EPA-AA-SDSB-82-11
Technical Report
Operational Characteristics Study
Instrumentation Systems
By
Glenn Thompson
Gayle Klemer
Terry Newell
Martin Reineman
June 1982
NOTICE
Technical Reports do not necessarily represent final EPA decisions
or positions. They are intended to present technical analysis of
issues using data which are currently available. The purpose in
the release of such reports is to facilitate the exchange of
technical information and to inform the public of technical
developments which may form the basis for a final EPA decision,
position or regulatory action.
Standards Development and Support Branch
Emission Control Technology Division
Office of Mobile Source Air Pollution Control
Office of Air, Noise and Radiation
U. S. Environmental Protection Agency
-------
-1-
The driving cycles used in the EPA test procedure for
light-duty vehicles were developed ten years ago. Since then,
significant changes have occurred which may affect vehicle use;
for example, fuel costs have increased, car pooling has increased,
and smaller vehicles are now more prevalent. In order to
investigate current vehicle use and to compare this use to the
test cycles used for exhaust emissions and fuel economy
measurements, EPA initiated the Operational Characteristics Study
(OCS).
The first stage of the Operational Characteristics Study was
to develop instrumentation to accurately monitor vehicle use. In
addition, it was necessary to develop the capability to transcribe
the data to a large computer system for analysis. This report
describes the three basic systems which have been developed: the
data collection system, the data transcription system, and the
analysis system. All documentation on the equipment and software
of these systems available at the present time has been collected
and is presented in the Appendices. In addition to the systems
descriptions, a brief evaluation of each system is presented, with
suggestions for improvements. It is intended that this report
will be a working document that is expanded as more information is
acquired and system modifications are made.
The instrumentation systems will be used in at-least two EPA
data collection programs, both in conjunction with EPA Emission
Factors programs. In an emission factors program EPA obtains
in-use vehicles and performs emissions tests on these vehicles
while providing the program participant with an alternate vehicle
for use during the test period. Consequently the Emission Factors
programs provide low cost access to vehicles for the OCS
instrumentation.
In the 1982 MVEL Emission Factors program several vehicles
will be equipped with the OCS instrumentation. If a program
participant is willing, the participant will be asked to take one
of the instrumented vehicles as the loan vehicle. This program
will act as a pilot study to test the durability of the
instrumentation and the capability of the analysis systems.
Problems with the instrumentation can be investigated and resolved
while the instrumentation is still under the direct operation and
control of MVEL personnel familiar with the systems.
The second application of the OCS instrumentation will be the
1983 Non-Detroit Emission Factors program. The location of this
program is not presently known because the contract has not yet
been awarded. In this program the instrumentation will be placed
in the vehicles of those Emission Factor Program participants who
are also willing to participate in the OCS program. This program
will provide data obtained directly from owner operated in-use
vehicles.
-------
-2-
I. Data Collection System
The data collection system was developed by MB Associates of
San Ramon, California under an EPA contract. This system consists
of the data sensors on the vehicle, a microprocessor which
modifies the sensor signals, and a tape recorder which stores the
information from the microprocessor. The signal interfacing is
accomplished through a control program stored in the read only
memory addressed by the microprocessor. A block diagram of the
instrumentation system is shown below:
Microprocessor
and Control
Program
Power Supplies and Other Support Electronics
A.
Vehicle Data Sensors
Five types of vehicle data are monitored: 1) vehicle
ignition switch state (on/off), 2) vehicle speed, 3) engine speed,
4) temperatures at six different locations on the vehicle, and 5)
the position of an urban/rural switch. In addition, a real-time
clock is maintained and monitored.
The ignition switch sensor simply monitors the voltage of the
ignition system. If voltage is present, the ignition switch is
obviously on, and the engine is presumed to be running.
The vehicle speed sensor is an optical encoder driven from a
"tee" in the speedometer cable. This encoder is a standard
industrial unit which generates 1000 pulses/revolution.
Unfortunately, the unit is large, and has been difficult to
install in one of the vehicles used in the pilot program. Also,
several speedometer cable failures have occurred after the encoder
was installed in this vehicle. These failures occurred in the
section of cable before the encoder, indicating that a standard
speedometer cable may not be strong enough to drive both the
encoder and the speedometer. We are presently improving the
encoder installation by removing this cable. In addition, one
encoder failure has occurred, either because of an initial
assembly problem, or because of the shock and vibration of the
automotive environment. This failure was repaired by the encoder
manufacturer.
