World Electric Vehicle Journal Vol.

4 - ISSN 2032-6653 - © 2010 WEVA Page000119

EVS25
Shenzhen, China, Nov. 5–9, 2010

Full-Scale Hardware-in-the-Loop Verification Environment

for Heavy-Duty Hybrid Electric Vehicles
Ari Hentunen, Jussi Suomela, Antti Leivo, Matti Liukkonen, Panu Sainio
Aalto University, P.O. Box 14300, 00076 AALTO, Finland
[email protected]

Abstract
This paper presents an implementation of a versatile full-scale hardware-in-the-loop (HIL) verification
environment for testing of heavy-duty hybrid electric vehicles and mobile work machines. The test
facility includes a full-scale hybrid system that can be loaded with either an electric motor dynamometer
or a programmable chassis dynamometer. Model-based software development tools and rapid control
prototyping hardware are used to implement control algorithms and other vehicle controller software as
well as to control other test equipment.
Hardware-in-the-loop simulation, rapid control prototyping, hybrid electric vehicle, non-road mobile machinery,
mobile work machine

1 Introduction sion limits are tightening tremendously in near

future for engines bigger than 37 kW. For exam-
The legislation and regulations for the exhaust ple, the introduction of Stage IIIB will cut the PT
emissions of the diesel engines used in mobile emission limit of engines in the power range of
work machines are tightening remarkably dur- 37–75 kW by 94 %, and the introduction of Stage
ing this decade. The European Union directives IV will cut the NOx limit of 56–130 kW engines
97/86/EC and 2004/26/EC set tight limits for the by 88 % and 130–560 kW engines by 94 %, re-
emissions of carbon monoxide (CO), hydrocar- spectively.
bons (HC), oxides of nitrogen (NOx), and partic-
ulates (PT) [1, 2]. Because of the tightening reg-
ulations, more aftertreatment of exhaust gas will
be needed, which in turn slows down the engine
dynamics, degrades performance, and increases
cost. Also fuel price is rising, which increases
operation costs of the machinery. In addition, the
EU has committed itself to expanding the use of
renewable energies by 20 %, achieving a 20 % in-
crease in energy efficiency and a 20 % reduction
in greenhouse gas emissions by 2020. To tackle
all these challenges, the original equipment man-
ufacturers (OEMs) of non-road mobile work ma-
chines have shown rising interest to hybridize the
drive train. Figure 1: European Union regulations for emission
The EU limits for emissions of variable-speed standards for diesel engines other than constant-speed
diesel engines to be installed in non-road mo- engines to be installed in non-road mobile machinery.
bile machinery according to [2] are illustrated in
Fig. 1. The type-approval dates for the engines
are one year earlier. The emission limits of the The internal combustion engine (ICE) of a con-
US EPA Tier regulations go in line with the EU ventional work machine is dimensioned to sup-
limits. As can be seen from the figure, the emis- ply the peak power during the anticipated work

EVS25 World Battery, Hybrid and Fuel Cell Electric Vehicle Symposium 1
World Electric Vehicle Journal Vol. 4 - ISSN 2032-6653 - © 2010 WEVA Page000120

cycle. Depending on the work machine type and sign and optimization, but the control strategies
its typical work cycle, the peak power demand become easily quite complex to implement. It
might occur e.g. during acceleration, driving up- must also be kept in mind that many OEMs do
hill with a full load, or driving against the pile not have strong experience with electric traction
when loading the bucket. The average power de- drives and other power electronic (PE) and high
mand of the work cycle is often close to a half power electrical equipment.
of the peak-power demand. In a hybrid elec- Testing of the control software is not a straight-
tric driveline, the ICE can be downsized drasti- forward task. The electric motors (EMs) rotate
cally due to the onboard energy storage that can with speeds up to 15 000 rpm with only a mil-
be used during peak-power demand to provide lisecond order of response times for full torque.
extra power and to store regenerative braking There may also occur failures e.g. in contactor
power. With big enough energy storage, e.g. a control that may result in very high inrush cur-
high power battery, the ICE can be dimensioned rents and possibly arcing and hardware failures.
to provide only the average power. Thus, with Therefore, it is a good practice to first simulate
hybridizing the power train, it is often possible the control algorithms against a plant model, and
to downsize the engine remarkably, and thus, to then validate and verify the control software in a
achieve better fuel economy as well as less emis- laboratory environment. [4, 5]
sions. [3] In addition, the emission limits of the Model-based software development (MBSD) and
downsized engine may be less tight compared to rapid control prototyping (RCP) hardware and
the original limits, because the downsized engine software are widely used in automotive indus-
may fit into a smaller power class. As can be seen try [6–14]. With the increasing amount of soft-
from Fig. 1, downsizing can be especially benefi- ware in the mobile work machines, the MBSD
cial if the engine power can be downsized below and RCP are getting more attention also in mo-
37 kW, since the Stages IIIB and IV do not con- bile work machine industry. With these tools it
cern engines that have net power below 37 kW. is straightforward to make early designs, validate
the controller against plant models, and imple-
ment the algorithms in a real RCP electronic con-
troller unit (ECU) by utilizing automatic code
generation (ACG). Furthermore, the controller
model can later be customized for the real pro-
G M
duction ECU, which shortens the gap between
the prototype and the final product.
The paper is organized as follows. The HIL
verification environment is described in detail in
section 2. Experimental results of HIL simula-
tion and efficiency measurements are presented
Figure 2: Layout of SHEV. in section 3. Section 4 concludes the paper.

