International Journal of Computer Applications (0975 – 8887)

Volume 99– No.12, August 2014

Performance Study based on Matlab Modeling for Hybrid

Electric Vehicles

Mihai-Ovidiu Nicolaica
PhD Student, Faculty of Electronics, Telecommunications & Information Technology,
"Gheorghe Asachi" Technical University, Iaşi, Romania

ABSTRACT became more advanced and the concept of simulation became

The present paper introduces an integrated method for a an actual design tool. Simulation is now the primary tool used
performance analysis of the hybrid electric vehicles. This before real word testing is possible [4]. The safety practices
method takes into consideration a set of parameters that and standards are becoming more regulated as industries
influence the system performance. The research is based on adopt a standardized set of practices for product design and
Matlab modeling and Hardware in the Loop approach. The test. Every manufacturer needs to provide evidence of
value added is given by the accuracy of the model that is part compliance with the latest regulations.
of the simulation together with the presented performance
analysis methods. A set of 42 case studies are proposed and
2. FUNCTIONAL SAFETY STANDARD
the efficiency of the method is tested through a series of ISO26262
simulation results. The motivation of the research is given by The complexity in the automotive industry is continuously
the fact that the automotive sector is in a continuous change, increasing together with the development of hybrid and
and in the last years the trend is towards electric vehicles. The electric vehicles. This also increases the effort in providing
advantage of this method is that the systems performance can safety-compliant systems. The Draft International Standard
be validated to a large extent from an early stage. The results (DIS) of ISO26262 was published in June 2009. Since the
can contribute to the practical need of developing hybrid and publication of the draft, ISO26262 has gained traction in the
electric vehicles at higher quality standards, faster and with a automotive industry. This standard provides recommendations
better cost. and regulations trough the entire development process with an
emphasis on providing requirements for validation and
Keywords confirmation measures in order to achieve an acceptable level
Hybrid Electric Vehicle, HIL, simulation, modeling of safety. Test is considered a critical component. It is
estimated that the cost of a failure decreases by 10 times when
1. INTRODUCTION the error is caught in production instead of in the field and
Research shows that the curves of conventional oil demand decreases 10 times again if it is caught in design instead of
and oil production are still in line until 2020 and by 2050 the production [5]. With a very good simulation and efficient
oil gap will dramatically increase till 50 Billion Barrels. This testing techniques the level of quality and safety can increase
issue of oil reduction in the world, and the air pollution and large efficiency gains can be generated. High cost
triggered many research to seek alternative ways to replace reduction would be another benefit. From the initial design
the combustion engine [1]. Hybrid and electric vehicles can be phase to the operational prototype a series of steps, such as
declared as the cars of the future, mainly because they Software in the Loop (SIL) followed by Hardware in the Loop
produce no pollution, are silent and have enhanced energy (HIL) validation of control algorithms and control hardware
efficiency. In the same time there are still some issues that should be performed. The SIL/HIL sequence has a high
still need to be solved, like limited driving range and long benefit in improving the safely level, also development cost
charging time [2]. These problems can be mitigated also by a and time, as the control systems can be validated to a large
very good parameterization and better control systems. This extent, before vehicle deployment [6].
can be achieved by developing a very good model of the
vehicle and testing the prototypes from the early concept 3. HARDWARE IN THE LOOP
phase. Research in the domain of Hybrid Electric Vehicles has Hardware in the Loop process, simulation and testing has
concentrated in the latest years in finding solutions for existed for more than 15 years. It was initially used in the
efficient motors and engines, reduced fuel emissions and Aviation Industry. Hardware in the Loop is a technique used
better battery lifetime. In the last years the share of software in the development and test of complex process systems.
controlled innovations in the automotive industry increased Hardware in the Loop simulation is the method used to test
from 20 percent to 80 percent. This is continuously growing functions, system integration, and communication of
and it is estimated that software will drive more than 90 Electronic Control Units (ECUs). ECUs are indispensable for
percent of the automotive systems in the next years. vehicles, airplanes and robots. The technical environment of
Considering this, the impact of software in electric vehicles the ECUs and interconnected system parts are simulated in
will be very high [3]. In order to have good achievements in HIL systems. The HIL tests are very systematic and also
the testing process, the test environment needs to be very completely safe, even when critical thresholds are exceeded.
realistic. In the last years the automotive industry is driven to The main goal is to detect errors in ECUs. Once detected, the
more complex electronic control systems. The number of situation that produces the error can be reproduced whenever
electronic control units (ECUs) is increasing and also the and however required [7]. It is a form of real-time simulation
complexity of communication networks between the ECUs. In with the addition of a real component in the loop. Usually the
the same time, mathematical modeling and simulation tools real component is the Electronic Control Unit (ECU). The role

