[Downloaded from www.aece.ro on Thursday, January 11, 2024 at 05:30:35 (UTC) by 58.169.10.68. Redistribution subject to AECE license or copyright.
Advances in Electrical and Computer Engineering
Improving the Energy Management of a
Solar Electric Vehicle
Muhammet Tahir GÜNEŞER, Erzat ERDİL, Mihai CERNAT, Turgut ÖZTÜRK
Electrical and Electronics Engineering Department, Karabük University, 78050 Karabük, Turkey
mtguneser@karabuk.edu.tr
1
Abstract—A solar electric vehicle (SEV) is an electric
vehicle (EV) with onboard photovoltaic cells charging a set of
batteries for extended driving range. This study aimed to
improve the energy management system of a SEV, called
YILDIZ, using a fuzzy logic control system (FLC). A MATLAB
based simulation model of three basic components of a solar
car: solar cell modules, batteries and motor drive system was
performed. An original FLC was developed. For proving its
applicability, the performances of the SEV were tested by
simulation, in accordance with the standard test drive cycle
ECE-15. The characteristics obtained with the original
Proportional Integral Fuzzy Logic Control (PI-FLC) were
compared with those obtained with a classical Proportional
Integral (PI) controller. Using the designed model, we calcu-
lated the range of YILDIZ with and without PV feeding which
gave us an opportunity to study and compare both SEV and
EV models on real race-track situation. Then the optimum
speed, at any time, which enabled the vehicle to reach a chosen
destination as quickly as possible, while fully using the
available energy, was calculated. Proposed solutions tested on
YILDIZ. Results of simulations were compared with YILDIZ
run on the Formula-G race track in Izmit, Turkey.
Index Terms—batteries, energy management, induction
motor, motion control, photovoltaic cells, solar energy.
I. INTRODUCTION
A solar electric vehicle (SEV) is an electric vehicle (EV)
with onboard photovoltaic cells charging a set of batteries
for extended driving range [1].
In this study, we developed a special fuzzy logic control
system aimed to improve the energy management of a solar
electric vehicle for extending its travelling range. In prin-
ciple, it is possible to use directly energy produced by solar
modules to drive a solar car, but using batteries as a stored
energy supply is a better and more efficient choice [2-3].
Basic components of a solar electric vehicle are [1-8]:
 solar cell modules;
 rechargeable energy storage system with batteries;
 electric drive motor and gear box;
 control system, which includes a maximum power
point tracker (MPPT);
 electrical distribution system;
 wheels and tires;
 body, chassis, suspension.
The block diagram of a typical SEV is shown in Fig. 1.
Solar modules and a Li-ion battery are used as energy
sources, via MPPT; the output voltage is compatible for
charging the battery and for supplying the electric motor
drive. Excess energy produced by solar modules is diverted
1
This work was supported in part by TUBITAK under Grant KBÜ-BAP-
13/1-DR-004.
Digital Object Identifier 10.4316/AECE.2015.04007
1582-7445 © 2015 AECE
Volume 15, Number 4, 2015
to charge batteries. Batteries also stored energy supplied by
the electric motor running on regenerative mode, during
braking and downhill motion of the vehicle [3, 5, 9].
The control system shares electrical energy between the
battery and the drive system, and commands the inverter.
The speed of the traction motor is regulated by the Field
Oriented Control (FOC) method [6].
Figure 1. The block diagram of solar electric vehicle
Solar cars can be designed for car races, considering the
special driving rules of that race [1]. The pilot, striving to
reach destination as soon as possible, needs to consider the
remaining energy stored in batteries, the level of solar cells
charging and the remaining distance to destination. Further-
more, the track conditions such as up- and down-hill, and
turns that effect energy consumption rate of a SEV need
consideration [3, 5, 6]. For example, since the acceleration
causes a higher rate of energy consumption, the pilot needs
to keep speed constant as long as possible without sacri-
ficing the spirit of the race [10-11]. Solar vehicles can also
be designed for transportation on short distances, as between
departments of an organization or golf areas.
This paper proposes an original Proportional-Integral
Fuzzy-Logic controller that increases the range of the solar
car by improving the energy management, assuming that the
meteorological conditions are relatively favorable. The
system automatically optimizes the speed at short intervals,
based on the total available energy i.e. the stored energy
remaining in the batteries, the energy produced by the solar
cells, taking in account the distance to destination. So it will
maximize the range. The proposed solution was tested on
our solar car on the Formula-G race track in Izmit.
The elaborated model of solar electric vehicle also can be
used to calculate the range of an electric vehicle with and
without PV feeding.
II. MATHEMATICAL MODELING
Modeling of certain parameters is required for describing
behavior and evaluating parameters of an electric vehicle.
Hence, the vehicle motion could be described by motion
equations and the power required moving the electric vehic-
le could be calculated with these equations. The retarding
forces: rolling resistance, aero-dynamic drag, uphill and
53
 [Downloaded from www.aece.ro on Thursday, January 11, 2024 at 05:30:35 (UTC) by 58.169.10.68. Redistribution subject to AECE license or copyright.]

