Review of Federal Test
Procedure Modifications
Status Report
February 11, 1993
Certification Division
Office of Mobile Sources
Office of Air & Radiation
U.S. Environmental Protection Agency
-------
Overview and Background
The cornerstone of the Clean Air Act (CAA) is the effort to
attain and maintain National Ambient Air Quality Standards
(NAAQS). Regulation of emissions from on-highway, area, and
stationary sources prior to enactment of the Clean Air Act
Amendments of 1990 has resulted in significant emission
reductions from these sources. However, due to factors such as
the growth in air pollution sources, including dramatic increases
in vehicle miles traveled (VMT), many air quality regions have
failed to attain the NAAQS, particularly those for ozone and
carbon monoxide (CO).
The Clean Air Act, as amended (CAA or Act) contains numerous
provisions that are intended to remedy these continuing air
quality problems. As part of this effort, Section 206(h) of the
CAA requires the Environmental Protection Agency (EPA) to "review
and revise as necessary" the regulations governing the Federal
Test Procedure (FTP) to "insure that vehicles are tested under
circumstances which reflect the actual current driving conditions
under which motor vehicles are used, including conditions
relating to fuel, temperature, acceleration, and altitude."
The FTP is the test procedure used to determine compliance
of light-duty motor vehicles with federal emission standards.1
The FTP is conducted on preproduction vehicles during the motor
vehicle certification process, used to establish that each
vehicle is designed to comply with the appropriate standards for
its full useful life. It is also used to test production line
and in-use vehicles for compliance with emission standards.
The procedure provides a way to consistently and
repetitively measure concentrations of hydrocarbons, oxides of
nitrogen, carbon dioxide/ and carbon monoxide emissions which
occur when a vehicle is driven over a simulated urban driving
trip.2 The principal elements of the test are designed to test
the evaporative and exhaust emissions under several simulated
situations. Evaporative emissions are tested after heating the
fuel tank to simulate heating by the sun (the diurnal test) and
again after the car has been driven and parked with a hot engine
(the hot soak test). Exhaust emissions are measured by driving
the vehicle (placed on a dynamometer) on a simulated urban
driving trip under two conditions: with a cold start designed to
1 The regulations that encompass the many aspects of the FTP are generally
contained in 40 CFR Part 86, Subparts A and B.
: For a detailed discussion of the development of this cycle, see: Kruse, Ronald
E., and Thomas A. Huls, SAE Paper #730553 "Development of the Federal Urban
Driving Schedule," 1973. A speed-time trace of this cycle is contained in 40 CFP.
Part 86, Appendix I.
- 2 -
-------
represent a morning startup after a long soak (a period of non-
use) and then following a hot start that takes place after the
cold start test while the engine is still hot. The FTP also
encompasses all factors relevant to vehicle testing, such as
fuel, vehicle' preconditioning, ambient temperature and humidity,
aerodynamic loss, and vehicle inertia simulations. In addition
to evaporative and exhaust emissions, the FTP is also used in
evaluating fuel economy.
This staitus report addresses the progress EPA has made to
date in complying with the CAA provision and the status of future
research efforts. The first section, "Areas of Potential
Concern," discusses the four general areas of concern with the
FTP noted in §206(h), the basis for each concern, and the
remedies that have been or are being implemented by EPA to date.
The remainder of this report discusses the research program being
conducted by the Agency regarding in-use driving behavior.
Areas of Potential Concern
It is a basic premise that motor vehicle emission levels
determined through the FTP should adequately reflect in-use
vehicle emissions. If in-use driving modes exist that generate
significant amounts of emissions that are not reflected on the
FTP, then the anticipated benefits from motor vehicle standards
are not being fully achieved.
It is also basic that no test procedure can reasonably
duplicate all in-use conditions. The overall goal of the
Agency's review of the FTP is to aid in determining whether or
not the FTP should be modified to reflect in-use conditions not
currently found in the test and, if so, what modifications should
be made. To meet this goal it is not enough to simply examine
factors such as ambient temperature ranges, in-use fuel
characteristics, or driving patterns. For example, qualitative
evidence has existed for years that certain types of actual
driving behavior are not represented on the FTP, such as high
acceleration rates. However, it would be counterproductive to
modify the FTP unless two conditions are met. First, the driving
behavior or other condition not represented properly by the FTP
should contribute a significant amount to motor vehicle
emissions. If it does not, then modifying the FTP would incur
substantial costs and disruption with little or no air quality
benefit. Second, any modification to the FTP should be expected
to promote design improvements to vehicles and thereby create
real improvements in controlling in-use emissions. If the
current FTP is already effective in reducing emissions during the
non-FTP driving or other conditions, then modifying the FTP would
again incur substantial costs with little or no air quality
benefit. Even if off-cycle emissions exist that are not properly
controlled by the FTP, it is critical to ensure that FTP
- 3 -
-------
modifications will actually promote the proper design
improvements. The Agency believes this approach is a reasonable
way to implement Section 206(h)'s requirements.
