Electrical Vehicle Design and Modeling

Udit Sangwan

Electrical Vehicle Design and Modeling

Udit Sangwan

visibility

…

description

479 pages

link

1 file

Abstract

Early version, also known as pre-print Link to publication from Aalborg University Citation for published version (APA): Schaltz, E. (2011). Electrical Vehicle Design and Modeling. In S. Soylu (Ed.), Electric Vehicles -Modelling and Simulations (1 ed., pp. 1-24). Croatia: INTECH.

Figures (424)

LU VV bc ee VRE SS 7 VULile UI ULC pees EEL Or Cu. LLUVYVE VEL, dit LLLLO a alls LLC GILILLOLluUic’
in Fig. 1 is chosen.

The purpose of the different components in Fig. 1 will here shortly be explained: The traction
power of the wheels is delivered by the three phase electric machine. The torque of the left
and right wheels are provided by a differential with also has a gear ratio in order to fit the high
speed of the electric machine shaft to the lower speed of the wheels. The torque and speed
of the machine are controlled by the inverter which inverts the battery DC voltage to a three
phase AC voltage suitable for the electric machine. When analyzing the energy consumption
of an electric vehicle it is important also to include the losses due to the components which
not are a part of the power chain from the grid to the wheels. These losses are denoted as
auxiliary loss and includes the lighting system, comfort system, safety systems, etc. During
the regenerative braking it is important that the maximum voltage of the battery is not
exceeded. For this reason a braking resistor is introduced. The rectifier rectifies the three
phase voltages and currents of the grid to DC levels and the boost converter makes it possible
to transfer power from the low voltage side of the rectifier to the high voltage side of the
battery. — LU VV bc ee VRE SS 7 VULile UI ULC pees EEL Or Cu. LLUVYVE VEL, dit LLLLO a alls LLC GILILLOLluUic’ in Fig. 1 is chosen. The purpose of the different components in Fig. 1 will here shortly be explained: The traction power of the wheels is delivered by the three phase electric machine. The torque of the left and right wheels are provided by a differential with also has a gear ratio in order to fit the high speed of the electric machine shaft to the lower speed of the wheels. The torque and speed of the machine are controlled by the inverter which inverts the battery DC voltage to a three phase AC voltage suitable for the electric machine. When analyzing the energy consumption of an electric vehicle it is important also to include the losses due to the components which not are a part of the power chain from the grid to the wheels. These losses are denoted as auxiliary loss and includes the lighting system, comfort system, safety systems, etc. During the regenerative braking it is important that the maximum voltage of the battery is not exceeded. For this reason a braking resistor is introduced. The rectifier rectifies the three phase voltages and currents of the grid to DC levels and the boost converter makes it possible to transfer power from the low voltage side of the rectifier to the high voltage side of the battery.

The forces which the electric machine of the vehicle must overcome are the forces due to
gravity, wind, rolling resistance, and inertial effect. These forces can also be seen in Fig. 2
where the forces acting on the vehicle are shown. — The forces which the electric machine of the vehicle must overcome are the forces due to gravity, wind, rolling resistance, and inertial effect. These forces can also be seen in Fig. 2 where the forces acting on the vehicle are shown.

Fig. 6. Electric circuit diagram of the boost converter.
The circuit diagram of the boost converter can be seen in Fig. 6. The losses of the boost — Fig. 6. Electric circuit diagram of the boost converter. The circuit diagram of the boost converter can be seen in Fig. 6. The losses of the boost

In order to utilize the three phase voltages of the grid vy, vy, and vy they are rectified by a
rectifier as seen in Fig. 7. In the rectifier the loss is due to the resistance Rp rf and threshold
voltage Vp thRE- — In order to utilize the three phase voltages of the grid vy, vy, and vy they are rectified by a rectifier as seen in Fig. 7. In the rectifier the loss is due to the resistance Rp rf and threshold voltage Vp thRE-

The three sub-processes in the “Simulation routine”-process are executed three times in order
to make sure that parameters converges to the same values for the same input. After the
“Simulation routine’-process is finish the “Calculate number of parallel strings’-process is
applied. In this process the number of parallel strings Ngatp is either increased or decreased.
When the minimum possible number of parallel strings that fulfills both the energy and power
requirements of the battery has been found the “Simulation routine’-process is executed in
order to calculate the grid energy due to the final number of parallel strings.
In principle all the energy of a battery could be used for the traction. However, in order to
prolong the lifetime of the battery it is usually recommended not to charge it to more than
90% of its rated capacity and not to discharge it below SoCgatmin = 20%, ie., only 70%
of the available energy is therefore utilized. In Fig. 10 it can be seen how the “Calculate
number of parallel strings’-process finds the minimum number of parallel strings Npatp
that fulfills both the energy and power requirements. This process is a part of the sizing
procedure shown in Fig. 9. In Fig. 10(a) the minimum state-of-charge min(SoCgat) is shown
and in Fig. 10(b) the maximum battery cell discharge current max(igat ce!) is shown. From
the figure it is understood that the first iteration is for Ngatp = 10. However, both the
minimum state-of-charge and maximum discharge current are satisfying their limits, i.e.,
SoCgatmin = 0.2 and [pat max,cell = 28 A, respectively. Therefore the number of parallel strings
is reduced to N Bat,p = 3 for iteration number two. However, now the state-of-charge limit is
exceeded and therefore the number of parallel strings is increased to Ngatp = 8 for iteration
three. This process continuous until iteration number six where the number of parallel strings
settles to Neatp = 6, as this is the minimum number of parallel strings which ensures that
both the state-of-charge and maximum current requirements are fulfilled. — The three sub-processes in the “Simulation routine”-process are executed three times in order to make sure that parameters converges to the same values for the same input. After the “Simulation routine’-process is finish the “Calculate number of parallel strings’-process is applied. In this process the number of parallel strings Ngatp is either increased or decreased. When the minimum possible number of parallel strings that fulfills both the energy and power requirements of the battery has been found the “Simulation routine’-process is executed in order to calculate the grid energy due to the final number of parallel strings. In principle all the energy of a battery could be used for the traction. However, in order to prolong the lifetime of the battery it is usually recommended not to charge it to more than 90% of its rated capacity and not to discharge it below SoCgatmin = 20%, ie., only 70% of the available energy is therefore utilized. In Fig. 10 it can be seen how the “Calculate number of parallel strings’-process finds the minimum number of parallel strings Npatp that fulfills both the energy and power requirements. This process is a part of the sizing procedure shown in Fig. 9. In Fig. 10(a) the minimum state-of-charge min(SoCgat) is shown and in Fig. 10(b) the maximum battery cell discharge current max(igat ce!) is shown. From the figure it is understood that the first iteration is for Ngatp = 10. However, both the minimum state-of-charge and maximum discharge current are satisfying their limits, i.e., SoCgatmin = 0.2 and [pat max,cell = 28 A, respectively. Therefore the number of parallel strings is reduced to N Bat,p = 3 for iteration number two. However, now the state-of-charge limit is exceeded and therefore the number of parallel strings is increased to Ngatp = 8 for iteration three. This process continuous until iteration number six where the number of parallel strings settles to Neatp = 6, as this is the minimum number of parallel strings which ensures that both the state-of-charge and maximum current requirements are fulfilled.

Fig. 10. Number of parallel strings Ngai,» due to the “Calculate number of parallel
strings”-process in Fig. 9. The numbers in the green and yellow boxes indicate the iteration
number of the design procedure. The yellow boxes are the first and last iteration number. (a)
Minimum state-of-charge SoCgat min. The red dashed horizontal lines indicates the minimum
allowed state-of-charge. (b) Maximum cell discharge current max(igat ce). The dashed red
horizontal line indicates the maximum allowed discharge current. — Fig. 10. Number of parallel strings Ngai,» due to the “Calculate number of parallel strings”-process in Fig. 9. The numbers in the green and yellow boxes indicate the iteration number of the design procedure. The yellow boxes are the first and last iteration number. (a) Minimum state-of-charge SoCgat min. The red dashed horizontal lines indicates the minimum allowed state-of-charge. (b) Maximum cell discharge current max(igat ce). The dashed red horizontal line indicates the maximum allowed discharge current.

Table 3. Specifications of UQM PowerPhase 75 (UQM, 2010) from UQM Technologies.
torque), and maximum speed are therefore — Table 3. Specifications of UQM PowerPhase 75 (UQM, 2010) from UQM Technologies. torque), and maximum speed are therefore

The efficiency of the machine for different torque-speed characteristics can be seen in Fig. 11.
It is seen that the efficiency is highest at continuous torque Ts,cont and nominal speed ng nom. A
common mistake in electric vehicle modeling is to assume a fixed efficiency of the components
and it can be understood from Fig. 11 that wrong conclusions therefore can be made if the
electric machines not is operating in a sufficient point of operation. The corner speed ns,corner,
nominal speed Ms nom, Maximum speed ns max, continuous torque Ts,cont, peak torque Ts,max,
and the torque contour Ts jimit are also shown in the figure. — The efficiency of the machine for different torque-speed characteristics can be seen in Fig. 11. It is seen that the efficiency is highest at continuous torque Ts,cont and nominal speed ng nom. A common mistake in electric vehicle modeling is to assume a fixed efficiency of the components and it can be understood from Fig. 11 that wrong conclusions therefore can be made if the electric machines not is operating in a sufficient point of operation. The corner speed ns,corner, nominal speed Ms nom, Maximum speed ns max, continuous torque Ts,cont, peak torque Ts,max, and the torque contour Ts jimit are also shown in the figure.