Advances in Electrical and Computer Engineering Volume 15, Number 4, 2015

acceleration forces, all being dependent on road conditions, B. Field Oriented Control and Fuzzy Logic Controller
need to be modelled. The electric motor and the battery- The control system shares electrical energy between the
bank have to be also modelled. Simulation helped to ensure battery and the drive system, and commands the inverter.
that equations were compatible with each other [3, 6]. The speed of the traction motor is regulated by the Field
Certain factors that could affect the motion of the vehicle, Oriented Control (FOC) method [6].
such as uphill-downhill slopes were considered. Additio- The energy management was optimized through a control
nally, the relationship between the electric machine, drive- system that used a Proportional Integral Fuzzy Logic Con-
train system, torque of the motor and the force transferred to troller (PI-FLC). We compared the performances of chosen
the wheels were modeled. control method with those that uses a Proportional Integral
The rolling resistance force F rr is directly proportional to (PI) controller. For this system PI-FLC was applied on FOC
the tire friction and the mass, as shown below as seen on Fig. 2.
Frr   rr m g , (1)
where μ rr is rolling resistance's coefficient.
The aero-dynamic drag force F ad caused by air friction is
Fad  0,5 Ac Cd v2 , (2)
where ρ is the air density, A c is the cross-section area, v is
the speed, and C d represents the constant drag coefficient.
The hill climbing force F hc is given by:
Fhc  m g sin () , (3) Figure 2. The block diagram of PI-FLC for FOC
where m is mass of the vehicle, g is the gravity constant, and
Where V d and V q are the applied voltages to the stator and
α is the slope angle.
i d and i q are the corresponding d and q axis, stator current
The traction force F la is:
and rotor currents. According to the working principle of
Fla  m a , (4)
FOC, two PI controllers calculate the voltage references V d
where a represents the linear acceleration. and V q from the reference currents. The third PI controller
The total traction force F te is: calculates the angular speed from reference values. The
Fte  Frr  Fhc  Fad  Fla . (5) rotor speed ω r is compared to rotor demand speed ω r * and
The total traction torque T te can be calculated by the resulting error is processed in the PI speed controller.
r , (6) But in this survey we replaced FLC instead of PI controller,
Tte  Fte
G which was speed controller, as seen on Fig. 2. The block
where r represents the wheel radius and G the gear ratio (a diagram of PI controller is seen on Fig. 3.
fixed one, in this paper).
The minimum traction power P te required to move the
vehicle with velocity v is given by
Pte  Fte v . (7)

A. Induction Motor Modeling

The induction motor is modelled by a system of differ- Figure 3. The block diagram of PI controller
ential electromechanical equations with currents and fluxes
of stator and rotor as state variables [12-14]: PI controller is the most used system for controlling speed
M2
of electrical motors, because it supplies a good control and it
Lr Rs  can be easily designed. However, when the mathematical
d id r M M 1
 id  g iq  d  r q  vd (8) model of the system is complex and nonlinear, PI controllers
dt  Lr Ls r  Lr Ls  Lr Ls  Ls
cannot provide an accurate output [18-20].
M2 Opposite, FLC can provide high performance for
Lr Rs 
diq r M M 1 complex nonlinear systems and can overcome nonlinearities
 iq  g id  q  r d  vq (9)
dt  Lr Ls r  Lr Ls  Lr Ls  Ls and uncertainties [21, 22]. FLC needs two different
d d M 1 (10) descriptions. First description consist of input and output
 id  d  (g  r ) q
dt r r variables, fuzzy rules structure, number and type of
d q M membership functions, the inference mechanism type, and
1 (11)
 iq   q  ( g  r )  d de-fuzzification. The second one consists of the parameters
dt r r of membership functions and fuzzy rules. The rule base,
2
dr 3 n p M Tf np data base, inference mechanism and de-fuzzification are
dt

2 J Lr
 J J

d iq  q id  r  TC  TL  (12)
four important components for FLC systems [23-25].
Our FLC system was modelled using data algorithms to
where the used notations are explained in Appendix B and
obtain suitable fuzzy sets and rules for optimum speed on
 r  Lr / Rr and   ( Lr Ls  M 2 ) /( Lr Ls ) are the time con-
the racetrack. It was applied to generate the torque demand
stant and the leakage constant, respectively. signal from the speed error.
The values and provisions of the variables are given in The inputs for the recommended fuzzy controllers were
Appendix B [13-17]. converted to fuzzy variables that ranged between -1 and 1.

54
 [Downloaded from www.aece.ro on Thursday, January 11, 2024 at 05:30:35 (UTC) by 58.169.10.68. Redistribution subject to AECE license or copyright.]

Advances in Electrical and Computer Engineering Volume 15, Number 4, 2015

The membership functions were the speed error (eν), the TABLE III. SPECIFICATIONS OF LITHIUM-ION BATTERIES [1]
difference of the speed error (d eν) and desired torque ( Te*). Capacity 12Ah
Nominal Voltage: 3,7 V
The membership functions are shown in Fig. 4 and Fig. 5.
Charge Max. Current 36 A
The membership of the fuzzy consisted of five triangular Condition Max. Voltage 4.2V ±0.03 V
functions that were named NL (Negative Large), NS
Continuous Current 60 A
(Negative Small), ZO (Zero), PL (Positive Large), and PS Discharge
Peak Current 240 A
Condition
(Positive Small) [22]. Cut-off Voltage 2,7 V

For modeling Li-ion batteries we used:

 Discharge model (i* > 0)
K Q0 * K Q0
f1 (it , i* , i )  E0  i  it  A exp( B it ) (13)
Q0  it Q0  it
Figure 4. The membership functions for eν and d eν  Charge model (i* < 0)
K Q0 K Q0
f 2 (it, i*, i)  E0  i*  it  Aexp(B it) (14)
(it  0,1Q0 ) Q0  it
where the following notations have been used:
E 0 - battery constant voltage;
K - polarization resistance (constant);
i* - low frequency dynamic current;
i- battery current;
Figure 5. The membership functions for Te*
it - extracted capacity (electric charge);
Q 0 - maximum battery capacity (electric charge).
The decision of the fuzzy inference consists of 25 input-
A, B - constants.
output rules for the 2-input fuzzy system. These rules are
The State of Charge (SOC) characterizes the electric
shown in Table I and are if-then expressions that identify the
charge that a battery can supply. SOC is the ratio of the
consequent output torque response for a given input require-
available charge at an arbitrary moment Q(t) to the
ment to obtain the desired speed.
maximum battery capacity (Q 0 ), where it represents total
C. Battery Modeling charge consumed until time t [26-28].
The state of battery charge, the energy consumption and Q(t ) Q0  i t (15)
SOC  
generation and the distance to destination, at any time, were Q0 Q0
also taken into consideration. Li-ion batteries offer flexible Using Eqs. (13-15), and data from Table III, charge-dis-
design and higher energy density as seen on Table II. charge diagram of the battery-bank was drawn (Fig. 6). An
increase in discharge current indicated an increase in speed
TABLE I. FUZZY RULE TABLE or climbing a hill; as seen on bottom graph of Fig. 6, higher
eν discharge currents cause faster decrease of the SOC.
NL NS ZE PS PL
d eν
Nominal Current Discharge Characteristic at 0.43478C (5.2174A)
NL T1 1 T1 2 T2 3 T2 4 T3 5 100
Discharge curve
NS T1 6 T2 7 T2 8 T3 9 T4 10 90 Nominal area
Voltage