Section 206(h) of the Act specifically requires that EPA
consider four potential areas of concern: fuel, temperature,
acceleration, and altitude. The Agency has identified several
other potential areas of concern relating to driving behavior:
speed, cold starts (frequency and driving behavior), trip length,
time between trips, and road grade.
Fuel
I. Gasoline.
The composition of the gasoline used for the FTP (commonly
referred to as indolene) was established by regulation over 20
years ago.3 While it was representative of in-use fuel at the
time, commercial or in-use fuel properties have changed
significantly since then, in some cases having a major impact on
vehicle emissions, both tailpipe and evaporative.4 Studies
conducted during the 1980's indicated that vehicles tended to
emit higher emissions using commercial gasoline than indolene,
particularly through evaporative losses. To address this
concern, EPA established volatility limits for gasoline and
alcohol blends. These regulations capped the allowable Reid
vapor pressure for commercial gasoline during the summer months.
The second phase of these controls became effective in the summer
of 1992.5 As a result of these actions, the emissions of a
vehicle fueled with indolene are more representative of the in-
use emissions results of a vehicle fueled with commercial
gasoline.
II. Diesel fuel.
The Agency has also taken steps to reduce the sulfur content
of in-use diesel fuel. Regulations were published on May 7, 1992
to reduce sulfur content in diesel fuel, to take effect on
October 1, 1993.'
III. Alcohol and other fuels.
The Agency promulgated regulations in 1989 which established
emission standards and test procedures for vehicles fueled with
methanol and proposed similar regulations in 1992 for vehicles
3 40 CFR Part 86 Section 86.113-94.
4 Evaporative emissions include diurnal, hot soak, refueling, and running losses.
5 55 FR 23658 (June 11, 1990).
6 57 FR 19535 (May 7, 1992).
- 4 -
-------
fueled with compressed natural gas and liquefied petroleum gases.
At this early stage of alternative fuel development, it is
impossible to know what the real-world fuel compositions will be
for any of these fuels when used in automotive applications. In
each of theses rulemakings, EPA has avoided adoption of narrow
fuel specifications, specifying instead that test fuels be
representative of typical in-use fuels.
These recent requirements established for in-use fuel appear
in general to satisfy §206(h)'s requirements regarding fuel use
in the FTP. In addition, the Agency is also addressing the use
of representative fuels in the Certification Short Test
rulemaking, for which a proposed rule is expected to be published
in January.
Temperature
The FTP is conducted between 68 and 86 degrees Fahrenheit,
and includes a cold start in its driving cycle.7 Vehicle
emissions after a cold start increase at colder temperatures as a
richer mixture is employed to ensure sufficient fuel vapor for
combustion to occur. In addition, colder temperatures lead to
longer warm-up times. This is not a major concern for ozone,
which is primarily a summertime phenomenon, but it is for CO.
Most CO exceedances occur from December to March and over half
occur at temperatures below 45 degrees Fahrenheit.
To reduce the emissions generated from motor vehicles during
cold temperature operation, EPA recently issued 20°F CO emission
standards and test procedure. These regulations were issued on
July 17, 1992 and are phased in beginning with the 1994 model
year.9 The regulations also established interim temperature
defeat device criteria to maintain proportional CO emission
control between the 20 degree standard and the warm temperature
-standards. These regulations insure that the Agency's test
procedures properly reflect the impact of temperature on CO
emissions. As the cold CO rules will prevent emission step-
functions just below 68 degrees that could also impact
hydrocarbon (HC) emissions, they will also insure that the FTP is
representative of HC emissions at colder temperatures.
At warmer temperatures the primary emission concern is
increased fuel evaporation. The Agency is in the process of
revising its evaporative test procedures to address a number of
concerns, including temperature. The final regulations are
expected to specify ambient test temperatures of 95°F. These new
test requirements should insure that vehicles can control
evaporative emissions for most in-use events.
The Agency believes that the above FTP changes appear in
1 An engine start is considered to be a "cold" start if it is preceded by a long
uninterrupted soak, such as those starts that occur after an overnight soak.
57 FR 31888 (July 17, 1992).
- 5 -
-------
general to satisfy §206(h)'s requirements regarding temperature
conditions in the FTP.
Altitude
It has long been recognized that at high altitude locations,
if there is no compensation for the lower air density, engines
tend to run rich more frequently and emit more HC and CO
emissions. Virtually all light-duty vehicles have been required
to meet emission standards at both low and high altitude without
adjustment or modification since the 1984 model year. Light-duty
trucks and light-duty vehicles have had separate high altitude
standards since the 1982 model year. Regulations published on
June 5, 1991 will require light-duty trucks to meet emission
standards at both low and high altitude without adjustment or
modification beginning with the 1997 model year.' The cold
temperature CO regulations require that both light-duty vehicles
and light-duty trucks meet the standard at both low and high
altitude without modification. The FTP does not specify an
altitude range in which the test must be conducted. In effect,
the regulations allow the FTP to be conducted at any altitude and
this, in fact, occurs.
