See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/271190872

Modeling aspects of an electric starter system for an internal combustion

engine

Conference Paper · October 2014

DOI: 10.1109/ECAI.2014.7090177

CITATIONS READS

4 1,354

3 authors:

Bogdan-Adrian Enache Luminita Mirela Constantinescu

Polytechnic University of Bucharest University of Pitesti
36 PUBLICATIONS   75 CITATIONS    29 PUBLICATIONS   149 CITATIONS   

SEE PROFILE SEE PROFILE

Emilian Lefter
University of Pitesti
13 PUBLICATIONS   15 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

STUDY OF VARIATION OF THE MECHANICAL PROPERTIES OF SUPERALLOYS INCONEL 718 AND RENE 41 UNDER THERMAL SHOCK View project

Experimental validation of a propulsion system with hydrogen fuel cell for a light vehicle - Demonstrator of Mobility with Hydrogen View project

All content following this page was uploaded by Bogdan-Adrian Enache on 21 January 2015.

The user has requested enhancement of the downloaded file.

 ECAI 2014 - International Conference – 5th Edition
Electronics, Computers and Artificial Intelligence
23 October -25 October, 2014, Bucharest, ROMÂNIA

Modeling aspects of an electric starter system

for an internal combustion engine
Bogdan-Adrian Enache, IEEE Member, Emilian Lefter
Lumini a-Mirela Constantinescu, IEEE ELECTROMET Research Center
Member University of Pite ti
Department of Electronics, Computing and Pite ti, Arge
Electrical Engineering [email protected]
University of Pite ti
[email protected]

Abstract – This paper presents a mathematical model of several subsystems and study the behavior of each one
an electric starter system for an internal combustion and the interactions between them. So [2] and [3]
engine. For achieving this we started from the main present some models for the starting relay, [4] models
elements that characterize the cranking process: torque, the on the one-way clutch mechanism of the CD and
speed, starter motor terminal voltage and the battery we are focused on developing a model for the DC
terminals voltage depending on the current drawn from electric starter, the battery that powers it up and their
it. Based on these values we developed a general model interactions with the ICE during cranking.
that reproduces the starter motor behavior during
cranking. To achieve our goal a set of preliminary tests were
performed on a SS that equips a 250 cc ICE. The data
Keywords: starter system; DC motor modeling; battery from these tests allowed us to determine the torque,
modeling. speed and power necessary for cranking. After this we
studied individually each component in order to
I. INTRODUCTION extract the parameters necessary for modelling them
(section 2).
The electric starter system (SS) for an internal
combustion engine (ICE), invented in 1912, was a Section 3 presents the development of each of the
strong argument in favor of the rapid generalization of component for the SS model. In this section it is
the ICE, on vehicles, over the electric motor. presented the model for the ES, the model for the
battery and the model of the interaction between the
Here's how an environmental friendly mean, the
ES and ICE during cranking.
electric motor that equips a SS led, indirectly, to a
direction that today seems to be less happier.
Therefore the powertrain suffers, especially in the last II. EXPERIMENTAL MEASUREMENTS OF THE SS
period, several changes such as hybrid variants, less BEHAVIOR
polluting. For establishing the operating parameters of the SS
With the development of new types of hybrid cars we have conducted a series of tests with it coupled to a
equipped with start-stop systems (Smart Idle Stop 250 cc ICE. The system used is presented in Fig. 1.
System, i -stop), the use of the SS during car Speed
functioning increased considerably. However, the SS Transducer
DC Electric Starter
used is the same for the new hybrid cars as for the Current Transducer Battery 12V 14Ah

classic ones. It is necessary, therefore, to study the

behavior of SS in these conditions of intensive use.
Toothed
Finite element analysis is extensively used for Wheel

studying the behavior of electric machines, but is very

time consuming and researchers are more interested in ICE – 250 cc

analytical methods that could also be used for LEGEND

DAQ
optimization purposes [1]. Data link
Electric link Computer
The main idea of this paper is to develop an Mechanic link

analytical model that is suitable for a wide range of SS

and could be used for studying and optimization too. Figure 1. The system used for the experimental measurements

An electric SS consists generally of a DC electric The monitored values were: the voltage across the
starter motor (ES), a coupling device (CD) and the battery, the voltage across the ES, the current drawn
starting relay. from the battery, speed during startup and torque.
Due to the combinations of electrical and These experimental data consist the baseline for
mechanical components an entire SS is difficult to establishing the interaction between the ES and ICE
model, so it is custom to break down the system in during cranking and also for validating the proposed
model.
 2 Bogdan-Adrian Enache, Lumini a-Mirela Constantinescu, Emilian Lefter

We summarize the data obtained, in Fig 2, which

presents the current drawn by the ES in an ICE
complete cycle. Fully charged voltage

End of nominal zone

90

80

End of exponential zone

70
current[A]

60

50

40

30
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 Figure 3. Battery discharge characteristic at nominal current
time[s]