15
 International Journal of Computer Applications (0975 – 8887)
Volume 99– No.12, August 2014

of the HIL system is to provide all electrical inputs to fully automatically manipulate them. Currently the most popular
exercise the ECU [8]. One of the key components of the HIL programming language used to develop Test Automation
system is the model that simulates the environment around the scripts in automotive and aerospace is Python. Automated
real ECU and drives the real time simulation hardware. The testing is usually performed by executing tests on a standard
block diagram of a classic HIL system is presented in Figure PC connected to the Hardware in the Loop (HIL) system.
1. Real time testing with Python scripts offers very high benefits.
They run on the processor board of the HIL system in real
time, synchronously with the model, so all test actions are
performed on a real-time basis – 100% reproducibly [12]. HIL
simulation begins with a system-level model that includes
embedded system algorithm and its operating environment.
This paper presents a case study using a Matlab model for a
hybrid electric vehicle. The model which can be easily
incorporated into a HIL system is used to analyze the
performance of hybrid vehicle architecture. A simplified
block diagram of one hybrid vehicle architecture is presented
in Figure 2.

Fig 1: HIL System Architecture.

A big advantage of Hardware in the Loop testing is the fact Fig 2: Hybrid Vehicle Architecture.
that extreme testing can be performed without the risk of
damaging any equipment. It is also recommended to test in a The Electronic Control Unit communicates with all main
simulated environment before the controller is applied on a components: Battery, DC/DC Converter, Electric
real target. HIL systems increase reliability and quality of the Motor/Generator, and Internal Combustion Engine (ICE). The
final product and bring innovation with a lower cost. Also the structure shown in Figure 2 can be modeled and the HEV
challenges of the testing process are mitigated by using HIL system architecture can be translated into Hardware in the
systems, like cost to test, cost of failure repeatability or Loop environment like the one presented in Figure 3.
availability [8]. HIL has become in the last years an
indispensable approach for prototyping and has allowed the
integration of hardware and software components to be
prepared earlier in the development cycle [9]. HIL initially
started only with ECU testing, but in the last years started to
incorporate additional devices like actuators or hydraulic
systems. This represents the transition from software in the
loop or model in the loop to system testing and integration.
This evolution improved the development cycle with 75
percent from time perspective. Components from different
industries are involved to create what is a mock-up of real life
equipment [10]. As HIL simulation can be done far earlier
than vehicle tests, it helps to improve quality at an early
development stage. A major Japanese automobile
manufacturer states that HIL simulation finds 90% of ECU
errors, and almost all the errors can be found before the Fig 3: HIL Test Environment
calibration phase [7]. One of the most important components The HIL simulation environment presented in Figure 3 can be
of Hardware in the Loop systems is the simulation model. The considered the intermediate step between the model in the
performance of the system is given by the accuracy of the loop analysis and actual HEV prototype. After the analysis is
model. Usually the real time model is built in performed in Matlab, the model can be compiled and loaded
Matlab/Simulink compiled and loaded on a simulator. A on a simulator. In this way the loop in the HIL system is
dSpace simulator can be used. A Simulator Mid-Size from closed. The Matlab model is suitable for a dSpace mid-size
dSpace is a ready-to-use, standard Hardware in the Loop simulator. The model inputs and outputs can be connected
simulator for testing automotive ECUs systematically and through a system interface to a real Electronic Control Unit.
automatically. The simulation model runs on the processor The ECU will have the expected inputs and outputs in a
hardware (single-processor or multiprocessor systems). The closed loop environment. In this way, the control software on
Simulator Mid-Size generates and measures I/O signals via the real unit can be tested, and the performance can be
the integrated dSpace I/O boards. The function range is assessed prior to the deployment on the real vehicle.
complemented by load and failure simulation [11]. User
interface software is able to control the simulator that runs the
simulation model. This system allows direct access to all
inputs and outputs of an EEC under test. The inputs and
outputs of an EEC in this test environment can be seen as an
item if an object oriented scripting language is used to

