlled by the  ity's computer system. Upon im-
plementation of this system for the present 68 intersections, improve-
ments in flow were observed.  The previous estimate of a 2-tnph speed  in-
crease was applied in this analysis.

Transit - Interim'park-ride facilities proved to be very successful
during EXPO, and a grant application has been submitted for permanent
development.  The extension of present transit patronage trends  to  1976
shows a reduction of some 3 percent in VMT for the 1976 base.

Car Pools - Although the enthusiasm (and potential) for carpooling  in
Spokane is less than in Seattle, a similar doubling of present employee
pooling forms a frame of reference.  An increase in average auto occu-
pancy from 1.19 to 1.24 is assumed as a measure of a successful  program
in Spokane.
                                111-13
                                               GCA/TECHNOLOGY DIVISION

-------
Table III-4.  1976 CO EMISSION ESTIMATES FOR LIGHT-DUTY
              GASOLINE VEHICLES BY MODEL YEAR, SEATTLE CBD














Model year

Pre-1968
1968
1969
1970
1971
1972
1973
1974
1975
1976
Total

Pre-1968
1968
1969
1970
1971
1972
1973
1974
1975
1976
Total
0)
bO
ca

O
ca
ex

»>»
bO
0)
4-)
ca
M
4-1
CO

r\
bO
a
*H *^*s
A! cd
S-i v-'
n)
ex a
o
(1) T-l
O 4J
3 ex
*"O O
0)
Pi

*%
M
a
•H f~ N

,M *— '

ex el
o
a) -H
O 4-J
P ex
T3 O
ai
' ^
te 4-1
3 4-i
o o
cd

ex 4J
a) -H







, 4J
S-< O
a) -H
[> M
•i-i i ?
r— 1 CO
<3J CU
P Pi
Light -duty autos
796
228
292
295
318
254
245
236
61
43
2,768
1,187
342
444
448
483.
386
372
359
94
65
4,180
111
222
285
288
310
248
239
230
60
42
2,701
794
228
291
295
317
253
245
235
61
43
2,762
790
226
288
291
315
251
242
233
61
42
2,739
698
199
254
257
211
222
214
206
54
38
2,419
793
227
290
293
316
253
244
234
61
43
2,754
Light -duty trucks
138
17
23
25
31
30
29
28
19
15
355
147
19
25
27
33
32
32
31
21
16
383
138
18
24
25
31
30
29
28
19
15
355
138
18
24
25
31
30
29
28
19
15
355
137
17
23
25
30
29
29
28
19
14
351
136
17
23
25
30
29
29
1 28
19
14
350
89
11
15
16
20
19
19
18
12
9
228
                         111-14
                                         GCA/TECHNOLOGY DIVISION

-------
Table III-5.  8-HOUR VMT ESTIMATES, SPOKANE CBD
1971 Base
1
I
1976 Base
Signal improvement
Transit program
W
f
Car pool (max.)
Sky walks
T
Parking management
-jo—Reduce 20% curb with
off-street
-»»- Flex-time
-Js*-Road pricing
-2s— Delivery restrictions
-&»-I and M
-s»— Exclude nonretrofit
-Js— Predictive exclusion
-j»~-Reduce parking
VMT
Avg.
speed
(raph)
13.9
13.9
15.9
15.9
15.9
15.9
15.9
15.9

16.0
16.0
16.1

See tej

Light -
duty
autos
53,586
56,710
56,710
54,828
53,786
53,786
53,786
53,625

53,786
44,610
53,786

ct

Light -
duty
trucks
3,823
4,121
4,121
4,121
4,121
4,121
4,121
4,121

4,121
4,121
3,442



Heavy-
duty
trucks
910
981
981
981
981
981
981
981

981
981
819



Diesel
1,032
1,112
1,112
1,112
1,112
1,112
1,112
1,112

1,112
1,112
929



Motor-
cycles
1,335
2,486
2,486
2,486
2,486
2,486
2,486
2,486

2,486
2,486
2,486



Total
60,686
65,410
65,410
63,528
62,486
62,486
62,486
63,325

62,486
53,310
61,462



                 111-15
                                 GCA/TECHNOLOGY DIVISION

-------
Parking Management - The difference between demand and the Spokane "lid"
(30 percent) is greater than in Seattle (6 percent).  Therefore, the
"lid" (and preconstruction review) has no effect on VMT or speed.

SkywaIks - This program has been implemented in Spokane, reducing pedes-
trian volumes in at least five intersections.  More inter-building
bridges are being considered.  Assuming that 20 percent of approaching
vehicles turn more smoothly due to reduced pedestrian interference, a
small share of CBD VMT is assumed to feel a 2-mph speed increase.  This
share is so small, however, that no effect occurs in the weighted average
speed.  Continued skywalk development would be expected to lead to a
significant impact, however.  Many benefits can be seen for skywalk de-
velopment and their construction should be encouraged.

Replace Curb Parking With Off-Street - If 20 percent of curb parking is
shifted in congested areas and a 1-mph increase is felt during peak
hours in the 15-mph range, the overall impact is so slight that it does
not affect the weighted average speed.  To achieve a significant impact
in Spokane, a larger share would have to be displaced, and no plans exist
for such a program.

Flex-time - As assumed for Seattle, a voluntary adjustment of work hours
around a core time period of mandatory attendance was envisioned leading
to a peak hoiir 1-mph speed increase but no change in VMT.

"Staggered days" were assumed to have no effect on VMT except on Monday
and Friday and no additional speed impact.

Road Pricing - Although there are more bridges closer to downtown
Spokane than in Seattle, the high share of through traffic creates addi-
tional complications in the imposition of tolls.  The technical problems
of automated charges for CBD street use exist as do policy problems of
special CBD  licenses.  As for Seattle, an impact of some type of pricing
                                111-16
                                                GCA/TECHNOLOGY DIVISION

-------
program (tolls, special licenses, etc.) leading to four times the auto-
mobile operating cost was assumed.  This assumption results in a 15-
percent VMT reduction.  The realism of such an increase in cost has not
been analyzed.

Delivery Restrictions - If merchant objections could be overcome, and a
35-percent share of truck delivery traffic be deleted,  VMT would be re-
duced and a slight speed increase seen during peak hours.   Because of
the high percentage of through truck traffic, the impact of this measure
amounts of a 16.5-percent reduction in total truck VMT.

Inspection and Maintenance - This measure does not affect  VMT or speed.

Exclusion of Nonretrofit vehicles - Again, as discussed for Seattle, the
reduction in Spokane depends upon the type of retrofit  program, the share
of vehicles covered, and enforcement.  One maximum value could be1 the
share of VMT contributed by pre-1968 vehicles.

Predictive Exclusion - A time exclusion, as compared to exclusion of
classes of vehicles, would probably be required to achieve high percent-
age reductions in VMT0  Although an exclusion between 12 noon and 6 p.m.
would delete some 40.5 percent of 8-hour VMT, a more complete shut-
down would occur by starting early enough to exclude employees.

Reduce Parking Sup_p_ly_ - This strategy has little application to Spokane.
If the "lid" were dropped to 80 percent of present value,  little or no
impact is estimated due to present excess of supply over demand.

CO Emission Estimates
Table III-6 shows the 8-hour carbon monoxide emissions computed from the
VMT-speed data given in Table III-5.  The fraction of annual miles
driven by vehicle age data used for these calculations are given in
Table III-7.

                               111-17
                                               OCA/TECHNOLOGY DIVISION

-------
                                  Table III-6.  8-HOUR CO EMISSION  ESTIMATES,  SPOKANE CBD
Q
O
n
;£

O
5
co
O
z
      I
      I—1
      CO

1971 Base
1
1
1976 Ease
^
?
Signal improvement
₯
!
Transit program
Car pool (max.)
Skywalks
Parking management
—SB— Replace 20% curb with
off-street
— 3»— Flex-time
— 5s»- Road pricing
-3s- Delivery restrictions
-a*- I and M
—is— Exclude nonretrofit
— 2a— Predictive exclusion
— 2s— Reduce parking
Emissions (kg/8 hours)
Light -
duty
autos
5,155
4,174
3,766
3,641
3,571
3,571
3,571
3,561
3,555
2,949
3,539

See tex


Light -
duty
trucks
493
385
349
349
349
349
349
349
347
347
288

i


Heavy-
duty
trucks
263
246
224
224
224
224
224
224
223
223
185




Diesel
30
32
32
32
32
32
32
32
32
32
27




Motor-
cycles
32
62
62
62
62
62
62
62
62
62
62




Total
5,973
4,899
4,433
4,308
4,238
4,238
4,238
4,228
4,219
3,613
4,101




Additional
reduction
needed
2,987
1,913
1,447
1,322
1,252
1,252
1,252
1,242
1,233
627
1,115





-------
         Table III-7.   FRACTION  OF ANNUAL MILES DRIVEN BY
                       VEHICLE AGE FOR SPOKANE, WASHINGTON-
Vehicle age
0
1
2
3
4
5
6
7
8
9
10
11
12
> 13
Light -duty autos
1971
0
0.063
0.105
0.113
0.106
0.093
0.093
0.092
0.070
0.060
0.046
0.030
0.028
0.101
1976a
0
0.082
0.107
0.100
0.099
0.096
0.089
0.083
0.075
0.063
0.048
0.035
0.028
0.095
Light -duty trucks
1971
0
0.068
0.103
0.113
0.093
0.082
0.080
0.070
0.050
0.041
0.031
0.022
0.025
0,222
1976a
0
0.097
0.114
0.096
0.089
0.083
0.073
0.065
0.054
0.044
0.033
0.025
0.022
0.205
Heavy-duty vehicles
1972b
0
0.130
0.133
0.123
0.124
0.085
0.068
0.058
0.050
0.040
0.031
0.024
0.017
0.117
      Average for calendar years 1971, 1972,  and 1973.
      Nationwide

As in Seattle, large reductions beyond those  provided by the basic
strategy package are required to meet the allowable emissions level.
Of the first four individual strategies, only road pricing is esti-
mated to have a really large impact.  The probable effect of an annual
inspection and maintenance program can be estimated at any step by
reducing light-duty automobile and truck emissions by 10 percent.   Table
III-8 provides a breakdown of emissions by model year for use in esti-
mating the effectiveness of various retrofit  and exclusion measures.
                               111-19
                                               GCA/TECHNOLOGY DIVISION

-------
Table III-8.  1976 CO EMISSION ESTIMATES FOR LIGHT-DUTY
              GASOLINE VEHICLES BY MODEL YEAR, SPOKANE CBD














Model Year

Pre-1968
1968
1969
1970
1971
1972
1973
1974
1975
1976
Total

Pre-1968
1968
1969
1970
1971
1972
1973
1974
1975
1976
Total
a)
60
cd

a
03


^
60
0)
4-1
C3

4-1
,0
S-l 4->
|I3 > 4-1
M O
OJ -H
> SH
•i-l 4-1
iH CO
a) a)
p J-l
Light -duty autos
1,247
305
366
347
357
283
280
269
69
48
3,571
1,243
305
365
346
356
283
279
269
69
48
3,563
1,241
304
364
345
354
282
278
268
69
48
3,553
1,030
252
302
286
294
234
231
222
58
40
2,949
1,236
303
362
343
353
281
276
266
69
48
3,537
Light-duty trucks
170
17
21
21
23
21
22
23
16
13
347
170
17
21
21
23
21
22
23
16
13
347
170
17
21
21
23
21
22
23
16
13
347
170
17
21
21
23
21
22
23
16
13
347
141
14
18
17
19
18
18
19
13
11
288
                        111-20
                                         OCA/TECHNOLOGY DIVISION

-------
PORTLAND

VMT Estimates

VMT's for Portland's GBD were calculated from 1972 counts.,  The area de-
fined by the CBD is larger than that used in Seattle or Spokane, and is
bounded by Hoyt Street, the Willamette River, and 1-405.  Three or four
average speeds are used to cover the range of speeds present in the
Portland CBD.  The estimated 8-hour VMT's for each speed level are
presented in Table III-9 for the two baseline periods and for the
various strategy combinations.  The rationale used in preparing the
VMT estimates follows.