Hybrid vehicles have at least two energy sources. 2 HIL Verification Environment
The primary power source is usually an ICE or a
fuel cell (FC) unit, while a battery or an electric
double-layer capacitor (EDLC) unit—also com- 2.1 Objectives
monly known as supercapacitor (SC) or ultra-
capacitor (UC) unit—acts as a secondary power The main objective is to make HIL simula-
source. A layout of a generic series hybrid elec- tions with full-scale power hardware for different
tric vehicle (SHEV) driveline is shown in Fig. 2. SHEV topologies. With full-scale power hard-
In SHEVs, all traction power is transferred in ware, it is also possible to get insight into all as-
electrical form. The ICE is only a source of elec- pects of the electrification of mobile work ma-
trical power. Because the vehicle speed is totally chines, including e.g.
decoupled from the engine speed, the ICE can be
controlled to trace the optimal efficiency or low- • control algorithm development and diag-
est emissions, or to optimize some other criteria. nostics
To avoid resistive losses, fairly high DC link volt-
age is usually chosen. If the battery is coupled • real interface electronics and time delays
directly to the DC link, then the battery voltage in command values, and communication
determines also the DC link voltage. However, through e.g. controller area network (CAN)
the DC–DC converter is often used to decouple
the ES from the DC link. Then it is possible to
control directly the ES current. Therefore, the • cabling, electromagnetic interference
use of a DC–DC converter also allows more so- (EMI), and safety issues with devices up to
phisticated power management strategies. 1 000 V
The advent of electric transmission poses also
new challenges for OEMs of mobile work ma- • noise vibration and harshness
chines. The OEMs need to develop new trac-
tion control algorithms as well as power man- • physical size and weight
agement strategies. Electrification offers more
performance and degrees of freedom in the de- • thermal behavior and management

EVS25 World Battery, Hybrid and Fuel Cell Electric Vehicle Symposium 2
World Electric Vehicle Journal Vol. 4 - ISSN 2032-6653 - © 2010 WEVA Page000121

The second objective is to validate the plant mod- and the left-hand side constitutes an EM dy-
els of the simulation platform that have been namometer for the traction motor. The frequency
made in recent years to simulate hybridization converter incorporates a brake chopper, which is
configurations for different kinds of mobile work connected to a brake resistor to allow regenera-
machines. This objective is actually a very im- tive braking when the switch S3 is open.
portant one, because good consistency between The power ratings of the hardware are shown in
the model and the real plant yields to accurate Table 1. The power rating of the UC unit and
and reliable simulation results, and on the other its dc–dc converter are for the nominal voltage,
hand, poor consistency gives unreliable informa- which is also very close the maximum voltage,
tion that is more or less useless. and thus, can be regarded as the power when the
In order to achieve good accuracy for the fuel UCs are fully charged. When the UCs are be-
consumption and to be able to develop control ing discharged, the available power decreases lin-
algorithms that work also in real vehicle—not early with the voltage. Thus, at the half of the
just in a simulator—detailed information about nominal voltage the continuous power is half of
the efficiency of the devices in the hybrid drive the power that is shown in Table 1. For the bat-
train is needed. Unfortunately, such information tery, the nominal and maximum power rating that
is rarely available from the subsystem suppliers. are shown in the table are for discharge opera-
It is common that the manufacturers give the ef- tion. Charging power must be remarkably lower.
ficiency only in the nominal point. Therefore,
one important objective is the ability to measure
the efficiency maps of EMs, hydraulic pumps and Table 1: Continuous power P and maximum power
motors, power electronic converters (PECs), etc. Pmax of the subsystems
That information can later be used in the plant
models as well as in the control algorithms. P Pmax
System Hardware
[kW] [kW]
2.2 Overview ICE 92 120
A full-scale power HIL verification environment Generator 120 130
has been implemented in the university facilities Gen. inverter 160 172
to research and demonstrate the electrification of Motor 67 150
mobile work machines. The layout of the power HEV Motor inverter 120 200
HIL hardware is shown in Fig. 3. The compo- Battery 72 144
nents are mostly off-the-shelf products that are Battery dc–dc 43 72
available on the market. The nominal dc link
voltage is 650 V, which is a common voltage rat- Ultracap. 59 234
ing for HEV buses and work machines. Ultracap. dc–dc 47 78
DC power AFE 310 340
Mains 400 V AC
Load motor 67 150
S1 S2 Dynamo Load inverter 120 200
Frequency Active Front-End Converter
Brake resistor 20 100
Converter