Six temperature probes are part of the OCS data collection
instrumentation. All of the temperature sensors are thermistors.
When supplied by MBA, these thermistors were covered with plastic
heat-shrink tubing. It was soon discovered that the heat shrink
tubing inadequately protected the thermistors, or its shrinkage
caused sufficient mechanical stress to destroy the wire
-------
-3-
connections to the thermistor beads. Removing the heat shrink
tubing and embedding the thermistors in a thermally conductive
epoxy has eliminated the thermistor breakage problem. The
thermistors and epoxy currently being used are specified in
Appendix A-l.
The thermistors themselves are marginally satisfactory, since
their temperature rating is from 0°C to 100°C. They can be used
somewhat outside this range without damage, but the temperature
signals will be non-linear. Unfortunately, oil temperatures can
easily exceed 100°C, and temperatures could be below 0°C during
winter operation in any northern U.S. city. We are currently
investigating alternate thermistors or other temperature sensors.
The final sensor is a switch located near the vehicle
driver's seat which the driver uses to indicate whether he
considers his driving to be urban or rural vehicle operation.
Installing the switch box is a minor mechanical problem because of
the need to firmly attach it so that the switch can easily be
operated by the driver with one hand. Also, installing the switch
box requires locating an acceptable access to the passenger
compartment.
All documentation on the data sensors, provided by MBA, or
researched by EPA, is presented in Appendix A-l.
B. Microprocessor and Control Programs
The microprocessor board used in the DCS instrumentation
system is a development board manufactured by RCA. This board is
one of several development boards in the RCA 1802 family. This
microprocessor family was chosen by MBA because it was the only
CMOS processor available when the project started. CMOS
integrated circuits are necessary for this project because they
require very little power.
The signal interfacing and most of the signal conditioning is
done on the main processor board in the area provided by RCA for
"user development." The documentation supplied by RCA and MBA for
this board, and for the 1802 processor itself is given in Appendix
A-2.
The microprocessor is controlled by the software program
stored in an Intel 2716 UV eraseable EPROM (electrically
programmable read-only memory). A data sheet on this EPROM is
included in Appendix A-2. The program is responsible for counting
the vehicle and engine speed pulses, applying a calibration factor
to them, converting the temperature data into calibrated units,
and most importantly the general formatting and transfer of the
data to the cassette tape recorder. This program was written by
Tom Pittman, a well-known microprocessor programmer, under a
subcontract to MBA.
-------
—4—
There are several notable aspects of the program. First, all
data are partially encoded. For example, vehicle speed can only
appear as a two byte word, and the first hexadecimal character of
the first byte must always be zero. Unfortunately, the speed is
not the only data that can appear with a leading zero; this can
also occur in the time, day, or vehicle ID strings. Fortunately
these data, which might be confused with vehicle speed should only
occur after an "FD" identifier code. However, anywhere from zero
to eight temperature data bytes can also appear following the "FD"
code. No code is provided to identify the number of the
temperature data to expect, which sensor the data come from, or
end of the temperature data string. Because of the complexity of
the data encoding and the variable length of some of the data
strings, it is very difficult to write automated transcription
decoding routines.
The software is used for almost all routine instrumentation
functions, such as pulse counting for the vehicle and engine
speeds. Consequently, it is complex because of the critical time
nature of these real-time functions. Unfortunately, the
documentation, which is presented in Appendix A-2, explains the
program in general terms, but not in sufficient detail that they
may be easily modified.
An example where a technically easy modification would be
beneficial is in the temperature data. The instrumentation
software was to convert the temperature sensor voltages into units
which were linearly related to the true temperature in degrees
Celsius. However, after the original program was written, MBA
changed the temperature sensors to thermistors having slightly
different characteristics but did not change the software. As a
result of this change, the conversion routine in the program is
incorrect, and the temperature data recorded on the tape are not
linear with true temperature. If the microprocessor programs were
well understood or thoroughly documented, it would be easy to
modify them so that true temperatures were recorded.