Figure 2. Current drawn by ES in an ICE complete cycle III. MODEL OL THE SS

The model of the SS includes a model of the
After these tests we broke the system apart and
permanent magnet DC motor, the mathematical model
studied the components individually.
of the battery which powers the motor and the load
For the DC electric starter we have conducted a that reproduces the development of torque and speed
series of bench test which consist of: tests with no load during cranking.
as generator, short-circuit tests as motor, tests with
load as motor. The values obtained are synthetized in A. Modeling the DC Motor
table I and II. The schematic diagram of a DC permanent magnet
connected to a DC voltage supply (Ua) is presented in
TABLE I. ES AS GENERATOR Fig. 4.
Nr. Generated Speed
Crt. voltage [V] [rad/s]
1 0.5935 62.796
2 0.75587 83.728
3 0.9367 104.66
4 1.1319 125.52

TABLE II. ES AS MOTOR

Nr. Current
Torque[Nm]
Crt. [A]
1 85 0.26
2 93 0.37
3 170 0.99
4 220 1.48

The tests carried on the battery started with

constant current discharges for establishing the
effective capacity. For this the test sequence from [12]
was performed. The sequence is composed by three
constant current discharge using currents I1 = C1/1, I2
= C2/2 and I3 = C3/3, where Ci is the battery capacity Figure 4. Schematic design of a DC permanent magnet motor
for i hours. The test sequence is repeated three times
and the effective capacity of the battery is considered The armature is represented by a back
determined if the difference between the three electromotive force (U0) in series with the rotor
discharges at I3 is less than 2%. In our case for the 14 winding resistance (Ra) and the winding self-
Ah battery used its effective capacity is 12.90 Ah. inductance (La).
After determining the effective capacity of the The time domain equations that model the
battery another discharge is been carried out at the behavior of the ES are [5,6]:
nominal current I = 0.7A which will allow us to
extract the battery model parameters – Fig 3.
dia (1)
U 0 + Ra ia + La = Ua
dt
 Modeling aspects of an electric starter system for an internal combustion engine 3

d ωm (2) where: U is the voltage across the battery, E0 is the

Tem − J − Bωm = TL
dt Open Circuit Voltage (OCV) of the battery, K is the
polarization coefficient, Q is the battery capacity, i is
where: the discharge current, A - the amplitude of the
exponential region of the discharge curve of the
battery and B is the coefficient corresponding to
TABLE III. SYMBOLS USED inverse of the time required maximum discharge
Symbol Meaning Expression Units capacity (Fig. 3).
ia Armature current - Amps In order to determine the coefficients A, B and K
Ra Armature resistance - Ohms
La Armature inductance - Henry
we used the discharge curve of the battery at nominal
τ Rotor time constant τ = La Ra current I = 0.7 A, from which we extract the
coordinates of three points: the point of maximum
Ua Supply voltage - Volts
voltage, the point from the end of the exponential area
U0 Back e.f.m. U 0 = kvωm Volts
and the point from the end of the nominal area (Fig.
ωm Angular velocity - Rad/s
J Moment of inertia - Kgm2
3). Based on this we calculated the coefficients as A =
B Damping coefficient - 0.87139 V, B = 0.2439 Ah-1, K = 0.0037549 V and E0
Tem Electromagnetic Tem = kt ia Nm = 11.7817 V.
torque
TL Load torque - Nm C. The interaction between the ICE and the SS
kv Velocity constant kv = ke ⋅ φe The interaction between the ICE and the SS is
produced mostly by the cranking torque. To model this
kt Torque constant kt = ke ⋅ φe we considered the current drawn by the ES during
cranking – Fig. 1 and the ES torque characteristic –
ke Motor constant - table II and established the ES load during cranking –
Φe Electromagnetic flux - Wb Fig. 5.

After we rearranged 1 and 2 we used the Laplace 0.35

transformation [7] and considered the initial conditions 0.3

[8] we obtained the transfer functions for the DC
motor [9]: 0.25
Torque [Nm]

0.2

−kvΩ m + U a (3)
Ia = 0.15
La s + Ra
0.1

0.05

− kt I a (4)
Ωm = 0
Js + B 0 0.05 0.1 0.15
time[s]
0.2 0.25 0.3 0.35

Considering the transfer equations all we have to Figure 5. ES load characteristic during cranking
do for establishing the DC motor model is to calculate
the values of the parameters from 3 and 4. The To integrate the load characteristic into the SS
armature resistance was measured directly and was model we used curve fitting techniques and we
found to be Ra = 0.027 Ω. For calculating the armature approximated it with an order 5 polynomial given by
inductance we measured the rotor time constant with (6):
the rotor blocked which was τ = 1.56 ms and so we
determined La = 0.000042 H. From tables I and II we
calculated the velocity constant kv = 0.00911 and T = −7.5541 ⋅ ( 5 ⋅ 102 ) ⋅ t 5 + 5.6872 ⋅ ( 5 ⋅ 102 ) ⋅ t 4 −
torque constant kt = 0.009037. The mechanical
(6)
parameters of the system were considered J = −1.4269 ⋅ ( 5 ⋅ 102 ) ⋅ t 3 + 0.1284 ⋅ ( 5 ⋅ 102 ) ⋅ t 2 −
0.000001 kgm2, B = 0.00045.
−0.0038 ⋅ ( 5 ⋅ 102 ) ⋅ t + 0.0005 ⋅ ( 5 ⋅ 102 )
B. Modeling the lead-acid battery
The model for the lead-acid battery is based on a IV. RESULTS AND DISCUSSION
modified Shepherd model [10].
With all the model’s parameters determined we
This model describes the variations of the battery used the Simulink/Matlab environment to create the
voltage depending on the discharge current and has the model for the entire SS. The model includes Simulink
following expression [11]: blocks for the DC machine and battery which hold the
values of the determined parameters and a custom
function block that models 6. The entire system model
Q (5) is presented in Fig. 6.
U a = E0 − K + A ⋅ exp ( − B ⋅ it )
Q −i⋅t
 4 Bogdan-Adrian Enache, Lumini a-Mirela Constantinescu, Emilian Lefter