As with fuel and temperature parameters for the FTP, EPA
believes that the above requirements appear in general to satisfy
§206(h)'s requirements regarding altitude conditions in the FTP.
Driving Behavior (including acceleration)
Current technology vehicles have achieved impressive
reductions in emissions during normal operation, primarily due to
catalyst technology development. Catalyst conversion
efficiencies (i.e. the rate at which HC and CO are oxidized into
carbon dioxide (C02) and water vapor, or oxides of nitrogen (NOx)
are reduced to nitrogen and oxygen) in a modern, properly
-operating, warmed-up vehicle can simultaneously exceed 98% for
HC, 99% for CO and 90% for NOx. This includes typical transient
urban traffic operation, such as that represented by the FTP.
However, these simultaneous catalyst conversion efficiencies are
only achievable with a three-way catalyst in a very narrow range
of fuel/air ratios around the minimum theoretical air requirement
for complete combustion (called stoichiometry). Thus, modern,
properly operating vehicles are designed to operate at
stoichiometry as much as possible during the FTP.
There are two types of operation that make it difficult to
operate an engine at stoichiometry. The first type of operation
is cold starts. Fuel must be vaporized with air to combust
properly. When the engine is cold there is not enough heat to
properly vaporize the fuel and additional fuel must be added for
proper operation. Cold start emissions are also increased due to
the lack of conversion activity in the catalyst until it heats up
56 FR 25724 (June 5, 1991).
- 6 -
-------
(little catalyst activity occurs below about 600 degrees
Fahrenheit). Thus, emission rates during cold starts can be 20-
100 times the emission rates during stoichiometric operation. In
fact, the vast majority of emissions from modern, properly
operating vehicles on the FTP occur during the first 10% of the
test, before the engine and catalyst have warmed up. Thus an
important question is whether the cold start portion of the FTP
properly reflects the proportion of time vehicles actually spend
in the warm-up mode.
The second type of operation that makes it difficult to
operate an engine at stoichiometry is high engine loads. High
loads on an engine running at stoichiometry can dramatically
increase engine and catalyst temperatures. These elevated
temperatures greatly increase NOx emissions and can cause engine
knocking and/or damage to the catalyst. The performance and
driveability of an engine under high load can be improved by
running with a richer mixture of fuel. Thus, to prevent
overtemperature damage to the catalyst and insure the best
possible driveability and performance, manufacturers often design
their vehicles to run rich at high loads. While this reduces NOx
emissions, it increases HC and CO by almost the same 20-100 times
factor as cold start operation. An important question then is
whether a significant amount of high load operation occurs in-use
that is not reflected on the FTP. Due to the nonlinear nature of
the emission rates, this amount of driving could actually be
fairly small and still have a significant emission impact.
There are a wide variety of in-use factors that impact the
amount of time vehicles spend in either a warm-up or high-load
mode. Warm-up factors include distributions of trip length, time
between trips (referred to as "soak time"), ambient air
temperature, initial idle time, and driving behavior. Factors
that can cause high loads on a vehicle include high acceleration
rates, high speeds, road grades, air conditioning operation, or
some combination of factors (such as moderate acceleration up a
moderate road grade). Complicating the assessment is the fact
that different vehicles have very different calibration
strategies. Thus, the impact on emissions of the exact same
driving behavior may vary widely from vehicle to vehicle.
To properly address the emission impact of driving behavior,
very detailed and statistically valid data are needed for both
actual driving conditions and the impact of this driving on
emissions from a wide variety of vehicles. When the CAA was
amended in 1990, very little information was available on any of
the driving behaviors or conditions discussed in this section, or
on their emission impacts. It was therefore necessary for EPA to
conduct a research program on actual driving behavior and the
emissions impact of such driving.
Raaearch on Driving Behavior
Past qualitative assessments have concluded that the FTP
effectively represents in-use emission reductions from vehicle
- 7 -
-------
emission standards (i.e. any off-cycle emissions did not occur
with enough frequency to have a significant impact) . However,
the larger the baseline emissions, the smaller the impact from
off-cycle emissions. Other CAA mandates such as Tier I emission
standards and longer useful life will all serve to reduce the
baseline emissions measured by the FTP. If adopted, Tier II
emission standards would have a like effect. Thus, any off-cycle
emissions will become relatively more important in the future.
The research program undertaken by the Agency is designed both to
quantify any emission impacts from off-cycle driving behavior and
to provide information needed to determine whether or not EPA
should make regulatory changes to the FTP.
