Simulation of a Series Hybrid Electric Vehicle based on Energetic Macroscopic Representation W. Lhomme1, A. Bouscayrol1, Member, IEEE, P. Barrade2, Member, IEEE , 1

L2EP Lille, USTL, 59 655 Villeneuve d'Ascq, France LEI, Ecole Polytechnique Fédérale de Lausanne, Switzerland, http://www.univ-lille1.fr/l2ep/, [email protected], [email protected] 2

Simulation of different HEVs could be made using EMR and Matlab-SimulinkTM for comparative studies. In this paper, this methodology is used to develop a simulation model of a whole HEV with a series configuration. In a first part, EMR and the deduced control structure are introduced. The second part deals with the EMR of the studied HEV. Section 3 is devoted to the maximum control structure of this HEV. In the last part, the EMR of the system is implemented into Matlab-SimulinkTM and simulation first results are provided.

Abstract— The simulation of a series HEV (hybrid electric vehicle) is suggested by using the Energetic Macroscopic Representation (EMR). Simplified models are used in a first step. They lead to the simulation of an overall HEV with its control. The system is decomposed in two parts: the charge and traction subsystems. Simulations with Matlab-SimulinkTM are provided for the simplified system. These first simulations can now be improved by using more complex models of some devices. Indeed, the EMR organizes systems as interconnected components, that leads to modularity and flexibility. Index Terms— Hybrid Electric Vehicle, modeling, power electronics, drive control.

II. EMR AND MCS OF POWER CONVERSION SYSTEMS

I. INTRODUCTION Environmental protection and energy conservation yields a progressive development of electric vehicle (EV) and hybrid electric vehicle (HEV) in the automotive industry [1]. Event if EVs lead to zero pollution emissions, they have a slight ability of energy storage of batteries. Thus, the autonomy of such vehicle is reduced in comparison with conventional cars. HEVs offer an interesting alternative between EV and ICE vehicles [2]. Most of automobile companies have developed HEV like Toyota (Prius), Honda (Insight), Renault (Kangoo), Ford, Nissan (Tino) [3, 4]. Nowadays, only a few of them are really on the market. HEVs are complex systems because there are composed of several devices of different nature (continuous, discrete, linear, non-linear…) and of different disciplines. Moreover, the are different configurations of HEV: series [5], parallel [6], series-parallel [7], structures with torque addition or speed addition… in order to compare different structures of HEV, simulation is a precious tool [8]. But, the simulation of an overall HEV is complex because it requires associations of devices with very different dynamics and characteristics. Most of them are often limited to traction system part [9], few propose a global model [8]. The Energetic Macroscopic Representation (EMR) has been developed to propose a synthetic description of electromechanical conversion system [10]. A Maximal Control Structure (MCS) can be deduced from this EMR through specific inversion rules. It leads to a control structure with a lot of operations and measurements. Practical control structures can be deduced from the MCS by some simplifications. This methodology has already been applied to an automatic subway [11], an electric vehicle [12], a highspeed railway train [13] and wind energy generation systems [14].

-1-

A. Energetic Macroscopic Representation (EMR) The Energetic Macroscopic Representation (EMR) is a synthetic graphical tool based on the principle of action and reaction between connected elements [15] of these components can be internally described by causal ordering graphs, or other descriptions as transfer functions, Petri nets, state models, bond-graphs.... Basic elements of an EMR — The EMR is based on three kinds of elements which describe the physical state of a component [10]. Source elements (green oval pictograms) produce state variables (outputs). They can be either generators or receptors. They are disturbed by reactions of other elements. Conversion elements ensure energy conversion without energy storage. They have tuning inputs to define the conversion between variables. Electrical conversions are depicted by orange square pictograms, electromechanical conversions by orange circular pictograms and mechanical conversions by orange triangular pictograms. Accumulation elements (orange rectangular pictograms with an oblique bar) connect other elements, thanks to energy storage, which induces at least one state variable. All these elements are connected through exchange vectors according to the principle of action and reaction. For energy conversion systems, specific association rules have been defined to built their EMR [10]. The names of variables are associated with the devices that they comes from (Tdcm as the torque produces by a DC machine for example). B. Maximum Control Structure (MCS) From the EMR of a system, one can deduce a control structure, which is composed of the maximum of control operations and measurements [14]. This method is so called

the maximum control structure (MCS). Continuous lines are associated with the inversion of action variables while dotted lines are related to the rejection of disturbance variables. All control blocks are depicted by blue parallelograms because they handle only information.