In addition to the minor problems, such as the temperature
nonlinearity, there appears to be at least two errors or "bugs" in
the instrumentation software. The most significant problem is the
inability of the software to accurately detect transmission gear
changes. The microprocessor tries to use the engine speed signal
and the vehicle speed signal to determine when a transmission gear
change (shift) occurs by computing the ratio between the engine
speed pulses and the vehicle speed pulses. This ratio is then
multiplied by a scaling factor and truncated so that the result
will be one hexadecimal number representing each gear. All the
numbers fall between 4 and E. Whenever this character changes the
new value is sent to the data tape, and should indicate a gear
change.
-------
-5-
At the present time more than one hexadecimal number often
appears for a single gear, and usually one number appears for two
different gears. Initially this problem was believed to be
related to electrical noise, and therefore the system was tested
using "clean" pulses from a signal generator. The gear change
sensing was not accurate even with the test pulses, therefore it
is concluded that the problem is most likely a precision or
truncation problem in microprocessor software. Until this problem
is resolved, the gear change data are not useful, and the engine
speed sensor has been disconnected to conserve data tape space.
The second, less critical, software error is in the routine
which counts the one second time pulses and converts the
accumulated total to "real clock time". As a result of some
error, whenever the hour is incremented, the most significant
digit of the day is decremented by three, unless this results in a
negative day, in which case no change is made. Unfortunately,
because of the poor documentation there is no easy way to locate
and correct this error in the program. Consequently the data will
be corrected during the data analysis.
C. The Cassette Recorder
The cassette recorder is a high-density Memodyne incremental
cassette tape recorder. The recorder is well suited for this
application; however, a special reader unit is required because an
uncommon data recording format is used to maximize the data
storage capacity of the tape. Documentation on the Memodyne
recorder is presented in Appendix A-3.
D. Operation of the Instrumentation
Operation of the instumentation is relatively simple, but
there are several aspects which must be carefully followed if
useful data are to be obtained.
Digital tape cassettes generally have a clear leader at the
ends of the magnetic tape. These tapes must be manually preset to
the beginning of the magnetic oxide coating. If this is not done,
the calibration symbols will not be recorded and there will be no
way to verify that the vehicle speed sensor calibration occurred.
The speed sensor must be calibrated every time the unit is
switched off. If this is not done an unknown random number,
generated by the microprocessor during the power-up cycle, will be
used as a calibration factor.
A second aspect that must be carefully followed is that the
reset and rerun switches of the unit must be activated twice (or
more) when starting. This is a bizarre quirk of the software
required because of the sophisticated but complex and problematic
method used to count the engine and vehicle speed pulses.
-------
-6-
Finally, when the unit is turned off, several minutes of
operation must be provided to "dump" any useful data stored in
internal buffers or these data will be lost. It is also desirable
to provide a similar time when the unit is powered up to allow for
possible initial loss by the reader.
Detailed instructions for tape initialization,
instrumentation operation, and speed calibration are provided in
Appendix A-4.
II. The Transcription System
The computer system used by EPA, the Michigan Terminal System
(MTS) of the University of Michigan, does not directly support
cassette tapes. Consequently a system is required to transmit the
data stored on the cassette tapes to MTS for analysis. This
system consists of several hardware components and software to run
the equipment.
A. Hardware
The first element of the data transcription system, a tape
reader unit, was supplied by MBAssociates as part of the EPA
instrumentation contract. This tape reader is based on the same
RCA microprocessor board that is used in the recording unit. The
reader microprocessor, however, is simply programmed to take
serial binary data from the cassette tape, combine these data into
8 bit bytes, separate each byte into two hexidecimal characters,
and then covert these characters into the American Standard Code
for Data exchange (ASCII). The ASCII code is then transmitted
serially through the interfaces of the reader. Both RS232 and 20
milliamp current loop interfaces are provided.
Several approaches were possible to interface the reader
signal to MTS. The reader could be reprogrammed to emulate a
computer terminal and then be directly connected to MTS, the data
could be transcribed to a medium directly supported by MTS, or the
data could be transcribed to a device which can communicate with
MTS.