Due to the minimum battery pack voltage requirement Nga; = 216 series connected battery
cells are required. The chosen vehicle is designed to be able to handle 14 repetitions of the
NEDC. From Fig. 10 it is understood that Ngatp = 5 parallel strings are demanded in order to
fulfill this requirement. This means that the battery pack has a capacity of — Due to the minimum battery pack voltage requirement Nga; = 216 series connected battery cells are required. The chosen vehicle is designed to be able to handle 14 repetitions of the NEDC. From Fig. 10 it is understood that Ngatp = 5 parallel strings are demanded in order to fulfill this requirement. This means that the battery pack has a capacity of

Fig. 13. Simulation results of the vehicle with 14 repeated NEDC cycles as input. (a) Battery
state-of-charge. (b) Battery current. (c) Battery voltage. (d) Power of the battery and grid. — Fig. 13. Simulation results of the vehicle with 14 repeated NEDC cycles as input. (a) Battery state-of-charge. (b) Battery current. (c) Battery voltage. (d) Power of the battery and grid.

Fig. 14. Energy distribution in the vehicle relative to the grid energy.

A VHDL-AMS model for the dynamic model is specified in an “architecture” description as
show in Listing 1.
The power available in the wheels of the vehicle is expressed by: — A VHDL-AMS model for the dynamic model is specified in an “architecture” description as show in Listing 1. The power available in the wheels of the vehicle is expressed by:

Figure 3 shows the description of the model of the PMSM in Simplorer 7.0 software.
The equation giving the angle by the motor can be written in the following form — Figure 3 shows the description of the model of the PMSM in Simplorer 7.0 software. The equation giving the angle by the motor can be written in the following form

To achieve an optimal control, which means a maximum torque, it is necessary to satisfy the
following condition:
The first part (A) in figure 1 illustrates the control strategy. It presents a first PI speed control
used for speed regulation. The output of the speed control is Iqres; its application to a second PI
current regulator makes the adjustment of phase and squaring currents. The outputs of current
regulators are Varer and Vorer; they are applied to a Park transformation block. Where, Varer and
Voret are the forcing function to decide the currents in d-q axis model which may be obtained
from 3-phase voltages (Va, Vp and V.) through the park transformation technique as: — To achieve an optimal control, which means a maximum torque, it is necessary to satisfy the following condition: The first part (A) in figure 1 illustrates the control strategy. It presents a first PI speed control used for speed regulation. The output of the speed control is Iqres; its application to a second PI current regulator makes the adjustment of phase and squaring currents. The outputs of current regulators are Varer and Vorer; they are applied to a Park transformation block. Where, Varer and Voret are the forcing function to decide the currents in d-q axis model which may be obtained from 3-phase voltages (Va, Vp and V.) through the park transformation technique as:

The relationship between the switching variable vector [a, b, c]t and the line-to-line output
voltage vector [Vab Vbc Vca]' and the phase (line-to-neutral) output voltage vector [Va Vb
Vc]t are given by the following relationships, where a, b, c are the orders of 51, S3, S5
respectively.
The structure of a typical three-phase VSI is shown in figure 6. As shown below, Va, Vb and
Vc are the output voltages of the inverter. S1 through S6 are the six power transistors IGBT
that shape the output, which are controlled by a, a’, b, b’, c and c’. When an upper transistor
is switched on (i.e., when a, b or c are 1), the corresponding lower transistor is switched off
(i.e., the corresponding a’, b’ or c’ is 0). The on and off states of the upper transistors, S1, Ss
and Ss, or equivalently, the state of a, b and c, are sufficient to evaluate the output voltage. — The relationship between the switching variable vector [a, b, c]t and the line-to-line output voltage vector [Vab Vbc Vca]' and the phase (line-to-neutral) output voltage vector [Va Vb Vc]t are given by the following relationships, where a, b, c are the orders of 51, S3, S5 respectively. The structure of a typical three-phase VSI is shown in figure 6. As shown below, Va, Vb and Vc are the output voltages of the inverter. S1 through S6 are the six power transistors IGBT that shape the output, which are controlled by a, a’, b, b’, c and c’. When an upper transistor is switched on (i.e., when a, b or c are 1), the corresponding lower transistor is switched off (i.e., the corresponding a’, b’ or c’ is 0). The on and off states of the upper transistors, S1, Ss and Ss, or equivalently, the state of a, b and c, are sufficient to evaluate the output voltage.

Electric Vehicles - Modelling and Simulations

Fig. 6. SIMULINK models for a vector control and his interaction in a chain of traction for
vehicle
Figure 6 details the vector control (Id=0 strategy) of the vehicle, implemented under
Matlab/simulink software. — Fig. 6. SIMULINK models for a vector control and his interaction in a chain of traction for vehicle Figure 6 details the vector control (Id=0 strategy) of the vehicle, implemented under Matlab/simulink software.

VHDL- -AMS descriptions were developed for each block of the electric vehicle including
structural models. The obtained blocks were connected in Simplorer 7.0 Software
environment to obtain a high level description our system as detailed on figure 8. The
exposed blocks include analogue/ digital electronic behavioural descriptions. It represents a
so complex multi-domain system (Fakhfakh et al., 2006). — VHDL- -AMS descriptions were developed for each block of the electric vehicle including structural models. The obtained blocks were connected in Simplorer 7.0 Software environment to obtain a high level description our system as detailed on figure 8. The exposed blocks include analogue/ digital electronic behavioural descriptions. It represents a so complex multi-domain system (Fakhfakh et al., 2006).

Modeling and Simulation of High Performance Electrical Vehicle Powertrains in VHDL-AMS

6.2 VHDL-AMS virtual prototype — Modeling and Simulation of High Performance Electrical Vehicle Powertrains in VHDL-AMS 6.2 VHDL-AMS virtual prototype

For each factor, we define three levels: low, center and high levels as detailed on table 3.
To optimize our control strategy, we have adopted an experimental design approach by
applying the Doehlert design. Six factors have been considered as shown on table 3: Ke, Ld,
Ts, E, R and rm. According to the number of factors, in order to limit the number of runs
and to take into account the major effects, a screening study is necessary. Consequently, a
first step of screening was conducted using a fractional factorial design. The last with six
factors is a design involving a minimum of 45 experiments (see appendix).

For each factor, we define three levels: low, center and high levels as detailed on table 3. — For each factor, we define three levels: low, center and high levels as detailed on table 3. To optimize our control strategy, we have adopted an experimental design approach by applying the Doehlert design. Six factors have been considered as shown on table 3: Ke, Ld, Ts, E, R and rm. According to the number of factors, in order to limit the number of runs and to take into account the major effects, a screening study is necessary. Consequently, a first step of screening was conducted using a fractional factorial design. The last with six factors is a design involving a minimum of 45 experiments (see appendix). For each factor, we define three levels: low, center and high levels as detailed on table 3.

Fig. 9. Dynamic response of the vehicle speed in Simplorer and Matlab
Electric Vehicles - Modelling and Simulations — Fig. 9. Dynamic response of the vehicle speed in Simplorer and Matlab Electric Vehicles - Modelling and Simulations

An optimal result of control strategy on the electrical vehicle is obtained with NEMRODW
software to obtain an optimal dynamic response. It is detailed on table 5. — An optimal result of control strategy on the electrical vehicle is obtained with NEMRODW software to obtain an optimal dynamic response. It is detailed on table 5.

Modeling and Simulation of High Performance Electrical Vehicle Powertrains in VHDL-AMS

The speed response shows the good dynamic suggested of our vehicle. The reference speed
is attained in 4.65 second. The direct current is equal to zero in the permanent mode, the
quadratic current present the image of the electromagnetic torque. The power is increased to
reach 58 kW. — The speed response shows the good dynamic suggested of our vehicle. The reference speed is attained in 4.65 second. The direct current is equal to zero in the permanent mode, the quadratic current present the image of the electromagnetic torque. The power is increased to reach 58 kW.

A hybrid electric vehicle is distinguee from a standard ICE driven by four different parts: a)
a device to store a large amount of electrical energy, b) an electrical machine to convert
electrical power into mechanical torque on the wheels, c) a modified ICE adapted to hybrid
electric use, d) a transmission system between the two different propulsion techniques.
Figure 1 shows the possible subsystems of a hybrid vehicle configuration [Chan, 2002],
[Ehsani, 2005] — A hybrid electric vehicle is distinguee from a standard ICE driven by four different parts: a) a device to store a large amount of electrical energy, b) an electrical machine to convert electrical power into mechanical torque on the wheels, c) a modified ICE adapted to hybrid electric use, d) a transmission system between the two different propulsion techniques. Figure 1 shows the possible subsystems of a hybrid vehicle configuration [Chan, 2002], [Ehsani, 2005]

The architecture of a hybrid vehicle is defined as the connection between the components
of the energy flow routes and control ports. Hybrid electric vehicles were classified into
two basic types: series and parallel. But presently HEVs are classified into four kinds:
series hybrid, parallel hybrid, series-parallel hybrid and complex . The primary power
source (steady power source) is made up of fuel tank and ICE and battery-electric motor
is taken as secondary source (dynamic power source).
A series hybrid drive train uses two power sources which feeds a single powerplant
‘electric motor) that propels the vehicle. In Figure 2 is shown a series hybrid electric
drive train where: fuel tank is an unidirectional energy source and the ICE coupled to an
electric generator is a unidirectional energy converter. The electric generator is connected
to an electric power bus through an electronic converter (rectifier). Electrochemical
oattery pack is the bidirectional energy source and is connected to the power bus by
means of a power electronics converter (DC/DC converter). Also the electric power bus
is connected to the controller of the electric traction motor. The traction motor can be
controlled either as a motor (when propels the vehicle) or as generator (to vehicle
oraking). A battery charger can charge batteries with the energy provided by an
electrical network.