Exponential area
ZE T2 11 T2 12 T3 13 T4 14 T4 15 80

70
PS T2 16 T3 17 T4 18 T4 19 T5 20
60
PL T3 21 T4 22 T4 23 T5 24 T5 25 0 0.5 1 1.5 2 2.5 3
Time (hours)
E0 = 88.1869, R = 0.067833, K = 0.038291, A = 6.9156, B = 5.0885
TABLE II. DATA OF DIFFERENT BATTERY TYPES [1] 100 6.5 A
Max. Battery Energy Density
Battery Chemistry 13 A
Mass (kg) (kW/kg) 90
Voltage

32.5 A

Sealed Lead acid 125 40 80

NiMH 70 80 70

Ni/Zn 75 60 0 0.5 1 1.5 2 2.5

Time (hours)
Ni/Fe 100 50
Figure 6. Charge-discharge diagram of the battery-bank
Li-ion 25 160
Li-polymer 25 200
D. Modelling of Solar Cells
LiFePO4 50 90
The current flow in a typical solar cell changes with
temperature as shown in Eq. (16), where T is temperature,
Li-ion batteries have lots of advantages such as high cell
IL photocurrent, e = 1.6 × 10−19 C electron charge, K =1.38
voltage, low self-discharge rate and long life-cycle [26-27].
× 10−23 J/K Boltzmann constant, I 0 saturation current, R s
A bank of 22 Li-ion batteries was used, characteristics of
serial resistant to represent internal losses, V output voltage
which are shown in Table III [1]. Series connection gave
and n number of series connected cells [28-31].
nominal voltage of 81,4 V, and nominal capacity of
Characteristics of the 402 pieces of solar cells mounted on
976,8 Wh (VAh).

55
 [Downloaded from www.aece.ro on Thursday, January 11, 2024 at 05:30:35 (UTC) by 58.169.10.68. Redistribution subject to AECE license or copyright.]

Advances in Electrical and Computer Engineering Volume 15, Number 4, 2015

SEV, are shown in Table IV. Data given in Table IV and 18

PV PANEL CURRENT-VOLTAGE

Eq. (16) were used to generate the PV Current-Voltage and 16

@1000W/m2 Data-25°C
@1000W/m2 Model-25°C
the PV Power-Voltage diagrams of the solar cells shown on @1000W/m2 Model-40°C
14
Fig. 7 (top) and Fig. 5 (bottom), respectively. @1000W/m2 Model-55°C

e V  I s Rs   .
@1000W/m2 Model-70°C

12

I s  I L  I0 1  exp (16)


Current [A]
10

 nKT 
8

6
TABLE IV. CHARACTERISTICS OF YILDIZ'S SOLAR CELLS
Unit Description Value 4

Ps Power 3,51 W 2

V Open circuit voltage 0,65 V 0

OC 0 10 20 30 40 50 60 70 80 90

I sc Short circuit current 5,40 A Voltage [V]

Nominal voltage (a)

Vnom 0,51 V PV PANEL POWER-VOLTAGE

I nom Nominal current 5,2 A 400

@1000W/m2 Data-25°C

Influence of temperature change to V nom -1,8 mV/ºC 350 @1000W/m2 Model-25°C

@1000W/m2 Model-40°C
Influence of temperature change rate to energy -0,32 %/ºC
300 @1000W/m2 Model-55°C
@1000W/m2 Model-70°C
The 402 pcs of Solar Cells were mounted in three homo- 250

Power [W]
geneous strings (P mp =470 W) each and connected to three 200

homogenous MPPT having input voltage (V OC ) of 87,1 V, 150

and output Voltage (V Batt ) of 96 V.
100
In this configuration, solar modules could supply the bat-
tery bank and the Induction Motor at the same time with a 50

current (I s ) of 16,2 A. Selected MPPT supports these con- 0

0 10 20 30 40 50 60 70 80 90
figurations as shown on Table V. The Current-Voltage and Voltage [V]

the Power-Voltage diagrams of solar modules formed with (b)

Figure 8. Current-Voltage and Power-Voltage diagrams of solar mo-dule
these cells, are shown on Fig. 8(a) and Fig. 8(b). for different ambient temperatures.
Thermal losses due to internal resistances in the cells are
reduced by air flow during the car motion; this allows III. COMPARISON OF SIMULATION RESULTS AND TEST RUNS
extending the range [29-31]. ON SOLAR ELECTRIC VEHICLE „YILDIZ”
Hence, both Eq. (2) and Eq. (16) are used to calculate, via
PI-FLC system, the optimum speed for minimal thermal and A. Modeling Results
aero-dynamic drag force losses. The race track, where SEV „YILDIZ” was tested is
PV CELL CURRENT-VOLTAGE 1950 m long. The first 975 m were uphill and the remaining
6
downhill. Since the race strategy was to have maximum
@1000W/m2 Data-25°C
available speed at any time, a speed control strategy is
Current [A]

4
@1000W/m2 Model-25°C

2
@1000W/m2 Model-40°C required.
@1000W/m2 Model-55°C
@1000W/m2 Model-70°C
Several driving cycle models are offered in the literature;
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 we chosen the widely used standard test drive cycle ECE-15
Voltage [V]
PV CELL POWER-VOLTAGE
[32] (Fig. 9). It has a significant advantage since it involves
4
periods of deceleration during which energy can be
@1000W/m2 Data-25°C
3
@1000W/m2 Model-25°C recovered. It has a maximum speed of 120 km/h, a duration
Power [W]