S3
DC Link 650 V
2.3 HEV System
As can be seen from Fig.3, the hybrid system re-
sembles the SHEV topology that has dc–dc con-
verters between energy storages and the common
dc link. The HEV system consists of
G
M M • ICE–generator set (gen-set) with a genera-
tor inverter
Mechanical
Brake Dynamometer or Hydraulic Traction Battery Ultracaps
Resistor Link Motor
ICE
• ultracapacitor module with a bidirectional
dc–dc converter
Figure 3: Power hardware layout of the HIL verifica-
tion environment. • battery with a bidirectional dc–dc converter
• traction motor and inverter
There are two switches in the system (S1 and S2 ) The system includes also an active front-end
that connect the ac–dc converters to the mains, (AFE) that can be used to provide dc power if
and a third switch S3 that interconnects the dc the gen-set and the battery are not used in the test
link to the intermediate dc link of the frequency setup. AFE actually acts like a battery in the dc
converter. In this paper, switches S1 and S2 are link, because it regulates the dc link voltage to a
assumed to be closed and S3 to be open unless desired value, and thus, provides the difference
otherwise stated. The right-hand side from the between the generated power and the consumed
switch S3 is the HEV system under test (SUT), power.

EVS25 World Battery, Hybrid and Fuel Cell Electric Vehicle Symposium 3
World Electric Vehicle Journal Vol. 4 - ISSN 2032-6653 - © 2010 WEVA Page000122