16
 International Journal of Computer Applications (0975 – 8887)
Volume 99– No.12, August 2014

4. CASE STUDY (MATLAB MODEL 45

Driving Cycle

FOR HIL) – ANALYSIS 40

Vehicle Speed
The block diagram of the model under analysis is presented in ICE Run
Figure 4 [13]. 35

30

Vehicle Speed(KM/h)
25

20

15

10

-5
0 20 40 60 80 100
Time (s)

Fig 5: Vehicle speed

The battery power behavior during the driving cycle for
different vehicle mass is presented in Figure 6. During motor
mode the power consumption increases, and a spike is visible
Fig 4: Model Block Diagram.
when the ICE starts. The power consumption increases until
The model has several parameters that can be configured, but the speed stabilizes, and while running on combustion engine
the analysis is concentrated on vehicle mass, battery voltage, the battery charges. The charging of the battery is represented
and engine power. The system control is also simulated for a by a negative value on the graph in Figure 6.
closed loop behavior. Several combinations of values for the Battery power vs Different vehicle mass
previous parameters are analyzed to generate a performance 50
matrix. Several simulations are executed for vehicle mass 600 Kg
700 Kg
from 600Kg to 1200Kg, for battery voltage from 200V to 40 800 Kg
300V, and for internal combustion engine from 43kW to 900 Kg
53kW. Different driving cycles are defined. A driving cycle is 1000 Kg
Battery Power (kW)

30
proposed in Table 1. 1100 Kg
1200 Kg

Table 1. First Driving Cycle 20

Time Accelerator Time Accelerator

10
0 0 56 0.03
7 0.2 60 0.03
0
15 0.2 65 0.03
25 0.2 70 0.15
30 0.05 75 0.15 -10
0 20 40 60 80 100
35 0.05 80 0.15 Time (s)
40 0.05 81 0.1
43 0.15 82 0.05 Fig 6: Battery power
50 0.15 85 0 During breaking the battery power goes in the negative area
55 0.15 90 0 (regenerative braking) and again a spike is visible when the
electric engine switches from generator mode to motor mode.
In motor mode the power consumption increases. This
The Accelerator in Table 1 is converted to engine speed
behavior is visible through the entire driving cycle at different
demand. A pedal deflection of 0.1 corresponds to 20km/hour,
scales depending on the speed value and the breaking
results in the speed demand of 800rpm, thereby starting the
gradient. It can be observed that the power consumption
engine. The simulated driving cycle has the duration of 90
depends on the vehicle mass while the generator mode
seconds with some accelerations and decelerations in order to
performance remains constant for different vehicle mass. This
see the behavior of the motor, the generator and the impact on
was regulated by the control unit. The results shown in Figure
the battery. The driving cycle has multiple acceleration and
6 are obtained with a battery rated voltage of 200V. The
braking sequences in a short timeframe. Also the acceleration
internal combustion engine has a maximum power of 43kW
and deceleration gradients are different in order to have a
with a maximum speed of 6000rpm. These simulation
wider simulation range. The speed corresponding to the
environment values are changed further in the investigation
driving cycle together with the activation or deactivation of
for a better overview. In the performance analysis is used an
the internal combustion engine (ICE) is presented in Figure 5.
ideal battery model for which the output voltage is given by
It is visible that the ICE start or stop is dependent on the speed
value. V  V0  I * R , where V0 is the rated voltage, I is the
output current, and R is the internal resistance. The charge
level is tracked, and a simulation error issued if it either
exceeds the ampere-hour capacity, or falls below 10% of the
ampere-hour capacity. The internal combustion engine (ICE)
that is part of the analyzed hybrid architecture is modeled as a