2 @1000W/m2 Model-40°C of 575 s, total distance of 7.7 km and an average speed of

@1000W/m2 Model-55°C
1
@1000W/m2 Model-70°C
62,6 km/h. Some speed rates are too high our SEV, YILDIZ,
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
which was only capable of 52 km/h maximum speed. So, we
Voltage [V] used only the first 2000 meters of the cycle with maximum
Figure 7. Current-Voltage and Power-Voltage diagrams of solar cells for speed not exceeding 50 km/h in that phase.
different ambient temperatures. 120

TABLE V. SPECIFICATIONS OF THE MPPT INSTALLED ON YILDIZ 100

Unit Description Value

ECE-15 Reference Speed (km/h)

T amb Maximum ambient air temperature 50°C 80

PV module short circuit current –

I sc-const 6A
constant 60

PV module short circuit current –

I sc-trans 8A
transient 40

P mp PV module peak power 0- 600 W

20
V oc PV module open circuit voltage 40 – 135 V
Efficiency for Is=6 A Vs=100 V, & 25ºC 98.00% 0
72, 96, 120, 144, 0 100 200 300 400 500 600
V Batt ( E 0 ) Battery Voltage (Selectable) Time (s)
168 V
Figure 9. The driving cycle ECE-15 [33].

56
 [Downloaded from www.aece.ro on Thursday, January 11, 2024 at 05:30:35 (UTC) by 58.169.10.68. Redistribution subject to AECE license or copyright.]

Advances in Electrical and Computer Engineering Volume 15, Number 4, 2015

SPEED OF ELECTRIC VEHICLE AND REFERENCE SPEED OF ELECTRIC VEHICLE AND REFERENCE

Electric Vehicle Speed Electric Vehicle Speed

40 Reference Speed 40 Reference Speed

Speed [km/h]
Speed [km/h]

20 20

0 0
0 50 100 150 200 250 0 50 100 150 200 250
Time [s] Time [s]
ERROR SPEED FOR PI CONTROL ERROR SPEED FOR PI-FUZZY CONTROL
0.8 0.015

Error Speed [km/h]

Error Speed [km/h]

0.6
0.01
0.4
0.005
0.2

0 0
0 50 100 150 200 250 0 50 100 150 200 250
Time [s] Time [s]

(a) (c)
STATE OF CHARGE STATE OF CHARGE

State of Charge [SOC %]

State of Charge [SOC %]

80 80

75 75

70 70
0 50 100 150 200 250 0 50 100 150 200 250
Time [s] Time [s]
CHANGE OF ENERGY CONSUMPTION ON BATTERY PACKAGE CHANGE OF ENERGY CONSUMPTION ON BATTERY PACKAGE
850 850

Change of Energy [Wh]

Change of Energy [Wh]

800 800

750 750
0 50 100 150 200 250 0 50 100 150 200 250
Time [s] Time [s]

(b) (d)
Figure 10. Control Performance and Error Speed of PI on the Driving Cycle (a); State of Charge (SoC) and Change of Energy Consumption on the Driving
Cycle (b); Control Performance and Error Speed of PI-Fuzzy on the Driving Cycle (c); State of Charge (SoC) and Change of Energy Consumption on the
Driving Cycle (d).
BATTERY-ELECTRIC MOTOR-PV PANEL CURRENT BATTERY-ELECTRIC MOTOR-PV PANEL CURRENT
60
40
50
38
40
36
30
34
Current [A]
Current [A]

20
32
10
30
0
28
-10
Battery Current 26
-20 Electric Motor Current
PV Panel Current 24
-30 39.5 40 40.5 41 41.5 42 42.5 43 43.5 44 44.5
0 200 400 600 800 1000 1200 1400 1600 1800 2000
Time [s]
Time [s]

(a) (c)
BATTERY-ELECTRIC MOTOR-PV PANEL CURRENT BATTERY-ELECTRIC MOTOR-PV PANEL CURRENT
60 22

20
50
18
40
16

30
Current [A]

14
Current [A]

20 12

10
10
8
0
6

-10 4
92 94 96 98 100 102 104 106
0 10 20 30 40 50 60 70 80 90 100
Time [s]
Time [s]

(b) (d)
Figure 11. DC Currents of PV modules, battery outputs and motor inputs by speed of 52 km/h (a); DC currents of PV and battery outputs and motor inputs
while first acceleration (b); DC currents of battery outputs and motor inputs at tracking point of demanded speed (c); DC currents of battery outputs and
motor inputs at top of hill (d).

57
 [Downloaded from www.aece.ro on Thursday, January 11, 2024 at 05:30:35 (UTC) by 58.169.10.68. Redistribution subject to AECE license or copyright.]

Advances in Electrical and Computer Engineering Volume 15, Number 4, 2015

BATTERY CURRENT FOR ELECTRIC MOTOR STATE OF CHARGE

60 100

State of Charge [SOC %]

50

50
40
Battery Current [A]

30
0
0 200 400 600 800 1000 1200
20 Time [s]
CHANGE OF ENERGY CONSUMPTION ON BATTERY PACKAGE
1000

Change of Energy [Wh]

10

0
500

-10

-20 0
0 200 400 600 800 1000 1200 0 200 400 600 800 1000 1200
Time [s] Time [s]

(a) (b)
SPEED OF ELECTRIC VEHICLE AND REFERENCE STATE OF CHARGE
100

State of Charge [SOC %]

Electric Vehicle Speed
40
Speed [km/h]

Reference Speed

50
20

0 0
0 200 400 600 8001000 1200 1400 1600 1800 2000 0 200 400 600 800 1000 1200 1400 1600 1800 2000
Time [s] Time [s]
ERROR SPEED FOR PI-FUZZY CONTROL CHANGE OF ENERGY CONSUMPTION ON BATTERY PACKAGE
0.02 1000

Change of Energy [Wh]