about 3.87%, but due to all the approximation we did

the model has a maximum error of 10.88%.

V. CONCLUSIONS
In this paper we presented the realization of a
general model for an electric starter system. The great
advantage of this model is that it uses only four
equations (3, 4, 5, 6) to describe the behavior of the
whole system. It can be easily customized, by
replacing the coefficients of the starter motor and the
Figure 6. ES model in Simulink battery, making it very useful in studying and
optimization of any electric starter system.
The output values of the model are: battery
voltage, armature current, speed and electric torque of ACKNOWLEDGMENT
the DC machine. These values are presented in Fig. 7.
The authors gratefully acknowledge the support
and facilities provided by the ELECTROMET
Research Center from University of Pite ti, Arge ,
România. Special thanks to Mr. Jumară Ion for his
valuable assistance in this work.

REFERENCES

[1] Hooshang Mirahki, Mehdi Moallem, Sayyed Abbas Rahimi,

Design Optimization of IPMSM for 42 V Integrated Starter
Alternator Using Lumped Parameter Model and Genetic
Algorithms, IEEE TRANSACTIONS ON MAGNETICS,
VOL. 50, NO. 3, MARCH 2014;
[2] Jawad Faiz, K. Moayed-Zadeh, Design of switched reluctance
machine for starter/generator of hybrid electric vehicle,
Electric Power Systems Research, 75 (2005), 153–160.
[3] Lajos Nagy, Tamás Szabó, Endre Jakab, Electro-dynamical
modeling of a solenoid switch of starter motors, Procedia
Engineering, 48 (2012), 445 – 452.
[4] Qi Ma, Sai SV Rajagopalan, Stephen Yurkovich, Yann G
Figure 7. ES model output Guezennec, A High Fidelity Starter Model for Engine Start
Simulations, American Control Conference June 8-10, 2005,
The validation of the model is made by direct Portland, OR, USA.
comparison of values obtained from simulation with [5] Clay S. Hearn, Damon A. Weeks, Richard C. Thompson,
the actual values used to model the system. For this we Dongmei Chen, Electric Vehicle Modeling Utilizing DC
Motor Equations, IEEE/ASME International Conference on
compared the current drawn by the ES in an ICE Advanced Intelligent Mechatronics Montréal, Canada, July 6-
complete cycle (Fig. 2) with the armature current from 9, 2010.
the simulation (Fig. 7). The result of the comparison is [6] Maamar Taleb, Matlab/Simulink Models For Typical Soft
presented in Fig. 8. Starting Means For A DC Motor, International Journal of
Electrical & Computer Sciences, IJECS-IJENS, Vol: 11, No:
02, 2011, p 25-30.
[7] Ali Bekir Yildiz, Electrical equivalent circuit based modeling
and analysis of direct current motors, Electrical Power and
Energy Systems, 43 (2012), 1043–1047.
[8] Abdulrahman A.A.Emhemed, Rosbi Bin Mamat, Modelling
and Simulation for Industrial DC Motor Using Intelligent
Control, Procedia Engineering, 41 ( 2012 ), 420 – 425.
[9] J. S. Valdez Martínez, P. Guevara López, J. J. Medel Juárez,
Series Wound DC Motor Modeling and Simulation,
Considering Magnetic, Mechanical and Electric Power
Losses, In proceeding of: Circuits and Systems, 2009,
MWSCAS '09, 52nd IEEE International Midwest
Symposium.
[10] Olivier Tremblay, Louis-A. Dessaint, Abdel-Illah Dekkiche,
A Generic Battery Model for the Dynamic Simulation of
Hybrid Electric Vehicles, In proceeding of: Vehicle Power
and Propulsion Conference, 2007, VPPC 2007.
[11] Olivier Tremblay, Louis-A. Dessaint, Experimental
Figure 8. Comparison between measured and modeled current
Validation of a Battery Dynamic Model for EV Applications,
World Electric Vehicle Journal, Vol. 3, 2009, p 1-10.
Analyzing the model values correlated with the [12] ***, USABC Electric Vehicle Test Procedure Manual,
measured ones we determined that the average error is Revision 2, 1996.

View publication stats