VOL. 11, NO. 4, FEBRUARY 2016 ISSN 1819-6608
ARPN Journal of Engineering and Applied Sciences
©2006-2016 Asian Research Publishing Network (ARPN). All rights reserved.
www.arpnjournals.com
2763
MODELING OF A SHELL ECO-MARATHON VEHICLE BASED ON
DRIVE-TRAIN CHARACTERISTIC AND DRIVER MODES TO
PREDICT FUEL CONSUMPTION OF THE VEHICLE ON A
SPECIFIC TRACK
Witantyo, Sutikno, Diva Aulia and Habibie Rahman
Mechanical Engineering Department of ITS, Sukolilo Campus, Surabaya, Indonesia
E-Mail: witantyo@me.its.ac.id
ABSTRACT
Shell Eco Marathon is a competition for fuel efficient vehicle organized by Shell annually for student around the
word. Every team should present a uniquely designed vehicle targeted to be driven to an extreme distance using 1 liter of
fuel. This study aimed to conduct a vehicle dynamic modelling by using Simulink program from Matlab to predict vehicle
fuel consumption. The Model is build based on vehicle data and drive-train characteristic. To model the vehicle, various
data such as body weight, tire/wheel weight and angular inertia, frontal area, drag coefficient and tire rolling resistance are
collected. To model the drive-train, transmission ratio, engine torque and specific fuel consumption curves, mechanical
efficiency of some rotating parts are also collected. Model of the vehicle is a close loop system in which engine as power
unit gave its torque to wheel to move the vehicle. More speed developed by the engine would produce more resistance of
vehicle dynamics. The calculations were conducted with changing vehicle speed, driver mode, and inclination of the track.
Predictions of accuracy were done by using competition data from Sepang, Malaysia circuits within 5% of error.
Keywords: shell eco-marathon, vehicle dynamic modeling, fuel efficiency.
INTRODUCTION
Every year, students from all over the world
compete in Shell Eco-marathon by designing vehicle to
get extreme distance on one liter of fuel. ITS Team
Sapuangin have been participated in Shell Eco-Marathon
(SEM) Asia since 2010 and always won a title in Urban
Concept category mostly in biodiesel and diesel fueled
vehicle. Urban Concept category is a new class developed
in 2003 in which the vehicles are designed to better
represent a conventional city car with certain
roadworthiness criteria required.
To accomplish an improved distance for the next
competition, vehicle dynamic modeling would be built to
represent the vehicle and the driver modes in the race
track to minimize fuel consumed. Vehicle dynamic
modelling for SEM car have been developed by Rahmanu
(2011) and Adeniyi and Mohammed (2012), both for
prototype category. While Rahmanu interested more on
the vehicle details, Adeniyi and Mohammed have more
concentration on the engine characteristics. On urban
concept car category, Grundiz and Jansson (2009) have
built a hybrid car modeling and Witantyo et al. (2013)
also developed a model to evaluate vehicle specification,
driver modes and track specific. In this model, drive-train
characteristics and driver modes are detailed more to get a
better simulation results.
VEHICLE MODELING
Vehicle modeling is built using Jeongwoo Lee
(2009) schemes as shown in Figure-1. In uphill track
when the engine is on, traction force (Ftrac) from drive
train would be resisted by the inertia force (Finertia) of the
vehicle, gravitational force (Fg), aerodynamic force (Faero)
and tire rolling resistance force (Frr). When the engine is
off, the vehicle would be gliding using inertia as the
power source and resisted by aerodynamic force and tire
rolling resistance force. In this condition, gravitational
force would accelerate the vehicle in downhill track and
vice versa.
Figure-1. Vehicle dynamics. (Jeongwoo Lee, 2009).
Aerodynamic force is a function of air density
(ρ), frontal area (A), drag coefficient (CD), and vehicle
speed (V). Drag coefficient actually is not a constant in
lower speed but its assumed constant in this model.
Faero=½.ρ.CD. A .V2 (1)
VOL. 11, NO. 4, FEBRUARY 2016 ISSN 1819-6608
ARPN Journal of Engineering and Applied Sciences
©2006-2016 Asian Research Publishing Network (ARPN). All rights reserved.
www.arpnjournals.com
2764
Rolling resistant force is a function of tyre
rolling resistant coefficient (Crr), vehicle mass (m), gravitation (g), and hill slope angle (θ). Crr is a function of
speed but its assumed constant in this model.