Error Speed [km/h]

0.015

0.01 500

0.005

0 0
0 200 400 600 800 1000 1200 1400 1600 1800 2000 0 200 400 600 800 1000 1200 1400 1600 1800 2000
Time [s] Time [s]

(c) (d)
RANGE OF EV (w/o PV Panel) RANGE OF EV (w/o PV Panel)

16
25
14

12 20

10
Range [km]

Range [km]

15
8

6 10

4
5
2

0 0
0 200 400 600 800 1000 1200 0 200 400 600 800 1000 1200 1400 1600 1800
Time [s] Time [s]

(e) (f)
Figure 12. Battery current of EV with the speed of 52 km/h (a); State of Charge (SOC) and Change of Energy Consumption with maximum speed -
minimum range of EV (b); Control Performance and Error Speed of 52 km/h of SEV (c); State of Charge (SoC) and Change of Energy Consumption with
maximum speed - minimum range of SEV (d); The range of EV with the speed of 52 km/h (e); The range of SEV with the speed of 52 km/h (f).

Both, PI and PI-FL control systems were used in the same seen on Fig. 11(b). Having reached the maximum speed at
driving cycle, as seen on Fig. 10, respectively, to record per- the 40th second, the control system stopped accelerating and
formance of YILDIZ with these simulation models. the load current decreased from 58 A to 38 A as shown on
Considering the SOC, which reflects consumption rate, Fig. 11(c). At the 93th s, it arrived top of the hill and the load
performance of both systems were good. But for the PI-FL current decreased to 19 A. During downhill stage, regenera-
control system, error speed values, i.e. instantaneous change tive braking charged battery-bank while PV modules partly
in speed, were very small. Small error speed indicates good supplied the motor. As seen on Fig. 11(d), the battery-bank
performance of the PI-Fuzzy controller to track the supplied only 6 A to complement PV production. All
reference speed. Hence PI-FLC was selected for this survey changes occurred in less than 0,1 seconds.
due to its high performance on tracking any speed change Detailed performance of PI-FLC is seen on Fig. 11(c) and
almost instantaneously. We applied PI-FLC simulation to Fig. 11(d) which are enlargements of Fig. 11(a). From the
Formula – G racetrack with maximum speed as 52 km/h and expanded views on Fig. 11(c) and Fig. 11(d) it can be seen
tried to observe, whether the controller could catch desired that that PI-FLC accurately controlled the speed changes
speed uphill and downhill. while driving uphill and downhill. So PI-FLC system was
The electric motor consumed more energy and drew 58 A performing well on ECE-15 driving cycle and Formula-G
current while first accelerating to reach speed to 52 km/h, as racetrack simulations as well.

58
 [Downloaded from www.aece.ro on Thursday, January 11, 2024 at 05:30:35 (UTC) by 58.169.10.68. Redistribution subject to AECE license or copyright.]

Advances in Electrical and Computer Engineering Volume 15, Number 4, 2015

SPEED OF MODEL AND SEV BATTERY CURRENT

55 60

50
50
45
40
40
30
35
Speed [km/h]

Current [A]
30 20

25 10

20
0
15
-10
10

5 Model Speed -20 Model Battery Current

SEV Speed SEV Battery Current
0 -30
0 200 400 600 800 1000 1200 1400 1600 1800 0 200 400 600 800 1000 1200 1400 1600 1800 2000
Time [s] Time [s]

(a) (c)
ELECTRIC MOTOR CURRENT BATTERY, ELECTRIC MOTOR and PV PANEL CURRENTS of SEV
70 70

60 60

50 50

40 40
Current [A]

Current [A]
30 30

20 20

10 10

0 0
Electric Motor Current
-10 Model Electric Motor Current -10 Battery Current
SEV Electric Motor Current PV Panel Current
-20 -20
0 200 400 600 800 1000 1200 1400 1600 1800 2000 0 200 400 600 800 1000 1200 1400 1600 1800
Time [s] Time [s]

(b) (d)
Figure 13. Acceleration and speed control comparison of MATLAB model and real-race tracks of SEV (a); Electric motor current comparison of MATLAB
model and real-race tracks of SEV (b); Battery current comparison of MATLAB model and real-race tracks of SEV (c); PV-Battery-Motor currents of real-
racetrack of SEV (d).

were fully compatible with the model, however the driving

B. The Simulation to Identify Range Difference between EV
and SEV
We simulated two runs on Formula-G race track with
constant speed, 52 km/h, which is maximum speed of
YILDIZ, to identify range difference between Solar Electric
Vehicle (SEV) and Electric Vehicle (EV), for both models.
It was performed the run of EV for a race model with
maximum speed–minimum range scenario. Until the energy
stored in the battery exhausted EV lasted 1210 s as seen on
Fig. 12(a) and Fig. 12(b).
Second run was simulation of SEV at same conditions.
SEV lasted 1960 s as seen on Fig. 10(c) and Fig. 10(d). The
range was covered 16,3 km by EV and 26.6 km as seen on
Fig. 12(e) and Fig. 12(f) respectively. This meant that
adding PV modules to EV could improve the range by 63%.
C. Proving the MATLAB Model
For verifying the model used in simulations, YILDIZ runs
on the Formula-G race track in Izmit, Turkey. The obtained
data are presented in Fig. 13. In real racetrack, YILDIZ
could not be driven at 52 km/h for so long. It was affected
by not intentional pilot interventions, road conditions and
other factors that caused losses, e.g. heating effect.
Additionally, an accident occurred on the 982th s, as
highlighted by brown arrow in Fig. 13(a). All these caused a
mismatch between the model and the race values of SEV
(Fig. 13(a)).
To test correlation between the simulation model and the
real vehicle, more detailed graphs were plotted as in Fig. 13.
As seen in Fig. 13(a), acceleration and slow down of SEV
style caused plenty of changes at the speed value. Electric
motor currents and battery charge-discharge stages are
shown in Fig. 13(b). Comparison of battery, PV panel and
motor currents for the MATLAB model and the real race
runs are shown in Fig. 13(c) and Fig. 13(d). These results
showed that simulation model closely follows performance
of YILDIZ on the Formula-G race track, what meant that the
designed PI-FLC system was acceptable for simulation of
SEV model. Accordingly, a range model could be uploaded
and SEV speed controlled could be programmed to reach
destination as quickly as possible.
D. The Simulations to Identify Improving Range of our SEV
We assumed two different simulation strategies to can
improve the range, first simulation determined as optimal
steady speed for desired range, latter is variable maximum
speed for desired range.
As known from modelling results, it can be seen that it
was possible to choose a speed where the energy consumed
during acceleration and uphill climb could be replaced by
downhill regenerative gain and PV production.
In first simulation scenario (optimal steady speed for
desired range); at the constant speed, which was calculated
to be 29,5 km/h; SEV arrived at the destination, 39 km, in
4740 s as seen on Fig. 14. As seen on Fig 14(a), battery and
motor currents are constant during uphill and downhill, the
changes are seen only top of hill and end of descent. And the
range graph, which is seen on Fig. 14(c), is linear because of
constant speed except starting and stopping times as seen
Fig. 14(d).