2.3.1 Gen-Set Unit of the 400 V unit is also still small enough to be
installed in the vehicle.
The diesel engine is AGCO SISU POWER 49 An HTM power series 390v module with a type
DTAG [15]. The continuous power rating of the code BMOD0018 P390 was selected as the UC
SG 115 gen-set is 92 kW at 1 500 rpm and lim- unit. It is a complete package with all the electri-
ited power rating is 102 kW at 1 500 rpm. How- cal connections, voltage balancing, thermal mon-
ever, the original generator has been replaced itoring, cooling, and cooling fans assembled in-
with a permanent magnet (PM) axial-flux gener- side the enclosure. The capacitance is 17.8 F and
ator manufactured by AXCO-Motors [16]. The the nominal voltage is 390 V. The unit comprises
generator has eight poles and its apparent power 146 series connected 2.7 V cells with 150 A nom-
rating is 125 kVA and electric power rating is inal and 900 A maximum current ratings, respec-
120 kW at 2 000 rpm. The generator is coupled tively. HTM power series 390v is not in produc-
directly to the crankshaft of the engine and it is tion anymore.
liquid-cooled.
Together the ICE and the generator constitute a
variable-speed diesel-generator (VSDG) set that 2.3.4 Battery
has a nominal speed of 2 000 rpm, nominal ap-
parent power of 125 kVA, and nominal electric Lithium-polymer battery consists of 98 cell that
power of 120 kW, and nominal electrical angular are connected in series. Battery manufacturer is
frequency of 133 rad/s. The ac from the gener- Kokam, and the cell type is SLPB 100216216H,
ator is converter to regulated dc voltage with a which is a 40 Ah cell [19]. The nominal volt-
frequency converter that is operating in dc link age is 3.7 V, minimum voltage 2.7 V, and max-
voltage control mode. The input ac is not con- imum voltage 4.2 V, respectively. Maximum
nected, but the dc link of the frequency converter charge current is 80 A, nominal discharge cur-
is connected to the common dc link. rent 200 A, and maximum discharge current for
The manufacturer of the frequency converter 10 s is 400 A, respectively.
is ABB and the type code of the converter is Battery cells are pre-assembled in modules that
ACSM1-04LS-260A-4+K457 [17]. The con- have seven cells connected in series and include
verter is liquid-cooled, and it is controlled via voltage measurements for each cell and four tem-
CAN bus. The continuous power rating is perature measurements, which are located be-
160 kW. The converter incorporates also a brake tween every other two adjacent cells. A variable-
chopper, and a brake resistor will be installed in speed cooling fan has been added in front panel
it later. of each module, and vent holes have been made
in sides of the modules. The battery comprises 14
modules that are connected in series. The nom-
2.3.2 Electric Traction Drive inal voltage of a module is 25.9 V, and thus, the
nominal voltage of the battery is 362 V. One extra
The electric motor and the inverter are from battery module is also available in the laboratory
Siemens electric drive system ELFA [18]. The to be tested separately.
motor type is 1PV5135-4WS28 and the inverter The battery management system (BMS) is man-
type is G650 D440/170/170 M7-1. The motor ufactured by Elithion [20]. It consists of a BMS
is an induction motor with four poles and maxi- controller and 14 cell boards, one for each bat-
mum speed of 10 000 rpm. The nominal torque tery module. Cell boards are located inside bat-
is achieved with 124 A current and is 160 Nm tery modules. The extra battery module has its
at 4 000 rpm. The maximum torque with 300 A own BMS.
current is 430 Nm at 2 500 rpm. With a little
lower maximum current of 250 A that yields to
360 Nm torque the constant torque region can be 2.3.5 DC–DC converter
extended to 4 000 rpm, and thus the maximum
power is 150 kW. The bidirectional three-phase boost converter
The inverter actually includes two three-phase in- topology is shown in Fig. 4. The converter
sulated gate bipolar transistor (IGBT) inverters is manufactured by MSc Electronics [21] and
and two brake choppers. However, only one in- its model is 200DCDC750A2SN. The converter
verter is in use at the moment. The nominal dc utilizes multi-phase interleaving, which reduces
link voltage is 650 V, output voltage 440 V ac, current ripple. The nominal ES side current of
and output current 170 A, respectively. The max- the converter is 120 A and the maximum current
imum output current is 300 A. Both the motor is 200 A.
and the inverter are liquid cooled. The converter is controlled with a current ref-
erence. Also hardwired tunable minimum and
maximum voltage limits for the input and the
2.3.3 Ultracapacitor output are incorporated into the control system
to prevent over- and undervoltages in the system.
The nominal voltage of the ultracapacitor unit The voltage of the energy storage (ES) needs to
was considered to be around 400 V. That is firstly be lower than the dc link voltage. If the dc link
because the UC voltage must be lower than the dc voltage decreases and goes below the ES voltage,
link voltage. 400 V unit can also store enough en- the diodes in the power stage begin to conduct
ergy so that it can be used as a peak-power buffer current from the ES to the dc link, and thus, the
for various kinds of work machines. The volume voltages stay equal until the dc link voltage starts

EVS25 World Battery, Hybrid and Fuel Cell Electric Vehicle Symposium 4
World Electric Vehicle Journal Vol. 4 - ISSN 2032-6653 - © 2010 WEVA Page000123

+ then AFE feeds the generated power back to the

mains.