The program developed by the Agency to evaluate driving
behavior contains three basic components. First, to determine
how vehicles are actually driven, an extensive amount of vehicle
monitoring was conducted. This is described in the next
section, "In-Use Assessment of Driving Behavior." Second, the
data from the vehicle monitoring is being analyzed to determine
cycle and trip information and the impacts of different factors
on driving behavior and to develop driving cycles that represent
the complete range of actual driving behavior. This part of the
program is described in the section, "Analysis of Driving
Behavior Data." The third part of the program involves assessing
the emission impact of the driving behavior. This is being
pursued both by development of a computer simulation model and by
vehicle testing, as described in the section, "Emission
Assessment of In-Use Driving Behavior."
These three sections are designed to describe the entire
research program from start to finish. The final section of this
report, "Status and Plans," identifies which portions of the work
have already been completed, which are currently in progress, and
which are still to be done. It outlines the status of the tasks
discussed in the sections on the research program and sets up a
general schedule for work that has not yet been completed.
In-Ua« Assessment of Driving Behavior
Outreach
From the start of the FTP study, EPA has made a concerted
effort to inform all interested parties of our plans to evaluate
in-use driving behavior and to solicit input and participation.
The first public meeting was held at the EPA National Vehicle and
Fuel Emissions Laboratory on December 20, 1990 to discuss EPA's
plans for the FTP study. This meeting was well attended and
generated considerable interest among motor vehicle
manufacturers. A second meeting was requested by auto
representatives to allow them to respond to several issues
discussed at the initial meeting. In these subsequent meetings,
the auto manufacturers demonstrated a willingness to work with
- 8 -
-------
EPA to ensure a thorough and successful study. As a result of
these meetings, an Ad Hoc Panel on the FTP was formed by the
Motor Vehicle Manufacturers Association (MVMA) and the
Association of International Automobile Manufacturers (AIAM).
As part of the outreach effort for the FTP Study, EPA held
discussions with the State and Territorial Air Pollution Program
Administrators (STAPPA) and the Association of Local Air
Pollution Control Officials (ALAPCO) members during the first
part of 1991. These talks led to the establishment of a
Cooperative Agreement with New York State's Albany Emission
Laboratory (AEL). The agreement called for AEL to conduct vehicle
testing to examine engine and catalyst cool-down temperatures and
their relationship to motor vehicle emissions.
Driving Behavior Data Acquisition
A centerpiece of EPA's approach to the FTP Study is the
examination of how vehicles are operated in the real world. Early
on, EPA reviewed existing data sources on current driving
behavior and found them deficient for our task. To truly
understand driving behavior EPA felt that it was critical to
collect real-time driving data on a representative sample of
drivers. The MVMA/AIAM Ad Hoc Panel on the FTP supported this
approach and talks were held to discuss a cooperative research
effort.
A meeting was held in Atlanta, Georgia in June of 1991 to
discuss options for conducting surveys of in-use driving
patterns. The meeting included experts from industry,
government, and academia. Two EPA contractors gave presentations
on alternative driving survey approaches. As a result of the
meeting, EPA concluded that two complementary approaches were
necessary; data were to be collected using both a chase car
approach and an instrumented vehicle approach. Each approach is
described below.
The next step was the selection of cities to survey.
Resource constraints limited the surveys to two cities. In
addition, the two driving survey methodologies restricted the
choices for possible cities. For example, to minimize potential
bias in the selection of drivers and vehicles, the instrumented
vehicle study required recruiting drivers from centralized
Inspection and Maintenance (I/M) stations. The chase car
approach dictated that a city possess an up-to-date
transportation network model. These requirements limited the
choices to a handful of prospective cities. The Agency selected
Baltimore, Maryland and Spokane, Washington. Baltimore
represents ai major urban ozone non-attainment area and is within
the Northeast corridor. In contrast, Spokane is characteristic
of a smaller, cold CO non-attainment area.
Additional information on in-use driving behavior has been
collected in Los Angeles, California and Atlanta, Georgia. The
California Air Resources Board (CARB) has utilized the chase car
approach to collect data in Los Angeles. The Agency and CARB
_ Q —
-------
have coordinated activities from the start of the FTP Study; the
GARB chase car study closely paralleled EPA's chase car efforts
and was conducted by the same contractor.
The Atlanta work was done in coordination with EPA's Air and
Energy Engineering Research Laboratory (AEERL) of the Office of
Research and Development (ORD). AEERL is conducting research in
Atlanta as part of a long-term project to improve mobile source
emissions inventories and emissions modeling. With financial
support from AEERL, an instrumented vehicle survey was conducted
in Atlanta during the summer of 1992 using the 3-parameter
dataloggers developed for the FTP study. This study, along with
the Los Angeles study, will improve the regional representation
of the original in-use driving surveys.
Instrumented Vehicle Study
The instrumented vehicle study consists of placing
instruments in individuals' privately-owned vehicles.
Datalogging devices are mounted inconspicuously under the
vehicle's hood and data are collected for about one week. To
minimize potential sampling bias, drivers are recruited randomly
from centralized I/M stations using a formal
recruitment/replacement protocol.