Signal Improvements - Computer control of signals was implemented in
1971.  The program has reduced the worst congestion points and improved
off-peak speeds.  A 2-mph speed increase has been assumed from base con-
ditions for the lower speed ranges.

Transit - Interim park-ride development is underway, new routes are
being implemented, and a new fare policy eliminating zone fares and pro-
viding free CBD riding (among other items) will start January 1975.
The coming downtown transit mall will facilitate system use.  Extrapo-
lation of present trends and mode use modeling both indicate that the
goal of doubling 1970 transit use in downtown is within reach.

Carpools - A vigorous carpool program exists in Portland.  If the number
of worker pools doubles (a rule-of-thumb), average occupancy could in-
crease from 1.36 to 1.44 with a consequent reduction in VMT.  The re-
duction shown in Table III-9 is based on this assumption.

Parking Management - The "lid" on parking supply can be monitored and
enforced through preconstruction review.  The lid itself should have no
effect on VMT and speed if other strategies are successful,   (Supply
                              III-21
                                               GCA/TECHNOLOGY DIVISION

-------
Table III-9.  8-HOUR VMT ESTIMATES,  PORTLAND CBD
1972
1
1976
1
Si
i
Tr
1
Ca
1
iaso
Base
gnal improvement
ansit program
f
r pool (max.)
Parking management
	 1
-5*~ Reduce 20% curb
with off-strcct

•*»- Flex- time
*»— I and M
VMT
Avg.
3 peed
(r.iph)
50.0
25.0
13.2
50.0
25.0
13.2
50.0
25.0
15.2
50.0
25.0
15.2
50.0
25.0
15.2
50.0
25.0
15.2
(a) 50.0
25.0
14.8
(b) 50.0
25.0
15.2
50.0
26.0
25.0
15.4

Light
cl u t y
autos
1,287
31,810
114,034
147,131
1,378
34,069
122,134
157,581
1,378
34,069
122,134
157,581
1,295
32,015
114,769
148,079
1,220
30,158
108,112
139,490
1,220
30,158
108,112
139,490
1,220
30,158
118,685
150,063
1,105
27,308
97,895
126,308
1,220
6,756
23,402
108,112
139,490
Light
d u f y
truck:;
105
2,387
9,274
11,966
114
2,826
10,130
13,070
114
2,826
10,130
13,070
114
2,826
10,130
13,070
114
2,826
10,130
13,070
114
2,286
10,130
13,070
114
2,286
10,130
13,070
114
2,286
10,130
13,070
114
2,826
10,130
13,070
Sec text
(leavy-
du ty
trucks
25
619
2,221
2,865
27
677
2,426
3,130
27
677
2,426
3,130
27
677
2,426
3,130
27
677
2,426
3,130
27
677
2,426
3,130
27
677
2,426
3,130
27
677
2,426
3,130
27
677
2,426
3,130

D to sc-l
28
692
2,432
3,202
31
756
2,711
3,498
31
756
2,711
3,498
31
756
2,711
3,498
31
756
2,711-
3,498
31
756
2,711
3,498
31
756
2,711
3,498
31
756
2,711
3,498
31
756
2,711
3,498

Motor-
c_yc Ics
29
729
2,612
3,370
60
1,473
5,279
6,812
60
1,473
5,279
6,812
60
1,473
5,279
6,812
60
1,473
5,279
6,812
60
1,473
5,279
6,812
60
1,473
5,279
6,812
60
1,473
5,279
6,812
60
1,473
5,279
6,812

Total
1,474
'36,437
130,623
168,534
1,610
39,801
142,680
184,091
1,610
39,801
142,680
184,091
1,527
37,747
135,315
174,589
1,452
35,890
128,658
166,000
1)452
35,890
128,658
166,000
1,452
35,890
139,231
176,573
1,337
33,040
118,441
152,818
1,452
6,756
29,134
128,658
166,000

                 111-22
                                 GCA/TECHNOLOGY DIVISION

-------
probably exceeds demand by 4 percent today, with demand decreasing.)
Otherwise, a VMT increase could occur as described in the next para-
graph.

Parking Supply Reductions^ - A 20-percent drop in the "lid" would cause
demand to exceed supply.  Short-range reduction in travel demand is not
likely.  Therefore, the impact of a parking shortage would depend upon
who is affected.

If the reduction hits the peak of demand, short-term parkers would be
impacted with a resulting increase in "searching" VMT at low speeds.
This impact appears in Table III-9, estimate (a).

If the reduction in spaces affects only employees (long-term spaces)
through rate increases to the short-term level, a reduction in light -
duty auto VMT to 1,105 at 50 mph, 27,308 at 25 mph, and 97,895 at 15
mph might be. achieved if employees shift to transit, park-shuttle, or
park-walk from outside the CBD.  This is shown in Table III-9, esti-
mate  (b).

Flex-time - Again, as with Seattle and Spokane, the "flex-time" ap-
proach to staggered hours is assumed because of its many advantages
over a rigid program.  The impact is estimated as a 1-mph speed in-
crease for peak hours only.

"Staggered days" are not included as an effective strategy because of
their limited impact (or no impact) on mid-week days.

Inspection and Maintenance - This program has no impact on VMT or speed.
                             111-23
                                              GCA/TECHNOLOGY DIVISION

-------
CO Emission Estimates

Table 111-10 shows the emissions calculated for the Portland CBD from
the VMT-speed data given in Table III-9.  The fraction of annual miles
driven by vehicle age data for Portland are given in Table III-ll.

Again, none of the strategy combinations tested reduce emissions to the
allowable level even after applying a 10 percent reduction to light-
duty vehicle emissions.  Table 111-12 gives the breakdown of emissions
by model year for various strategy combinations.

SUMMARY COMMENTS ON METHODOLOGY

Many of the characteristics and impacts of the transportation control
measures that have been analyzed have been assumed to be similar for
all three cities.  Thus, to avoid repetition, procedures used in de-
veloping VMT and speed change estimates have been discussed in more
detail for the first city studied, Seattle, than for Spokane or
Portland.  Through necessity, these analyses have made use of many
gross factors and assumptions, and the possibility for substantial
errors exists.  Even more detailed planning techniques may involve
errors up to 15 percent.

Current deficiencies in data and projection techniques point out the
need for increased planning, transportation modeling, and monitoring.
CBD network modeling would allow strategy impacts on speeds to be
tested more thoroughly, and modal split modeling with regional net-
works would allow a better evaluation of transit.  Specific programs,
such as parking reductions or predictive exclusion, need to be laid
out in more detail and reviewed for feasibility so that impact analy-
sis can be more realistic.  Also, the economic and social impacts of
candidate strategies should be reviewed in detail as well as their
impact on VMT and air quality.
                              111-24
                                              GCA/TECHNOLOGY DIVISION

-------
                                  Table 111-10.   8-HOUR CO EMISSION ESTIMATES,  PORTLAND CBD
o
o
— (
m
n
IT


O

5
     M
     M
     Ul

1972 Base
w
1976 Base
t
S ignal improvement
W
Trans it program
₯
f
Car pool (max.)
y
Parking Management
— ^-Reduce 207° curb with
off-street

— 2s>- Flex- time
— S»- I and M
Emissions (kg/ 8 hours)
Light -
duty
autos
12,754
9,814
8,960
8,420
7,931
7,931
(a) 8,742
(b) 7,182
7,845
See text
Light-
duty
trucks
1,241
1,022
935
935
935
935
949
935
927

Heavy-
duty
trucks
760
1,049
683
683
683
683
693
683
678

Diesel
92
100
100
100
100
100
100
100
100

Motor-
cycles
104
211
211
211
211
211
211
211
211

Total
14,951
12,196
10,889
10,349
9,860
9,860
10,695
9,111
9,761

Additional
reduction
needed

5,468
4,161
3,621
3,132
3,132
3,967
2,383'
3,033

NOTE:  Allowable emissions, assuming a 55 percent reduction from 1972 base levels, are

       6728 kg/8 hours.
g
O
"7

-------
  Table III-ll.  FRACTION OF ANNUAL MILES DRIVEN BY VEHICLE
                 AGE FOR PORTLAND, OREGON
Vehicle age
0
1
2
3
4
5
6
7
8
9
10
11
12
> 13
Light -duty autos
1972
0
0.116
0.129
0.109
0.115
0.098
0.085
0.077
0.071
0.054
0.042
0.028
0.018
0.058
1976a
0
0.114
0.132
0.114
0.106
0.098
0.086
0.078
0.067
0.056
0.041
0.028
0.021
0.059
Light -duty trucks
1972
0
0.118
0.107
0.132
0.138
0.085
0.076
0.061
0.050
0.038
0.028
0.020
0.015
0.132
1976a
0
0.110
0.140
0.125
0.117
0.098
0.073
0.058
0.048
0.038
0.028
0.020
0.017
0.128
Heavy-duty vehicles
1972b
0
0.130
0.133
0.123
0.124
0.085
0.068
0.058
0.050
0.040
0.031
0.024
0.017
0.117
Average for calendar years 1971, 1972, and 19/3,
Nationwide.
                         111-26
                                         GCA/TECHNOLOGY DIVISION

-------
Table 111-12.  1976 CO EMISSION ESTIMATES FOR LIGHT-DUTY
               GASOLINE VEHICLES BY MODEL YEAR, PORTLAND CBD













Model year


Pre-1968
1968
1969
1970
1971
1972
1973
1974
1975
1976
Total

Pre-1968
1968.
1969
1970
1971
1972
1973
1974
1975
1976
Total

^ CO
(-1 ^
cfl
ft 0
0
0) -H
O J-i
P ft
'i} O
cu
F*J


60
a
•H s*~\
^ &
J-) 
-------
Because of the lack of accurate predictive techniques, expanded moni-
toring during program implementation  is needed.  Suggested parameters
and items for detailed monitoring include:  numbers of bus passengers,
budget items, VMT and speed, meteorology, and pollutant concentrations.
If plans go awry, or standards and deadlines change, appropriate modi-
fications to the control measures can be made.   Regular trend vehicle
counts and speed measurements are beginning to  be made.  If these are
carefully tabulated by route and speed, any new or better data could
be included as it became available and used to  update impact calcula-
tions.  CBD cordon counts identifying hour, traffic class (including
motorcycles), occupancy, and through travel would be useful on a regu-
lar schedule.  Employment inventory data and parking demand checks on
an annual basis would be helpful.  Carpool follow-up surveys and counts
are needed.