[he possible operating modes of series hybrid electric drive trains are [Ehsani, 2005]: 1. — The architecture of a hybrid vehicle is defined as the connection between the components of the energy flow routes and control ports. Hybrid electric vehicles were classified into two basic types: series and parallel. But presently HEVs are classified into four kinds: series hybrid, parallel hybrid, series-parallel hybrid and complex . The primary power source (steady power source) is made up of fuel tank and ICE and battery-electric motor is taken as secondary source (dynamic power source). A series hybrid drive train uses two power sources which feeds a single powerplant ‘electric motor) that propels the vehicle. In Figure 2 is shown a series hybrid electric drive train where: fuel tank is an unidirectional energy source and the ICE coupled to an electric generator is a unidirectional energy converter. The electric generator is connected to an electric power bus through an electronic converter (rectifier). Electrochemical oattery pack is the bidirectional energy source and is connected to the power bus by means of a power electronics converter (DC/DC converter). Also the electric power bus is connected to the controller of the electric traction motor. The traction motor can be controlled either as a motor (when propels the vehicle) or as generator (to vehicle oraking). A battery charger can charge batteries with the energy provided by an electrical network. [he possible operating modes of series hybrid electric drive trains are [Ehsani, 2005]: 1.

Electric Vehicles - Modelling and Simulations

The bulk capacitor C, models the cell’s open circuit voltage, and R.q is included to represent
the self-discharge of the cell.

Voltages and currents describing the characteristics of the model shown in Fig.4.c are given
by equations — Electric Vehicles - Modelling and Simulations The bulk capacitor C, models the cell’s open circuit voltage, and R.q is included to represent the self-discharge of the cell. Voltages and currents describing the characteristics of the model shown in Fig.4.c are given by equations

The control strategies for hybrid and electric vehicles are based on SOC knowledge. The
batteries behavior can be modeled with different electrical circuit structures and different
linear or nonlinear mathematical models. The different estimation methods can be used for
the batteries parameters determination. The method based on continue time model and LIF
algorithm assures a good accuracy for parameters estimation. The estimated parameters can
be used for the battery SOC determination. — The control strategies for hybrid and electric vehicles are based on SOC knowledge. The batteries behavior can be modeled with different electrical circuit structures and different linear or nonlinear mathematical models. The different estimation methods can be used for the batteries parameters determination. The method based on continue time model and LIF algorithm assures a good accuracy for parameters estimation. The estimated parameters can be used for the battery SOC determination.

Fig. 5. a. Characteristics of traction motors ; b. Tractive effort characteristics of an ICE
vehicle
The electric motors can operate in two modes: a) as motor which convert electrical
energy taken from a source (electric generator, battery, fuel cell) into mechanical energy
used to propel the vehicle, b) as generator which convert the mechanical energy taken
from a motor (ICE, the wheels during vehicle braking, etc..) in electrical energy used for
charging the battery. The motors are the only propulsion system for electric vehicles.
Hybrid electric vehicles have two propulsion systems: ICE and electric motor, which can
be used in different configurations: serial, parallel, mixed. Compared with ICE electric
motors has some important advantages: they produce large amounts couples at low
speeds, the instantaneous power values exceed 2-3 times the rated ICE, torque values are
easily reproducible, adjustment speed limits are higher. With these characteristics ensure
good dynamic performance: large accelerations and small time both at startup and
braking. — Fig. 5. a. Characteristics of traction motors ; b. Tractive effort characteristics of an ICE vehicle The electric motors can operate in two modes: a) as motor which convert electrical energy taken from a source (electric generator, battery, fuel cell) into mechanical energy used to propel the vehicle, b) as generator which convert the mechanical energy taken from a motor (ICE, the wheels during vehicle braking, etc..) in electrical energy used for charging the battery. The motors are the only propulsion system for electric vehicles. Hybrid electric vehicles have two propulsion systems: ICE and electric motor, which can be used in different configurations: serial, parallel, mixed. Compared with ICE electric motors has some important advantages: they produce large amounts couples at low speeds, the instantaneous power values exceed 2-3 times the rated ICE, torque values are easily reproducible, adjustment speed limits are higher. With these characteristics ensure good dynamic performance: large accelerations and small time both at startup and braking.

The steady state characteristics of induction machines can be derived from its equivalent
circuit. In order to develop a per phase equivalent circuit of a three-phase machine, a wound
rotor motor as shown in Figure 6.a. is considered here. In case of a squirrel cage motor, the
rotor circuit can be replaced by an equivalent three-phase winding. When three-phase
balanced voltages are applied to the stator, the currents flow in them. The equivalent circuit,
therefore is identical to that of a transformer, and is shown in Figure 6. b. Here R, is the
stator winding resistance, L; is self inductance of stator, L, is self inductance of rotor
winding referred to stator, R, is rotor resistance referred to stator, Ly, is magnetizing
inductance and s is the slip. The parameters of the equivalent circuit are the stator and rotor
leakage reactances X,, and X,,, magnetizing reactance X;,, and the equivalent resistance

1l-s

or’

oo 7 ixvlaa als. Ama Am bo Blok
end or rotor by end rings. In normal running there 1s no ailrerence between a cage type or
wound rotor machine as for as there electrical characteristics are concerned.

When the stator is energized from a three phase supply a rotating magnetic field is
produced in the air gap. The magnetic flux from this field induces voltages in both the stator
and rotor windings. The electromagnetic torque resulting from the interaction of the
currents in the rotor circuit (since it is shorted) and the air gap flux, results in rotation of
rotor. Since electromotive force in the rotor can be induced only when there is a relative
motion between air gap field and rotor, the rotor rotates in the same direction as the
magnetic field, but it will not run at synchronous speed. An induction motor therefore
always runs at a speed less than synchronous speed. The difference between rotor speed
and synchronous speed is known as slip. The slip s is given by — The steady state characteristics of induction machines can be derived from its equivalent circuit. In order to develop a per phase equivalent circuit of a three-phase machine, a wound rotor motor as shown in Figure 6.a. is considered here. In case of a squirrel cage motor, the rotor circuit can be replaced by an equivalent three-phase winding. When three-phase balanced voltages are applied to the stator, the currents flow in them. The equivalent circuit, therefore is identical to that of a transformer, and is shown in Figure 6. b. Here R, is the stator winding resistance, L; is self inductance of stator, L, is self inductance of rotor winding referred to stator, R, is rotor resistance referred to stator, Ly, is magnetizing inductance and s is the slip. The parameters of the equivalent circuit are the stator and rotor leakage reactances X,, and X,,, magnetizing reactance X;,, and the equivalent resistance 1l-s or’ oo 7 ixvlaa als. Ama Am bo Blok end or rotor by end rings. In normal running there 1s no ailrerence between a cage type or wound rotor machine as for as there electrical characteristics are concerned. When the stator is energized from a three phase supply a rotating magnetic field is produced in the air gap. The magnetic flux from this field induces voltages in both the stator and rotor windings. The electromagnetic torque resulting from the interaction of the currents in the rotor circuit (since it is shorted) and the air gap flux, results in rotation of rotor. Since electromotive force in the rotor can be induced only when there is a relative motion between air gap field and rotor, the rotor rotates in the same direction as the magnetic field, but it will not run at synchronous speed. An induction motor therefore always runs at a speed less than synchronous speed. The difference between rotor speed and synchronous speed is known as slip. The slip s is given by

The dynamic model of ac machine can be developed [Ehsani, 2005], [Husain, 2003], using
the concept of “space vectors”. Space vectors of three-phase variables, such as the voltage,
current, or flux, are very convenient for the analysis and control of ac motors and power
converter. A three-phase system defined by ya(t), ya(t), and yc(t) can be represented uniquely
by a rotating vector y(t) in the complex plane. — The dynamic model of ac machine can be developed [Ehsani, 2005], [Husain, 2003], using the concept of “space vectors”. Space vectors of three-phase variables, such as the voltage, current, or flux, are very convenient for the analysis and control of ac motors and power converter. A three-phase system defined by ya(t), ya(t), and yc(t) can be represented uniquely by a rotating vector y(t) in the complex plane.

active switching state vectors Uz... Us and two zero vectors Up and U7 as shown in Figure
8.b. — active switching state vectors Uz... Us and two zero vectors Up and U7 as shown in Figure 8.b.

Fig. 8. a. Switched three-phase waveforms ; b. Switching state vectors
Electric Vehicles - Modelling and Simulations

5. Control strategies — Fig. 8. a. Switched three-phase waveforms ; b. Switching state vectors Electric Vehicles - Modelling and Simulations 5. Control strategies

Fig. 9. a. Torque-speed characeristics under V/f control; b. VSI induction motor drive V/f
controlled
When frequency is increased, the flux will decrease, and the torque developed by the motor
will decrease as shown in Figure 9.a. When frequency is decreased, the flux will increase
and may lead to the saturation of magnetic circuit. Since in PWM inverters the voltage and
frequency can be controlled independently, these drives are fed from a PWM inverter.

The control scheme is simple as shown in Figure 9.b with motor being supplied by three-
phase supply dc-link and PWM inverter. — Fig. 9. a. Torque-speed characeristics under V/f control; b. VSI induction motor drive V/f controlled When frequency is increased, the flux will decrease, and the torque developed by the motor will decrease as shown in Figure 9.a. When frequency is decreased, the flux will increase and may lead to the saturation of magnetic circuit. Since in PWM inverters the voltage and frequency can be controlled independently, these drives are fed from a PWM inverter. The control scheme is simple as shown in Figure 9.b with motor being supplied by three- phase supply dc-link and PWM inverter.