To date, a major accomplishment of the FTP study is the
cooperation and support of both MVMA and AIAM. The instrumented
vehicle component of the FTP study in Baltimore and Spokane was a
joint effort with MVMA and AIAM. After EPA designed the basic
program and established a contract with Radian Corporation to
conduct the testing, the MVMA/AIAM Ad Hoc Panel on the FTP
contributed funds and specialized instrumentation to the
contractor to augment our base program. The MVMA/AIAM Ad Hoc
Panel also agreed to allow EPA to manage the entire data
.collection effort. The joint goal was to instrument 144 vehicles
•in each city, for a total of 288 vehicles, as described below:
100 EPA sponsored 3-parameter vehicles
98 MVMA/AIAM sponsored 3-parameter vehicles
90 MVMA/AIAM sponsored 6-parameter vehicles
288 Total
The 3-parameter instrumentation measures vehicle speed,
engine speed in revolutions per minute (rpm), and manifold
absolute pressure. The 6-parameter instrumentation also collects
information on coolant temperature, throttle position, and the
air/fuel ratio. The latter instrumentation packages were custom-
designed by each manufacturer. The 3-parameter instrumentation
was developed by Radian Corporation. At the conclusion of the
program, the 55 3-parameter dataloggers were turned over to EPA
for future use.
A pilot study was conducted in Spokane on January 6-10,
1992. Alternative solicitation and incentive strategies were
- 10 -
-------
evaluated and 4 vehicles were instrumented. Data collection for
the full study in Spokane began February 3 and was completed the
first week of March. The Baltimore study followed, with data
collection completed in early April, 1992.
Chase Car Study
The traditional chase car approach typically involves
driving an instrumented vehicle in a manner that simulates the
driving behavior of the vehicle being "chased." The methodology
used by the Agency in this project is an enhanced version of this
traditional approach. The new chase approach uses a chase car
which is instrumented with a grill-mounted laser rangefinder.
With the laser rangefinder, it is possible to accurately
calculate the speed of a target vehicle without the chase car
having to emulate the target car's driving behavior. The chase
car is driven over representative road routes, which are
generated using a transportation network model. The strengths of
this approach are its ability to collect driving patterns data
for a large sample of vehicles and the virtual elimination of the
bias introduced by the drivers knowing that their driving
behavior is being monitored.
The laser rangefinder was a modification of a new
technology/ a hand-held laser gun used by police for identifying
speeders. Pilot tests of the laser-equipped chase car were
conducted in the summer and fall of 1991. Data collection for
Baltimore study began in November of 1991; the contractor, Sierra
Research, drove a total of 248 routes finishing December 20.
Following enhancements to the laser and other on-board electronic
equipment, Sierra Research carried out the CARB-sponsored Los
Angeles chase car study in the Spring of 1992. The Spokane chase
car study was completed by the end of July, 1992.
AEL Research
The current FTP implicitly assumes that all starts are
either "hot" starts (represented by an engine off time of 10
minutes before restart) or "cold" starts (represented by an
overnight soak before restart). However, almost half of all
restarts in-use occur after a soak period of 10 minutes to 4
hours, for which the engine and catalyst may be in intermediate
temperature conditions. The New York State Automotive Emission
Laboratory, under a grant from EPA, conducted testing to study
the effect of soak time and ambient conditions on engine and
catalyst temperatures. To date, AEL has drafted 4 progress
reports which presented results of the testing. This information
will provides insight into the condition of the engine and
catalyst during starts that fall into these intermediate
temperature conditions.
Analysis of Driving Behavior Data
- 11 -
-------
Data Summary
The instrumented vehicle and chase car studies have produced
very large data sets that must be synthesized to permit useful
study. Both surveys collected second-by-second measurements of
several variables, resulting in millions of individual data
points. Following data collection, EPA's contractors edited
these data in order to validate their accuracy and adjust for
known limitations of the measurement equipment. Data suspected
as being significantly flawed were withheld from subsequent
analyses pending more detailed examination.
In the first phase of this data synthesis, EPA constructed
lists of summary statistics for the two sets of survey data based
on a review of past driving studies and on an assessment of the
current project's requirements. These lists were provided to the
individual survey contractors, who developed computer programs to
generate the requested values. These outputs were furnished to
the agency in the form of paper hardcopy as well as electronic
data files. The data summary phase offers an initial overview of
the survey outcomes and provides inputs to further, more in-depth
analyses. In this way, a number of questions can be addressed
using a greatly reduced data set and considerably fewer computer
resources.
The data summary was designed to comprehensively describe
recorded driving behavior and to anticipate the study of
emissions impacts. The basic description centers on three
aspects of driving: overall speed and acceleration patterns,
instrumented vehicle trip or chase car route characteristics, and
variation in individual vehicle/driver behavior. Factors
suspected to have a disproportionate impact on emissions are
given special emphasis. These include the quantity of high
acceleration driving and the proportion and type of driving that
occurs under cold engine conditions (discussed below).
