GPS Based Methodology for Drive Cycle Determination

Author(s): Anand V. Kulkarni, Rekha R. Sapre and Chirag P. Sonchal

Source: SAE Transactions, Vol. 114, Section 7: JOURNAL OF PASSENGER CARS:
ELECTRONIC AND ELECTRICAL SYSTEMS (2005), pp. 233-239
Published by: SAE International
Stable URL: https://www.jstor.org/stable/44682424
Accessed: 08-10-2021 00:57 UTC

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms

SAE International is collaborating with JSTOR to digitize, preserve and extend access to SAE
Transactions

This content downloaded from 46.13.104.98 on Fri, 08 Oct 2021 00:57:34 UTC
All use subject to https://about.jstor.org/terms
 2005-01-1060

GPS Based Methodology for Drive Cycle Determin

Anand V. Kulkarni, Rekha R. Sapre and Chirag P. So
Engineering Research Centre
TATA Motors Limited

Copyright © 2005 SAE International

ABSTRACT: depending on the type and application of the

vehicle under test. Some of these are
standardized cycles and serve as regulation
Drive cycles are an important tool to a vehicle
assessment
development team. In order to achieve good tests for vehicle certification.
design tunes, an accurate drive cycle is Drive cycles are primarily classified as
important. It is essential that errors occurring Modal cycles and HYZEM cycles. Modal cycles
out of drive cycles be contained within the level are characterized by significant portions of
of data scatter obtained from the use of a drive steady speeds and linear accelerations. As
cycle. For example: fuel consumption. Only an such they tend to be artificial (not truly
accurate drive cycle will be able to highlight the representative of real life) and may be biased
small differences associated with minor towards a particular speed or acceleration
changes to vehicle specification that affect level.fuel
However since they offer easy
consumption. definitions, they can easily be adopted as
A methodology has been established to derive
standard tools and find wide usage in emission
a drive cycle representative of real life development.
usage.
This has been done using a GPS based Since modal cycles do not represent
system. Recombination methodology was usedreal driving patterns, an additional group of
driving cycles has been developed: the
to recreate a synthetic drive cycle from actual
HYZEM cycles. HYZEM cycles are transient
driving conditions. Correlation was determined
using synthesis of sequences as observed in cycles. The parts at which the speed is
constant are much smaller than in modal
actual data. Results show that the drive cycle
data matches very well with actual values of cycles. The HYZEM cycles are derived fr
fuel consumption under real life conditions. real driving patterns. Hence, they are a be
representation for actual driving conditi
INTRODUCTION: than the Modal cycles. The cycle that has b
constructed in the current exercise would thus
Drive cycles have always been considered get as classified as HYZHEM.
important tools to validation engineers. It
diminishes scatter and provides a cohesive Every (geographical) area will have a
and repeatable method for evaluating changes very specific driving pattern. This pattern is
made to the design tunes. It provides a dependent upon the infrastructure of the
normalized platform for comparison of various location, the vehicular density as well as the
specifications against each other. It also enforcement issues in that place.
enables a quantitative assessment of any For vehicle manufacturers, it may be of great
parameter under consideration for a vehicle interest to study the vehicle behavior in a
under test. specific region. It can then serve to provide
A 'drive cycle' means a standardized driving significant inputs pertaining to the existing
pattern [1]. Driving cycles are typically designs and make the market offering better.
represented as a vehicle speed versus time However, not always is it feasible to dedicate
trajectory. Many driving cycles currently exist,233 extensive resources to undertake an elaborate