the differential part will be not taken into account and, thus, an equivalent single wheel will be considered. These elements will be inserted in a second step. B. EMR of the battery charge subsystem.

Inversion of elements — As explained in the causal ordering graph theory [10], a control structure has to inverse the global function of the power system. In the MCS methodology, the global control structure is decomposed into several control blocks. Each block has to inverse one power element of the EMR. Conversion elements are inverted directly. Accumulation elements need controllers in order to solve the inversion problem of their state variables. All the inputs of power elements, which are not used in the inversion chain, become disturbance inputs. So they are directly rejected in order to minimize their influence. MCS of conversion system — First, a tuning chain is defined from the technical requirements. It connects the chosen tuning input of the global system to the wished action output through action inputs of power elements. The other inputs become so disturbances. In most electric drive applications, the static converter is chosen as the tuning element. Then a control chain is obtained by inversion of the tuning one: from the reference variable to the tuning one. All the variables are initially considered as measurable. So, the control chain links the control blocks, which are inversions of power elements connected by the tuning chain. It is obvious that the MCS is the most complete control strategy where the variables can be measured. In real cases, control structures can be deduced from the MCS by simplifications or by taking into account estimations of non-measurable variables. III. EMR OF THE STUDIED HEV A. Description of the studied HEV The studied HEV has a series configuration (Fig. 1). The internal combustion engine (ICE) is used at constant rotation speed in order to charge the battery. The electric motor ensures all the traction of the HEV from the energy stored in the battery. Because the efficiency of an electric machine is higher than an ICE, and because the ICE is used at an optimum speed, the fuel consumption and the pollution emissions are reduced in comparison with classical vehicles. This HEV can be decomposed into two parts: the battery charge and the traction system. The battery charge is composed of a tank, a combustion engine, a DC generator and a chopper (voltage adaptation). The traction system is composed of a chopper, a permanent magnets DC machine, a gearbox (speed adaptation), a mechanical differential, two wheels and the vehicle chassis. Of course, these two parts are strongly coupled and a current can directly flow from the first chopper to the second one. In this preliminary study, this coupling is not taken into account and simplified models are used. Moreover DC machines with choppers have been chosen (Fig. 2). They will be replaced by AC machines and voltage source converters later. Moreover,

-2-

Equivalent sources — This subsystem leads to the conversion of fuel into electrical energy. Three equivalent sources can be defined. Thus, they are represented by green oval pictograms (Fig. 3). The first source is a Chemical Source (CS), which is associated with the fuel stored in the tank. It delivers a gasoline flow dgas to the ICE. The engine reacts by a pressure pice. The battery has to store electrical energy and can be considered as an Electrical Source (ES). The charge is ensured by a current icharge, in order to give a battery voltage ubat. In order to take into account the charge and discharge phenomena, the battery is firstly simulated as a capacitor with an equivalent time constant: C cap

d u bat + Rcap u bat = ich arg e dt

(1)

where Rcap and Ccap are the parameters of the equivalent capacitor. The battery could thus be considered as an accumulation element. Note that more complex models of battery have to be used in next studies [9]. The traction part is assumed to be an equivalent electrical source (ESeq)to the charge part. It leads to le discharge current ichop2. This equivalent source and the chemical source are linked by a parallel connection, which induces a direct transfer from the charge part to the traction part, without going through the battery. This connection is represented by a coupling electrical converter (overlapping squares) and expresses the parallel connection: u1 = u 2 = ubat  ich arg e = ichop1 − ichop 2

(2)