Frr=Crr.m.g.cos(θ) (2)
Gravitational force is a function vehicle mass
(m), gravitation (g), and hill slope angle (θ).
Fgrad=m.g.sin(θ) (3)
Inertia force is a function vehicle mass (m), and
acceleration (a). Rotational inertia of wheels is neglected
in this model.
Finertia=m.a (4)
At any acceleration, traction force should be in
balance with sum of all resistance forces. Power traction
is traction force multiply by vehicle velocity, and vehicle
velocity should be up-dated continuously.
Ftrac = Fdrag+Frr+Fgrad+Finertia (5)
Ptrac = ( Fdrag+Frr+Fgrad+Finertia)V (6)
V(t)= V0 + at
(7)
DRIVE TRAIN MODELING
Drive train model is built using engine
characteristic curve from Yanmar in Figure-2. Traction
force is calculated using torque curve and fuel consumed
is calculated using specific fuel consumption curve.
��௡��௡� = ݂ሺ����௡��௡�ሻ (8)
���௧௨�� = ��௡��௡� ∗ ���௨௧�ℎ (9)
Figure-2. Diesel engine characteristic curves.
(www.yanmar.com)
Figure-3. Clutch efficiency modeling.
Clutch model is added to connect engine
rotational speed to vehicle speed especially when the
vehicle starts to move. The centrifugal clutch works
automatically to connect engine torque at higher speed.
Clutch power transfer efficiency at increasing engine
speed is modeled in Figure-3.
In engine stop mode, it was modelled that the
clutch was also work as a one-way power transmission to
eliminate inertia losses due to engine brake. Therefore, it
will disconnect the transmission immediately when
engine rotational speed is lower than vehicle speed. Gear
ratio is also added in the model to get actual vehicle
speed.
݊ℎ௢௨௦� = ௩ೡ೐ℎ�೎೗೐ଶ�∗௥೏�೙�೘�೎ ∗ 60 ∗ ݃݁ܽݎ ݎܽݐ�݋ (10)

VOL. 11, NO. 4, FEBRUARY 2016 ISSN 1819-6608
ARPN Journal of Engineering and Applied Sciences
©2006-2016 Asian Research Publishing Network (ARPN). All rights reserved.
www.arpnjournals.com
2765
Gear ratio is also used to get actual force from
the engine to the wheel. Then, engine power used is also
calculated.
��௡��௡� = ��೎೟ೠ�೗௥೏�೙�೘�೎ × ݃݁ܽݎ ݎܽݐ�݋ (11)
��௡��௡� = ��௡��௡� ∗ �௩�ℎ���� (12)
TRACK MODELING
Track is mapped using Google earth to get the
elevation of every dot in the track. Then, by using the
distance of every dot, inclination and declination along
the track is calculated. Example of track mapping is
shown in Figure-4 and the result of track mapping is
shown in Figure-5.
Figure-4. Mapping of Sepang International circuit
(Rahmanu, 2011).
Figure-5. Sepang SEM Asia circuit, elevation vs distance
(Rahmanu, 2011).
DRIVER MODES
Because engine power is bigger than required
power to move the vehicle along the circuit, therefore the
driver should start and stop the engine at several point
along the track to minimize fuel consumed. This condition
was modelled as driver modes and become the important
point to variate the vehicle speed to achieve less than
maximum time limit and minimizing fuel consumed.
Figure-6. Driver modes to start and stop engine
at several location of the circuit.
To simulate driver behaviour in this model, it
was assuming that the driver will start the engine and
accelerate until a certain upper speed limit of the vehicle,
then the engine will be stopped. At this time, the vehicle
will be gliding and decelerate due to resistance forces.
When lower speed limit is attained, the driver will start
the engine and accelerate again. This cycle is repeated
until near the finish line. At certain distance of the finish
line, the engine cannot be started. Upper and lower speed
limit were adjusted accordingly to achieve the least fuel
used. However, the adjustment of these speed limits
should be careful not to get over the lap time limit of the
competition that is 315 seconds for one race lap. Base on
experiment, starting the engine consumes a certain
amount of fuel without giving any power to be used.
Therefore, number of engine starting should also be
counted to calculate fuel used.
FUEL CONSUMPTION CALCULATION
Fuel consumption of the vehicle is calculated by
using specific fuel consumption data from Figure-2 and
engine rpm data. Then, absolute fuel consumption (be)
and total fuel mass required (m) could be calculated.