A number of measures are contained within the basic
descriptive categories. In addition to speed and acceleration, a
measure known variously as specific power or positive kinetic
energy (pke) was compiled. As a composite of the first two
values, this measure has been suggested as a useful predictor of
high emission episodes/ a claim supported by preliminary analysis
of emission test data collected as part of this study.
For the instrumented vehicle survey, trip characteristics
identified as needing further analysis include:
time and distance of a trip,
soak time between trips,
time in idle versus moving, and
average speed and specific power.
.»
Vehicle/driver measures needing analysis consist of:
average daily driving time,
driving distance,
number of trips day,
average stops per hour,
- 12 -
-------
fractions of time spent at high levels of speed, and
acceleration and specific power.
In the chase car study, route measures are similar to the
trip measures given above. In addition, statistics are
calculated on the portion of survey time during which the laser
was locked on a target vehicle.
Several types of summary measures were requested from EPA's
contractors. Measures of location and dispersion (mean and
standard deviation), minimum, and maximum capture the essential
features of a. particular sample. Greater detail is found in
frequency distributions and their associated graphical display.
Because of the concern with the emissions impact of non-FTP high
speed, acceleration, and power, the upper percentiles of these
distributions are computed explicitly to support more detailed
study of these "extremes" of driving behavior. Finally, the
combined distribution of speed and acceleration will be tabulated
to enable more detailed study of engine load patterns.
The driving behavior surveys collected or identified
different factors that could influence driving behavior, such as
type of vehicle, location, and road congestion. The summarized
data describesd above will also be broken down by these factors.
Analysis of these breakdowns enables the assessment of potential
bias as well as the relative influence of measured factors on
driving. For the instrumented vehicle study, criteria needing
examination for impact on driving behavior include:
vehicle age,
vehicle performance,
transmission type,
time of day and week,
driver age,
recruitment site, and
observation phase (first day or later).
Criteria collected by the chase car study needing study include:
road type,
road grade,
road congestion level, and
target vehicle performance type.
Survey Bias
In order to judge possible bias induced by the data
acquisition methodologies, the study includes several types of"
comparisons. For the instrumented vehicles, the breakdown by
observation phase is used to examine: (1) whether survey
participant driving behavior is influenced by the presence of t:he
datalogger/ and (2) if any such influence diminishes over time.
Another potential concern with the instrumented vehicle data
involves driver refusal to participate. Drivers of high-
performance or luxury vehicles were less likely to agree to have
their vehicles instrumented. If these drivers and vehicles have
- 13 -
-------
different operating characteristics than the population at large,
it could have a significant impact on the results of the study
due to the disproportionate impact of brief periods of high
acceleration .on emissions. While vehicle solicitation was
designed to replace drivers who refused to participate with
drivers and vehicles with similar characteristics, the data will
also be analyzed to try to determine if bias occurred and, if so,
how to analytically correct for the bias.
There are three primary concerns with potential bias from
the chase car study. The first is the representativeness of
driving behavior on the selected routes to overall driving
behavior. The second is the accuracy of the laser rangefinder/
i.e., does it properly reflect the speed and acceleration of the
target vehicles. The third is whether aggressive drivers are
properly represented in the data base. While the methodology
used to select vehicles to follow was carefully constructed to
ensure the proper selection of all types of drivers/vehicles, the
chase car is more likely to lose aggressive drivers prematurely.
Analysis of whether or not this occurred, and determination of
analytical methods to correct for this if it did, are in
progress.
Another potential source of bias is the locations selected
for the driving surveys. Study of the surveys'
representativeness to other areas and analyses of the differences
in driving behavior for the surveyed areas will permit adjustment
of estimates to reflect target populations. With the
instrumented vehicle survey, detailed vehicle information can be
accessed to enable weighing of summary statistics on vehicle
characteristics. For the chase car data, it is possible to
compare route features with those of the population. The driver
aspect of this analysis is restricted due to the limited
-collection of driver demographics.
Cold Start Analv«i«
Because it is known that driving under cold engine
conditions contributes much higher emissions than after warm-up,
special attention is being given to analyzing cold start driving.
The basic issue concerns differences in driving behavior under
cold and warm engine conditions, including the initial idle time.
This poses a number of problems in estimating the portions of the
survey driving that occurred in the cold state. Warm-up time
varies with the vehicle, pre-start soak time, initial idle time,
and type of driving, making it difficult to classify engine and
catalyst temperature. Data from the six-parameter instrumented
vehicles include coolant temperature which will be studied for
insights into the problem. The work done by the New York State
Auto Emissions Laboratory will also help classify starts into
cold, warm, and hot categories.
Cycle Dev+lopmant
To assess the impact of non-FTP driving, it is necessary to
- 14 -
-------
estimate the difference between emissions predicted by the FTP
cycle and emissions that occur in actual driving. Using either
computer models or dynamometer testing, this assessment requires
the development of one or more driving cycles that are
representative of the real world. The driving survey data
discussed above will serve as the primary input to this component
of the project.