udcl 2.6 Mechanics

+
The mechanical environment is built on high in-
ues sulation rubber floor using T-slot beams with
weight of 500 kg each. Also all fastening com-
ponents have been overdimensioned by purpose
to have a very rigid mounting and to avoid the
Figure 4: Bidirectional three-phase boost converter. need for unnecessary flexible couplings. The
couplings used are jaw-type couplings, thus al-
lowing axial plug-in, good dynamics, vibration
reduction, and failsafe behavior. In case of over-
to rise. The minimum current of the converter is load, they do allow some torsional flexibility.
20 A.
2.7 Measurement Equipment
2.4 DC power supply
The mechanical torque and speed of the trac-
Vacon Active Front-End (AFE) with tion motor are measured with a KTR Dataflex
NXA 0460 5 input module [22] is used as a 42/1000 transducer. The transducer is installed
dc power supply. The nominal ac current is in the mechanical link in the coupling between
460 A and the continuous dc output power rating the motors.
is 310 kW. AFE includes a three-phase IGBT Norma D6100 power analyzer has been used to
bridge, an LCL line filter, and a pre-charge measure electrical power and efficiency of the
circuit. AFE regulates the dc link voltage the de- EMs, PECs, and other EUT. Measurement data
sired setpoint. The recommended output voltage are transferred through RS232 link to the host
range is about 600–700 V dc for 400 V ac mains computer.
voltage. The input power factor can be set to A new Hioki 3390 power analyzer with motor
unity, thus no reactive power is consumed. testing option and D/A output option has been
AFE can be controlled and monitored via CAN recently acquired to the laboratory. It has four
bus or from the panel. It is possible to set a posi- channels and the motor testing option, which
tive and a negative current limit. However, if the calculates the mechanical power from the mea-
consumed power is more than the current limit sured torque and speed information. Currents
allows, the dc link voltage decreases, and at a cer- are sensed with ac/dc clamp on CT 9278 current
tain point, the diodes of the IGBT bridge begin to clamps, which have a maximum continuous in-
conduct and keep the dc link voltage at rectified put current of 350 A for a bandwidth of 19 kHz,
mains voltage of 540 V. 200 A for 40 kHz, and 100 A for 100 kHz, re-
spectively.
Ultracapacitor voltage is measured with LEM
2.5 Load dynamometer AV100-750 transducer and current with LEM LA
An electric motor dynamometer is used to load 305-S transducer.
the traction motor. The motor manufacturer and
model are the same as the traction motor, which
is attached to the test bench (subsection 2.3.2). 2.8 Other Equipment
However, the original speed sensor has been re-
placed with SICK CKS36 Encoder, which has A programmable chassis dynamometer for two
resolution of 1 024 bit/rev.The dynamometer can wheels and a cold/warm chamber, which has a
be used to test also other traction motors by re- temperature range of 50... + 50 C, are avail-
placing the equipment under test (EUT) with an- able in the laboratory facilities. The size of the
other EM. chamber is big enough to drive a van inside. The
Vacon NXP frequency converter with an inverter programmable chassis dynamometer is located
unit INU 0300 5 is used to control the EM. Fre- next to the dc power supply and energy storages.
quency converter can be controlled and moni-
tored via CAN bus. The motor can be controlled
either with a torque reference with a speed limit 2.9 Cooling Circuit
or with a speed reference with a torque limit.
When the dynamometer is braking the traction A common rail liquid cooling circuit with six in-
motor, and hence, acting as a generator, the brak- dependent branches that are controlled with elec-
ing power is dumped to the brake resistor through trically operated valves is used to provide proper
the brake chopper. The dc links of the frequency cooling for liquid-cooled subsystems. An elec-
converter and AFE can also be interconnected by tric pump from EMP is used to control the flow
closing the switch S3 . Then, the generated power in the common rail, and electrical valves are used
flows back to the dc link and can be used e.g. to to open and close branches. Flow, pressure, and
provide power to the traction motor. If the trac- temperature data from each branch are measured
tive power is provided by the gen-set or battery, and sent to the ECU.

EVS25 World Battery, Hybrid and Fuel Cell Electric Vehicle Symposium 5
World Electric Vehicle Journal Vol. 4 - ISSN 2032-6653 - © 2010 WEVA Page000124

From the cold chamber circulation there is also a 3 Experimental Results

separate 20 C circulation of water-glycol mix-
ture available, if extensive cooling circulation is
needed.
3.1 Efficiency Measurements