ICE modeling — A very simple model is used for the internal combustion engine. It leads to the produced torque Tice and pressure pice from the gasoline flow and rotation speed Ωshaft [9]:  pice = mice K ice Ω shaft η ρ Pc with K ice =  Ωice max Tice = mice K ice d gas

(3)

where η is the ICE efficiency, ρ the density of gasoline, Pc the calorific value of gasoline and Ωice_max the maximum rotation speed of the engine. The control of the machine is ensured by the mice ratio, which defines the actual flow of gasoline through a specific actuator. As a quasi-stationary and steady state model is used in this first step, the engine is depicted as a conversion element because no energy accumulation is considered.

wheel

=

Electric Generator

ICE

=

Electric Motor

Battery =

= Mechanical differential

Tank

wheel

Fig. 1. Configuration of the studied HEV

ICE

tank

shaft

dgas

Tdcg

Tice Ωshaft

pice

DC generator chopper 1 battery chopper 2

Ωshaft

idcg

ichop1

uchop1

ubat

DC machine gearbox wheel envir.

icharge uchop2 idcm

ichop2

mice

vvehic

Tdcm T gear

Fres

Ωgear Ω wheel

Fig. 2. Electromechanical scheme of the studied HEV

ICE

tank

shaft Ωshaft

Tice

dgas

battery chopper 2

DC generator chopper 1 edcg

uchop2

ubat

ichop1

idcg

DC machine idcm

gearbox Tdcm

wheels

Tgear

environ.

vehicle

Fwheel

vvehic SM

CS pice

Tdcg

Ωshaft mice

idcg

ubat ubat

uchop1 mchop1

ichop2

icharge

idcm

edcm

Ωgear

Ωwheel

Fres

vvehic

mchop2

ES

Fig. 3. EMR of the studied HEV

Tice

dgas

Ωshaft

edcg

idcg

uchop2

ubat

ichop1

Tdcm

idcm

Tgear

Fwheel

vvehic SM

ESeq

CS pice

Ωshaft mice

Tdcg

idcg

uchop1

ichop2

ubat

idcm

mchop1 Tdcg_ref

edcm

Ωgear

Ωwheel

vvehic

Fres

mchop2

idcg_ref uchop1_ref

uchop2_ref idcm_ref

Tdcm_ref

Tgear_ref

Fwheel

vvehic_ref

CM Tice_ref

Ωshaft_ref Fig. 4. MCS of the studied HEV

tank

ICE

shaft

DC generator

windings

battery

chopper1

ES

battery traction

chopper2

windings

DC machine

gearbox

wheel

chassis

ES

MS

Environment

MS

INIT INIT

Trajectory CM

Fig. 5. Matlab-Simulink model of the charge subsystem

Fig. 6. Matlab-Simulink model of the traction subsystem

-3-

The environment of the vehicle is considered as a mechanical source (MS). It yields a resistive force to the motion Fres from the vehicle velocity:

Shaft modeling — The classical relationship of the shaft yields the rotation speed as the state variable. This speed is thus produced from an interaction between the engine and generator torques, Tice and Tdcg: J

dΩ shaft dt

+ f Ω shaft = Tice − Tdcg

2 Fres = Froll + k areo vvehic

(4)

where Froll is the rolling force and kaero the aerodynamic coefficient. Other phenomena could be taken into account.

where f and J are the friction coefficient and moment of inertia of the shaft. As the shaft induces a state variable, it has to be represented by an accumulation element.

DC motor modelling — Another DC machine with permanent magnets is used for the traction. It could be modelled with the same relationships than the DC generator. The motor current idcm is the state variable of the accumulation element:

DC generator modeling — The DC machine has permanent magnets. Two parts can be considered. First, an electromechanical conversion yields an electromotive force edcg and the machine torque from the rotation speed and the machine current idcg: Tdcg = kφg idcg  edcg = kφg Ω shaft

Ldcm

(5)

d idcg + Rdcg idcg = edcg − uchop1 dt

Tdcm = kφm idcm  edcm = kφm Ω gear

(9)

( 10 )

where kφm is the flux coefficient.