Several approaches to cycle development currently are under
review. These vary considerably in level of subjectivity. One
approach, used in developing the current FTP, is to splice
together segments of real speed patterns that are selected from
the survey data. A final cycle is obtained by matching summary
features of the resulting speed-time trace with those of the full
sample. A virtue of this and related approaches is their basis
in real driving experience that can be reproduced in dynamometer
testing. The choice of segments and matching criteria are
potential difficulties.
A more directly quantitative approach to cycle development
is to generate a vehicle speed-time trace using Monte Carlo
simulation. Simulated second-by-second values are chosen
according to statistical criteria derived from the survey data.
Cycles are subjected to matching criteria in order to screen out
unsatisfactory candidates. This is likely to be a more efficient
method of producing different cycles, but these cycles are wholly
"unreal" in comparison to the splicing approach described above.
Emission Assessment of In Use Driving
Approach
In analyzing data from the in-use driving surveys it is
-essential to consider the emissions impact of the real-world
"driving patterns that are not represented by the current FTP
driving cycle. As discussed, above, this requires assessment of
a wide range of driving behavior, factors influencing emissions,
and manufacturer calibration strategies. In order to perform
these large scale assessments, EPA is developing a computer model
which simulates vehicle emissions over any desired driving cycle.
EPA is using the modeling approach because it affords flexibility
in analyzing the emission impact of the driving survey data and
could allow us to conduct a smaller vehicle testing program. A
simulation model will allow the emission assessment of a number
of unedited and/or composite driving cycles over a large number
of vehicles with relative ease. Conversely, a strict vehicle
testing-based approach for an initial assessment of the emission
impact of in-use driving behavior would limit the assessment to a
small number of composite cycles over a relatively small sample
of vehicles, and would not allow the needed flexibility for the
type of large-scale assessment desired by EPA. Vehicle testing
will be used during the course of the emission assessment effort
- 15 -
-------
in order to validate the results of the computer simulation
model. Contingency testing plans to gather basic emission impact
information are also being prepared in case the model proves to
be too inaccurate for qualitative analysis.
Emission Simulation Model
The simulation model computes instantaneous fuel and
emission rates based on instantaneous vehicle speed. This model
is currently being developed as two components, known as VEHSIM
and VEMISS.
The VEHSIM component was originally developed by GM and
later revised by the Department of Transportation. The VEHSIM
model takes instantaneous (generally second-by-second) vehicle
speed inputs and calculates instantaneous engine speed and load.
These calculations of engine speed and load are performed
utilizing vehicle information regarding vehicle aerodynamics,
drivetrain, transmission, and engine accessories stored in a
database known as a part library.
The second component of the model, VEMISS, was developed by
EPA to provide fuel and emission rate calculations based on the
engine speed and load inputs produced by VEHSIM. VEMISS uses a
series of lookup tables, known as engine maps, to simulate the
fuel and emission rates for a particular vehicle. An engine map
contains fuel and emission rates over a matrix of engine speed
and load. VEMISS implements an interpolation method with the
engine map in order to calculate fuel and emission rates for an
instantaneous engine speed and load.
Upon completion of VEHSIM and VEMISS, these components will
be linked to produce a fully functioning model capable of
simulating instantaneous fuel and emission rates based on an
inputted speed/time trace.
Engine Map Development
EPA's goal is to determine the emission impact of actual
driving behavior on technology that is available today and will
likely be available in the next few years. To this end, EPA
tested a fleet of 29 late model, current technology, low mileage
vehicles which cover a broad range of vehicle types (both car and
light truck). The objective is to enable EPA to match the
driving characteristics of any vehicle in the driving survey
sample with a representative vehicle from the 29 vehicle fleet.
The Agency has completed the testing and development of
engine maps for each of the 29 vehicles. Currently, EPA is
working to compile the part libraries for these vehicles so that
the VEHSIM/VEMISS model will be able to simulate fuel and
emission rates using any speed/time trace input for each vehicle
in the fleet.
Warm Modal Validation
Before the model can be used to confidently assess the
emission impact of in-use driving, a validation of the model must
- 16 -
-------
be performed. This validation effort is in progress and
currently focuses on the component of the model which simulates
emissions under warm operating conditions. The purpose of the
first phase validation is to assess the accuracy of the warm
model and highlight the refinements necessary to improve the
accuracy of the model. This is being done by comparing the model
output to actual data over a series of test cycles, including the
FTP and a high acceleration cycle. Future validations will
include comparisons of model results to test results run using
new test cycles on a single large-roll electric dynamometer
(which should much more accurately reflect in-use emissions).
Refinements may include improvements to the engine maps, part
libraries, and the model itself. Validation will continue in an
iterative fashion for each vehicle as improvements are made.