59
 [Downloaded from www.aece.ro on Thursday, January 11, 2024 at 05:30:35 (UTC) by 58.169.10.68. Redistribution subject to AECE license or copyright.]

Advances in Electrical and Computer Engineering Volume 15, Number 4, 2015

BATTERY-ELECTRIC MOTOR-PV PANEL CURRENT RANGE OF EV (w/o PV Panel)

50
Battery Current 40
40 Electric Motor Current
PV Panel Current 35

30
30

20
25

Range [km]
Current [A]

10
20

0 15

-10 10

-20 5

-30 0
0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000 0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000
Time [s] Time [s]

(a) ( c)
STATE OF CHARGE RANGE OF EV (w/o PV Panel)
100
State of Charge [SOC %]

39

50 38.98

38.96
0

Range [km]
0 500 1000 1500 2000
2500 3000 3500 4000 4500 5000
Time [s] 38.94
CHANGE OF ENERGY CONSUMPTION ON BATTERY PACKAGE
1000
Change of Energy [Wh]

38.92

500 38.9

38.88
0
0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000 4720 4725 4730 4735 4740
Time [s] Time [s]

(b) (d)
Figure 14. DC Currents of PV modules, battery outputs and motor inputs during race cycle with constant speed (steady speed for desired range) (a); State of
Charge (SoC) and Change of Energy of SEV Consumption with constant speed (b) ; Stopping time and range value of the scenario steady speed for desired
range (c); Stopping time and range value (d).
SPEED OF SOLAR ELECTRIC VEHICLE AND REFERENCE BATTERY-ELECTRIC MOTOR-PV PANEL CURRENT
60 100
Solar Electric Vehicle Speed Battery Current
Reference Speed Electric Motor Current
Speed [km/h]

40 80
PV Panel Current

20 60

0 40
Current [A]

0 500 1000 1500 2000

2500 3000 3500 4000 4500 5000
Time [s]
ERROR SPEED FOR PI-FUZZY CONTROL
20
1.5
Error Speed [km/h]

1 0

0.5 -20

0 -40
0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000 0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000
Time [s] Time [s]

(a) (c)
STATE OF CHARGE RANGE OF SEV (w/o PV Panel)
100 40
State of Charge [SOC %]

35

50 30

25

0
Range [km]

0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000 20
Time [s]
CHANGE OF ENERGY CONSUMPTION ON BATTERY PACKAGE 15
1500
Change of Energy [Wh]

10
1000
5

500
0

0 -5
0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000 0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000
Time [s] Time [s]

(b) (d)
Figure 15. Acceleration and speed control of SEV with PI-FLC (a): State of Charge (SoC) and Change of Energy of SEV Consumption with variable speed
(optimum speed – desired range) (b) ; DC Currents of PV modules, battery outputs and motor inputs during race cycle with variable speed (c); Stopping
time and range value of variable speed (d).

60
 [Downloaded from www.aece.ro on Thursday, January 11, 2024 at 05:30:35 (UTC) by 58.169.10.68. Redistribution subject to AECE license or copyright.]