Detailed methods for such monitoring and evaluation of the effective-
ness of transportation control plans are now being developed for the
U.S. Environmental Protection Agency by GCA/Technology Division.  When
completed, this work will include procedures for using traffic data -
such as the vehicle counts and speed measurements suggested here - to
provide an early and continuing indicator of how well the control plan
is working.
                             111-28
                                              GCA/TECHNOLOGY DIVISION

-------
                               SECTION IV
             STATISTICAL ANALYSIS OF EMISSIONS DATA FROM THE
                 PORTLAND INSPECTION/MAINTENANCE PROGRAM
INTRODUCTION

The Department of Environmental Quality of the State of Oregon has been
conducting a voluntary emissions testing program for gasoline-powered
vehicles.  The purpose of the program is to assess the effectiveness of
the inspection/maintenance program adopted by the State as part of its
Transportation Control Plan.  According to Oregon's present plan,  this
program would include all motor vehicles registered in three Oregon
Counties - Clackamas, Multnomah, and Washington.  The emissions inspec-
tion would be performed using idle and loaded (dynamometer) emissions
tests at State inspection stations.  Large fleets may be inspected at
stations privately operated under State supervision.

The voluntary test program operated by the Department of Environmental
Quality has tested over 12,000 vehicles to date, using measurements of
emissions during idle, 2500 rpm, and 30 and 50 mph light load conditions.
Emissions measurements were volumetric - ppm of hydrocarbons, percent of
carbon monoxide,  and percent of carbon dioxide.  When a vehicle failed
the prescribed emissions test, the need for corrective action was  sug-
gested by the inspector, and the driver was encouraged to return for a
follow-up inspection.  Data on vehicle characteristics and testing re-
sults were collected for each vehicle, keypunched, then made available
on computer tape.
                                IV-1
                                               GCA/TECHNOLOGY DIVISION

-------
This section presents an analysis of the data collected for a sample of
5900 vehicles.  Approximately 400 of the 5900 vehicles returned for a
retest, so the sample contained 5900 initial tests and 400 retests -
6300 tests in all.  It was difficult to determine which cars retested
had received maintenance between the first and second tests.  Some of
the retested cars passed the first test and required no maintenance.  Of
the retested cars which failed the first test, some did not receive
maintenance.  Others did receive maintenance but the only indication of
this was the reported repair cost.  It is possible that a repair cost
was not given for some vehicles even though they had gone in for repairs.
The method used to determine some of the retested cars which had received
maintenance was to select those cars having a repair cost greater than
zero - there were 104 in this subsample.

Table  IV-1 shows the information which was collected for each vehicle
and the location of the information in each line of data on the tape.
Table  IV-2 shows the transformations and computations which were per-
formed on some of the data to prepare them for analysis.

The program package .SPSS  (Statistical  Package  for  the Social Sciences)
was used to summarize and evaluate the data.  The package contained pro-
grams which permitted the following computations and analyses to be
performed:
     1.   descriptive statistics for the entire sample
     2.   frequency tables and histograms of categorical data
     3.   n-way frequency tables of categorical data
     4..   basic statistics and analysis of variance for sub-
          groups
     5.   t-test
     6.   stepwise regression
     7-   Pearson correlation and nonparametric correlations
                                IV-2
                                               OCA/TECHNOLOGY DIVISION

-------
Table IV-1.  VEHICLE DATA FORMAT
Columns
1
2-6
7-13
14-15
16-19
20-23
24-29
30


31

32-35
36-39
40-43
44-47
48-51
52-55
56-59
60-62
63-65
66-68
69-71
72-74
75-77
1
2-35
36
Variable name
Format number
DATE
LICENSE
MODELYR
MAKE
LINE
MILEAGE
EXHAUST


RE TEST

REPRCOST
HC1
HC2
HC3
HC4
HC5
HC6
C01
C02
C03
C04
COS
C06
Format number
Same as line 1
Registration
Values
1. First of two lines
Year and Julian day

Model year
Manufacturer


1. Single
2. Dual, left
3. Dual, right
1. Retest
Blank not re test
Repair cost
Hydrocarbons at idle
Hydrocarbons at 2500 rpm
Hydrocarbons at idle
Hydrocarbons at 30 mph
Hydrocarbons at 50 mph
Hydrocarbons at idle
Carbon monoxide at idle
Carbon monoxide at 2500 rpm
Carbon monoxide at idle
Carbon monoxide at 30 mph
Carbon monoxide at 50 mph
Carbon monoxide at idle
2. Second of two lines

0. Available
1. Not available
            IV-3
                            GCA/TECHNOLOGY DIVISION

-------
Table IV-1 (Continued).  VEHICLE DATA FORMAT
Columns
37-38
39
40-42
43-46
47
48
49
50-56







57-61



Variable name
COUNTY
SMOKE
NOISE
ENGDISPL
CYLIN
TRANSM
WGTCLASS
Under hood inspection
50 PCV
51 AIRPUMP
52 EGR
53 EVAP
54 DISTRIB
55 PREHEAT
56 CATMUF
Engine modification
inspection
57 GARB
58 INMANI
59 EXH
Values
3. Clackamas
26. Multnomah
34. Washington
99. Out of state
and others
1. Fail
0. Pass

Engine displacement in cubic
.centimeters or cubic inches.
Number of cylinders - 2, 4,
6, or 8.
Transmission type
1. Manual
2. Automatic
Weight class
1. Under 2800
2. 2800-3800
3. Over 3800
0. Not present
1. One present
2. Present and disconnected
3. Both 1 and 2
Positive crankcase ventilation

Exhaust gas recirculation

Distributor

Catalytic muffler
0. No tampering
1 . Tampering
Carburetor
Intake manifold
Exhaust
                  IV-4
                                  GCA/TECHNOLOGY DIVISION

-------
              Table IV-1 (Continued).  VEHICLE DATA FORMAT
Columns


62-64
65-67
68-70
71-73
74-76
77-79
Variable name
60 ADDON
61 OTHER
C021
C022
C023
C024
C025
C026
Values
Addon devices

Carbon dioxide at idle
Carbon dioxide at 2500 rpm
Carbon dioxide at idle
Carbon dioxide at 30 mph
Carbon dioxide at 50 mph
Carbon dioxide at idle
                   Table IV-2.  DATA TRANSFORMATION
Variable
                    Transformations
Make



Modelyr

Modelyr

Engdispl

Mileage
The 145 different values were receded to 9 values:
(1) GM  (2) Ford  (3) Chrysler   (4) Japanese
(5) Other foreign   (6) Volkswagen   (7) AMC
(8) Other domestic   (9) Other.
Receded to age:  (0) 75  (1) 74  (2) 73  (3)  72   (4)  71
(5) 70  (6) 69  (7) 68  (8) 67   (9) pre 67.
Receded to control:  (0) pre 68  (1) 68-69   (2)  70-71
(3) 72-75.
Receded to CI  (cubic inches):  (1)  less than  150
(2) 150-250  (3) 250-350   (4) greater than 350.
Receded to 5 values:  (1)  less than 20,000
(2) 20,000-40,000   (3) 40,000-60,000  (4) 60,000-
80,000  (5) greater than 80,000.
                                IV-5
                                               GCA/TECHNOLOGY DIVISION

-------
DESCRIPTION OF SAMPLE VEHICLE CHARACTERISTICS

The vehicle characteristics of the sample are described by determining
the distributions within the categories of age, make, weight class, en-
gine displacement, number of cylinders, and county.  Figures IV-1 through
IV-6 show the numerical distributions of vehicles among the categories.
Table IV-3 is a two-way frequency distribution of vehicles in age and
make categories.

In addition to describing the characteristics of the sample, the fre-
quency distributions of the sample vehicles can be compared with in-
formation available for the population of vehicles in the three counties -
Clackamas, Multnomah, and Washington - from which the sample was drawn.
It should then be determined if the sample is representative of the pop-
ulation and what inferences about the population can be drawn.
Tables IV-4 through IV-6 show, respectively, the age distribution,  the
make distribution, and the control by make distribution of the  vehicle
                               2
population and the sample.  A X  goodness of fit test was performed com-
paring the number of sample vehicles in each age category (observed)
with the number of the population vehicles in the same category (ex-
                                 2
pected).  In the same manner, a X  goodness of fit was performed on the
sample and population distributions in the make categories and  the  con-
trol by make categories.  Results showed that the sample was signifi-
cantly different from the population for all three characteristics.  In
general, there were more vehicles built in recent model years in the.
sample than there were in the population.  This would tend to decrease
the average emissions of .the whole sample.  Some makes of cars  were
overrepresented in the sample while others were underrepresented.  The
effect on the average emissions of the whole sample, then, depends on the
difference in emissions among makes and age groups which will be dis-
cussed in EMISSIONS OF GROUPS.
                                IV-6
                                              GCA/TECHNOLOGY DIVISION

-------
  2000
   1500
o
z
UJ
rD
o
UJ
1C
u.
1000
   500
                                   5


                                  AGE
                                                        >8
              Figure  IV-1.   Sample  age  distribution
                              IV-7
                                             GCA/TECHNOLOGY DIVISION

-------
  2000
   1500
o
Z
Ul

O

{£  1000
    500
           GM
FORD CHRYSLER
OTHER   VW
FOREIGN
                           JAPANESE
AMC   OTHER  OTHER
      DOMES-
      TIC
                                  MAKE
              Figure IV-2.   Sample  make  distribution
                             IV-8
                                            GCA/IECHNOLOGY DIVISION

-------
         3000
       o
       z
       tu
       ^>
       o
       UJ
       cc
       U.
2000
         1000
                UNDER    2800-   OVER

                2800     3800    3800


                     WEIGHT CLASS
    Figure IV-3.  Sample weight class distribution
      3000
    v
    o
    z
    Ul

    o
    Ul
    cc
      2000
      1000
    UNDER

    150
150-

250
250-

350
                                      350-
                  ENGINE DISPLACEMENT

                      (CUBIC INCHES)




Figure IV-4.  Sample engine displacement distribution
                       IV-9
                                       GCA/TECHNOLOGY DIVISION

-------
     4000
     3000
  o
  2
  U
  z>
  o
  LU
  OC
  U-
2000
      1000
               2468

                NUMBER OF CYLINDERS.