17
 International Journal of Computer Applications (0975 – 8887)
Volume 99– No.12, August 2014

simple gasoline fuel engine model with speed governor. The Representative parameters are the DC bus voltage, the DC bus
throttle input signal lies between zero and one and specifies current and the power of the engine, motor and generator.
the torque demanded from the engine as a fraction of the These graphs are affected by different vehicle mass values. In
maximum possible torque. This signal also indirectly controls order to extend the analysis 42 test scenarios are developed
the engine speed. If the throttle signal rises above one or falls where not only the vehicle mass is varied but also the battery
below zero, the engine limits the signal to one or zero voltage (200V to 300V) and ICE power (43kW to 53kW). All
respectively. If the engine speed rises above the maximum results vectors for generator power and motor power are
speed, the engine torque drops to zero [14]. A better overview plotted on the same figure to see the distribution over the
on the Battery power distribution during acceleration is entire simulation range. The simulation result is visible in
represented in Figure 7. Figure 9.
Battery power vs Different vehicle mass
Generator Power distribution

Generator Power (kW)

10

40
600 Kg
700 Kg
0
Battery Power (kW)

30 800 Kg
900 Kg
1000 Kg -10
20 0 20 40 60 80 100
1100 Kg Time (s)
1200 Kg
Motor Power distribution
10 Speed 50

Motor Power (kW)

Eng Run

0
0

0 2 4 6 8 10 12
Time (s) -50
0 20 40 60 80 100
Fig 7: Battery Power during acceleration Time (s)

The Engine Run signal represents the internal combustion

Fig 9: Generator and Motor Power distribution
engine which starts and powers the car from the moment
Engine Run becomes true. The efficiency is directly The graph shape is similar over all test cases with differences
depending on vehicle mass as demonstrated in Figure 7. For in signal values. The positive and negative values of the
the same driving cycle and the same model parameterization a battery power are considered for an efficiency analysis. For all
set of important parameters are also displayed in the plots in test cases the percentage of battery charge time and the
Figure 8. percentage of discharge time are calculated over the same
Engine Power vs Different vehicle mass
driving cycle that was previously presented in Table 1 above.
12 600 Kg
The charge time percentage varies between 44.1753% and
700 Kg 49.2528% over all test scenarios. The discharge time
10 800 Kg percentage varies between 50.7472% and 55.8247% over all
900 Kg
test scenarios. A graphical representation is presented in
Engine Power (kW)

1000 Kg
8
1100 Kg Figure 10.
1200 Kg
6

4 50
Charge mode
Percentage

2 48

0
46
-2
0 20 40 60 80 100
44
Time (s) 0 10 20 30 40 50
Generator Power vs Different vehicle mass
10 56
600 Kg
700 Kg
Percentage

800 Kg
54
Generator Power (kW)

900 Kg
5
1000 Kg
1100 Kg
1200 Kg 52
0 Discharge mode
50
0 10 20 30 40 50
-5 Test Case
Fig 10: Charging percentage distribution – DC 1

-10 The best efficiency over this driving cycle is obtained with a
0 20 40 60 80 100 vehicle mass of 600Kg, a battery voltage of 200V and an
Time (s) internal combustion engine of 53KW. For a better
understanding and a comparative analysis against the first set
Fig 8: Engine and Generator Power
of results, another driving cycle is considered as per Table 2.