Modification of the MBA unit was rejected primarily because
the limited documentation available on the reader program would
make its modification difficult and uncertain. Even if
modifications were successful it would be an expensive approach
because the reader operates at such a slow speed that long
computer connection times would be required for data
transmission. Transcription to a medium directly supported by
MTS, such as 9-track computer tape, was rejected because of the
high cost of 9-track tape drives ($5,000 - $18,000). Also, this
approach would not provide any easy method to preview data for
general equipment malfunctions. The final method, data
transcription to equipment which then rapidly transmits to MTS,
-------
-7-
was chosen. This approach had the advantages that some of the
necessary equipment already existed at EPA, and the additional
transcription step would provide an opportunity to observe the
data and perform preliminary data analysis.
The system selected includes a mirocomputer for data
manipulation, a video monitor for data display, and a Techtran
tape recorder to receive the data. The Techtran recorder can be
connected to the MTS to transmit the reformatted data tapes. The
documentation on this equipment is provided in Appendix B-l.
In the simplest system configuration the data tapes are
directly transcribed from the reader onto the Techtran tape. The
contents of the tape are displayed using the RCA terminal keyboard
and the RCA mirocomputer monitor. This display provides a rapid,
quiet and low cost method of previewing the data and detecting
equipment malfunction. This is the approach being used for the
emission factors pilot program.
One idiosyncrasy of the reader unit should be noted. The
reader responds to the beginning of tape (BOT) and end of tape
(EOT) marks (holes) present on many digital cassette tapes, while
the recorders in the data collection systems do not. Consequently
if the cassette tapes are not manually advanced beyond the BOT
hole, about 18 inches into the tape, the recorder will record over
the tape marker, but the reader will not read beyond this point.
The tape can be manually advanced beyond the BOT marker and the
reader restarted at this point, but the data in the vicinity of
the tape marker will always be lost. Consequently, tapes without
EOT/EOT holes should always be used.
In addition to simple transcription of the data to a Techtran
tape, the transcription system contains a microcomputer, an RCA
Microboard Computer Development jjystem CDP18S694, which can be
used for preliminary analysis. The MCDS is also based on the RCA
1802 microprocessor. This was selected so that knowledge learned
from the transcription processes would be directly applicable to
the data recording instrumentation and the tape reader.
B. Software
In the single direct transcription process, the equipment
provides the necessary interfacing and no control programs are
necessary. A decoding program is still required to convert the
data into engineering units, but this program runs only on the MTS
computer. This program has been written and is presented in
Appendix B-2 along with examples of the program input and outputs.
There are two disadvantages of the direct data
transcription. First, and most significant, it is very difficult
to preview the data since it is both in the hexidecimal number
system and encoded by the data collection system. The other
-------
-8-
disadvantage is that the current decoding program is written in
FORTRAN which is relatively inefficient for the individual bit
data manipulation required for the data decoding. This
inefficiency makes the program expensive to run using the
relatively large data sets anticipated for the project.
Data decoding can be accomplished during the transcription
process by the MCDS. In this case, the microprocessor program
translates the encoded data written by the data collection
instrumentation into standard engineering units. This greatly
facilitates review of the data and early detection of equipment
errors. In addition, it reduces the external computing time
required by the MTS system and thereby reduces the program cost.
The decoding program for the microcomputer is relatively
complex because, as discussed earlier, different data elements are
encoded with different systems. Also, some parameters such as
engine on/off, gear changes, and urban/rural switch changes can
occur almost anywhere in the data field. This decoding program is
not yet completely operational, however the current program is
presented in Appendix B-2. The remaining problems are in
interface areas and should be resolved before the non-Detroit
Emission Factor Program begins. When these problems are resolved
the decoding program will transcribe all data into engineering
units and reformat it. In its new format the data will be
recorded on to a tape on the Techtran recorder, then that unit
will be used to put the data into an MTS line file.
III. The Analysis System
The final data analysis will be done under MTS on the
computer of the University of Michigan. The computer is an Amdahl
470, a large mainframe system. The details of the computing
equipment are unimportant to the user, however, since all analysis
programs are written in standard FORTRAN and executed from a
terminal.
The analysis programs serve two basic purposes. First, the
data are analyzed for many trip parameters: average speed, time
in various speed bands, etc. Finally, the statistics of these
parameters will be calculated and compared with the statistics of
the same parameters computed from the current EPA driving cycle.