(6)

Mechanical transmission modeling — A simplified mechanical transmission is considered: an equivalent wheel replaces the differential double-wheel system; the slip phenomenon [16] is not yet taken into account. These assumptions lead to consider this mechanical transmission as only three parts by using association rules [10]. The gearbox has to adapt the rotation speeds between the machine and the wheel. It yields the gear torque Tgear and the speed of the machine shaft from the machine torque and the wheel speed Ωwheel:

where Rdcg and Ldcg are the resistance and inductance of the armature windings. This second element is an accumulation element, because it stores kinetic energy in the inductance. Chopper modeling — The chopper ensures a DC-DC conversion to adapt the charge of the battery from the ICE and DC generator. It yields the charge current ichop1 and the voltage applied to the machine from the machine current and the voltage battery: uchop1 = mchop1 ubat i  chop1 = mchop1 idcg

d idcm + Rdcm idcm = uchop 2 − edcm dt

where Rdcm and Ldcm are the resistance and inductance of the armature windings. An electromechanical element yields the machine torque Tdcm and the electromotive force edcm:

where kφg is the flux coefficient. Thus this function is depicted by an orange circle. The armature windings yield the machine current as state variable from the electromotive force and the supplied voltage, uchop1: Ldcg

(8)

Tgear = k gear Tdcm  Ω gear = k gear Ω wheel

(7)

( 11 )

where kgear is the gearbox ratio. This device is thus depicted by an orange triangular pictogram. The wheel has to produce a linear motion from a rotational motion. It yields the traction force Fwheel and the wheel rotation from the gear torque and the vehicle velocity vvehic:

where mchop1 is the modulation function of the chopper, which is a combination of the commutation orders of its power switches [10]. An averaged model will be used in simulation. This device is thus depicted by an orange square pictogram (electrical conversion without energy accumulation).

 Fwheel = Tgear / Rwheel  Ω wheel = vvehic / Rwheel

C. EMR of the traction subsystem Equivalent sources — This subsystem leads the conversion of electrical energy into mechanical energy. The charge part is assumed to be an electrical equivalent source for the traction part. It provides a supplied voltage ubat to this traction part. The reaction current ichop2 leads to a discharge of the battery.

( 12 )

where Rwheel is the wheel radius. The wheel is thus a mechanical converter (orange triangle pictogram). The vehicle velocity is obtained with the classical dynamics relationship from the traction and resistive forces:

-4-

Fwheel − Fres = M

d vvehic dt

( 13 )

where M is the mass of the vehicle. The vehicle chassis is then an accumulation element with the velocity as state variable.

velocity control loop is added in the MCS. Thus, the control of the traction part has to define the chopper modulation function to obtain the wished vehicle velocity. Control of the vehicle velocity — The control structure of the traction subsystem is deduced from the EMR by using inversion rules. This MCS part requires velocity and current control loops (Fig. 4).

IV. MCS OF THE STUDIED HEV A maximum control structure of the simplified HEV is deduced from its EMR. This MCS will be used in simulation to validate the global modeling of the system. All variables are considered as directly measurable in this preliminary study. More practical control structures can be deduced from the MCS with simplifications and estimations of non-measurable variables [14].

V. SIMULATION OF THE STUDIED HEV A. Transposition of EMR to Matlab-SimulinkTM The EMR of the system is directly converted into the Matlab-SimulinkTM model (Fig. 5 and Fig. 6). Indeed, the action-reaction organization yields the block description of this software by choosing appropriated inputs and outputs.