Fig. 10. Block diagram of the rotor flux oriented control of a VSI induction machine drive

The structure of the experimental model of the hybrid vehicle is presented in Figure 11. The
model includes the two power propulsion (ICE, and the electric motor/generator M/G)
with allow the energetically optimization by implementing the real time control algorithms.
The model has no wheels and the longitudinal characteristics emulation is realized with a
corresponding load system. The ICE is a diesel F8Q of 1.91 capacity and 64[HP]. The
electronic unit control (ECU) is a Lucas DCN R04080012J-80759M. The coupling with the
motor/ generator system is assured by a clutch, a gearbox and a belt transmission.
The electric machine is a squirrel cage asynchronous machine (15kW, 380V, 30.5A, 50H:
2940 rpm) supplied by a PWM inverter implemented with IGBT modules (6KM200GB122D
The motor is supplied by 26 batteries (12V/45Ah). The hardware structure of th
motor/ generator system is presented in Figure 12. The hardware resources assured by th
control system eZdsp 2808 permit the implementation of the local dynamic contr
algorithms and for a CAN communication network, necessary for the distributed contre
used on the hybrid electric vehicle, [Livint et all 2008, 2010]

With the peripheral elements (8 ePWM channels, 2x8 AD channels with a resolution of 1
bits, incremental transducer interface eQEP) and the specific peripheral for th — The structure of the experimental model of the hybrid vehicle is presented in Figure 11. The model includes the two power propulsion (ICE, and the electric motor/generator M/G) with allow the energetically optimization by implementing the real time control algorithms. The model has no wheels and the longitudinal characteristics emulation is realized with a corresponding load system. The ICE is a diesel F8Q of 1.91 capacity and 64[HP]. The electronic unit control (ECU) is a Lucas DCN R04080012J-80759M. The coupling with the motor/ generator system is assured by a clutch, a gearbox and a belt transmission. The electric machine is a squirrel cage asynchronous machine (15kW, 380V, 30.5A, 50H: 2940 rpm) supplied by a PWM inverter implemented with IGBT modules (6KM200GB122D The motor is supplied by 26 batteries (12V/45Ah). The hardware structure of th motor/ generator system is presented in Figure 12. The hardware resources assured by th control system eZdsp 2808 permit the implementation of the local dynamic contr algorithms and for a CAN communication network, necessary for the distributed contre used on the hybrid electric vehicle, [Livint et all 2008, 2010] With the peripheral elements (8 ePWM channels, 2x8 AD channels with a resolution of 1 bits, incremental transducer interface eQEP) and the specific peripheral for th

Fig. 13. Emulation system of the longitudinal dynamics characteristics of the vehicle

Fig. 14. The functional structure of a CANopen slave

Fig. 15. The Simulink model assigned to the slave CANopen communication node

Fig. 19. The estimated torque from mechanical load emulator
Fig. 18. a) The active current from electric traction motor b) The reference torque for
Sinamics system — Fig. 19. The estimated torque from mechanical load emulator Fig. 18. a) The active current from electric traction motor b) The reference torque for Sinamics system

2.1.2 Design of VSC ASR controller
The meanings of the parameters can be found in the literature !.
VSC with sliding mode has good robustness to the input signals so that this strategy has
advantage to the ASR control system which needs the vehicle velocity observation and
signal identification. But there is always high-frequency-chattering on the sliding surface. In
the following text a VSC controller, which doesn’t depend on the identification of the
optimal slip ratio, is designed and its performance will be analyzed through simulation.

In order to make the VSC possess excellent robustness to the additional uncertainties and
interferences, the control law adopted here is equivalent control with switching control.
Hence, the output torque of the e-motors can be expressed as [2!: — 2.1.2 Design of VSC ASR controller The meanings of the parameters can be found in the literature !. VSC with sliding mode has good robustness to the input signals so that this strategy has advantage to the ASR control system which needs the vehicle velocity observation and signal identification. But there is always high-frequency-chattering on the sliding surface. In the following text a VSC controller, which doesn’t depend on the identification of the optimal slip ratio, is designed and its performance will be analyzed through simulation. In order to make the VSC possess excellent robustness to the additional uncertainties and interferences, the control law adopted here is equivalent control with switching control. Hence, the output torque of the e-motors can be expressed as [2!:

Vehicle Dynamic Control of 4 In-Wheel-Motor Drived Electric Vehicle

During driving process, the slip ratio of the wheel can be expressed as: — Vehicle Dynamic Control of 4 In-Wheel-Motor Drived Electric Vehicle During driving process, the slip ratio of the wheel can be expressed as:

Fig. 2.1-3. Slip ratio-Longitudinal adhesion coefficient on different road surface

From the simulation results we can get that the vehicle can keep away from skipping and
the acceleration performance is good when it starts. But the slip ratio occurs fluctuation
when it’s among 0 to 0.3 and the e-motor’s output torque also fluctuates near 300Nm. In
reality, big fluctuation is harmful to the e-motor and sometimes the e-motor can’t fully
realize what the controller requires. Therefore, there are some defects in this method. — From the simulation results we can get that the vehicle can keep away from skipping and the acceleration performance is good when it starts. But the slip ratio occurs fluctuation when it’s among 0 to 0.3 and the e-motor’s output torque also fluctuates near 300Nm. In reality, big fluctuation is harmful to the e-motor and sometimes the e-motor can’t fully realize what the controller requires. Therefore, there are some defects in this method.

The standard model of MFC is got under the condition that the slip ratio is set to 0. It means
that the road’s adhesion force isn’t fully utilized and the driving performance will be bad. So
this control strategy is not perfect. Secondly, MFC hasn’t good robustness to the input
signals. Especially when the angular speed is disturbed, deviation of the controller will
happen.
According to the research results from Tokyo University 5], when the tire is completely
adhered, the vehicle’s equivalent mass is equal to the sum of the sprung mass and non-
sprung mass. When the tire slips, the angular speed changes significantly. During
acceleration, the angular speed is obviously smaller than the ideal value which is outputed
by the standard model. In light of this principal, the tire’s angular speed should be
compared to the angular speed from the standard model at any time. And then the
difference is used as the basis for a correction value through a simple proportional control to
adjust the e-motor’s output torque. So that the tire can avoid slipping.

MFC strategy only requires the e-motor’s output driving torque and the tire’s angular speed
signal to put ASR into practice. Consequently the estimation of the longitudinal velocity and
the optimal slip ratio identification can be ignored. Therefore, this strategy is practical. The

te ce I a eB Ne ee ee ek — The standard model of MFC is got under the condition that the slip ratio is set to 0. It means that the road’s adhesion force isn’t fully utilized and the driving performance will be bad. So this control strategy is not perfect. Secondly, MFC hasn’t good robustness to the input signals. Especially when the angular speed is disturbed, deviation of the controller will happen. According to the research results from Tokyo University 5], when the tire is completely adhered, the vehicle’s equivalent mass is equal to the sum of the sprung mass and non- sprung mass. When the tire slips, the angular speed changes significantly. During acceleration, the angular speed is obviously smaller than the ideal value which is outputed by the standard model. In light of this principal, the tire’s angular speed should be compared to the angular speed from the standard model at any time. And then the difference is used as the basis for a correction value through a simple proportional control to adjust the e-motor’s output torque. So that the tire can avoid slipping. MFC strategy only requires the e-motor’s output driving torque and the tire’s angular speed signal to put ASR into practice. Consequently the estimation of the longitudinal velocity and the optimal slip ratio identification can be ignored. Therefore, this strategy is practical. The te ce I a eB Ne ee ee ek

From Fig.2.2-4, we can see that, on a low adhesion coefficient road surface, the vehic!
doesn’t slip. The slip ratio is in the ideal scope. Comparing with the above mentioned VS!
strategy, the fluctuation of the slip ratio for combined control is improved. The fluctuatio
time continues 2.5s before stable convergence range occurs and the peak of the fluctuation ¢
the slip ratio is 0.5. With the work of the combined control strategy the fluctuation scope |
narrowed and the same to the e-motor’s output torque. The drive performance for th
combined control strategy is also excellent. On the low adhesion surface, the longitudin.
velocity can reach 17m/s after 10s from starting.

lable 1 shows the driving performance for different control methods on road surface wit
low adhesion coefficient (1=0.2). — From Fig.2.2-4, we can see that, on a low adhesion coefficient road surface, the vehic! doesn’t slip. The slip ratio is in the ideal scope. Comparing with the above mentioned VS! strategy, the fluctuation of the slip ratio for combined control is improved. The fluctuatio time continues 2.5s before stable convergence range occurs and the peak of the fluctuation ¢ the slip ratio is 0.5. With the work of the combined control strategy the fluctuation scope | narrowed and the same to the e-motor’s output torque. The drive performance for th combined control strategy is also excellent. On the low adhesion surface, the longitudin. velocity can reach 17m/s after 10s from starting. lable 1 shows the driving performance for different control methods on road surface wit low adhesion coefficient (1=0.2).

Fig. 2.2-5. Start on the jump p road (from p=0.2 top=0.7 )

Fig.2.2-6 shows the simulation results with MFC strategy which is on the low adhesion
coefficient road surface. In this simulation test, the wheel speed is disturbed that is manually
offset by white noise(0.1kw) in order to verify the effectiveness to the disturb of the velocity

signal. — Fig.2.2-6 shows the simulation results with MFC strategy which is on the low adhesion coefficient road surface. In this simulation test, the wheel speed is disturbed that is manually offset by white noise(0.1kw) in order to verify the effectiveness to the disturb of the velocity signal.

Where V is the vehicle longitudinal velocity and Vw is the wheel velocity. Vw=Rw, where R,
w are the wheel radius and angular velocity respectively. — Where V is the vehicle longitudinal velocity and Vw is the wheel velocity. Vw=Rw, where R, w are the wheel radius and angular velocity respectively.

Fig. 3.1-2. Simulation Result of the Hybrid-ABS with the wheel velocity as the control
parameter
represents the low adhesive road. The top output torque of the electric motor is 136Nm and
the delay time due to the physical characteristic of the electric motor 5 ms.

Fig. 3.1-2 shows the simulation result using the wheel velocity Vw as the control parameter.
The braking distance is apparently decreased. The slip ratio is restrained under 20%. The
unexpected increased amplitude of the slip ratio is mainly due to the delay of the electric
motor’s output, which can be proved in Fig. 3.1-2 (b). This can cause contradiction in the
braking process. Fig. 3.1-2 (c) shows longitudinal vehicle velocity and wheel velocity under
this control parameter. — Fig. 3.1-2. Simulation Result of the Hybrid-ABS with the wheel velocity as the control parameter represents the low adhesive road. The top output torque of the electric motor is 136Nm and the delay time due to the physical characteristic of the electric motor 5 ms. Fig. 3.1-2 shows the simulation result using the wheel velocity Vw as the control parameter. The braking distance is apparently decreased. The slip ratio is restrained under 20%. The unexpected increased amplitude of the slip ratio is mainly due to the delay of the electric motor’s output, which can be proved in Fig. 3.1-2 (b). This can cause contradiction in the braking process. Fig. 3.1-2 (c) shows longitudinal vehicle velocity and wheel velocity under this control parameter.