ܾ� = ݏ݂ܿ ∗ ��௡��௡� ∗ ଵଷ6଴଴଴଴଴ (13)
∫ ܾ� ݀ݐ = ݉௧଴ (14)
Distance (s) and fuel density is needed to get the
vehicle fuel consumption in kilo meter per liter.
Fuel consumption = sm/ρ (15)
MODEL OVERVIEW
Total Simulink model was built as shown in
Figure-7. Upper part of the model shown was dedicated to
vehicle dynamic and track characteristic. Lower part of
the model was driver behaviour and drive train modelling.
Upper and lower speed limit should be chosen
appropriately to give best fuel consumption.
-15
-10
-5
0
5
0 1000 2000 3000
e
le
v
a
ti
o
n
(
m
)
distance (m)
VOL. 11, NO. 4, FEBRUARY 2016 ISSN 1819-6608
ARPN Journal of Engineering and Applied Sciences
©2006-2016 Asian Research Publishing Network (ARPN). All rights reserved.
www.arpnjournals.com
2766
Figure-7. Complete model of vehicle, drive train, circuit and driver modes.
RESULTS AND DISCUSSIONS
Validation was done using SEM Asia 2012
vehicle and driver data at Sepang International Circuit. At
that time the achievement record is 167 km per liter using
4 times start stop cycles. Similar race situation was
simulated using the model in Simulink. Given the track
model and all vehicle parameter inserted, the model did
the simulation of the race.
Figure-8 shows the vehicle speed and engine on-
off mode at upper speed of 35 km per hour and lower
speed of 25 km per hour. At these speed limit there are
four times engine starting in one lap.
Figure-9 shows the time required to race in the
track with previous start and stop cycles. The time is just
slightly less than 315 seconds for one lap race. Figure 10
shows a resulted record of 174 km per liter. A very close
approximation with less than 5% of error from the real
record of 167 km per liter.
Figure-8. Vehicle speed and engine on-off mode.
Figure-9. Time and distance traveled in one lap race.
Figure-10. Simulation results of fuel consumption.
0
0.5
1
1.5
-20
0
20
40
60
80
-1000 0 1000 2000 3000
o
n
-o
ff

sp
e
e
d
(
k
m
/h
)
distance (m)
speed
on-off
-1000
0
1000
2000
3000
0 100 200 300 400
d
is
ta
n
ce
(
m
)
time (sec)
VOL. 11, NO. 4, FEBRUARY 2016 ISSN 1819-6608
ARPN Journal of Engineering and Applied Sciences
©2006-2016 Asian Research Publishing Network (ARPN). All rights reserved.
www.arpnjournals.com
2767
CONCLUSIONS
Even though the effect of vehicle speed on the
drag coefficient and tire rolling resistance is neglected, the
simulation result shows that the model almost represents
vehicle and driver performance in real race condition. The
model also could be used as source of information in
evaluating the vehicle to improve its performance for the
next SEM Asia competition.
REFERENCES
I. Rahmanu, Pemodelan dan Simulasi Performa Sapu
Angin 1 Menggunakan Matlab Simulink, Tugas Akhir,
Sepuluh Nopember Institute of Technology, Surabaya,
2010.
AAG. Adeniyi, A. Mohammed, Eco-marathon car driving
pattern and miles per gallon, AU J.T. 15(4): 246-252
(Apr. 2012).
E. Grunditz, E. Jansson, Modeling and simulation of a
hybrid electric vehicle for Shell Eco-marathon, MSc
Thesis in Electric Power Engineering, Chalmers
University of Technology, Goteborg, 2009.
Witantyo, et al., Optimasi metoda pengemudian untuk
meminimalkan konsumsi BBM dengan gabungan
pemodelan karakteristik kendaraan dengan lintasan,
Simposium Nasional RAPI XII FT UMS Solo, 2013.
L. Jeongwoo, Vehicle Inertia Impact on Fuel
Consumption of Conventional and Hybrid Electric
Vehicles Using Acceleration and Coast Driving Strategy,
Virginia Polytechnic Institute and State University,
Blacksburg, Virginia, 2009.
Yanmar L48V Performance Curve, [Online]
http://www.yanmarengines.co.uk/detail.php?
product_id=29.
Sepang International Circuit, North Track, [Online]
http://www.sepangcircuit.com.my/story/sepang-circuits-
north-track.