2.10 Control System The efficiency maps of the traction motor, in-
verter, dc–dc converter, and ultracapacitor unit
Almost all laboratory equipment is controlled have been measured with Norma D6100 power
with a dSPACE MicroAutoBox (MABX) analyzer. Mechanical torque and speed mea-
DS1401/1505/1507 RCP ECU. It has four CAN surements from torque transducer were used to
buses and lot’s of analog and digital I/O. The calculate mechanical power of the EM. During
voltage level of the digital I/O is 5 V, analog the measurements of the traction motor and in-
inputs 5 V, and analog outputs 4.5 V, respec- verter, the switch S from Fig. 3 was set to closed-
tively. However, because voltage levels of some position and S1 to open-position, respectively.
devices is exceed those, some additional analog This allows the braking power of the dynamome-
and digital I/O has been implemented using ter to be used to provide power for the traction
Beckhoff’s industrial automation I/O modules. motor. Thus, only losses are taken from the
Beckhoff’s I/O modules communicate with mains.
MABX via CAN bus. The efficiency map of the traction motor and in-
MBSD is utilized to produce code for the verter were done by controlling the dynamome-
RCP ECU. Models are made with MAT- ter with a speed reference and the EUT with a
LAB/Simulink/Stateflow. Also Real-Time torque reference. The speed range was measured
Workshop and dSPACE RTI blockset are needed at speed intervals of 200 rpm. In each speed set-
to generate C code from the Simulink model. point, torque range was measured at torque inter-
The compiled code is then uploaded into the vals of 20 Nm. The measurements were made af-
MABX, after which the controller starts ex- ter a steady-state had been achieved. The power
ecuting the real-time program. The ECU is analyzer can calculate the electrical power of a
connected to the host computer via high speed three-phase symmetrical load from two line-to-
link. All measurements as well as other signals line voltages and two phase currents. Therefore,
can be monitored online from the host computer it was possible to measure the input and output
through dSPACE ControlDesk. A graphical power of the inverter simultaneously by measur-
user interface (GUI) has been made for every ing the dc link voltage and current, and two line-
device that is controlled with the ECU. It is to-line output voltages and two phase currents.
also possible to tune online e.g. the constant With the mechanical torque and speed measure-
parameters and the gains of the original model. ments, both the EM and the inverter efficiency
This makes it easy to implement manual controls maps were achieved. In the simulation models it
of the laboratory equipment directly from the is enough to have a combined map for the trac-
host computer. tion motor and inverter. The measured combined
efficiency map is shown in Fig. 5. It can be
seen that the best efficiency can be achieved with
speeds beyond the nominal point of the motor.
2.11 Simulation Platform
A flexible simulation platform has been created
that includes a library of easily parametrizable
plant models of HEV systems, e.g. EMs, invert- 350
ers, dc-dc converters, battery, ultracaps, gears,
ICE, etc. With the simulation platform it is easy 300

and fast to make early simulations of different

kind of hybridization configurations for mobile 250
Torque [Nm]

work machines and to develop and evaluate con- 200

trol strategies for them. Simulation time-step is
0.85
0.5

0.8
0.7

0.9

1 ms. Backward functional modeling from the 150 0.92

imposed load cycle is utilized in simulating the

HEV system. The load cycle can be a mechanical 100

load for EMs or an electrical load that is formed 50

e.g. from the measured traction power data of a
conventional machine. 0
0 1000 2000 3000 4000 5000 6000 7000 8000
The accuracy of the models has been defined to Speed [rpm]

resemble large signal behavior and to neglect fast

transients and all switching frequency character- Figure 5: Combined measured efficiency map of
istics of PE converters and EMs. This is because Siemens ELFA 1PV5135-4WS28 traction motor and
simulated driving cycles are long and detailed G650 D440/170/170 M7-1 inverter.
models that include switching behavior take very
long time to execute and need detailed informa-
tion about the plant.

EVS25 World Battery, Hybrid and Fuel Cell Electric Vehicle Symposium 6
World Electric Vehicle Journal Vol. 4 - ISSN 2032-6653 - © 2010 WEVA Page000125

3.2 HIL Simulation of SHEV Load 150

Sharing Algorithm 100

Simulation model validation is demonstrated 50

with a HIL simulation, where an UC stack is used

Voltage [V]
0
to provide peak power during short power peaks.
Thus, the primary power source—e.g. an ICE or
−50

a fuel cell—dynamics need not be as fast, and the −100

primary power source can be downsized. In this −150 Measurement

HIL simulation, the dc power supply was used Simulation

to emulate the primary power source. The load −200

0 50 100 150 200
power was made with the EM dynamometer. The Time [s]

control stategy, plant models, and experimental (a) DC–DC converter current.
results are described in more detail in [23].
Two modeling principles for the ultracapacitor 380
were used in the simulation: constant capaci-
tance (constant C) with equivalent series resis- 360

tance (ESR), and variable capacitance (variable

C) with ESR. In the constant C model, the nomi-
340

Voltage [V]
nal capacitance value was used. In the variable C 320
model, the capacitance has been calculated from
the constant current measurements, and a capac- 300