A. Global control of the system The EMR shows that the system has three freedom degrees: the engine coefficient mice, and the modulation functions of both choppers, mchop1 and mchop2. The engine coefficient is used to impose a constant rotation speed of the charge subsystem. The modulation function of the charge-chopper is used to charge the load torque for the ICE. Indeed, the ICE has a better efficiency and smaller pollution emissions for the rated speed and rated torque. The modulation function of the traction-chopper is used to produce the wished torque required by the driver through the acceleration pedal. B. MCS of the charge subsystem

B. Simulation results The time constants of both subsystems are very different. The charge and discharge of the battery require a long simulation time (several hours). But the vehicle dynamics has shorter time constants. For this reason, the simulation is decomposed into two parts, one for the charge subsystem, and the second for the traction subsystem. The presented simulations have only to validate the overall modeling of the studied HEV. Practical scenarios have to be defined when the simulation is improved by more complex models [4, 9, 17]. A very simple trajectory is used in these first simulations. Simulation of the charge operation — This simulation is made with a great sample time to enable a long simulation time without a too long computation time. The discharge current is defined from the other simulation part: the current ichop2 deduced from the traction cycle (see Fig. 11) is stored in a look-up table and repeated several times. In order to reduce the computation time, the time constant of the battery charge has been reduced to 3 hours. A charge test is simulated with an upper limit of the battery voltage of 270 V and a lower limit of 200 V. The charge and discharge of the battery is function of the current imposed by the traction subsystem (Fig. 7). The engine is starts when the lower limit is reached and stops when the upper limit is reached (Fig. 8).

Control of the ICE speed — The chain linking the ICE parameter kice and the rotation speed Ωshaft has to be inverted. It leads to the engine control to keep a constant speed in a charge operation. This algorithm requires a speed controller (Fig. 4). Control of the ICE torque — The chain linking the modulation function of the charge-chopper mchop1 to the torque of the DC generator Tdcg has to be inverted. It leads to the current control loop of the DC generator (Fig. 4). Charge management — A management block is used to define the reference rotation speed Ωshaft_ref and the load torque of ICE Tdcg_ref. This management is very simple: if the battery voltage is smaller than a lower limit, the ICE starts (rated references are imposed); if the battery voltage is greater than an upper limit, the ICE stops (zero references are imposed). This strategy requires the measurement of the battery voltage (Fig. 4). Note that the voltage measurement is a real problem for battery and estimation algorithms are used in practical controls.

Simulation of the traction operation — This simulation is made with a short sample time in order to take the smallest dynamics into account (dynamics of the current loop). As the chosen velocity trajectory has a short duration, the battery voltage is assumed to have a constant value in this test. A start-up is imposed to reach a 50 km/h velocity. After 35 s the HEV stops and reverses at 5 km/h to park. The velocity follows the reference (Fig. 9). The torque of the DC machine indicates that four quadrant operations are well suited (Fig. 10). This cycle leads to a specific current injected in the battery (Fig. 11). This cycle has been used to evaluate the operations of the charge subsystem.

C. MCS of the traction subsystem Velocity or torque control? — In a practical way, the driver gives a torque reference by acting on the accelerator pedal. The control of the vehicle velocity is ensured by the driver itself. But, in order to simulate a traction cycle, the

-5-

300

VI. CONCLUSION

300

An overall series HEV has been simulated with MatlabSimulinkTM by using EMR. Some components have been modeled by very simple relationships and the electrical coupling between choppers and the battery has been neglected. The simulation results validate the suggested representations. This first step will be completed by connecting both parts and using more precise models of some components: internal combustion engine, battery, AC machines instead of DC machines; mechanical differential… The relationships inside each EMR element will thus be changed without changing the global structure of the simulation.

200

VII. REFERENCES

200 battery voltage ubat (V) 100 time (min) 0 0

40

80

120

Fig. 7. Battery voltage during the charge test

ICE speed Ωshaft (rad/s)

[1]

100 [2]

00 0

40

80

120 time (min)

[3]

Fig. 8. Speed of ICE during the charge test

60

[4]

velocity (km/h) [5]

40

vvehic

20

[6]

vvehic_ref

[7] [8]

0

time (s) 0

10

20

30

40

50

[9]

Fig. 9. Velocity of the HEV [10]

torque Tdcm (Nm)

200

[11]

0

[12]

-200

time (s) 0

10

20

30

40

50

[13]