      Figure IV-5.  Sample cylinders distribution


  4000
  3000
o
o
UJ
o:
  2000
   1000
            3       26       34      99     OTHER

                          COUNTY



       Figure IV-6.  Sample county distribution




                        IV-10
                                        CCA/TECHNOLOGY DIVISION

-------
                                      Table IV-3.  SAMPLE AGE BY MAKE DISTRIBUTION
Age /make
0
1
2
3
4
5
6
7
8
9 +
Total
GM
0
105
195
208
167
191
202
187
175
771
2201
Ford
3
92
172
210
158
158
155
156
133
373
1610
Chrysler
1
63
120
111
81
83
108
96
74
273
1010
Japanese
0
49
89
85
78
31
21
11
11
10
385
Other
foreign
0
23
28
35
46
33
35
16
23
70
309
Volkswagen
0
19
24
25
34
28
37
30
29
89
315
AMC
0
11
30
26
20
24
24
36
15
108
294
Other
domestic
0
5
22
25
3
10
6
9
6
55
141
Total
5
367
683
726
592
560
592
541
469
1763
6298
o
o
-H
m
O
X


O

5
o
O

-------
         Table IV-4.   AGE DISTRIBUTION OF SAMPLE AND POPULATION
Age
1
2
3
4
5
6
7
8
9+
3 county
July, 1974
population
0.036
0.09
0.086
0.067
0.063
0.072
0.065
0.061
0.46
Sample
0.058
0.109
0.115
0.094
0.089
0.094
0.086
0.075
0.28
Observed
367
683
726
592
560
592
541
469
1763
6293
Expected
227
566
541
421
396
453
409
384
2895
6293
X  = 459.6
                   0.95
        Table IV-5.   MAKE DISTRIBUTION OF SAMPLE AND  POPULATION
Make
GM
Ford
Chrysler
Japanese
Other foreign
Volkswagen
AMC
Other domestic
Multnomah county
1972 population
0.39
0.278
0.126
0.073
0.045
0.041
0.032
0.015
Sample
0.351
0.257
0.161
0.061
0.049
0.050
0.047
0.024
Observed
2201
1610
1010
385
309
315
294
141
6265
Expected
2443
1742
789
457
282
257
200
63
6265
X2 =
263.7
                        - 14.07
                                IV-12
                                               GCA/TECHNOLOGY DIVISION

-------
 Table IV-6.  CONTROL  BY MAKE  DISTRIBUTION OF SAMPLE AND POPULATION
Control
0
0
0
0
0
1
1
1
1
1
2
2
2
2
2
3
3
3
3
3

Make
1
2
3
7
0
1
2
3
7
0
1
2
3
7
0
1
2
3
7
0

Population
0.205
0.115
0.067
0.026
0.065
0.067
0.038
0.027
0.006
0.036
0 . 049
0.041
0.021
0.005
0.048
0.065
0.044
0.022
0.066
0.048

Sample
0.151
0.081
0.055
0.020
0.047
0.062
0.050
0.033
0.010
0.026
0.057
0.050
0.026
0.007
0.042
0.081
0.076
0.047
0.011
0.068

Observed
946
506
347
123
293
389
311
204
60
165
358
316
164
44
263
508
• 477
295
67
429
6265
Expected
1284
720
420
163
407
420
230
169
38
226
307
257
132
31
301
407
276
138
38
301
6265
X  = 740.929
                     0.95
-30'1
Note:  Control:   (0) pre  68   (1)  68
       Make:   (1) GM   (2) Ford   (3)
               (0) Other foreign  and
          •69   (2) 70-71   (3) 72-75.
          Chrysler   (7) AMC
          domestic.
                                IV-13
                                               GCA/TECHNOLOGY DIVISION

-------
The sample frequency distributions can also be compared with those of
certain subsamples which are of interest.  The age, make, weight class,
cylinder, and control distributions of the subsample of cars which
failed at least one of the carbon monoxide, hydrocarbons, and smoke
tests are shown and compared with those distributions of the entire
sample in Tables IV-7 through IV-11.  The fail subsample differed sig-
nificantly from the sample with respect to age, make, weight class, and
number of cylinders.  There were more older cars, more of Japanese and
other foreign make, more cars under 2800 pounds, and more four-cylinder
cars in the fail subsample than were expected when comparing it with
the entire sample.  The fail subsample did not differ significantly
from the sample with respect to control.  Control is a grouping of cars
by age according to the level of emissions controls present in cars of
a particular model year - from no controls, pre-68, to extensive con-
trols, 72 to 75, with intermediate increases from 68 to 69 and 70 to
71.  Failure levels were set according to the control categories so that
approximately 50 percent of the cars in a particular control category
would fail the emissions tests.  Therefore, it is expected that the
sample and the fail subsample would not differ significantly with re-
spect to control.

DESCRIPTION OF SAMPLE EMISSIONS

During emissions testing, readings of carbon monoxide,  hydrocarbons,
and carbon dioxide were taken at either three or six modes.  The modes
are:
     1.   idle
     2.   2500 rpm
     3.   idle
     4«   30 mph, light load
     5.   50 mph, light load
     6.   idle
                                IV-14
                                               GCA/TECHNOLOGY DIVISION

-------
       Table IV-7.   AGE DISTRIBUTION OF SAMPLE AND FAIL SUBSAMPLE
Age
1
2
3
4
5
6
7
8
9
Sample
0.058
0.109
0.115
0.094
0.0.89
0.094
0.086
0.075
0.28
Sub sample
fail
6.045
0.100
0.132
0.095
0.099
0.089
0.091
0.071
0.278
Observed
134
298
394
283
.296
265
271
211
830
2982
Expected
173
325
343
280
265
280
256
224
835
2982
X  = 24.743
                  , 0.95
          = 15'51
      Table IV-8.  MAKE DISTRIBUTION OF SAMPLE AND FAIL SUBSAMPLE
Make
GM
Ford
Chrysler
Japanese
Other foreign
Volkswagen
AMC
Other domestic

Sample
0.351
0.257
0.161
0.061
0.049
0.050
0.047
0.023

Sub sample
fail
0.335
0.260
0.158
0.068
0.059
0.053
0.041
0.019

Observed
1001
776
473
203
177
159
122
	 57.
2968
Expected
1042
763
478
181
145
148
139
68
2964
X  = 16.298
A, 0.95 = 14'°7
                                IV-15
                                               GCA/TECHNOLOGY DIVISION

-------
  Table IV-9.  WEIGHT CLASS DISTRIBUTION OF SAMPLE AND FAIL SUBSAMPLE

under 2800
2800 - 3800
over 3800
Sample
0.169
0.247
0.584
Sub sample
fail
0.187
0.234
0.578
Observed
524
656
1619
2799
Expected
473
691
1635
2799
X2 = 7.429    X2
                2, 0.95
   Table IV-10.  CYLINDER DISTRIBUTION OF SAMPLE AND FAIL SUBSAMPLE

2
4
6
8
Sample
0.006
0.174
0.177
0.643
Subsample
fail
0.005
0.197
0.155
0.644
Observed
13
544
435
1810
2812
Expected
17
489
498
1808
2812
X  = 17.553
                  , 0.95
    Table IV-11.  CONTROL DISTRIBUTION OF SAMPLE AND FAIL SUBSAMPLE

pre 68
69 - 70
70 - 71
72 +
Sample
0.354
0.180
0.183
0.283
Subsample
fail
0.349
0.180
0.194
0.278
Observed
1041
536
579
829
2985
Expected
1057
537
546
845
2985
X  = 2.542
                   0.95
' 7'81
                                IV-16
                                               GCA/TECHNOLOGY DIVISION

-------
All of the cars were tested at the first three modes - the idle test.
About one-third of the sample was tested in addition at the second three
modes - the dynamic test.  The third mode  (second  idle) emissions were
compared with the fail criteria to determine passing or failure.  See
Table IV-12.

                       Table IV-12.  FAIL  CRITERIA
Control
pre 68

68-69

70-71

72-74

Fail Criteria (idle mode)
HC > 1200 ppm
CO > 6 percent
HC > 600 ppm
CO > 5 percent
HC > 500 ppm
CO > 4 percent
HC > 350 ppm
CO > 3 percent
The carbon monoxide, hydrocarbons, and carbon dioxide emissions are
summarized for each mode  in Figures  IV-7 through IV-9.  Each figure
shows  for a  particular  pollutant  the arithmetic mean, geometric mean,
and 95  percent confidence  interval for the six modes.  It can be seen
that the carbon monoxide  and hydrocarbons arithmetic means are higher
for the idle modes and  lower for  the dynamic modes while the reverse is
true for the carbon dioxide means.  The three idle mode means of carbon
monoxide increase from  first to third idle while the hydrocarbons idle
mode means increase from  first to second idle but decrease from second
to third idle.  Of the  three carbon monoxide dynamic mode means, the
lowest  is at 2500 rpm and  the other  two, 30 mph and 50 mph, are nearly
the same.  The means of the hydrocarbons dynamic modes show a decrease
from 2500 rpm to 30 mph to 50 mph.  Only 3 percent of the vehicles
failed  the exhaust smoke  test.
                                IV-17
                                               GCA/TECHNOLOGY DIVISION

-------
   14
   I 3
   12
   1 1
   10
o
o
    8
H
2
UJ
0  7
K  I
ID
O.
              ARITHMETIC  MEAN
	GEOMETRIC  MEAN
 I  95 PERCENT CONFIDENCE
    INTERVAL
       ii    II    I
                                      -t
                          II    I)    1
       COI    C02    C03   C04   C05   COS
Figure IV-7.  Carbon monoxide mean  emissions
                   IV-18
                                   GCA/TECHNOLOGY DIVISION

-------
900


800


700


600
5   500
O
I
400


300


200


100
                 ARITHMETIC MEAN
           ---- GEOMETRIC  MEAN
             I   95  PERCENT CONFIDENCE
                 INTERVAL
                                       •i
                     it
                           ii
                                 ii
                                       i
         HCI
            HC2   HC3   HC4   HC5   HC6
 Figure IV-8.  Hydrocarbons mean emissions
                  17-19
                                  GCA/TECHNOLOGY DIVISION

-------
   16


   15

   14

   13

   12

    II

   10
 H  9
 tu
 o
 LU  8
 Q.
  OJ
 8  7
      ARITHMETIC  MEAN
          I  95 PERCENT  CONFIDENCE
             INTERVAL
        ii    ii	L_l	LJ	L_L
C02I   C022  C023
                                 C025   C026
Figure IV-9.  Carbon  dioxide mean emissions
                  IV-20
                                   GCA/TECHNOLOGY DiVISION

-------
Only carbon monoxide and hydrocarbons results will be considered further
in this discussion since carbon dioxide is not considered a serious ve-
hicle emission.  Also, the discussion of results will look primarily at
the emissions readings for the third mode (second idle) because this was
the mode used for determining failure and because a car that failed was
more likely to fail at one of the idle modes due to generally higher
emissions for the idle modes.  The third hydrocarbons reading will be
referred to as HC3 and the third carbon monoxide reading as COS.

The geometric mean of a distribution is always less than the arithmetic
mean.  The degree of difference between them is a good indication of
the degree of skewness of the data.  As can be seen in Figures IV-7 and
IV-8 the geometric means are in general much less than the corresponding
arithmetic means.  This means that a number of outlying readings are
tending to increase the arithmetic means so that the arithmetic means
as a measure of central tendency are high.  This is borne out by the
skewness measure which was calculated for each mode for each pollutant.