itance mapping has been defined based on the 280

Measurement

calculations, i.e. the capacitance depends on the

Simulation, variable C
Simulation, constant C
voltage and current. 260

Validation measurements are shown in Fig. 6. It

0 50 100 150 200
Time [s]
can be seen that the simulated and measured cur-
rents and voltages match quite well. However, (b) Ultracapacitor voltage.
there are some abnormalilities especially in the
currents. These are due to several reasons, one Figure 6: Comparison of measured and simulated ul-
being the dc–dc converter’s minimum current of tracapacitor voltage.
20 A, which causes differences in near zero cur-
rent, even though that functionality is modeled in
the plant model. Another cause for differences
is that the exact behavior of the dc power sup- Acknowledgments
ply’s voltage regulator is hard to model. How-
ever, the mean error is only -0.32 A and the rms
error 5.7 A. For the UC voltages, the variable C This study has been carried in HybDrive and
model gives more accurate results with mean er- HybLab projects funded by the Finnish Funding
ror of 1.7 V and rms error 3.0 V. Agency for Technology and Innovations (Tekes)
and Multidisciplinary Institute of Digitalization
and Energy (MIDE) of Aalto University, respec-
tively.
4 Conclusion
A versatile full-scale HIL verification environ-
ment for HIL simulations of heavy-duty hybrid References
electric vehicles and work machines was pre-
sented. The environment includes a full-scale hy- [1] “Directive 97/68/EC of the European Parlia-
brid electric power system with a gen-set, a bat- ment and of the Council of 16 December
tery, an ultracapacitor unit, and a traction drive. It 1997 on the approximation of the laws of the
also includes an EM dynamometer, a high-power
dc power supply, a closed-loop cooling circula- Member States relating to measures against
tion, and a versatile and powerful control system the emission of gaseous and particulate pol-
with automatic code generation and RCP hard- lutants from the internal combustion engines
ware and software. to be installed in non-road mobile machin-
The HIL verification environment is located at ery,” Journal of the European Union L 59 of
university facilities and it can be used for re-
search and education purposes as well as for February 1998, page 1 et seq. [Online]. Avail-
making case-study HIL simulations for work ma- able: http://ec.europa.eu/enterprise/sectors/
chine manufacturers. The test facility has been mechanical/non-road-mobile-machinery
used e.g. to measure the efficiencies of PECs,
EMs, and hydraulic pumps and motors, to char- [2] “Directive 2004/26/EC of the European Par-
acterize devices and validate plant models, to liament and of the Council of 21 April 2004
make experimental HIL simulations to verify and amending Directive 97/68/EC on the approxi-
validate HEV control algorithms, to provide load
cycles for energy storage testing, and to provide mation of the laws of the Member States relating
high and controllable electric power for various to measures against the emission of gaseous and
other research equipment. particulate pollutants from internal combustion

EVS25 World Battery, Hybrid and Fuel Cell Electric Vehicle Symposium 7
World Electric Vehicle Journal Vol. 4 - ISSN 2032-6653 - © 2010 WEVA Page000126