There are five analysis tasks: 1) cross-field and range
checking, 2) calculation of trip and trip-segment statistics, 3)
calculation of driver statistics, 4) analysis of these statistics,
and 5) graphic display generation. The first three of these are
programmed in FORTRAN, the analysis is implemented in the Michigan
Interactive Data Analysis System (MIDAS), and most of the graphics
will be generated using CALCOMP, an MTS-supported package of
FORTRAN subroutines. A few of the graphic displays may also be
generated using MIDAS.
-------
-9-
A. Crossfield and Range Checking; OCSPGM1
The program OCSPGM1 is used to conduct certain crossfield and
range checks on the data. The program has been written in a
modular construction, that is, subroutines are extensively used to
allow modifications or the addition of other crossfield or range
checks if necessary.
There are seven subroutines in OCSPGM1, one each for input
and output plus five data checking routines. The primary purpose
of these subroutines is to check for erroneous data, test for
abnormal values of speed and temperature, for abnormal changes in
these parameters and for abnormal differences between related
parameters. In addition, a check is made to verify that time
always increases. Whenever an "error" condition is detected these
subroutines output detailed information on the error location and
the number of times it occurred to one of the output files. For
example, each time that the subroutine DELTMP finds a temperature
record in which the difference in the ambient and fuel tank
temperature is greater than 20°C (36°F), the following information
is written to a single line in this files
- occurrence number within trip (1, 2, 3,...)
- time (given as number of minutes after time that trip
began: 7 min., 103 min., etc.)
- ambient and fuel tank temperature (°C)
- absolute value of difference in above temperatures.
In addition to testing for abnormal data values, each
subroutine calculates the extreme value of the checked variable
(usually the maximum value) and the number and frequency of the
occurrence of the extremes.
A listing of OCSPGM1 appears as Appendix C-l. This program
has been tested on a "dummy" data set formed by inserting time and
temperature data at one-minute intervals through the LA-4 driving
cycle. Erroneous speed data were also substituted for several of
the speeds specified by the LA-4 cycle. All of the subroutines
correctly detect the "error" conditions described above.
B. Calculation of Trip Statistics - OCSPGM2
The program OCSPGM2, like the crossfield and range checking
program, consist of a calling program and a number of
subroutines. These routines read in the data obtained from one
tape (one study participant) and calculate a number of statistics
describing the trips and between-trip periods.
-------
-10-
Statistics calculated for each trip Include: duration of
trip, total distance, mean speed, maximum speed, amount of time at
idle, mean and maximum acceleration and deceleration, mean of
absolute values of all accelerations and decelerations, percent
time in acceleration, cruise and deceleration modes, number and
mean duration of full stops (idles) within trip, hot or cold
start, and times of day trips began and ended.
Each trip will be divided into modes of operation and trip
segments. The modes of operation are acceleration, cruise, and
deceleration. A trip segment is defined as a subset of the speed
data that begins and ends with a full stop. Three-minute duration
trip segments will also be formed from the first three
three-minute blocks of time in a trip. Most of the statistics
calculated for entire trips are also calculated for these trip
segments and modes of operation.
The program OCSPGM2 is still under development. While none
of the statistical measures are difficult to calculate,
implementing all of the desired calculations in an optimal manner
and creating a well-designed data base are not trivial problems.
Appendix C-2 contains a listing of the draft OCSPGM2 program.
Later revisions of this document will contain the final version.
C. Operator Statistics
The trip statistic output files from OCSPGM2 are used as
input to OCSPGM3. This FORTRAN program determines the multi-trip
mean values and frequencies of the parameters analyzed for each
study participant. Such "participant summary," or multi-trip
statistics include trips per day, mean distance per trip, miles
travelled per day, and other variables based on all of the data
obtained from one study participant.
OCSPGM3 is still in relatively early development. As more
OCS data become available significant improvements in the design
structure of this program may be desirable and may result in
significant program revisions.
Appendix C-3 is reserved for OCSPGM3 and it will be presented
in future revisions of this document.
D. Analysis of the Statistics; OCSPGM4
OCSPGM4 merges the analysis of the vehicle operation data
with the participant's answers to the questionnaire. This
combined data set is then analyzed to observe the distributions of
driving characteristics for the entire program.