The 95 percent confidence interval around the sample mean of a pollutant
at a particular mode indicates that there is 95 percent confidence that
the population mean lies within that interval.  In general,  the 95 per-
cent confidence intervals are very narrow, especially for the means of
the first three modes.  This is due mostly to the large sample size -
about 6300 vehicles were tested in the first three modes and about one
third of these were also tested in the second three modes.  A larger
sample size decreases the length of the confidence interval around the
mean, thus increasing confidence in the mean.

The normality of the emissions data can be studied by the standard de-
viation, kurtosis, and skewness.  The standard deviations for the pol-
lutants measured at the different modes were in general about equal to
the mean values.  This indicates that there is a large amount of spread
to the data.  Kurtosis is a measure of the peakedness of a distribution
of data.  The data distributions of each pollutant at each mode were,

                               IV-21
                                              GCA/TECHNOLOGY DIVISION

-------
for the most part, more peaked than the normal distribution; i.e.,
leptokurtosis.  A distribution is considered to be skewed when there
is a considerably larger number of extreme cases on one side of the dis-
tribution curve than on the other.  When the skewness measure is posi-
tive, the distribution is skewed to the right.  In general, skewness
indicated that most of the data distributions were markedly skewed to
the right.  Figures IV-10 and IV-11 are relative frequency graphs of
the third hydrocarbons (HC3) mode data and the third carbon monoxide
(C03) mode data, respectively.  The hydrocarbons data distribution is
very definitely skewed to the right, with 13.2 percent of the data
greater than 1000 ppm and 1.6 percent greater than 2000 ppm.  The car-
bon monoxide data distribution is also skewed to the right with 1.2 per-
cent of the data greater than 10 percent CO and, in addition, is bimodal.
The bimodality of the carbon monoxide distribution could not be broken
down into two separate distributions by control, weight class, cylinders,
or engine displacement.  The emissions tests data are not normally
distributed, which has been found in other studies.

The log normality of the emissions was studied to determine, if the log
of the data would produce a normal distribution.  Figures IV-12 and
IV-13 are relative frequency graphs of the log of the data from the
third hydrocarbons mode and the third carbon monoxide mode.  Both dis-
tributions appear more normal and skewness and kurtosis are decreased.
But it appears that the log distributions are not sufficiently more
normal than the data distributions to warrant using them.  In addition,
the relatively large sample size reduces the sampling error and in-.
creases confidence that the data distributions are representative of
the population distribution.

As discussed in DESCRIPTION OF SAMPLE VEHICLE CHARACTERISTICS, the sample
is significantly different from the population with respect to age, make,
and control by make due to bias.  Since emissions also tend to differ
significantly between age groups, between make groups, and between control
                                IV-2 2
                                              GCA/TECHNOLOGY DIVISION

-------
    40
    30
o
z
UJ
rj
o
UJ
oc
    20
     10
       0    100  200  300  400   500   600  700  800  900  1000

                    PPM  HYDROCARBONS  (IDLE MODE)



  Figure IV-10.  Rela.tive  frequency of hydrocarbons (idle mode)


       40,	
      30 -
   o
   z
   UJ
   ID
   o
   UJ
   o:
                        34567

                       PERCENT CO (IDLE MODE)
                                                          10
Figure IV-11.  Relative frequency  of  carbon  monoxide (idle mode)
                             IV-23
                                             GCA/TECHNOLOGY DIVISION

-------
Q
O
—i
m
n
n:
O
5
g
<
en
O
z
     H
     <
     N3
     •P-
                                                             HYDROCARBONS (PPM)

                                                                   100
                                                                      1000
                                                           10,000
O
2
LJ
Z)
O
LJ
CC
                     1 0-
          0.25  0.5   0.75   1.0
1.25   1.5   1.75  2.0   2.25  2.5   2.75   3.0   3.25  3.5

    LOG  HYDROCARBONS (IDLE  MODE)
3.75    4
             Figure IV-12.   Relative  frequency  of log hydrocarbons (idle  mode)

-------
                                  PERCENT CO
 z
 UJ


 o
 LU
 CH
     '10 -
           •0,75 -0.5  -0.25   0   0.25  0.5   0.75

                           LOG CO, idle mods
25   1.5    1.75   2.0
Figure IV-13.   Relative frequency of log carbon monoxide (idle mode)
                                IV-25
                                                 GCA/TECHNOLOGY DIVISION

-------
by make groups (see EMISSIONS OF GROUPS), the sample mean emissions may
not be representative of the population mean emissions.  This is de-
termined by weighting emissions, as shown in Tables IV-13 through IV-15.
Table IV-13 shows that age adjusting increases the COS mean emissions
slightly and the HC3 mean emissions considerably.  The over-representation
of newer cars in the sample apparently has affected the sample mean,
especially for hydrocarbons.  Table IV-14 shows that adjusting for the
different make composition of the sample does not appreciably affect
either C03 or HC3 mean emissions.  It appears that the effects of over-
representation of some vehicle makes have been cancelled out by the
effects of higher emissions of other vehicle makes.  In Table IV-15,
there is little effect on C03 but a considerable effect on HC3 from
adjusting for control by make.  Apparently,  some control by make groups
with higher emissions have been underrepresented in the sample.
                    Table IV-13.  WEIGHTING  FOR AGE
Age groups
1
2
3
4
5
6
7
8
9+
Proportion of
population
0.036
0.09
0.086
0.067
0.063
0.072
0.065
0.061
0.46
Proportion
of sample
0.058
0.109
0.115
0.094
0.089
0.094
0.086
0.075
0.28
Mean emissions
COS
3.1
2.8
3.5
4.2
4.3
4.6
4,6
5.5
5.4
HC3
208.7
271.1
285.8
353.9
378.4
418.5
470.2
679.6
764.4
 Sample  mean emissions
 C03        4.4
 HC3      485.3
Population adjusted mean emissions
             4.7
           557.8
                                IV-26
                                              GCA/TECHNOLOGY DIVISION

-------
                  Table IV-14.   WEIGHTING FOR MAKE
Make
groups
1
2
3
4
5
6
7
8
Proportion of
population
0.39
0.278
0.126
0.073
0.045
0.041
0.032
0.01
Proportion
of sample
00351
0.257
0.161
0.061
0.049
0.050
0.047
0.024
Mean emissions
C03
4.4
4,5
4.4
3.7
4.9
4.3
4.5
4.0
HC3
526.8
371.7
422 . 1
414.1
630.4
625.8
387.2
518.7
Sample mean emissions
C03       4.4
HC3     466.4
Population adjusted mean emissions
             4.4
           466.4
                                IV-2 7
                                               GCA/TECHNOLOGY DIVISION

-------
             Table IV-15.  WEIGHTING FOR CONTROL BY MAKE
Control
group
0
0
0
0
0
1
1
1
1
1
2
2
2
2
2
3
3
3
3
3
Make
group
1
2
3
7
0
1
2
3
7
0
1
2
3
7
0
1
2
3
7
0
Proportion
of population
0,206
0.115
0.067
0.026
0.065
0.067
0.038
0.027
0.006
0.036
0.049
0.041
0.021
0.005
0.048
0.065
0.044
0.022
0.006
0.048
Proportion of
sample
0.151
0.081
0.055
0.020
0.047
0.062
0.050
0.033
0.010
0.026
0.057
0.050
0.026
0.007
0.042
0.081
0.076
0.047
0.011
0.068
Mean emissions
C03
5.5
5.3
4.6
5.4
6.0
4.7
4.9
4.3
4.5
4.2
4.4
4.0
4.4
4.6
4.2
2.2
3.8
4.1
2.7
2.9
HC3
874.7
557.0
594.2
558.5
933
471.5
386.9
412.8
317.9
561
354.2
290.5
368.0
308.3
477
204.5
277.9
272.1
249.1
314
Sample mean
C03       4-
HC3     485
 emissions
.4
.5
Population adjusted
             4.6
           548.3
mean emissions
                               IV-28
                                              GCA/TFCHNOLOGY DIVISION

-------
EMISSIONS OF GROUPS

The sample can be divided into several different kinds of groups accord-
ing to certain vehicle and testing characteristics.  It is interesting
to compare emissions between the groups within each characteristic.
Table IV-16 shows the results of a one way analysis of variance per-
formed on HC3 and C03 results within the following characteristics -
age, make, weight class, control, test and cylinders.  The groups being
compared within each characteristic are shown at the bottom of the table
Both HC3 and C03 showed a significant difference in variance between age
groups, control groups, and test groups.  HC3 showed a significant dif-
ference in variance between make groups while C03 did not and C03 showed
a significant difference in variance between weight class groups and
cylinder groups while HC3 did not.  It was expected that there would
be significant differences in hydrocarbons and carbon monoxide between
the groups of some of these characteristics.  It appears that weight
class and number of cylinders do not significantly affect hydrocarbons
emissions levels; also, that make does not significantly affect carbon
monoxide emissions levels.

Figures IV-14 through IV-21 show mean HC3 and C03 emissions  for age
groups, make groups, control groups and mileage groups.   In  general,
mean emissions are higher for older cars, for cars with  higher mileage,
and for cars with no or early emissions controls.  C03,  however, is
higher for the 1974 cars.  HC3 mean emissions were especially high for
cars of Volkswagen and other foreign makes,  while C03 mean emissions  did
not vary significantly between makes.

Table IV-17 shows the results of a t-test performed on HC3 and G03 re-
sults between exhaust groups, between retest groups,  and between result
groups.  There were three exhaust groups - single, dual, left, and dual,
right - but only the latter two were compared by the t-test.   Neither
                               IV-29
                                              GCA/TECHNOLOGY DIVISION

-------
                Table IV-16.   ANALYSIS OF VARIANCE


HC3
C03

AGE
S
S

MAKE
S
N
WEIGHT
CLASS
N
S

CONTROL
S
S

TEST
S
S

CYLIN
N
S
Note:  S - significant
       N - not significant
    Characteristic
                   Groups
         Age
         Make

     Weight class
        Control
         Test

       Cylinders
0, 1, 2, 3, 4, 5, 6, 7, 8, 8+
GM, Ford, Chrysler, Japanese, Other foreign,
AMC, VW, Other domestic.
Under 2800, 2800-3800, over 3800.
pre-68, 68-69, 70-71, 72-74.
One test only, first of two tests  (before
retest), second of two tests (retest)
2, 4, 6, 8
                        Table IV-17.  T-TEST

HC3
C03
EXHAUST
N
N
RETEST
S
S
RESULT
S
S
Note: S - Significant
      N - Not significant
    Characteristics
                    Groups
        Exhaust
        Retest
        Result
Dual, left versus dual, right
Before retest versus retest
Pass versus fail
                               IV-30
                                               OCA/TECHNOLOGY DIVISION

-------
o
o
m
n
:E
z
O

5
g


S}
O
     M

     <
      I
o
5
ui 4
Q
O
UI
-1
Q
- 3
o
o
f-
2
LJ
O
cc
UI 9
Q. c
1
-



_






-




























































































































































































74 73 72 71 70 69 68 67 PRE-67
AGE
                                     Figure IV-14.  Carbon monoxide  (idle mode)  by age