In practical engineering applications, the chattering may appear when sign function is used.
Therefore the Saturation function ‘sat ()’ is used to substitute for sign function. — In practical engineering applications, the chattering may appear when sign function is used. Therefore the Saturation function ‘sat ()’ is used to substitute for sign function.

Fig. 3.3-4. Simulation results on the road with yz =0.9

Fig. 3.3-5. Simulation results on the road with wu = 0.2
Vehicle Dynamic Control of 4 In-Wheel-Motor Drived Electric Vehicle — Fig. 3.3-5. Simulation results on the road with wu = 0.2 Vehicle Dynamic Control of 4 In-Wheel-Motor Drived Electric Vehicle

Fig. 3.3-6. The road adhesion coefficient changes from y= 0.2 to w= 0.9 at the 1st second

Table 3. Braking distance and braking time on the different road

Fig. 4.1-1. Vehicle dynamic control structure

Fig. 4.2-2. Arc-tangent function vs. magic model

Here, m represents the mass of the vehicle, J, represents the yaw rotational inertia of the
vehicle, c,; and cy are the fitting parameters for the front wheel, c,; and c,, are the fitting
parameters for the rear wheel, |; is the distance from the gravity point to the front axle and
I, is the distance from the gravity point to the rear axle, V is the gravity point velocity of the
vehicle, 6; is the steering angle for the front wheel.

Based on non-linear model mention above, we can design yaw-rate follow controller. In our
case. the dvnamic function of vaw rate is second-order svstem:
Vehicle Dynamic Control of 4 In-Wheel-Motor Drived Electric Vehicle — Here, m represents the mass of the vehicle, J, represents the yaw rotational inertia of the vehicle, c,; and cy are the fitting parameters for the front wheel, c,; and c,, are the fitting parameters for the rear wheel, |; is the distance from the gravity point to the front axle and I, is the distance from the gravity point to the rear axle, V is the gravity point velocity of the vehicle, 6; is the steering angle for the front wheel. Based on non-linear model mention above, we can design yaw-rate follow controller. In our case. the dvnamic function of vaw rate is second-order svstem: Vehicle Dynamic Control of 4 In-Wheel-Motor Drived Electric Vehicle

Vehicle Dynamic Control of 4 In-Wheel-Motor Drived Electric Vehicle

Fig. 4.4-3 presents the behaviors of several state values of the vehicle during such operation.
Among them the yaw rate response can match the desired value well. Supposing on level
and smooth road, when the peak value of the lateral acceleration is close to 1.0g, the vehicle
has been working under critical condition. The 4~ 4 phase trajectory indicates that the
vehicle can keep steady even when the slip angle reaches 8 degree.
Vehicle Dynamic Control of 4 In-Wheel-Motor Drived Electric Vehicle — Fig. 4.4-3 presents the behaviors of several state values of the vehicle during such operation. Among them the yaw rate response can match the desired value well. Supposing on level and smooth road, when the peak value of the lateral acceleration is close to 1.0g, the vehicle has been working under critical condition. The 4~ 4 phase trajectory indicates that the vehicle can keep steady even when the slip angle reaches 8 degree. Vehicle Dynamic Control of 4 In-Wheel-Motor Drived Electric Vehicle

Fig. 4.4-4. Estimated Cornering Stiffness of Tire

Figure shows the adjustment of switch function value during operation; there is obvious
chattering when simulation time is near 12s and 14s, but that causes no severe fluctuate to
general yaw moment. Figure illustrates the actual yaw moment can realize the general
control requirement basically, which guarantees the achievement of motion follow control.
Fig. 4.4-4 shows the estimated values of the cornering stiffness in the double lane change
simulation, the vehicle lateral motion characteristic adjusts acutely during lane change. If
the LOR controller were designed according to the fixed value of the cornering stiffness, the
control effect would get worse along with the fluctuation of the cornering stiffness. Fig. 4.4-5
illustrates how the actual yaw moment follows the requirement of the control during the
whole control process. It’s clear that the optimize allocation algorithm can finely meet the
requirement of the stability control even under the critical condition. — Figure shows the adjustment of switch function value during operation; there is obvious chattering when simulation time is near 12s and 14s, but that causes no severe fluctuate to general yaw moment. Figure illustrates the actual yaw moment can realize the general control requirement basically, which guarantees the achievement of motion follow control. Fig. 4.4-4 shows the estimated values of the cornering stiffness in the double lane change simulation, the vehicle lateral motion characteristic adjusts acutely during lane change. If the LOR controller were designed according to the fixed value of the cornering stiffness, the control effect would get worse along with the fluctuation of the cornering stiffness. Fig. 4.4-5 illustrates how the actual yaw moment follows the requirement of the control during the whole control process. It’s clear that the optimize allocation algorithm can finely meet the requirement of the stability control even under the critical condition.

Generally, the nonlinear interrelationships between the slip ratio 2 and friction coefficient
u formed by tire’s dynamics can be modeled by the widely adopted Magic Formula
(Pacejka & Bakker, 1992) as shown in Fig. 2. — Generally, the nonlinear interrelationships between the slip ratio 2 and friction coefficient u formed by tire’s dynamics can be modeled by the widely adopted Magic Formula (Pacejka & Bakker, 1992) as shown in Fig. 2.

Electric Vehicles - Modelling and Simulations

The concept of MTTE approach for vehicle anti-slip control is firstly proposed in (Yin et al.,
2009). The MITE approach can achieve an acceptable anti-slip control performance under
common operation requirements. However, the MTTE approach is sensitive to the varying
of the wheel inertia. If the wheel inertia varies, the anti-slip performance of the MTTE will
deteriorate gradually. This paper is devoted to improve the anti-slip performance of the
MTTE approach under such concerned abnormal operations. An advanced MTTE approach
with fault-tolerant performance is then proposed. Based on the MITE approaches, the
following considerations are concerned.

1. No matter what kind of tire-road condition the vehicle is driven on, the kinematic

tn ata Ae nein meters FAA mizrlacel amd baa eh aceite Jeo a lsizvacre Fat Sas lea, — Electric Vehicles - Modelling and Simulations The concept of MTTE approach for vehicle anti-slip control is firstly proposed in (Yin et al., 2009). The MITE approach can achieve an acceptable anti-slip control performance under common operation requirements. However, the MTTE approach is sensitive to the varying of the wheel inertia. If the wheel inertia varies, the anti-slip performance of the MTTE will deteriorate gradually. This paper is devoted to improve the anti-slip performance of the MTTE approach under such concerned abnormal operations. An advanced MTTE approach with fault-tolerant performance is then proposed. Based on the MITE approaches, the following considerations are concerned. 1. No matter what kind of tire-road condition the vehicle is driven on, the kinematic tn ata Ae nein meters FAA mizrlacel amd baa eh aceite Jeo a lsizvacre Fat Sas lea,

Fig. 5. Disturbance estimation based on closed-loop observer.
Fig. 4. Disturbance estimation based on open-loop observer. — Fig. 5. Disturbance estimation based on closed-loop observer. Fig. 4. Disturbance estimation based on open-loop observer.

A Robust Traction Control for Electric Vehicles Without Chassis Velocity

Obviously, from Eq. (11), the anti-slip performance of MTTE will be enhanced when AM is a
positive value and reduced when AM is a negative value. Additionally, in common vehicles,
the MITE approach is insensitive to the varying of Mn. Since passenger and driving
resistance are the primary perturbations of M,, the MTTE approach reveals its merits for
general driving environments. The fact shows that the MTTE control scheme is robust to the
varying of the vehicle mass M.
Model uncertainty and sensor fault are the main faults concerned in this study. Since the
conventional MTTE approach is based on the open-loop disturbance estimation, the system
is hence sensitive to the varying of wheel inertia. If the tires are getting flat, the anti-slip
performance of MTTE will deteriorate gradually. Figure 8 illustrates the advanced MTTE
scheme which endows the MTTE with fault-tolerant performance. The disturbance torque — Obviously, from Eq. (11), the anti-slip performance of MTTE will be enhanced when AM is a positive value and reduced when AM is a negative value. Additionally, in common vehicles, the MITE approach is insensitive to the varying of Mn. Since passenger and driving resistance are the primary perturbations of M,, the MTTE approach reveals its merits for general driving environments. The fact shows that the MTTE control scheme is robust to the varying of the vehicle mass M. Model uncertainty and sensor fault are the main faults concerned in this study. Since the conventional MTTE approach is based on the open-loop disturbance estimation, the system is hence sensitive to the varying of wheel inertia. If the tires are getting flat, the anti-slip performance of MTTE will deteriorate gradually. Figure 8 illustrates the advanced MTTE scheme which endows the MTTE with fault-tolerant performance. The disturbance torque

Fig. 8. Advanced MTTE control system.
Tg comes from the operation friction. When the vehicle is operated on a slippery road, it
causes the Ty to become very small, and due to that the tires cannot provide sufficient
friction. Skidding often happens in braking and racing of an operated vehicle when the tire’s
adhesion cannot firmly grip the surface of the road. This phenomenon is often referred as
the magic formula (i.e., the 4 - / relation). However, the uw - 4 relation is immeasurable in
real time. Therefore, in the advanced MTTE, the nonlinear behavior between the tire and
road (i.e., the magic formula) is regarded as an uncertain source which deteriorates the
steering stability and causes some abnormal malfunction in deriving. — Fig. 8. Advanced MTTE control system. Tg comes from the operation friction. When the vehicle is operated on a slippery road, it causes the Ty to become very small, and due to that the tires cannot provide sufficient friction. Skidding often happens in braking and racing of an operated vehicle when the tire’s adhesion cannot firmly grip the surface of the road. This phenomenon is often referred as the magic formula (i.e., the 4 - / relation). However, the uw - 4 relation is immeasurable in real time. Therefore, in the advanced MTTE, the nonlinear behavior between the tire and road (i.e., the magic formula) is regarded as an uncertain source which deteriorates the steering stability and causes some abnormal malfunction in deriving.