Advances in Electrical and Computer Engineering Volume 15, Number 4, 2015

Second simulation was made for the scenario variable ACKNOWLEDGMENT

maximum speed for desired range (Fig. 15). The control We would like to thank members of Solar Car Team of
system was programmed such that the vehicle reached the Karabük University (KUGAT), as well as Karabük
destination as quickly as possible by optimal using the University, Turkey.
available energy.
So, SEV accelerated in 30 s to maximum speed 52 km/h APPENDIX A
as seen on Fig. 15(a); after this the control system changed VEHICLE PARAMETERS
the speed according to the State of Charge of the battery as Parameter Name Value
seen on Fig. 15(b). Vehicle mass (m) 270 kg
The trip was of 4498 s, when the battery charge was Frontal area (A c ) 0.8 m2
Aero-dynamic drag coefficient (C d ) 0.19
exhausted. This was faster than the steady speed scenario, Rolling resistance coefficient (μ rr ) 0.048
and the SEV just managed to reach destination as seen on Air density (ρ) 0.23 kg/m3
Fig. 15(d). Some times during the trip, the speed decreased Wheel Radius (r) 0.32 m
Gravitational acceleration (g) 9.81 m/s2
due to lowering of the SOC, and consequently the motor
current also decreased, as seen on Fig.15(c).
APPENDIX B
IV. CONCLUSIONS MOTOR PARAMETERS
Parameter Name Value
This study proposed an improvement of the energy Power rating (P) 3 kW
management system of a solar electric vehicle using a fuzzy Number of poles (n p ) 2
logic control system. The performances of the solar electric Stator resistance (R s ) 0.2147 Ω
Rotor resistance (R r ) 0.2205 Ω
vehicle were tested by simulation, in accordance with the Stator inductance (L s ) 0.065181 H
standard test drive cycle ECE-15. The characteristics Rotor inductance (L m ) 0.065181 H
obtained with the original Proportional Integral Fuzzy Logic Mutual inductance (M) 0.0641 H
Moment of inertia (J m ) 0.102 kgm2
Control (PI-FLC) were compared with those obtained with a Friction torque coefficient (T f ) 0.009541Nms
classical Proportional Integral (PI) controller. The PI-FLC
control system can manage more accurately the speed and REFERENCES
get better performances of Solar Electric Vehicle (SEV). [1] D.T. Wisniewski, “Solar flair: An open-road challenge,” IEEE poten-
Further using the designed model, the range of an electric tials, January-February 2010, pp. 6-10, 2010. [Online]. Available:
vehicle with and without PV feeding, was calculated. Propo- http://dx.doi.org/10.1109/mpot.2009.935608
[2] J. Connors, “On the subject of solar vehicles and the benefits of the
sed models were tested by simulation on our electric ve- technology,” in Proc. of International Conference on Clean Electrical
hicle, called YILDIZ, with Solar Cells (SEV) and without Power, ICCEP '07, pp. 700-705, 2007. [Online]. Available:
solar cells (EV). http://dx.doi.org/10.1109/iccep.2007.384287
[3] O. Ustun, M. Yilmaz, C. Gokce, U. Karakaya, R.N. Tuncay, “Energy
The simulation results were compared with those obtained
management method for solar race car design and application,” in
by YILDIZ on the Formula-G race track in Izmit, Turkey. A Proc. of IEEE International Electric Machines and Drives Conference,
good concordance of the simulation results with measured IEMDC '09, pp. 804-811, 2009. [Online]. Available: http://dx. doi.org
ones validates the proposed model. /10.1109/iemdc.2009.5075296
[4] R. Al Zaher, S. de Groot, H. Polinder, P. Wieringa, “Comparison of
Then the optimum speed, at any time, which enabled the an axial flux and a radial flux permanent magnet motor for solar race
vehicle to reach a chosen destination as quickly as possible, cars,” in Proc. of XIX. International Conference on Electrical Mach-
while fully using the available energy, was calculated. ines ICEM 2010, pp. 1-6, 2010. [Online]. Available: http://dx.doi.org/
10.1109/icelmach.2010.5608195
First scenario for testing by simulation the behavior of [5] B. Wu et al., “A management strategy for solar panel –battery –super
SEV was called optimal steady speed. The speed was capacitor hybrid energy system in solar electric vehicle,” in Proc. of
calculated to be 29,5 km/h; SEV arrived at destination in 8th International Conference on Power Electronics (ECCE Asia),
Korea, pp. 1682-1687, 2011. [Online]. Available: http://dx.
4740 s. doi.org/10.1109/ icpe.2011.5944438
Second scenario for testing by simulation the behavior of [6] P. Pudney, P. Howlett, “Critical speed control of a solar car,” Optimi-
SEV was called optimum variable speed – desired range. zation and Engineering, Kluwer Academic Publishers, vol. 3, pp. 97-
107, 2002.
SEV was programmed such that the vehicle reached the [7] T. Sarkar, M. Sharma, S.K. Gawre, “A generalized approach to design
destination as quickly as possible by optimal usage of both the electrical power system of a solar electric vehicle,” in Proc. of
types of available energy. SEV arrived at destination in 2014 IEEE Students' Conference on Electrical, Electronics and
Computer Science (SCEECS), pp. 1-6, 2014. [Online]. Available:
4498 s, 242 s faster than on the previous scenario. This http://dx.doi. org/10.1109/sceecs.2014.6804490
control system was applied as automatic car control to the [8] A.G. Calderón, G.G. Ruiz, A.C.G. Bohórquez, “GPRS telemetry sys-
solar electric vehicle YILDIZ. So pilot mismatches were tem for high-efficiency electric competition vehicles,” Proc. of 2013
avoided and performance of SEV was improved. World Electric Vehicle Symposium and Exhibition (EVS27), pp. 1-7,
2013. [Online]. Available: http:/dx.doi.org/10.1109/evs.2013.6914788
The proposed system can motivate EV designers to [9] Z. Juda, “Advanced batteries and super-capacitors for electric vehicle
consider adding PV modules to new design EV’s for propulsion systems with kinetic energy recovery,” Journal of KONES
improving range or for decreasing the battery-bank stack. Power Train and Transport, Institute of Aviation, Poland, vol. 18, no.
4, pp. 165-171, 2011.
With minimal changes, the elaborated model was used for [10] Y. Yang, S.C. Down, “Optimal design and control of a wheel motor
EV and we calculated the range of SEV at unfavorable for electric passenger cars,” IEEE Trans. on Magnetics, vol. 43, no. 1,
meteorological conditions when PV feeding falls down. pp. 51-61, 2007. [Online]. Available: http://dx.doi.org/10.1109/tmag.
2006. 886153
[11] C.C. Chan, “The state of the art of electric, hybrid, and fuel cell
vehicles,” in Proc. of IEEE, vol. 95, no. 4, pp. 704-718, 2007.
[Online]. Available: http://dx.doi.org/10.1109/jproc.2007.892489