This content downloaded from 46.13.104.98 on Fri, 08 Oct 2021 00:57:34 UTC
All use subject to https://about.jstor.org/terms
study. Under such cases is it possible to Chase Car methodology:
extract salient characteristics of the road
conditions existing in the target market Theby
instrumented chase car driver identifies a
undertaking a comparatively simple and target car according to the protocol, records
uncomplicated study? the details of the target car & initiates data
This paper discusses the approach adopted for logging. The trip details are acquired into the
such a study and the reasonable success software program including the start & stop
achieved. location. Thus, each trip data gets collected in
a database. The application developed (as
detailed in appendix I) enables the derivation of
METHODOLOGY: statistical information from the trip data for
further use. This has been elaborated in the
The objective was to create the drive following sections.
conditions existing in the real world on the test
track. In order to collect data describing driving
behavior 'chase car' [2] concept was used forREAL-WORLD DRIVING DATA
the purpose of this exercise. The overall
methodology for the GPS based drive cycle The driving patterns are completely different on
determination is shown schematically in highways and in the urban driving condition
figure 1 . Collection of drive patterns on the highwa
posed some initial problems. Since the driv
Summary of GPS based Drive Cycle durations were expected to be high, the dat
Determination files would be huge. Also since we were
interested in finding average trip lengths as
one of the outputs, it would be difficult to
manage to accurately capture a start point and
Trip Details end point of drivers. In a city, within a known
geographical region, it was considered
Jl
reasonable to capture such data, which could
Chase Car Data be subjected to statistical analysis, without
Acquisition distorting any data. Also driving on highways
does not display a large variety of modes since
JI most of the driving is determined by speed
limitations rather than traffic conditions.
Data Analysis using
Since speed and time histories are
Application Software
fundamental to define drive cycles, the data
logging primarily involved recording speed &
li time histories.

Derive the boundary

condition values
MEASUREMENT PARAMETERS

i
The parameters of primary interest (Speed
Construct
distance, & time histories) could then be
driving eve
subjected to various statistical analyses to be
the bounda
able to reconstruct the drive cycle under
controlled conditions later.
ir It was decided to capture the drive patterns of
Validate the Drive Cycle various drivers in different parts of the city
(region) at different times. Approaching a
Figure 1 : Summary of GPS based drive cycle cross-section of people who would be willing to
determination help collect data on their drive profiles was
234

This content downloaded from 46.13.104.98 on Fri, 08 Oct 2021 00:57:34 UTC
All use subject to https://about.jstor.org/terms
considered. However this posed immediate
problems. Firstly, approaching a sufficiently
large group who would be willing to participate
was tedious. Secondly, and most importantly,
even if that could be made possible, these
people were likely to inadvertently alter their
driving habits owing to knowledge that their
drive patterns were being recorded. This would
introduce an unrectifiable bias into the data,
possibly distorting the entire exercise.
In order to overcome this problem, we decided
to adopt the Chase car protocol. In this, a
chase car installed with the specifically
developed GPS application collected second-
by-second speed-time profiles by following
identified target vehicles. The instrumented
chase car driver would identify a target car
according to the protocol, record the details of
the target car and mimic the drive pattern of
the target vehicle and thereby acquire
information about the drive pattern.
BOUNDARY CONDITIONS OF A DRIVE
CYCLE
The acquired speed & time histories were
further differentiated into the instantaneous there was no loss of drive data and memory
sizes of the data collected were not
constituents, e.g. acceleration level of a drive.
Consequently the boundary conditions that humungous.
define a drive cycle were decided. They were
as follows:
a) Speed
b) Time
c) Distance (trip length)
d) % Speed distribution
e) Acceleration levels
Acceleration levels were further classified as:
i) Normal acceleration (0.01 to 1.46 m/s2)
ii) Hard acceleration (>1.46 m/s2)
iii) Cruise (+/- 0.01 m/s2)
iv) Normal deceleration (-1.46 to -0.01
m/s2)
v) Hard deceleration (< -1 .46 m/s2)
INSTRUMENTATION OPTIONS
In order to collect data with a chase car, two
options were available. The first one was to
235
This content downloaded from 46.13.104.98 on Fri, 08 Oct 2021 00:57:34 UTC
All use subject to https://about.jstor.org/terms
use a radar system on the vehicle to capture
data of the target vehicle. The second one was
to use a Datron optical sensor system. Both
these systems suffered from drawbacks of
either being a hindrance to the traffic or being
conspicuous. Conspicuity inherently had
potential for detection, which could have led to
aberrant changes in the target car driver's
driving pattern. A system, which was capable
of good data collection as well as being subtle,
was the need. A GPS system was devised
wherein the only visible part is the antenna,
which is common place.
SYSTEM DESCRIPTION
The GPS equipment is capable of giving
location coordinates & thereby, speeds &
distance. A specific algorithm was written to
enable the data from the GPS to be analyzed
statistically and user defined outputs to be
generated. A more specific description of this
algorithm is enclosed below in this paper
(appendix 1). In order to maintain ease of data
analysis, the GPS data acquisition frequency
was set to 1 Hz. This frequency was
considered to be appropriate for ensuring that
DATA ACQUISITION SET UP
nDC _ T~ I Laptop (Software in
nDC Ul-'o Keceiver NMEA Visual Basic 6.0 and MS
■<
0183
Z RS 232 port Z
USB
EMS
* * Interface
Interface
▲ X
Vehicle
Battery
Figure 2: Data acquisition setup schematic
DATA COLLECTION PROTOCOL 2. Average speeds = 32.8 km/hr
3. Average trip duration= 1 1 .2 minutes
4. Acceleration
The data was collected in the city of Pune, levels
INDIA. In order that the data was free of all 4.1 Cruise = 8.2%
biases, an elaborate protocol was setup for the 4.2 Rapid Accel =1.0%
data collection. 4.2 Normal Accel =46.7%
a) Data would be collected in different 4.3 Rapid Decel = 2.0%
parts of the city at different times in 4.4 Normal Decel = 42%
order to capture the various modes 5 Speed Level at percentages