A Robust Traction Control for Electric Vehicles Without Chassis Velocity

Now consider the affection of model uncertainty AJ,, to wheel inertia J,,. It yields
Tw = Nw + Jn - Since the mass of vehicle is larger than the wheels, in most of the commercial
vehicle, a@Mr* > AJ,,+ Ju, is always held. Especially, the mass of passengers can also
increase M to convince the condition of Eq. (18). Since the varying of J,, caused by AJ,,
cannot affect the anti-slip control system so much. This means that the proposed control — A Robust Traction Control for Electric Vehicles Without Chassis Velocity Now consider the affection of model uncertainty AJ,, to wheel inertia J,,. It yields Tw = Nw + Jn - Since the mass of vehicle is larger than the wheels, in most of the commercial vehicle, a@Mr* > AJ,,+ Ju, is always held. Especially, the mass of passengers can also increase M to convince the condition of Eq. (18). Since the varying of J,, caused by AJ,, cannot affect the anti-slip control system so much. This means that the proposed control

In order to implement and evaluate the proposed control system, a commercial electric
vehicle, COMS3, which is assembled by TOYOTA Auto Body Co. Ltd., shown in Fig. 11 was
modified to carry out the experiments’ requirements. As illustrated in Fig. 12, a control
computer is embedded to take the place of the previous Electronic Control Unit (ECU) to
operate the motion control. The corresponding calculated torque reference of the left and the
right rear wheel are independently sent to the inverter by two analog signal lines. Table 1
lists the main specifications. — In order to implement and evaluate the proposed control system, a commercial electric vehicle, COMS3, which is assembled by TOYOTA Auto Body Co. Ltd., shown in Fig. 11 was modified to carry out the experiments’ requirements. As illustrated in Fig. 12, a control computer is embedded to take the place of the previous Electronic Control Unit (ECU) to operate the motion control. The corresponding calculated torque reference of the left and the right rear wheel are independently sent to the inverter by two analog signal lines. Table 1 lists the main specifications.

Fig. 11. Experimental electric vehicle and setting of slippery road for experiment.
Table 1. Specification of COMS3. — Fig. 11. Experimental electric vehicle and setting of slippery road for experiment. Table 1. Specification of COMS3.

. the experiments, the relation factor of MTTE scheme is set as a = 0.9. The time constant
f LPFs in the comparison experiment are set as 1, =7, =0.05. It is known that thi
assenger’s weight varies. Hence, this paper adopts the PI compensator as the kernel o
isturbance estimation. The PI gains are set as K, =70, and K; =60. As shown in Fig. 11
1e slippery road was set by an acrylic sheet with a length of 1.2m and lubricated wit!
ater. The initial velocity of the vehicle was set higher than 1m/s to avoid th
nmeasurable zone of the shaft sensors installed in the wheels. The driving torque delay i
1e advanced MTTE approach is exploited to adjust the phase of the estimated disturbance
nder a proper anti-slip control, the wheel velocity should be as closed to the chassi
elocity as possible. As can be seen in Fig. 13, the advanced MTTE cannot achieve any anti
ip performance (i.e. the vehicle is skidded) if the reference signal is no delayed. Figure 1
iso shows the measured results, and obviously, the digital delay of motor driver ha
gnificant affections to the advanced MTTE. According to the practical tests of Fig. 13, wit!
roper command delay of 20ms, the advanced MTTE can achieve a feasible performance
lence, in the following, all experiments to the advanced MTTE utilize this delay parameter.
The MTTE-based schemes can prevent vehicle skid. These approaches compensate th
reference torque into a limited value when encountering a slippery road. Based on th
experimental result of Fig. 14, the reference torque of MTTE-based approaches i
constrained without divergence. Figure 14 is evaluated under the nominal wheel inertia. A
can be seen in this figure, both the conventional MTTE and advanced MTTE are with goox
anti-slip performance. Nevertheless, as indicated in the practical results in Fig. 15, the anti
slip performance of MTTE impairs with the varying of wheel inertia. In addition, Fig. 1
shows the same testing on the advanced MTTE. Apparently, the advanced MTTE overcome
this problem. The advanced MTTE has fault-tolerant anti-slip performance against th — . the experiments, the relation factor of MTTE scheme is set as a = 0.9. The time constant f LPFs in the comparison experiment are set as 1, =7, =0.05. It is known that thi assenger’s weight varies. Hence, this paper adopts the PI compensator as the kernel o isturbance estimation. The PI gains are set as K, =70, and K; =60. As shown in Fig. 11 1e slippery road was set by an acrylic sheet with a length of 1.2m and lubricated wit! ater. The initial velocity of the vehicle was set higher than 1m/s to avoid th nmeasurable zone of the shaft sensors installed in the wheels. The driving torque delay i 1e advanced MTTE approach is exploited to adjust the phase of the estimated disturbance nder a proper anti-slip control, the wheel velocity should be as closed to the chassi elocity as possible. As can be seen in Fig. 13, the advanced MTTE cannot achieve any anti ip performance (i.e. the vehicle is skidded) if the reference signal is no delayed. Figure 1 iso shows the measured results, and obviously, the digital delay of motor driver ha gnificant affections to the advanced MTTE. According to the practical tests of Fig. 13, wit! roper command delay of 20ms, the advanced MTTE can achieve a feasible performance lence, in the following, all experiments to the advanced MTTE utilize this delay parameter. The MTTE-based schemes can prevent vehicle skid. These approaches compensate th reference torque into a limited value when encountering a slippery road. Based on th experimental result of Fig. 14, the reference torque of MTTE-based approaches i constrained without divergence. Figure 14 is evaluated under the nominal wheel inertia. A can be seen in this figure, both the conventional MTTE and advanced MTTE are with goox anti-slip performance. Nevertheless, as indicated in the practical results in Fig. 15, the anti slip performance of MTTE impairs with the varying of wheel inertia. In addition, Fig. 1 shows the same testing on the advanced MTTE. Apparently, the advanced MTTE overcome this problem. The advanced MTTE has fault-tolerant anti-slip performance against th

Fig. 14. Practical comparisons between MTTE and advanced MTTE to nominal Jw.
electric vehicles can be further enhanced.
wheel inertia varying in real time. Figures 17 and 18 show the performance tests of MTTE
and advanced MTTE against different vehicle mass. It is no doubt that the MTTE-based
control schemes are robust in spite of different passengers setting in the vehicle. From
experimental evidences, it is evident that the advanced MTTE traction control approach has
consistent performance to the varying of wheel inertia J,, and vehicle mass M. As shown in
these figures, the proposed anti-slip system offers an effective performance in maintaining
the driving stability under more common situations, and therefore the steering safety of the
electric vehicles can be further enhanced. — Fig. 14. Practical comparisons between MTTE and advanced MTTE to nominal Jw. electric vehicles can be further enhanced. wheel inertia varying in real time. Figures 17 and 18 show the performance tests of MTTE and advanced MTTE against different vehicle mass. It is no doubt that the MTTE-based control schemes are robust in spite of different passengers setting in the vehicle. From experimental evidences, it is evident that the advanced MTTE traction control approach has consistent performance to the varying of wheel inertia J,, and vehicle mass M. As shown in these figures, the proposed anti-slip system offers an effective performance in maintaining the driving stability under more common situations, and therefore the steering safety of the electric vehicles can be further enhanced.

Fig. 15. Experimental results of MTTE to different Jw.

Fig. 18. Experimental results of advanced MTTE to different M
Fig. 17. Experimental results of MTTE to different M. — Fig. 18. Experimental results of advanced MTTE to different M Fig. 17. Experimental results of MTTE to different M.

Fig. 1. Configuration of the traction system proposed
The proposed traction system is an electric vehicle with two drives, Fig.1. Two machines
thus replace the standard case with a single machine and a differential mechanical. The
power structure in this paper is composed of two permanent magnet synchronous motors
which are supplied by two three-phase inverters and driving the two rear wheels of a
vehicle through gearboxes, Fig. 2. The traction system gives different torque to in-wheel-
motor, in order to improve vehicle stability. However, the control method used in this
chapter for the motors is the fuzzy direct torque control with will give the vehicle a dynamic
behaviour similar to that imposed by a mechanical differential (Arnet, 1997; Hartani, 2005;
Hartani, 2007). The objective of the structure is to reproduce at least the behaviour of a
mechanical differential by adding to it a safety anti skid function. — Fig. 1. Configuration of the traction system proposed The proposed traction system is an electric vehicle with two drives, Fig.1. Two machines thus replace the standard case with a single machine and a differential mechanical. The power structure in this paper is composed of two permanent magnet synchronous motors which are supplied by two three-phase inverters and driving the two rear wheels of a vehicle through gearboxes, Fig. 2. The traction system gives different torque to in-wheel- motor, in order to improve vehicle stability. However, the control method used in this chapter for the motors is the fuzzy direct torque control with will give the vehicle a dynamic behaviour similar to that imposed by a mechanical differential (Arnet, 1997; Hartani, 2005; Hartani, 2007). The objective of the structure is to reproduce at least the behaviour of a mechanical differential by adding to it a safety anti skid function.