Fig. 10. Torque of the traction DC machine [14]

current ichop2 (A)

100

[15]

0 [16]

time (s) -100

0

10

20

30

40

50 [17]

Fig. 11. Absorbed current from the traction part

-6-

G. Maggeto, "Advanced drive systems and infrastructure for electric or hybrid busses, van and passenger car", EPE Journal, vol. 2, no. 2, pp. 71-76, June 1992. V. Wouk, "Hybrids: then and now", IEEE Spectrum, July 1995, pp. 16-21. B. Bates, "Getting a Ford HEV on the road", IEEE Spectrum, July 1995, pp. 22-25. R. Trigui, F. Badin, B. Jeanneret, F. Harel, G. Coquery, R. Lallemand, J. P. Ousten, M. Castagné, M. Debest, E. Gittard, F. Vangraefshepe, V. Morel, L. Baghli, A. Rezzoug, J. Labbé, S. Biscaglia, “Hybrid light duty vehicles evaluation program” International Journal of Automotive Technology, vol. 4, no. 2, June 2003, pp. 65-75. A. Di Napoli, F. Crescimbini, L. Solero, F. Caricchi, F. Caponni, "Multiple-input DC-DC power converter for power-flow management in Hybrid Vehicles", IEEE-IAS'02, October 2002, CD-ROM. S. Delprat, T. M. Guerra, J. Rimaux, "Optimal control of a parallel powertain: from global optimization to real time control strategy", EVS'18, Berlin, October 2001, CD-ROM. D. Hermance, S. Sasaki, "Hybrid electric vehicles take to the streets", IEEE Spectrum, November 1998, pp. 48-52. R. Trigui, B. Jeanneret, F. Badin, "Systemic approach for the prediction of dynamic performances and energy consumption of hybrid vehicles, construction of the model library VEHLIB", RTS revue, Elsevier, to be published in 2004. B. K. Powell, K. E. Bailley, S. R. Cikanek, "Dynamic modelling and control of hybrid electric vehicle powertrain systems", IEEE Control Systems Magazine, October 1998, pp 17-33. A. Bouscayrol, X. Guillaud, J. P. Hautier, Ph. Delarue, "Macromodelling for electromechanical conversions: application to electrical machine controls", (text in French) RIGE, vol. 3, no 2, pp. 257282, June 2000. J. C. Mercieca, J. N. Verhille, A. Bouscayrol, “Energetic Macroscopic Representation of a subway traction system for a simulation model ”, IEEE-ISIE'04, Ajaccio (France), May 2004, CD-ROM. J. Pierquin, P. Escané, A. Bouscayrol, M. Pietrzak-David, J. P. Hautier, B. de Fornel, "Behavior model control of a high speed railway traction system", EPE-PEMC'2000, vol. 6, pp. 197-202, Kosice (Slovak Republic), September 2000. J. Pierquin, B. Vulturescu, A. Bouscayrol, J. P. Hautier, "Behaviour model control structures for an electric vehicle", EPE-01, Graz (Austria), August 2001, CD-ROM. A. Bouscayrol, Ph. Delarue, “Simplifications of the Maximum Control Structure of a wind energy conversion system with an induction generator", International Journal of Renewable Energy Engineering, vol. 4, no. 2, pp. 479-485, August 2002. A. Bouscayrol, B. Davat, B. de Fornel, B. François, J.P. Hautier, F. Meibody-Tabar, M. Pietrzak David, "Multi-machine Multi-converter Systems for drives: analysis of coupling by a global modeling", IEEE-IAS'2000, Rome, October 2000, CD-ROM. Y. Hori, Y. Toyoda, Y. Tsukuoka, “Traction control of electric vehicle: basic experimental results using the test EV UOT Electric March”, IEEE Trans. on Industry Applications, vol. 34, no. 5, September/October 1998, pp. 1131-1138. V. Talon, N. Venuti, P. Higelin, "SI engine modeling using a Bond Graph formalism", ICE'03, Capri (Italy), September 2003, CD-ROM.