18
 International Journal of Computer Applications (0975 – 8887)
Volume 99– No.12, August 2014

Table 2. Second Driving Cycle understanding. The model used can be incorporated in
Hardware in the Loop test system for additional
Time Accelerator Time Accelerator measurements. The parameterization of the model can be
0 0 56 0.2 extended and further analysis can be done in order to improve
7 0.2 60 0.2 the model performance. An objective for further research is to
15 0.2 65 0.2 extend and perfect additional simulation models for electric
25 0.2 70 0.2 vehicles that can be used for model in the loop simulations
30 0.2 75 0.2 and that can be easily incorporated in HIL systems.
35 0.2 80 0.2
40 0.2 81 0.2 6. REFERENCES
43 0.2 82 0.2 [1] Bambang, S. K., Soebagio, Mauridhi, H. P., Design and
50 0.2 85 0 Development of Small Electric Vehicle using
55 0.2 90 0 MATLAB/Simulink, International Journal of Computer
Applications (0975 – 8887) Volume 24 – No.6, June
2011, pp.19-23
The new driving cycle has the same time duration but only an
acceleration of 7 seconds and a deceleration of 5 seconds [2] Schaltz, E., Electrical Vehicle Design and Modeling,
comparing to the first driving cycle where several acceleration Aalborg University Denmark
and decelerations were performed. The drive speed is
constant. The electric motor is for a longer time in generator [3] Bringmann, E., Krämer A., Model-based Testing of
mode and the vehicle is powered by the ICE most of the time. Automotive Systems, 2008, International Conference on
The charge time percentage varies between 87.4340% and Software Testing, Verification, and Validation, pp. 485
88.8062% over all test scenarios. The discharge time to 493
percentage varies between 11.1938% and 12.5660% over all [4] King, J. C., Model-Based Design of a Plug-In Hybrid
test scenarios. A graphical representation is presented in Electric Vehicle Control Strategy, 2012, MS Thesis in
Figure 11. Mechanical Engineering, Virginia Polytechnic Institute
and State University, Blacksburg, VA
89
[5] What is the ISO 26262 Functional Safety Standard, 2012,
www.ni.com
Percentage

88 [6] Wilhelm, E., Fowler, M.W., Fraser, R.A., Stevens, M.B.,

Hardware in the Loop Platform Development for Hybrid
Charge mode Vehicle
87
0 10 20 30 40 50 [7] ECU Testing with dSPACE HIL Systems
http://www.dspace.com/en/pub/home/products/systems/e
13 cutest.cfm
Discharge mode
Percentage

[8] Halvorsen, H. P., Hardware-in-the-Loop Simulation,

12 Telemark University College Faculty of Technology
[9] Vasiliu C., Vasile N., Innovative HIL Architecture for
11
Electric Powertrain Testing, U.P.B. Sci. Bull., Series C,
0 10 20 30 40 50 Vol. 74, Iss. 2, 2012, pp. 269-278
Test Case
Fig 11: Charging percentage distribution – DC 2 [10] Vasiliu C., Vasile N., Hardware-In-The-Loop Simulation
for Electric Powertrains, Rev. Roum. Sci. Techn.
The analysis shows that the best performance over the second Électrotechn. et Énerg., 57, 2, p. 212–221, Bucarest,
driving cycle is obtained with a vehicle mass of 600Kg, a 2012
battery voltage of 200V and an internal combustion engine of
53kW. Very similar energy efficiency is obtained in case of [11] dSPACE Simulator
vehicles with a higher mass over the second driving cycle. http://www.dspace.com/en/pub/home/products/systems/e
This underlines the fact that the vehicle mass has a higher cutest/hil_simulation_dspace_sim.cfm
impact in driving cycles where many acceleration and [12] Real-Time Testing
breakings are performed. https://www.dspace.com/en/inc/home/products/sw/test_a
utomation_software/automated_test_execution_in_re.cf
5. CONCLUSION m
Hardware in the Loop and Model in the Loop analysis are
very powerful methods for system design and validation. This [13] Real-Time Simulation of a Hybrid–Electrical Vehicle
paper offered an overview about these techniques applied in http://www.mathworks.de/matlabcentral/fileexchange/26
hybrid and electric vehicles. A model based analysis of hybrid 351
electric vehicle architecture was performed. A set of test cases
[14] Help Matlab Simulink
were executed using this model that helped the system

IJCATM : www.ijcaonline.org
19