Fig. 3. Components of the traction system proposed

The PMSM model can be described in the stator reference frame as follows (Pragasen, 1989):

Vehicle Stability Enhancement Control for Electric Vehicle Using Behaviour Model Control

2.4.2 Modeling of the transmission gearbox-wheel
Modeling of the transmission gearbox-wheel is carried out in a classical way which is given by: — Vehicle Stability Enhancement Control for Electric Vehicle Using Behaviour Model Control 2.4.2 Modeling of the transmission gearbox-wheel Modeling of the transmission gearbox-wheel is carried out in a classical way which is given by:

Fig. 7. Detailed EMR of the mechanical coupling

Fig. 10. Membership functions for the FLC2
Fig. 11. Fuzzy logic controller structure — Fig. 10. Membership functions for the FLC2 Fig. 11. Fuzzy logic controller structure

Fig. 12. System diagram of a PMSM-fuzzy DTC drive system with stator PI resistance
estimator — Fig. 12. System diagram of a PMSM-fuzzy DTC drive system with stator PI resistance estimator

Fig. 13. PI resistance estimator for FDTC-PMSM drive system

A Matlab/Simulink model was built in order to verify the performance of the PI resistance
estimator. The system performances with and without the resistance estimator were then
compared. Fig. 14 shows the result of flux, torque, and amplitude of current of the system
with PI stator resistance estimator. — A Matlab/Simulink model was built in order to verify the performance of the PI resistance estimator. The system performances with and without the resistance estimator were then compared. Fig. 14 shows the result of flux, torque, and amplitude of current of the system with PI stator resistance estimator.

Fig. 16. Modulation of the traction effort: (a) COG and (b) EMR.

tig. 18. Inversion of the converter CR: (a) COG; (b) EMR
Fig. 17. Control structure deduced from the inversion — tig. 18. Inversion of the converter CR: (a) COG; (b) EMR Fig. 17. Control structure deduced from the inversion

Vehicle Stability Enhancement Control for Electric Vehicle Using Behaviour Model Control

Electric Vehicles - Modelling and Simulations

3.5.2 Application of the BMC control to the traction system
The first step to be made is to establish a behaviour model. In this case, we choose a
mechanical model without slip, which will be equivalent to the contact wheel-road in the
areas known as pseudo-slip. This model can be considered as an ideal model. However, the
inertia moments of the elements in rotation and the vehicle mass can be represented by the
total inertia moments Jp nog of each shaft motor which is given by: — Electric Vehicles - Modelling and Simulations 3.5.2 Application of the BMC control to the traction system The first step to be made is to establish a behaviour model. In this case, we choose a mechanical model without slip, which will be equivalent to the contact wheel-road in the areas known as pseudo-slip. This model can be considered as an ideal model. However, the inertia moments of the elements in rotation and the vehicle mass can be represented by the total inertia moments Jp nog of each shaft motor which is given by:

Vehicle Stability Enhancement Control for Electric Vehicle Using Behaviour Model Control

The influence of the disturbance on the wheel speeds in both controls is shown in Fig. 21. An
error is used to compare the transient performances of the MCS and the BMC. This figure
shows clearly that the perturbation effect is negligible in the case of BMC control and
demonstrates again the robustness of this new control. — Vehicle Stability Enhancement Control for Electric Vehicle Using Behaviour Model Control The influence of the disturbance on the wheel speeds in both controls is shown in Fig. 21. An error is used to compare the transient performances of the MCS and the BMC. This figure shows clearly that the perturbation effect is negligible in the case of BMC control and demonstrates again the robustness of this new control.

Vehicle Stability Enhancement Control for Electric Vehicle Using Behaviour Model Control

EE ————E——E— «=O

Initially, we suppose that the two wheels are not skidding and are not disturbed. Then, a
80 km/h step speed is applied to our system. We notice that the speeds of both wheels and
vehicle are almost identical. These speeds are illustrated in the Fig. 24(a) and (b). Fig. 24(c)
shows that the two motor speeds have the same behaviour to its model. The difference
between these speeds is represented in the Fig. 24(d). From Fig. 24(e) we notice that the slips
da and 2 of both wheels respectively, are maintained in the adhesive region and the traction
forces which are illustrated by the Fig. 24(f) are identical, due to the same conditions taken
of both contact wheel-road. The motor torques are represented in Fig. 24(g) and the imposed
torques of the main controller and the behaviour controllers are shown in Fig. 24(h). The
resistive force of the vehicle is shown by the Fig. 24(i). — Vehicle Stability Enhancement Control for Electric Vehicle Using Behaviour Model Control EE ————E——E— «=O Initially, we suppose that the two wheels are not skidding and are not disturbed. Then, a 80 km/h step speed is applied to our system. We notice that the speeds of both wheels and vehicle are almost identical. These speeds are illustrated in the Fig. 24(a) and (b). Fig. 24(c) shows that the two motor speeds have the same behaviour to its model. The difference between these speeds is represented in the Fig. 24(d). From Fig. 24(e) we notice that the slips da and 2 of both wheels respectively, are maintained in the adhesive region and the traction forces which are illustrated by the Fig. 24(f) are identical, due to the same conditions taken of both contact wheel-road. The motor torques are represented in Fig. 24(g) and the imposed torques of the main controller and the behaviour controllers are shown in Fig. 24(h). The resistive force of the vehicle is shown by the Fig. 24(i).

5. Conclusion
Electric Vehicles - Modelling and Simulations — 5. Conclusion Electric Vehicles - Modelling and Simulations

Fig. 1. Some advantages of FPGA use when controlling multiple electric motors.

Another important feature of the powertrain library is the energy loss minimization block of
induction motors (IMs). It is a well known fact that IMs, although robust and cheap, are less
efficient than other electric motors, like the permanent magnets motors (Zeraoulia et al., 2006).
To partially attenuate this drawback, our IP core library incorporates a loss minimization
algorithm (Araujo et al., 2008) which, based on the IM losses model, finds the flux current
setpoint that minimizes the energy losses for each operating point (torque, speed) of the
motor. Experimental studies conducted in our prototypes showed that, in urban driving
cycles, energy savings between 10% and 20% can be achieved, which contribute to increase
the vehicle efficiency and range per charge metric (Araujo et al., 2008). Since the motor
control and the energy minimization are model based, the effectiveness of these algorithms
relies on good estimates of the motor parameters. To address this limitation, our team
developed an additional IP core for performing the identification of the IM model parameters,
whose output data is employed to tune the motor controller and the loss minimization
algorithm (Cerqueira et al., 2007). — Another important feature of the powertrain library is the energy loss minimization block of induction motors (IMs). It is a well known fact that IMs, although robust and cheap, are less efficient than other electric motors, like the permanent magnets motors (Zeraoulia et al., 2006). To partially attenuate this drawback, our IP core library incorporates a loss minimization algorithm (Araujo et al., 2008) which, based on the IM losses model, finds the flux current setpoint that minimizes the energy losses for each operating point (torque, speed) of the motor. Experimental studies conducted in our prototypes showed that, in urban driving cycles, energy savings between 10% and 20% can be achieved, which contribute to increase the vehicle efficiency and range per charge metric (Araujo et al., 2008). Since the motor control and the energy minimization are model based, the effectiveness of these algorithms relies on good estimates of the motor parameters. To address this limitation, our team developed an additional IP core for performing the identification of the IM model parameters, whose output data is employed to tune the motor controller and the loss minimization algorithm (Cerqueira et al., 2007).

throttlemap module, where the throttle signal is mapped on a current reference (task 2), and
the torque allocation task (3) to define the current (or torque) level that must be imposed to
each electric motor in the uCar. After completing task 3, the soft processor sends the current
references to the motor control IP cores that command the power converters to achieve the

motor torque and flux regulation.

DE ee ee Se a a eS ee ee): i a, ee ee Ce, ee a a — throttlemap module, where the throttle signal is mapped on a current reference (task 2), and the torque allocation task (3) to define the current (or torque) level that must be imposed to each electric motor in the uCar. After completing task 3, the soft processor sends the current references to the motor control IP cores that command the power converters to achieve the motor torque and flux regulation. DE ee ee Se a a eS ee ee): i a, ee ee Ce, ee a a

Table 1. Resource utilization of the main IP cores (Note: the design tool was the ISE WebPack
8.2.03i, FPGA family: Spartan 3, Speed Grade: -5). — Table 1. Resource utilization of the main IP cores (Note: the design tool was the ISE WebPack 8.2.03i, FPGA family: Spartan 3, Speed Grade: -5).

Fig. 5. Latency introduced by the MC sub-modules (the main clock in the FPGA has a
frequency of 50MHz, thus 2500cyces = 50us)
FPGA Based Powertrain Control for Electric Vehicles — Fig. 5. Latency introduced by the MC sub-modules (the main clock in the FPGA has a frequency of 50MHz, thus 2500cyces = 50us) FPGA Based Powertrain Control for Electric Vehicles

GO a ee—=S————_—_—_e__ eee ee ee “_

compromising the bandwidth of other modules.

A summary of the resource utilization in the IP cores implementation, such as slices, dedicated
multipliers and Block Ram (BRAM), is presented in Tables 1 and 2. The two Motor Controllers
instantiated in control unit are the most demanding on the FPGA resources, requiring 44% of
the slices and 92% of the dedicated multipliers available on the chip. Although there are a
considerable number of slices available (39%), the low number of free multipliers prevents the
inclusion of additional MC, presenting a restriction for future improvements in this FPGA; in
other words, such improvements would need an FPGA with more computational resources,
thus more costly. In addition to the MC, there are also others modules to perform auxiliary
functions (sensor interface, protections, soft processor), described in the previous section, and
which consume 17% of the FPGA area. — GO a ee—=S————_—_—_e__ eee ee ee “_ compromising the bandwidth of other modules. A summary of the resource utilization in the IP cores implementation, such as slices, dedicated multipliers and Block Ram (BRAM), is presented in Tables 1 and 2. The two Motor Controllers instantiated in control unit are the most demanding on the FPGA resources, requiring 44% of the slices and 92% of the dedicated multipliers available on the chip. Although there are a considerable number of slices available (39%), the low number of free multipliers prevents the inclusion of additional MC, presenting a restriction for future improvements in this FPGA; in other words, such improvements would need an FPGA with more computational resources, thus more costly. In addition to the MC, there are also others modules to perform auxiliary functions (sensor interface, protections, soft processor), described in the previous section, and which consume 17% of the FPGA area.