b) The instrumented car would wait at Sr. Speed range Percentage

major shopping areas, multiplexes, No. Km/hr

industrial estates, housing colonies, 5.1 0-10

offices etc. A car moving out of the 5.2 10-20

parking lot would be identified as the 5.3 20-30
target car. 5.4 30-40
c) This car would be followed until the end 5.5 40-50
of its travel and this drive would
5.6 50-60
constitute a trip length. 5.7 60-70
d) Second-by-second data would be 5.8 I 70-80
collected and the driver in the test car
6 % Total time spent a
would try and maintain the same 7 Max speed = 79 km/h
distance at all times (a pre-requisite for
mimicking the drive pattern of the target Thus the 18 parameters as above were
car) determined as the boundary conditions
e) Information about the driver would be adequate to define the drive cycle.
archived. For e.g. male/ female, age, car
model, registration no, trip details etc.
f) Time of the day and the area would also
be recorded.
g) In case the driver in the target car got
suspicious of the following car and
attempted to shake off the tail, data was
discarded.
In this manner data from 150 target cars was
collected. This comprised people of various
age groups from different parts of the city at
different times of the day. No data was
collected on public holidays and weekends Figure 3: Constructed Pune city cycle
since traffic conditions were considered to be
significantly different than normal weekdays.
The data thus collected, was subjected to
1 . Cumulative averages
2. Level at speed intervals
3. Acceleration levels classification

RESULTS:

After the analysis of the data, the following

results were obtained for the city of Pune,
INDIA
Figure 4: Pune city cycle Level at values
1 . Average trip length = 6.2 km
236

This content downloaded from 46.13.104.98 on Fri, 08 Oct 2021 00:57:34 UTC
All use subject to https://about.jstor.org/terms
SYNTHESIS OF THE DRIVE CYCLE