-------
   700
   600
   500
i   400
ui
Q
O
2
UJ
Q
-  300
   200
    I 00
            74       73       72
 71
AGE
70      69      68
         Figure  IV-150   Hydrocarbons  (idle  mode)  by age
                             IV-32
                                              GCA/TECHNOLOGY DIVISION

-------
UJ
Q
O
UJ
_l
O
O
O
z
LjJ
O
cc
LU
0.
         GM
FORD  CHRYSLER
OTHER

FOREIGN
                                                   VW
AMC
OTHER

DOMESTIC
                                JAPANESE
                                     MAKE
           Figure IV-16.  Carbon monoxide (idle mode) by make
                                 IV-33
                                                 GCA/TECHNOLOGY DIVISION

-------
1 \J\J
600
500
5
Q_
Q.
I 400
UJ
o
LJ
_1
- 300
o"
200
100
-
-

-

-
-
















































GM FORD CHRYSLER










































-





























OTHER VW AMC OTHER
FOREIGN DOMESTIC
                    JAPANESE
                         MAKE
Figure IV-17.  Hydrocarbons (idle mode) by make
                    IV-34
                                    GCA/TECHNOIOGY DIVISION

-------
5
4
LU
a
O
2
LU
0 3
O
o
PERCENT
ro
1
-
-

-
-










































PRE- 68-69 70-71 72-75
1968
CONTROL
Figure IV-18.  Carbon monoxide (idle mode) by control
                       IV-3 5
                                       GCA/TECHNOLOGY DIVISION

-------
   800



   700



   600
Q.
Q.
,   500



   400
UJ
o
I
   300
   200
    00
               PRE-
                68
68-69     70-71


    CONTROL
                                              72-75
     Figure IV-19.  Hydrocarbons (idle mode) by control
                           IV-3 6
                                          GCA/TECHNOLOGY DIVISION

-------


5
LJ 4
o
o
s
UJ
_j
0
o 3
o
PERCENT
ro
1
1 0 TO 20,000
2 20,000 TO 40,000
3 40,0
4 60, 0(
5 > 80
-
-
-







00 TO 60
DO TO 80
,000









,000
,000


























                           MILEAGE
Figure IV-20.  Carbon monoxide (idle mode) by mileage
                       IV-3 7
                                       GCW TECHNOLOGY DIVISION

-------
( VJ 80,000
-


-
-














































                    234
                          MILEAGE
Figure IV-21.  Hydrocarbons  (idle mode) by mileage
                     IV-3 8
                                     GCA/TECHNOLOGY DIVISION

-------
HC3 nor COS mean emissions were significantly different between the two
dual exhaust groups, indicating that only one side of a dual exhaust
need be sampled to determine the emissions of a vehicle.  The result
t-tests showed that both COS and HC3 mean emissions were significantly
different between the pass and fail groups, as expected.  For only those
vehicles coming in for two tests, the retest t~test compared the emis-
sions of the first test with the second test.  Both HC3 and COS mean
emissions were significantly different between these two groups.

Looking at retest again, Figures IV-22 and IV-23 show the mean emissions
of the sample, the before retest subsample, and the retest subsample
for each mode of hydrocarbons and carbon monoxide.  For hydrocarbons at
all modes, the mean emissions of the before retest group are higher than
for the whole sample.  A decrease occurs so that the mean emissions of
the retest group are less than the before retest group and on the order
of the sample mean.  This would be expected if maintenance has occurred
between the two tests.  For carbon monoxide, except at the first mode,
the opposite case occurs - the mean emissions of the retest group are
considerably higher than the sample mean and the mean, of the before re-
test group.  It is not known how many vehicles of the two test group
actually received corrective maintenance between the two tests or how
much time elapsed between the two tests.  This increase in carbon monox-
ide mean emissions, unexpected when compared to the decrease in hydro-
carbons, could be due to deterioration in low emissions from continued
aging rather than reduction from high emissions to low brought about by
maintenance, or maintenance could be faulty.  The contrast between the
carbon monoxide and the hydrocarbons trends makes it difficult to guess
what happened between the first and second tests and to draw any con-
clusions.

Figures IV-24 and IV-25 are similar to Figures IV-22 and IV-23 except
that the mean emissions of only those cars in the retest group whose
repair cost is greater than zero are compared with the sample mean
                               IV- 3 9
                                               GCA/TECHNOLOGY DIVISION

-------
o
o
m
Q
Z
O
5
5
g
c/5
O
     M
     <
  1000


   900


   800

   700


   600
3=
£,  500
o
a:
   400


   300
                      100
                                MEANS  OF
                                I  SAMPLE
                                2 BEFORE  RETEST
                                3 RETEST
                                HCI
                         HC2
HC3
HC4
HC5
HC6
                         Figure IV-22.  Hydrocarbons mean emissions - sample, before retest,  retest

-------
Q
O
n
z
c
6
     -p-
                      H-
                      2
                      UJ
                      UJ
                      0.
                                MEANS  OF
                                I  SAMPLE
                                2 BEFORE RETEST
                                3 RETEST
        COI
C02
C03
C04
CO 5
C06
CO
o
z
Figure IV-23.  Carbon monoxide mean emissions - sample, before retest, retest

-------
O
O
                    Q_
                    CL
                    O
                    ac
        800



        700 -



        600 -



        500



        400



        300



        200



        100
                                                                         MEANS  OF
                                                                         I  SAMPLE
                                                                         2 BEFORE  RETEST
                                                                         3 RETEST  REPRCOST >0
b
                  HCI
HC2          HC3         HC4         HC5          HC6
c/2
O
z
Figure IV-24.  Hydrocarbons mean emissions - sample,  before retest,  retest  and  repair  cost  <0

-------
o
o
Q

2
O
5

3
     OJ
                          10
                           9 -
                           8 -
                       O   6
                       o
         o
         a:
         LJ
         a.   4
                   CO I
C02
C03
C04
                                                                   MEANS  OF

                                                                   I  SAMPLE

                                                                   2 BEFORE RETEST

                                                                   3 RETEST REPRCOST>0
C05
COG
GO

b
Figure IV-25.  Carbon monoxide mean emissions  - sample, before retest, retest and  repair  cost  <0

-------
emissions and  the before retest mean emissions.  The  hydrocarbons  mean
emissions for  this group are comparable  to the entire retest group.  The
carbon monoxide mean emissions, however, are lower.  In fact, the pat-
tern noticed in Figure IV-23 of retest mean carbon monoxide emissions
being higher than before retest is not as evident among those retest
cars with repair cost greater than zero.  In Figure IV-25, for the two
idle modes, C03 and G06 , the mean emissions of the retest, repair cost
greater  than zero group are less than the before retest mean emissions.
The three dynamic modes still show mean  emissions of the retest, repair-
cost greater than zero group to be greater than the before retest mean
emissions but  for COS, the difference is less.  The changes could be due
to the fact that it is more likely that  the cars in this subgroup have
actually received maintenance.

The cumulative frequency of C03 for the whole sample is compared with the
cumulative frequencies of C03 of the before retest and retest subsamples
in Figure IV-26.  The sample curve is the lowest and the before retest
curve is the highest, as expected.  But  the before retest and retest curves
cross at the far end, indicating that something is affecting the higher
cases.  The cumulative frequencies of the sample and the before retest
and retest subsamples of HC3 are compared in Figure IV-27.  The sample
curve is lowest as expected, but the before retest and retest curves are
reversed, the retest curve being highest.  It appears that HC3 has
increased over time.

Figures IV-28 and IV-29 are cumulative frequency curves of C03 and HC3,
respectively, broken down into.control groups.  The vehicles with no
emissions controls (pre-68) are the highest curves.  As emissions controls
increase (from 68-69 to 70-71 to 72-74), the curves drop but remain
parallel.  A line has been drawn at the  50 percent cumulative frequency
point on both graphs.  This can be used  to determine emissions levels
•to be used as fail criteria corresponding to a 50 percent failure rate -
see Table IV-18.  Similarly, cumulative  frequency curves for carbon
monoxide and hydrocarbons at the dynamic modes - 2500 rpm, 30 mph, and
50 mph - were drawn (Figures IV-30 through IV-35).  But it was found that,
                                IV-44
                                               GCA/TECHNOLOGY DIVISION

-------
O
O
m
r>
IE

O

5
     i
     •P-
     01
               .10



                9



                8



                7
 o
 o:
 LJ  4
                   SAMPLE

                   BEFORE RETEST


                   RETEST
                       i	i	i	I	I	I	I
                                                                                                  7

                           10
                                                                  i	i	|	i	i	i	i	i	i
                        20
30
40        50        60


CUMULATIVE FREQUENCY
                                                                                     70
80
90
                                                                                                                  100
2
<
CO

O
Figure  IV-26.   Cumulative  frequency of carbon monoxide (idle mode)  - sample, before  retest, retest

-------
o
o
n
_i_
O
5
Q
<
C/}
O
     ON
  1000


   900


   800


   700


^  600
2
CL
£:  500
o

X  400-


   300


   200


   100
                             SAMPLE

                             BEFORE RETEST
                             RETEST
                       I	[	I	I	1	I	I
                                                     J	I	I	I	1	I	I	I	1_    I
                            10
                        20
30
40        50        60

 CUMULATIVE FREQUENCY
70
80
                                                         90
                                                                                                                  100
  Figure  IV-270   Cumulative frequency of hydrocarbons  (idle mode)  - sample, before retest,  retest

-------
O
O
rn



O

5

3
O
     H

     <


     -P-
                10




                 9
              HI
              o
              O
              UJ
o  5
o


£  4
UJ
o
oc  3
u
a.