Cold Modal Development
Cold start emission simulations also need to be developed to
estimate the impact of cold start driving behavior and soak time.
As the cold start simulation will calculate emissions using the
warm engine-out emission maps as the starting point, development
of the cold start module is being sequenced behind the warm
component of the model. Once the warm component is
satisfactorily validated, the cold operation component will be
developed based on existing test data on the 29 vehicles,
integrated into the model, and validated. Upon completion, the
model will be able to simulate fuel rate and emissions for an
entire vehicle trip.
NPRM Development
The Agency will use the analyses described above in
determining whether or not the driving cycle or other aspects of
the FTP should be revised to properly represent vehicle emissions
during actual driving conditions. However, in any proposal to
revise the FTP, EPA would also need to consider various other
issues, including:
- Technology assessment
- Type of revision needed
- Lead time
- Cost and cost effectiveness analyses
The technology assessment includes determining the changes
needed for manufacturers to reduce emissions during the
identified off-cycle condition, the level of reduction achievable
with different technologies and/or calibration strategies, and
the feasibility of making the technology changes. Closely
related issues are cost and lead time, as greater levels of
technology change or added component requirements will increase
the cost of the regulation and, possibly, increase the lead time
needed for manufacturers to implement the changes.
- 17 -
-------
The type of revision to the FTP would also have to be
considered. For example, revisions to the existing cycle would
impact the usefulness of the vast array of historical data and
would require assessments of the impact on CAFE and on the
stringency of the emission standards. It may be much more cost
effective, instead, to establish a new cycle and standard (much
as was done for cold temperature CO emissions, where a new 20
degree cycle was established with a separate standard). If the
emission benefits prove to be relatively minor, or if it appears
that emissions could be effectively reduced without standards, it
might be desirable to simply promulgate stronger defeat device
requirements. The basic strategy to improve the FTP would have
to be evaluated, as well as the impacts on costs and emission
benefits.
Evaluation of this wide array of control strategies,
technology requirements, standard stringency, costs, and benefits
is a complex task. The level of complexity will be significantly
impacted by the results of the study in regards to the level of
off-cycle emissions and the type of driving generating the
emissions.
Status and Plans
The above sections describe the programs being implemented
by EPA to assess the emissions impact of driving behavior. This
section outlines the tasks completed to date, work in progress,
and plans to complete the study. In order to keep this outline
reasonably simple and understandable, only very brief statements
are made to describe each item. The purpose of this section is
simply to show how all the pieces come together to complete the
study. Detailed descriptions of each piece have already been
presented in the previous sections of this report.
- 18 -
-------
Driving Behavior Assessment
Emission Assessment
Tasks Completed to Date
Vehicles instrumented:
Spokane - 102 3-parameter
42 6-parameter
113 3-parameter
37 6-parameter
(ORD) - 110 3-parameter
Baltimore -
Atl.
Chase car:
Spokane - 249. routes
Baltimore - 248 routes
L.A. (ARE) - 202 routes
Data processed (Baltimore and
Spokane)
Summary statistics defined
1st draft data summaries:
Spokane instrumented veh
Baltimore instrumented veh
Baltimore chase car
Cool-down study (N.Y. AEL):
6 vehicles tested
4 progress reports
29 vehicles mapped:
1990-92, MPI
40-55 steady state points
7 cold start steady state
FTP
Calif, accel cycle
Data cownloaded to PC
Engine torque calcualted
Warm emission module written
(VEMISS)
1st cut evaluation of maps
Work in Progress
Trip definition
Bias analyses/correction
Recovery of suspect vehicles
Process Atlanta data
Phase I warm speed/accel
analysis (irist. veh.)
Trip definition
Torque verificaiton
Soakl time emission impact
Compile parts library
Validate warm modeling/maps
(2 vehicles)
Summarize test data - all veh.
- 19 -
-------
By January, 1993
Complete cooldown analysis (AEL) Begin cold start model
Analyze: development
Trip length and soak time Enhance part library data
Cold start driving behavior Validate VEHSIM/part library
Warm driving behavior methodology
Compare modal data to VEMISS
predictions
Assess modeling effectiveness,
inc. testing validation
Develop contingency test plan
By March, 1993
Develop warm driving cycles Complie cold start database
Analyze effects of driving 1st cut warm speed/accel
behavior influences on: emission evaluation
trips and soak time Compile/validate part libraries
cold driving - all vehicles
warm driving
Compile draft of driving
behavior study for review
Hold public workshop on analysis
of driving behavior study
By May, 1993
Publish preliminary technical Validate warm VEMISS/maps on
report; request public rest of vehicles
comment
By S«pt«mt*r, 1993
Complete enhancements to model
Complete cold start module
Validate model with test data
Assess emission impact of all
driving behavior
By Nov*mb«r, 1993
Final study report, including recommendations, ready for internal
Agency review/ begin development of an NPRM to revise the FTP or a
notice of intent to not revise the FTP.
- 20 -
------- |