engines to be installed in non-road mobile ma- [14] Y. Cheng, J. V. Mierlo, P. Lataire, and
chinery,” Journal of the European Union L 146 G. Maggetto, “Test bench of hybrid electric ve-
of April 2004, page 1 et seq. [Online]. Avail- hicle with the super capacitor based energy stor-
able: http://ec.europa.eu/enterprise/sectors/ age,” Vigo, Spain, Nov. 2007, pp. 147–152.
mechanical/non-road-mobile-machinery
[3] T. Lehmuspelto, M. Heiska, A. Leivo, and [15] AGCO Sisu Power homepage. [Online]. Avail-
A. Hentunen, “Hybridization of a mobile able: http://www.agcosisupower.com/
work machine,” World Electric Vehicle Journal,
vol. 3, 2009. [Online]. Available: http://www.
[16] AXCO-Motors homepage. [Online]. Available:
evs24.org/wevajournal/
http://www.axcomotors.com/
[4] M. J. Marcel, T. A. Haskew, and K. A. Williams,
“Test facility for a hybrid fuel cell electric vehi-
[17] ABB homepage. [Online]. Available: http:
cle,” Proc. IEEE, pp. 734–739, Mar. 2007.
//www.abb.com/
[5] R. M. Schupbach and J. C. Balda, “A versatile
laboratory test bench for developing powertrains
[18] Siemens electric drive system ELFA. [Online].
of electric vehicles,” in Proc. IEEE Vehicular
Available: http://www.siemens.com/elfa/
Technology Conference 2002 Fall, vol. 3, Van-
couver, Canada, Sep. 2002, pp. 1666–1670.
[6] J. Weber, Automotive Development Processes: [19] Kokam homepage. [Online]. Available: http:
Processes for Succesful Customer Oriented Ve- //www.kokam.com/
hicle Development. Berlin, Germany: Springer,
2009. [20] Elithion homepage. [Online]. Available: http:
[7] S. C. Nagaraj and B. Detrick, “HIL and RCP //www.elithion.com/
tools for embedded controller development in
hybrid vehicles,” Dearborn, MI, Sep. 2009, pp. [21] MSc Electronics homepage. [Online]. Avail-
896–902. able: http://www.mscelectronics.fi/
[8] A. Wagener, P. Seger, C. Koerner, and H. Kabza,
“Simulation-based automatic code-generation [22] Vacon homepage. [Online]. Available: http:
for ECUs in distributed control systems, applied //www.vacon.com/
in a testbed for a hybrid vehicle drivetrain,”
vol. 2, Cholula, Puebla, Mexico, Dec. 2000, pp.
[23] M. Liukkonen, A. Hentunen, and J. Suomela,
643–648.
“Validation of quasi-static series hybrid electric
[9] A. Leivo, J. Suomela, and A. Hentunen, “Model vehicle model,” Lille, France, Sep. 2010. (in
reusability and cooperation in model based press).
HEV control system development,” Stavanger,
Norway, May 2009.
[10] C. Lin and L. Zhang, “Hardware-in-the-loop
simulations and its application in electric vehi-
cle development,” Harbin, China, Nov. 2008.
[11] H. Hu, G. Xu, and Y. Zhu, “Hardware-in-the-
loop simulation of electric vehicle powertrain Authors
system,” in Proc. Power and Energy Engineer-
ing Conference, Wuhan, China, Mar. 2009. Ari Hentunen
Ari Hentunen received his M.Sc.
[12] A. Bouscayrol, W. Lhomme, P. Delarue, (Tech.) degree in electrical en-
B. Lemaire-Semail, and S. Aksas, “Hardware- gineering from the Helsinki Uni-
in-the-loop simulation of electric vehicle trac- versity of Technology, Finland, in
tion systems using energetic macroscopic repre- 2005. Currently he is working as a
sentation,” in Proc. IEEE 32nd Annual Confer- researcher at the Aalto University,
ence on Industrial Electronics (IECON 2006), Department of Electrical Engineer-
Paris, France, Nov. 2006, pp. 5319–5324. ing. His main research interests are
[13] A. Bouscayrol, “Different types of hardware-in- in the field of hybrid electric work
the-loop simulation for electric drives,” Cam- machines and modeling of lithium-
bridge, UK, Jun. 2008, pp. 2146–2151. ion batteries.

EVS25 World Battery, Hybrid and Fuel Cell Electric Vehicle Symposium 8
World Electric Vehicle Journal Vol. 4 - ISSN 2032-6653 - © 2010 WEVA Page000127

Jussi Suomela Jussi Suomela re-

ceived the M.Sc., Lic.Sc., and
D.Sc. degrees from Helsinki Uni-
versity of Technology (TKK), Es-
poo, Finland, in 1992, 2001 and
2004, respectively. Suomela has
worked in the University since
1991 in various positions. His
main research areas are hybrid
electric transmission in off-road
mobile machines and field and ser-
vice robotics. At the moment
Suomela is professor of automa-
tion technology.

Antti Leivo Antti Leivo received

his M.Sc. degree in automation
and systems technology from the
Helsinki University of Technology,
Finland, in 2005. He has been
working since 2005 as a researcher
at the Helsinki University of Tech-
nology. His main research projects
are in the field of model based soft-
ware development of hybrid elec-
tric work machines.

Matti Liukkonen Matti Liukko-

nen (S09) was born on July 19,
1983. He received the M.Sc.
(Tech.) degree in electrical engi-
neering from the Helsinki Univer-
sity of Technology (HUT), Espoo,
Finland, in 2009. His major fields
of studies are in power electron-
ics, hybrid power trains and elec-
tric drives.

Panu Sainio Panu Sainio obtained

M.Sc in 1997, and Lic.Sc in 2006.
He has been participating in a num-
ber of national and international
projects of vehicle engineering and
testing. His primary research inter-
ests are tire-road contact and heavy
hybrid vehicles.

EVS25 World Battery, Hybrid and Fuel Cell Electric Vehicle Symposium 9