   2



    I
              PRE-68


              68 TO 69


              70 TO 71


              72 TO 75
                                                          I
                                                               I
                                                                   j_
                                                              J_
                            10
                       20
30
40        50       60


CUMULATIVE FREQUENCY
70
80
90
                         Figure IV-28.  Cumulative frequency  of carbon monoxide  (idle mode) by control
100

-------
O
O
Q
~J—
6
5
     co
  1000

   900 -

   800

   700 -

   600 -
S  500
UJ
a  400

$£  300

   200

   100
                            PRE-1968
                            68-69
                            70-71
                            72-75
                                                          i_
                                                               t
                                                                    i
                             10
                         20
30
40        50        60
 CUMULATIVE FREQUENCY
70
80
90
GO
b
              Figure IV-29.  Cumulative frequency of hydrocarbons (idle mode)  by control

-------
10
 o
                5
                
-------
Q
O
— i
m
D

Z
O

5
                600
                500
              a.
              (T
              o
              o
              in
              e\j

              ^r 30°
              "5
              CL
              Q.
              ^ 200
              o
              I
                 100
PRE-1968

68-69

70-71

72-74
                             10
         20
30
40        50        60

 CUMULATIVE  FREQUENCY
70
80
90
                                                                                      100
Q
O
                              Figure IV-31.   Cumulative frequency of hydrocarbons  (2500 rpm) by control

-------
O
O
m
O
IE
Z
O
5
S
Q
g
     H

     Ln
               O  4
               o
               o
               ^  „
               0.  2
	 PRE-68
	 68-69
	70-71
	72-74
      10
                                                              I
                                                                   I
                                                                        I
20
30
40       50        60
CUMULATIVE FREQUENCY
                                                              70
80
90
100
                          Figure IV-32.  Cumulative frequency of  carbon monoxide (30 mph) by  control

-------
o
o
1C
O
5
g
<
CO
O
z
     H
     N>
CL
Q.
                  O
                  X
600


500


400


300


200


100
           	 PRE-1968
           	68- 69
           	70- 71
           	72 - 74
                        10
               20
                                                     I
                                                              j_
                                           30
                                40       50        60

                                CUMULATIVE  FREQUENCY
                                                                                 70
                                                                         80
90
100
           Figure IV-33.   Cumulative frequency of hydrocarbons (30 mph) by control

-------
                              PRE-1968
Q
O
z
O
5
5
2
<
c/2
O
     f
     Ul
     CO
               O
               u
               LJ
               O
               a:
               UJ
               0.
                        	— 68-69
                        	70-71
                        	72-74.
                                                          i
                                                                    i
                                                                         i
                             10
                     30
40        50         60

CUMULATIVE FREQUENCY
70
          80
SO
Figure IV-34.   Cumulative  frequency of earbon monoxide (50 mph) by control
(00

-------
o
o
m
n
O
5
g
O
      i
     Ul
                 600




                 500 -



                 400 -
              3
              a.
              Q.
               - 300
o
X
                 200



                 100
                  PRE-68
           ---  70-71

           ---  72-74
                                                          JL
                                                               _L
                                                                    JL
                                                          J_
               10
20
30
40        50        60

CUMULATIVE FREQUENCY
70
80
                                                                   90
                                                                  100
                            Figure  IV-35.   Cumulative frequency of hydrocarbons  (50 mph)  by control

-------
           Table IV-18.   FAILURE RATE AND FAILURE LEVELS

Pollutant
CO (idle)



HC (idle)



Control
pre 68
68-69
70-71
72-74
pre 68
68-69
70-71
72-74
Emissions levels for failure rates of
40 %
5.3 %
4.8
4.1
3.2
680 ppm
340
290
210
50 %
4.8 %
4.0
3.5
2.1
560 ppm
290
260
180
60 %
4.1 %
3.4
2.8
1.4
480 ppm
260
200
180
for emissions at 30 mph and 50 mph, more than half of the vehicles had
hydrocarbons emissions less than 20 ppm and more than half had carbon
monoxide emissions less than 0,2 percent.  The cumulative frequency
curves can be used for determining failure levels corresponding to fail-
ure rates for the Portland vehicle population.  Even though the sample
and population differed significantly with respect to control, this
variable has been eliminated as an effect by splitting the sample into
control groups.

CORRELATION AND' REGRESSION
Some of the variables believed to affect emissions were studied using
stepwise regression.  Eight different dependent variables were chosen
for the stepwise regression - HC3 (third mode-second idle), HC4 (fourth
mode - 30 mph light load), C03 (third mode), C04 (fourth mode), and the
log of each of these, LOGHC3, LOGHC4, LOGC03, and LOGC04.  The independ-
ent variables used to predict a dependent variable in a stepwise regres-
sion were age, make, mileage, weightclass, and deterioration factors for
hydrocarbons (DETERHC) and for carbon monoxide (DETERCO).  The deteriora-
tion factors, obtained from AP-42, are  factors by which  the emissions of
                                IV-55
                                              GCA/TECHNOLOGY DIVISION

-------
a light duty car when new are multiplied to predict emissions after
time has elapsed.  The deterioration factors depend on the model year
and the age of the car.

Table IV-19 shows for each dependent variable the order of the independ-
ent variables entered in a stepwise fashion into the regression equa-
                      2
tion.  The parameter r , the proportion of the variance in the depend-
ent variable accounted for by the regression equation, is shown in
                                            2
parentheses.  For each dependent variable, r  increases as new independ-
ent variables are entered into the regression equation.  An independent
variable is entered only if the significance of the resulting regression
equation was greater than mere chance.  The most important variable
in all of the regression equations was AGE, with the deterioration
factors and WGTC1ASS next in importance.  MAKE and MILEAGE were the
                               2
least important.  In general, r  values were low, indicating that much
of the variance of the dependent variables remained unexplained after
                                              2
the independent variables were entered.  But r  values were higher for
the LOG variables so it appears that the logarithms of emissions values
are better predicted by the regression equations.
The correlations between the emissions levels of a pollutant at differ-
ent modes are shown in Table IV-20 in six correlation matrices - idle
mode CO, idle mode HC, dynamic mode CO, dynamic mode HC, idle mode CO
by dynamic mode CO, and idle mode HC by dynamic mode HC.  In general,
correlation is better when the emissions are from like modes - both
idle or both dynamic.  Correlation of idle and dynamic hydrocarbons
is better than for carbon monoxide.  More than half of the correlation
coefficients are greater than 0.5, more than two thirds are greater than
0.4.   Correlation is positive for all pairs,  as expected.   It appears
that,  in general, a high reading at one mode indicates high readings at
other modes.
                              IV-56
                                              GCA/TECHNOLOGY DIVISION

-------
Table IV-19.
STEPWISE REGRESSION
Dependent
variable
C03
LOGC03
C04
LOGC04
HC3
LOGHC3
HC4
LOGHC4
2
Independent variables (r )
1
AGE
(0.098)
AGE
(0.121)
AGE
(0.036)
AGE
(0.214)
AGE
(0.144)
AGE
(0.278)
AGE
(0.168)
AGE
(0.378)
2
DETERCO
(0.098)
WGTCLASS
(0.123)
DETERGO
(0.039)
DETERCO
(0.227)
DETERHC
(0.160)
WGTCLASS
(0.287)
DETERHC
(0.174)
WGTCLASS
(0.388)
3
WGTCLASS
(0.098)
DETERCO
(0.125)
MAKE
(0.041)
MAKE
(0.232)
WGTCLASS
(0.168)
DETERHC
(0.296)
WGTCLASS
(0.176)
MILEAGE
(0.389)
4
'MILEAGE
(0.099)
MILEAGE
(0.126)
WGTCLASS
(0.042)
WGTCLASS
(0.234)
MAKE
(0.169)
MAKE
(0.297)
MAKE
(0.176)
MAKE
(0.389)
5

MAKE
(0.126)
MILEAGE
(0.042)
MILEAGE
(0.235)
MILEAGE
(0.169)
MILEAGE
(0.297)
MILEAGE
(0.176)
DETERHC
(0.389)
               IV-57
                             GCA /TECHNOLOGY DIVISION

-------
Table IV-20. CORRELATION COEFFICIENT MATRICES
Idle\ Idle
mode \mode
C01
COS
C03
0.403
—
C06
0.429
0.773
Idle\ Idle
mode \ mode
HC1
HC3
HC3
0.935
—
HC6
0.869
0.895
Dynamic \ Dynamic
mode \ mode
C02
C04
C04
0.393
—
C05
0.199
0.706
Dynamic \ Dynamic
mode \ mode
HC2
HC4
HC4
0.576
—
HC5
0.508
0.687
Idle \ Dynamic
mode \ mode
C01
C03
C06
C02
0.375
0.202
0.255
C04
0.090
0.498
0.478
COS
0.088
0.557
0.506
Idle \ Dynamic
mode \ mode
HC1
HC3
HC6
HC2
0.551
0.577
0.508
HC4
0.605
0.689
0.632
HC5
0.546
0.525
0.576
                   IV-58
                                  GCA/TECHNOLOGY DIVISION

-------
The correlations between the emission levels of carbon monoxide and
hydrocarbons at the same modes were also determined.  Correlation was
positive and low - the correlation coefficient r was never greater than
0.5.  Thus, a high level of one pollutant does not necessarily indicate
a high reading of the other.

TAMPERING DATA

Among the data recorded for each vehicle tested was information concern-
ing the condition of certain under-the-hood and engine parts,  that is,
whether or not there was tampering.  Table IV-21 records the percentages
of vehicles for each engine characteristic where there was tampering.

                    Table IV-21.  ENGINE TAMPERING

CARBURETOR
INTAKE MANIFOLD
'EXHAUST
ADDON
OTHER
Tamper ing
1.9 %
1.3
1.3
2.2
2.4
Table IV-22 shows the percentages of vehicles for each under-the-hood
characteristic within the following categories - not present, one pre-
sent, one present and disconnected, and two present and one disconnected,

RECOMMENDATIONS FOR FUTURE DATA COLLECTION

If data collection is to be continued when the inspection/maintenance
program becomes mandatory, certain improvements could be made to in-
crease the usefulness of the data.  In order to determine emissions
reductions after maintenance and deterioration of emissions reductions
                                IV-59
                                              GCA/TECHNOLOGY DIVISION

-------
                  Table IV-22.  UNDER HOOD TAMPERING

PCV
AIRPUMP
EGR
EVAP
DISTRIBUTOR
PREHEAT
CATALYTIC MUFFLER
Not
present
10.1 %
90.3
87.2
70.6
83.2
42.3
98.8
One
present
88.3 %
8.6
12.5
28.8
16.0
52.3
1.1
Present and
disconnected
1.0 %
1.0
0.2
0.3
0.6
4.0
0.0
One present
one disconnected
0.5 %
0.2
0.1
0.3
0.2
1.3
0.1
over time, it would be desirable to collect all information on a vehicle
into one case, using the license plate number to identify all information
on one vehicle and the date to order all the lines of information.
Background information could be collected on one card for each vehicle
coming in for its first test.  A second card would include maintenance
information and testing results.  Maintenance information might include
when and what kind of maintenance was performed, the cost, and whether
the maintenance was performed by a service station or by the vehicle
owner.  The information from each subsequent test - both maintenance
information and testing results - could be entered on another card and
the cards sorted by license plate number arid date.  The usefulness of
collecting C09 emissions data and information on vehicle tampering should
be studied to determine if this information should continue to be
collected in the future.

There should be specific instructions on filling out the vehicle  testing
data sheets.  All missing values should be filled in with the same code.
Where there are several possible responses to a question, they  should be
clearly defined and coded with numbers.  Someone should be made respons-
ible for editing the data sheets before keypunching.
                                IV-60
                                               GCA/TECHNOLOGY DIVISION

-------
CONCLUSIONS

Comparing the age, make, and control by make distributions of the sample
with the population, it was found that the sample was significantly
different from the population with respect to all three characteristics.
Since emissions also tend to differ significantly between age groups,
between make groups, and between control by make groups, the sample
mean emissions may not be representative of the population mean emissions.
This can be checked by weighting,  Population weighting of the sample for
age and for control by make increased mean idle hydrocarbons emissions.
Weighting for make did not make a difference in idle hydrocarbon mean
emissionso  Idle carbon monoxide mean emissions were not appreciably
affected by weighting for any of the three characteristics.