The results thus obtained revealed distinct city

drive modes as can be seen from the speed
distribution percentages as shown in figure 4.
This information was analyzed for specific
sequences that were seen to be common
across different data files. Using this segments
information an elementary drive cycle was
constructed. When this was run on the test
track, the software program developed enabled
a summary to be generated of all the 18
parameters listed. A comparison with the gross
average of the actual city data (of the 18
parameters under consideration) highlighted
the gap between the actual data & the derived
results. Subsequently finer segmentation was
attempted after looking at smaller sequences.
This helped achieve finer correlation with the Chart 1: Fuel consumption
actual data. Lastly, the constructed drive cycle
was modified in incremental steps to match the
desired boundary conditions. Thus the final CONCLUSIONS
drive cycle consisting of different driving events
akin to city drive modes was arrived at. This 1. In this exercise, a general methodology
drive cycle showed very close matching with the drive cycle determination has been
the actual data on the 18 parameters. The final established using a GPS based system.
drive cycle determined is as shown in figure 3.
2. Currently a city drive cycle for the city of
Pune has been established & validation results
VALIDATION OF THE CYCLE FOR are matching the actual results very closely.
CORRELATION
3. Hence, this drive cycle has shown a good
The cycle so constructed needed correlation
correlation with the actual real-world conditions
before it could be put to use as a from which it was derived.
developmental tool. In order to do so, fuel
consumption was chosen as the parameter for
validation. Fuel consumption was monitored on
REFERENCES
a test vehicle in the city and the same was
checked on the synthesized cycle on the track.
1. ISO definitions of drive cycles: Wh
The results are as shown in Table 1 . paper
2. "Data collection for driving cycle
development: Evaluation of data
Vehicle I Actual City I "sute onfracķ"
collection protocols" by Niemeier et al.
With Without With Without
3. N. Dembski, Y. Guezennec and A.
Soliman et al. "Analysis and
Hatchback 1 12.66 15.51 12.55 15.53
Experimental Refinement of Real-Wo
Saloon 1 11.5 13.6 11.76 13.86 Driving Cycles" SAE Paper 2002-01-
0069.
Saloon 2 9.6 11.6 9.43 11.84
4. SAEFL Report "Real world drive cycles
Table 1: Drive cycle validation testmeasurement:
for emission results ARTEMIS
and Swiss cycles" by Infras
237

This content downloaded from 46.13.104.98 on Fri, 08 Oct 2021 00:57:34 UTC
All use subject to https://about.jstor.org/terms
CONTACT:

Anand V. Kulkami
Project Manager
Vehicle Testing
Engineering Research Centre
TATA Technologies Limited
B.E. [Automobile Engineering]
avkulkarni@tatatechnoloqies.com

Ms. Rekha R. Sapre

Senior Engineer
Electronics Division
Engineering Research Centre
TATA Motors Limited
B. E. [Electronics & Telecommunication
Engineering]
rekha.sapre@tatamotors.com

Chirag P. Sonchal
Senior Engineer
Vehicle Testing
Engineering Research Centre
TATA Motors Limited
B. E. [Mechanical Engineering]
chiraq@tatamotors.com

238

This content downloaded from 46.13.104.98 on Fri, 08 Oct 2021 00:57:34 UTC
All use subject to https://about.jstor.org/terms
 Appendix I
Figure 5: Flow Chart

Start
* i

Initialize GPS

receiver to NMEA

Id, vehicle type, Driver details,

Location Information, start Time and

EXIT START Enable Timer of 1 -second interval

I s. Button Button

( END ) N
YT N 'y

ç ' Capture Data from GPS Receiver

Log in MS Access Database:

N Multiple Velocity, Acceleration, Latitude,
. Trip y Longitude, Altitude, and Distance

Y '|/ /STOpY n
I

Enter

Y 'T/
Display comparison graph

of speed profiles Record Stop Time, Distance

Enter Trip Id for analysis - Plot Charts speed v/s

Display Trip Details like percentage time below,

vehicle type, Distance

traveled, avg. spe

duration

Compute % time and % N. ^

below in speed range 0 to

170 in steps of 10 kmph tři> Print

Classify Acceleration N BACK

V n. Button BACK

239

This content downloaded from 46.13.104.98 on Fri, 08 Oct 2021 00:57:34 UTC
All use subject to https://about.jstor.org/terms