OCSPGM4 is not a true program in the usual FORTRAN sense, but
is a series of MIDAS (Michigan Interactive Data Analysis System)
commands in a control file. Use of MIDAS greatly simplifies the
-------
-11-
programming effort required. Also since MIDAS is a relatively
efficient system many analyses can be conducted quickly at a
relatively low computer time cost.
This final analysis control file is the most straightforward
of the programming tasks. However, it is the last to be used and
since it requires complete knowledge of the structure of the data
files, it has not been developed yet. Appendix C-4 is reserved
for inclusion of a listing of OCSPGM4 when it is completed.
E. Generation of Graphic Displays
Most of the final data analysis results can best be
understood in the form of graphic displays. These will include
histograms, distributions and displays of variable regressions.
Initially, these histograms, distributions, and regressions will
be produced using MIDAS.
Previewing these MIDAS outputs will allow parameters to be
changed, if appropriate, before generating the final graphics
using the MTS CALCOMP System.
CALCOMP provides great user control and produces high quality
graphics. The programs for creating graphics through CALCOMP are
FORTRAN programs consisting primarily of calling statements to the
many CALCOMP subroutines. As with the statistical analysis, the
programming will be straightforward, but require detailed
information of the data analysis files. Therefore, these programs
will not be developed until data have been collected and the data
analysis files created. Appendix C-5 is reserved for listings of
these programs and sample outputs.
IV. System Evaluation - Recommended Improvements
A. Data Collection System
The MBA instruments are performing well. Very few problems
have arisen with the basic unit even though occasional accidental
user abuse, such as shorting out the main power regulators has
occurred. In contrast, most of the vehicle data sensors provided
by MBA have failed or have been unsatisfactory in some manner.
The thermistor probes had to be imbedded in epoxy to prevent
breakage, the optical encoder used for the vehicle speed sensor
has caused speedometer cable failures, and the gear change sensing
circuit does not function reliably. In general, the electronic
apsects of the unit function far better than the automotive
interfaces. The first, and most important, recommendation is that
the automotive sensors be improved. This is essential if long
term reliability of the data collection is expected. Efforts are
in progress to improve the temperature sensors and the
installation of the vehicle speed sensors.
-------
-12-
A major problem area is the detection of gear changes. The
selection of gear shift points is an important aspect of the EPA
test procedure and area of past and present concern. .Because this
area is important, it is recommended that efforts be made to
investigate the present difficulty and to develop a solution.
This could be accomplished in several months using auxiliary
equipment and would not delay the present program. Total
equipment cost would be between $500 and $1,000.
Numerous software improvements are possible. Unfortunately,
even though these changes may be simple they will be time
consuming because of the lack of detail in the software
documentation provided by Tom Pittman/MBA. It is recommended that
improvements in this area be attempted only after the gear change
sensor problem is resolved. If the gear change problem is
satisfactorily solved, then software changes in the
instrumentation program will probably be necessary to implement
the solution and other improvements can be made at that time.
B. Data Transcription System
The data transcription system is working satisfactorily.
That is, tapes can be transcribed and analyzed. A significant
improvement would be to provide decoding and preliminary analysis
of the data during the transcription process. This improvement is
well underway and should be functioning shortly. Since this
entire system was developed at EPA, we are able to easily
incorporate changes and improvements in whenever they appear
beneficial.
C. Data Analysis
The data analysis programs cannot be considered completed
until sufficient data have been collected by the project to verify
their capability. Therefore, the data analysis programs must be
considered as still under development. Even for the programs
which have been written and tested, it is known that some changes
in input and output formats will be required.
When this sytem is used and desired modifications become
apparent this will be the easiest system to modify since it was
developed by EPA and since all of the programs are written in
FORTRAN or are part of a University of Michigan interactive
command system.
-------
-13-
Appendix A-l
1. EPA Operational Characteristics Study Instrumentation
Phase III Final Report. Prepared for EPA by
MBAssociates
2. Blueprints of instrumentation
3. Manufacturer's data sheet - optical encoder
4. Manufacturer's data sheet - epoxy used to encase the
thermistors
------- |