The sample was also compared with the subsample of vehicles which failed
one or more of the tests.  The fail subsample differed significantly
from the sample with respect to age, make, weight class, and number of
cylinders.

When carbon monoxide and hydrocarbons mean emissions are compared bet-
ween modes, it can be seen that the means are higher for the idle modes
and lower for the dynamic modes.  The geometric means of each pollu-
tant at each mode are in general considerably less than the correspond-
ing arithmetic means, indicating that a number of outlying readings are
tending to increase the arithmetic means.  The emissions tests data
were shown by measures of standard deviation, kurtosis, and skewness
not to be normally distributed.  The large sample size decreases the
length of the 95 percent confidence interval, reduces sampling error,
and increases confidence in the observed data distributions.

Idle carbon monoxide and idle hydrocarbons mean emissions were compared
between the groups within age, make, weight class, control, test, cyl-
inder, exhaust, and retest characteristics.  Both differed significantly
                               IV- 61
                                              GCA/TECHNOLOGY DIVISION

-------
between age groups, control groups, test groups, and retest groups; idle
carbon monoxide mean emissions differed significantly between weight class
groups and cylinder groups; idle hydrocarbons mean emissions between make
groups.  Neither was significantly different between the two dual exhaust
groups, indicating that only one side of a dual exhaust need be sampled to
determine the emissions of a vehicle.

When hydrocarbons mean emissions at the different modes are compared
between the before retest and the retest groups and between the before
retest and the retest, repair cost greater than zero groups, the before
retest mean emissions are higher at all modes.  This would be expected
if maintenance had occurred between the two tests.  For carbon monoxide,
however, the retest group's mean emissions are higher than the before
retest group except at the first mode, which would seem to indicate
that deterioration in emissions rather than reduction through maintenance
had occurred,  This may be a result of not being able to identify
clearly all cars which had received maintenance between the two tests.
That this is possible is indicated by the change in the pattern when
comparing carbon monoxide mean emissions between the before retest and
the retest, repair cost  greater than  zero groups  - the mean emissions of
the three idle modes are less than the corresponding before retest mean
emissions, while the mean emissions of the three dynamic modes remain
greater than the before retest mean emissions.

Cumulative frequency curves of carbon monoxide and hydrocarbons at idle,
2500 rpm, 30 mph, and 50 mph, broken down into control groups, have
been drawn.   These curves can be used for the Portland vehicle popula-
tion,  to determine the emissions levels which correspond to different
failure rates.

Some of the variables believed to affect emissions were studied by using
them in regression equations.  Age was the most important variable, with
deterioration factors and weight class next in importance.  But in
general, less than 40 percent of the variance of any dependent variable
was explained by entering all of the independent variables.
                                IV-62
                                                GCA/TECHNOLOGY DIVISION

-------
The correlations between the emissions levels of a pollutant at different
modes and between pollutants at the same mode were studied.  Correlation
was positive for all pairs and higher when the emissions were from like
modes.  A high reading at one mode appears to indicate high readings at
other modes of the same pollutant, but a high reading of one pollutant
does not necessarily indicate a high reading of the other pollutant at
the same mode-

Of the 6300 tests, only 104 could be identified as possibly occurring
after maintenance had been performed - these tests indicated a repair
cost greater than zero.  Therefore, emissions reductions were not cal-
culated for such a small group.

Data on under hood tampering and engine tampering showed that only a
small percentage of the vehicles had been tampered with - in all cases,
less than 6 percent and in most cases, less than 2 percent.

APPLICATIONS TO AN INSPECTION/MAINTENANCE PROGRAM

Emissions Testing

Comparing the emissions of the left and right sides of a dual exhaust
showed that there was no significant difference between the two sides.
It was concluded that, during emissions testing, it would be necessary
to test only one side of a dual exhaust to determine the emissions of
a vehicle.  However, there may be  some dual exhaust vehicles where .the
emissions of one side are considerably higher than the other side.  This
would result in errors of omission if the high emitting side were not
the side tested.  It is advisable  to determine the rate of  errors of
omission which would occur if only one side of a dual exhaust were tested.
One should decide whether or not the rate of errors of omission is
acceptable in view of the savings  of time which would result.
                               IV-63
                                              GCA/TECHNOLOGY DIVISION

-------
In the analysis of the Portland inspection/maintenance program emissions
data, emissions levels of the idle modes were higher than for the
dynamic modes, so most vehicles which failed, failed due to high emis-
sions in the idle modes.  It would appear from this, that tests of idle
mode emissions would be sufficient to identify high emitting vehicles.
Since the Federal Test Procedure is the standard of measure for
defining vehicular emission levels during typical urban driving patterns,
an effective inspection test should pass only those vehicles which
would pass the FTP.

An analysis of the correlation between the vehicular emission levels
measured by the FTP and the emissions measured by several different
                                         2
short test procedures has been performed.   In general, the results
showed that the best correlation is always achieved for the loaded mode
tests which provide a measure of the mass emission rate.  (Table IV-23)-
The idle mode volumetric measurement consistently achieved the lowest
degree of correlation with the FTP.  The poor correlation between the
idle measurements and the FTP should be considered when comparing the
idle mode and loaded mode inspections.

One comparison of the effects of an idle mode emission inspection and a
dynamic (loaded) mode emission inspection has shown that the tests would
                                                     2
be equally effective in a vehicle inspection program.   Another study
indicated that loaded steady state inspections were more effective than
                                                                        3
idle inspections in reducing carbon, monoxide and hydrocarbons emissions*

Measurements of idle emissions do not provide as much diagnostic informa-
tion so maintenance may be excessive.  Certain kinds of maintenance may
be performed which reduce idle mode emissions but which do not effect,
                                    2
or may increase, the true emissions.   Therefore, art inspection/main-
tenance program would probably be most effective if a dynamic mode test
procedure were used.
                               IV-64
                                              GCA/TECHNOLOGY DIVISION

-------
Table IV-23.  CORRELATION OF FTP AND SHORT TEST PROCEDURES'
Test procedure
Federal short cycle
Seven mode cycle
Key mode (multiple regression)
Steady state modes
(multiple regression)
Idle mode
Seven mode cycle
Key mode (multiple regression)
Steady state modes
(multiple regression)
Idle mode
Emission
measurement
mass
mass
mass
mass
mass
volumetric
volumetric
volumetric
volumetric
Correlation coefficient
HC CO
0.94
0.91
0.96
0.06
0.80
0.57
0.79
0.81
0.35
0.81
0.70
0,81
0.82
0..62
0.77
0.68
0.68
0.50
                         IV-65
                                        GCA/TECHNOLOGY DIVISION

-------
The Portland inspection/maintenance program measured the concentration of
carbon monoxide, hydrocarbons, and carbon dioxide in the exhaust mixture.
Another method, Constant Volume Sampling (CVS), provides an integrated
measure of the total exhaust flow over the test cycle and the sample
which is analyzed provides an integrated measure of the average pollu-
tant concentration during the entire cycle.  The mass emission rate as
a function of vehicle usage can be directly calculated from this.
Table IV-23 shows that the best correlation with the FTP occurs when the
mass emission rate, rather than the volumetric concentration, is meas-
ured.  Therefore, an inspection/maintenance program would probably be
more effective if results were measured as mass emission rate.

The cumulative frequency of emissions levels can be used to determine
failure levels corresponding to different failure rates.  Figures IV-28.
and IV-29 and Table IV-18 were determined from the Portland inspection/
maintenance program idle emissions data.  The curves may be applied to
the Portland vehicle population.

Percent reductions in idle emissions may be estimated using the failure
levels determined from these cumulative frequency distributions, see
Table IV-24.  For each control group, the critical level for a 50 percent
failure rate and the median emissions level for the failing vehicles are
shown; these were determined from Figures IV-28 and IV-29.  Calculation
of the percent reduction from the median emissions value of the failing
vehicles (75 percent on the cumulative frequency curves) assumes that,
on the average, the emissions of the failing vehicles will be reduced
to at least the critical level.  Some vehicles after maintenance will
achieve reductions to emissions less than the critical level, but this
will probably be balanced by other vehicles which cannot achieve the crit-
ical level without excessive cost.  The emissions reductions for the
failing vehicles in each control group are. then adjusted to represent the
reductions for all vehicles in that group (for a 50 percent failure rate,
multiply by 0.5) .  The idle emissions reductions for the sample and for
the population are determined using the proportions of vehicles in the

                               IV-6 6
                                              GCA/TECHNOLOGY DIVISION

-------
                                    Table IV-24.
ESTIMATION OF IDLE EMISSIONS REDUCTIONS

FOR A 50 PERCENT FAILURE RATE
co

O


Pollutant
CO


HC




Control
pre 68
68-69
70-71
72-74
pre 68
68-69
70-71
72-74
Critical level
for 50%
failure rate
(idle emissions)
4.8%
4.0
3.5
2.1
560 ppm
290
260
180
Median level
of failing
vehicles
(idle emissions)
6.5%
5.8
5.0
4.4
1000 ppm
510
410
290
% reduction
from fail
median to
critical level
26%
45
30
52
23%
43 '
38 ^
38

% reduction
for entire
sample
13%
22.5
15
26
11.5%
21.5
19
19




CO
HC

Sample
idle emissions
reductions
17.7%
16.8%
Population
weighted
idle emissions
reductions
17 . 3%
15.4%


40 CFR 51
Appendix N
10%
11%


Portland
TCP
20%
25%
o
o
-H
m
n

z
o

5

-------
control groups.  The population weighted idle emissions reductions for
carbon monoxide and for hydrocarbons are estimated to be 17.3% and
15.4%, respectively.  The assumption on which the estimates are based -
that, on the average, the emissions of the failing vehicles will be re-
duced to at least the critical level - should be checked in future data
analyses to determine the validity of the estimates.  The estimates are
intermediate between the emissions reductions presented in 40 CFR 51,
Appendix N and those presented in the Portland Transportation Control
     4 5
Plan. '   They suggest that a considerable reduction in the vehicle
population's emissions is possible through an inspection/maintenance
program.

Data Collection

The recommendations made in Section G would enable emissions reductions
and deterioration to be determined.

It is important to know the characteristics of a sample and to compare
it with the population.  Bias may have been involved in selecting the
sample.  In that case, weighting of the sample results may be important
in forming inferences about the population.
                               IV-68
                                              GCA/TECHNOLOGY DIVISION

-------
                              REFERENCES


1.   Compilation of Air Pollutant Emission Factors.   AP-42.   EPA.
     Office of Air and Water Programs.

2.   Control Strategies for In-Use Vehicles.   EPA Office  of Air  and
     Water Programs.  Washington, D.C.  (1972).

3.   Effectiveness of Short Emission Inspection Tests in  Reducing Emis-
     sions Through Maintenance.  EPA Office of Air and Water  Programs,
     Ann Arbor, Michigan.   (1973).
4.   38 Federal Register 15197, June 8,  1973.

5.   Portland Transportation Control Plan.
                               LV-69
                                              OCA-'TECHNOLOGY DIVISION

-------