61
 [Downloaded from www.aece.ro on Thursday, January 11, 2024 at 05:30:35 (UTC) by 58.169.10.68. Redistribution subject to AECE license or copyright.]
Advances in Electrical and Computer Engineering
[12] Y. Zhang, L. Heping, W. Huabin, “Torque control strategy for parallel
hybrid electric vehicles using fuzzy logic,” WSEAS Trans. on
Systems, vol. 10, pp. 116-125, 2011.
[13] J. Wang, X. Zhang, D. Kang, “Parameters design and speed control of
a solar race car with in-wheel motor,” Proc. of 2014 IEEE Transporta-
tion Electrification Conference and Expo (ITEC), pp. 1-6, 2014.
[Online]. Available: http://dx.doi.org/10.1109/itec.2014. 6861798
[14] A. Sevinc, “Speed sensorless control of induction motors,” Ph.D.
Thesis, Department of Electrical Engineering, University of Bristol,
UK, 2001.
[15] A.A. Bohari et al, “Improved FOC of induction motor with online
neural network,” WSEAS Trans. on Power Systems, vol. 9, pp. 136-
142, 2014.
[16] J. Guzinski, H. Abu-Rub, “Sensorless induction motor drive for elec-
tric vehicle application,” International Journal of Engineering,
Science and Technology, vol. 2, no. 10, pp. 20-34, 2010.
[17] A. Damiano, G. Gatto, I. Marongiu, M. Porru, A. Serpi, “Real-time
control strategy of energy storage systems for renewable energy
sources exploitation,” IEEE Trans. on Sustainable Energy, vol. 5, no.
2, pp. 567-576, 2014. [Online]. Available: http://dx.doi.org/10.1109/
tste.2013.2273400
[18] C. Dedid, M. Soebagio., M.H. Purnomo, “Induction motor speed con-
trol with fast response using the Levenberg Marquardt method for
electric cars,” International Journal of Computer Applications, vol.
42, no. 13, pp. 14-18, 2012.
[19] P. Brandstetter, T. Krecek, “Speed and current control of permanent
magnet synchronous motor drive using IMC controllers,” Advances in
Electrical and Computer Engineering, vol. 12, no. 4, pp. 69-76, 2012.
[Online]. Available: http://dx.doi.org/10.4316/aece.2012.04001
[20] P. Palacky et al., “Control algorithms of propulsion unit with
induction motors for electric vehicle,” Advances in Electrical and
Computer Engineering, vol. 14, no. 2, pp. 69-76, 2014. [Online].
Available: http://dx. doi.org/10.4316/aece.2014.02012
[21] M. Khan, N.C. Kar, “Speed tracking performance of fuzzy based
vector control of induction motor drives for hybrid electric vehicles,”
in Proc. of Canadian Conference on Electrical and Computer
Engineering, CCECE 2008, pp. 607-610, 2008. [Online]. Available:
http://dx.doi.org/10.1109/ccece.2008.4564606
[22] F. Khoucha, M.E.H. Benbouzid, A. Kheloui, “An optimal fuzzy logic
power sharing strategy for parallel hybrid electric vehicles,” Vehicle
Power and Propulsion Conference (VPPC), pp. 1-5, 2010.
[23] K. Laroussi, M. Zelmat, M. Rouff, “Implementation of a fuzzy logic
system to tune a PI controller applied to an induction motor,”
Advances in Electrical and Computer Engineering, vol. 9, no. 3, pp.
62
Volume 15, Number 4, 2015
107-113, 2009. [Online]. Available: http://dx.doi.org/10.4316/aece.
2009.03019
[24] Y.C. Luo, W.X. Chen, “Sensorless stator field orientation controlled
induction motor drive with a fuzzy speed controller,” Computers and
Mathematics with Applications, vol. 64, pp. 1206-1216, 2012.
[25] I. Birou, V. Maier, S. Pavel, C. Rusu, “Indirect vector control of an
induction motor with fuzzy-logic based speed controller,” Advances
in Electrical and Computer Engineering, vol. 10, no. 1, pp. 116-120,
2010, [Online]. Available: http://dx.doi.org/10.4316/aece.2010.01021
[26] M. Karadeniz, C. Sezenoğlu, A. Balıkçı, “An active balancing system
for electric car batteries (in Turkish),” in Proc. of. ELECO 2012:
Symposium on Electrical and Electronics Engineering, Paper 184, pp.
450-454, 2012.
[27] A. Mohamed, V. Salehi, T. Ma, O.A. Mohammed, “Real-time energy
management algorithm for plug-in hybrid electric vehicle charging
parks involving sustainable energy,” IEEE Trans. on Sustainable
Energy, vol. 5, no. 2, pp. 577-586, 2014. [Online]. Available:
http://dx.doi.org/10.1109/tste.2013.2273400
[28] M. Ceylan, T. Sarıkurt, A. Balıkçı, “A novel Lithium-Ion-Polymer
battery model for hybrid/electric vehicles,” Proc. of. 23rd International
Symposium on Industrial Electronics (ISIE), pp. 366-369, 2014.
[Online]. Available: http://dx.doi.org/10.1109/isie.2014.6864640
[29] C. Xiao, X. Yu, D. Yang, D. Que, “Impact of solar irradiance inten-
sity and temperature on the performance of compensated crystalline
silicon solar cells,” Solar Energy Materials & Solar Cells, An Interna-
tional Journal Devoted to Photovoltaic, Photo thermal, and Photoche-
mical Solar Energy Conversion, vol. 128, pp. 427-434, 2014.
[30] S. Armstrong, W.G. Hurley, “A thermal model for photovoltaic
panels under varying atmospheric conditions,” Applied Thermal Engi-
neering, vol. 30, pp. 1488-1495, 2010.
[31] M.G. Villalva, J.R. Gazoli, E.R. Filho, “Comprehensive approach to
modeling and simulation of photovoltaic arrays,” IEEE Trans. on
Power Electronics, vol. 24, no. 5, 2009. [Online]. Available: http://
dx.doi.org/10.1109/tpel.2009.2013862
[32] J.K. Shiau, Y.C. Wei, M.Y. Lee, “Fuzzy controller for a voltage-regu-
lated solar powered MPPT system for hybrid power system applica-
tions,” Energies, vol. 8, no. 5, pp. 3292-3312, MDPI-Open Access
Publishing, 2015. [Online]. Available: http://dx.doi.org/10.3390/en
8053292
[33] A. Poursamad, M. Montazeri, “Design of genetic-fuzzy control stra-
tegy for parallel hybrid electric vehicles,” Control Engineering Prac-
tice, vol. 16, no. 7, pp. 861-873, 2008. [Online]. Available: http://
dx.doi.org/10.1016/j.conengprac.2007.10.003