(a) Motor controller and SVPWM configuration

Fig. 7. uCar electric vehicle prototype.
FPGA Based Powertrain Control for Electric Vehicles — Fig. 7. uCar electric vehicle prototype. FPGA Based Powertrain Control for Electric Vehicles

Fig. 9. Detailed view of the uCar (left motor) results during accelerating, field weakening and
regenerative braking.
To further validate the EV control unit performance, Fig. 9 show the detailed results of the
left motor controller for tree different operating modes: acceleration, field weakening and
regenerative braking. The data depicted in these figures was acquired with the controller
internal datalogger, which enable us to keep track of the most relevant EV variables, such
as: mechanical (motor speed), energy source (voltage, current and power) and the motor
controller ( torque (i7) and flux (iz) currents, modulation index (m) and the slip frequency
(fstip)) variables. During the acceleration mode (Fig.9(a), 9(c)), performed with the throttle at
100%, the ig and ig currents are set at the maximum value in order to extract the maximum
motor torque and vehicle acceleration (2.2km/h/s). When the EV reaches 18km/h the motor
voltage saturates at 83% and the flux current is reduced to allow the vehicle to operate in
the field weakening area, with a power consumption of 2.5kW per motor. In fact, analyzing
the evolution of the power supplied by the batteries during the experimental driving cycles
(Fig. 8), it is interesting to note that the electric motors spend most of the time operating in
this field weakening zone. Fig. 9(b) and 9(d) shows the detailed results of third EV operation
FPGA Based Powertrain Control for Electric Vehicles — Fig. 9. Detailed view of the uCar (left motor) results during accelerating, field weakening and regenerative braking. To further validate the EV control unit performance, Fig. 9 show the detailed results of the left motor controller for tree different operating modes: acceleration, field weakening and regenerative braking. The data depicted in these figures was acquired with the controller internal datalogger, which enable us to keep track of the most relevant EV variables, such as: mechanical (motor speed), energy source (voltage, current and power) and the motor controller ( torque (i7) and flux (iz) currents, modulation index (m) and the slip frequency (fstip)) variables. During the acceleration mode (Fig.9(a), 9(c)), performed with the throttle at 100%, the ig and ig currents are set at the maximum value in order to extract the maximum motor torque and vehicle acceleration (2.2km/h/s). When the EV reaches 18km/h the motor voltage saturates at 83% and the flux current is reduced to allow the vehicle to operate in the field weakening area, with a power consumption of 2.5kW per motor. In fact, analyzing the evolution of the power supplied by the batteries during the experimental driving cycles (Fig. 8), it is interesting to note that the electric motors spend most of the time operating in this field weakening zone. Fig. 9(b) and 9(d) shows the detailed results of third EV operation FPGA Based Powertrain Control for Electric Vehicles

Fig. 1. The PM and steel characteristics used as the active part’s materials of the PMSM.
The | permanent magnet synchronous machine (PMSM) has to provide a maximum power
density. For that, good quality materials should be used. The permanent magnet (PM)
material is of Nd-Fe-B type, with 1.25 T remanent flux density. The steel used for the
construction of PMSM is M530-50A. The material characteristics are presented in Fig. 1. — Fig. 1. The PM and steel characteristics used as the active part’s materials of the PMSM. The | permanent magnet synchronous machine (PMSM) has to provide a maximum power density. For that, good quality materials should be used. The permanent magnet (PM) material is of Nd-Fe-B type, with 1.25 T remanent flux density. The steel used for the construction of PMSM is M530-50A. The material characteristics are presented in Fig. 1.

Fig. 2. The PMSM: (a) geometry; (b) fractioned type winding configuration; (c) the resultant
voltage vectors. — Fig. 2. The PMSM: (a) geometry; (b) fractioned type winding configuration; (c) the resultant voltage vectors.

Fig. 3. Phasor diagram for PMSM in motor-load conditions.

Table 3. Comparison of obtained results for the designed PMSM.
2.3 Drive modeling for controlling the PMSM
The power factor for the PMSM is quite reduced (as it can be seen in Table 3). In order to
increase the power factor, it is possible to use capacitor battery connected to stator winding.
This solution is expensive and non-reliable. On the other hand, the current, and finally the
power factor are depending on angle 6 and @ (see Fig. 3).

It is possible to rewrite the d,q-axis currents by imposing B = (6 — @). Thus, the currents are
Iq = —I,-sin(B) and I, = 1, -cos(B). If one will choose the q axis as phase origin, in Fig. 2,
the electric motor will operate to unity power factor (cos p = 1) if B = 6.
Electric Vehicles - Modelling and Simulations — Table 3. Comparison of obtained results for the designed PMSM. 2.3 Drive modeling for controlling the PMSM The power factor for the PMSM is quite reduced (as it can be seen in Table 3). In order to increase the power factor, it is possible to use capacitor battery connected to stator winding. This solution is expensive and non-reliable. On the other hand, the current, and finally the power factor are depending on angle 6 and @ (see Fig. 3). It is possible to rewrite the d,q-axis currents by imposing B = (6 — @). Thus, the currents are Iq = —I,-sin(B) and I, = 1, -cos(B). If one will choose the q axis as phase origin, in Fig. 2, the electric motor will operate to unity power factor (cos p = 1) if B = 6. Electric Vehicles - Modelling and Simulations

The purpose of this subsection is to introduce the control modeling of a PMSM and the
controllability of the motor at unity power factor.

First of all, the motor model will be presented. After a short review of the vector control
technique, the results of the PMSM control are presented.

From the PMSM equivalent circuit (Fig. 4), one can obtain the machine’s mathematical
model. The model takes into account the iron loss. The voltage equations, as a matrix, are: — The purpose of this subsection is to introduce the control modeling of a PMSM and the controllability of the motor at unity power factor. First of all, the motor model will be presented. After a short review of the vector control technique, the results of the PMSM control are presented. From the PMSM equivalent circuit (Fig. 4), one can obtain the machine’s mathematical model. The model takes into account the iron loss. The voltage equations, as a matrix, are:

Neglecting the phase resistance, the internal angle tangent becomes:

ig. 7. PMSM simulation results: (a) electrical performances; (b) mechanical performances.
Electric Vehicles - Modelling and Simulations — ig. 7. PMSM simulation results: (a) electrical performances; (b) mechanical performances. Electric Vehicles - Modelling and Simulations

Fig. 8. PMSM simulation results: energetic performances.

Supplementary constraints were considered for the mechanical outputs (torque and power)
and electrical (supplied current) characteristics, see Table 5. — Supplementary constraints were considered for the mechanical outputs (torque and power) and electrical (supplied current) characteristics, see Table 5.

The goal of the optimization process is to maximize the power density (power/mass ratio) -
this is our objective function. The parameters to be varied, in the optimization process, are:
the length of the machine, the air-gap length, the PM length (on the magnetization
direction), the inner stator diameter, the stator slot’s height, the tooth width, the stator yoke
height and isthmus height. The initial values and the boundaries are given in Table 4. — The goal of the optimization process is to maximize the power density (power/mass ratio) - this is our objective function. The parameters to be varied, in the optimization process, are: the length of the machine, the air-gap length, the PM length (on the magnetization direction), the inner stator diameter, the stator slot’s height, the tooth width, the stator yoke height and isthmus height. The initial values and the boundaries are given in Table 4.

Fig. 10. FEM results of PMSM in motor regime.

The resume of the optimized results of the PMSM are given in Table 6.
(In order to validate the obtained optimization results, the author has introduced again the
modified parameters into the FEM software to re-compute the performances of the designed
machine. The optimization results have been confirmed by the FEM analysis, like the results
presented in Section 3.) — The resume of the optimized results of the PMSM are given in Table 6. (In order to validate the obtained optimization results, the author has introduced again the modified parameters into the FEM software to re-compute the performances of the designed machine. The optimization results have been confirmed by the FEM analysis, like the results presented in Section 3.)

Fig. 11. Optimization results of PMSM: source current and mechanical performances
constraints. — Fig. 11. Optimization results of PMSM: source current and mechanical performances constraints.

Fig. 12. Optimization results of PMSM: evolution of geometrical and energetic parameters.

Efficient Sensorless PMSM Drive for Electric Vehicle Traction Systems

The commutation process has to ensure that the action of switching the current direction is
synchronized with the movement of the flux in the air gap, and so the motor must have a
sensor for measuring the position of the flux wave relative to that of the stator windings.
Simple Hall-effect sensors are used with BLDCM in order to manage the commutation
sequence and form the appropriate current waveform. On the other hand, a high
resolution encoder or resolver is necessary for the PMSM control mode to generate
sinusoidal currents. — Efficient Sensorless PMSM Drive for Electric Vehicle Traction Systems The commutation process has to ensure that the action of switching the current direction is synchronized with the movement of the flux in the air gap, and so the motor must have a sensor for measuring the position of the flux wave relative to that of the stator windings. Simple Hall-effect sensors are used with BLDCM in order to manage the commutation sequence and form the appropriate current waveform. On the other hand, a high resolution encoder or resolver is necessary for the PMSM control mode to generate sinusoidal currents.

The estimation here depends exclusively on the motor speed and the sampling time. So,
more attention should be paid to the sampling time in high speed operation
particularly. For the test motor, the frequency of the Hall signals goes beyond 1.4 kHz at
rated speed; therefore relatively fast sampling time should be used for the estimation
(100 sus).
Electric Vehicles - Modelling and Simulations — The estimation here depends exclusively on the motor speed and the sampling time. So, more attention should be paid to the sampling time in high speed operation particularly. For the test motor, the frequency of the Hall signals goes beyond 1.4 kHz at rated speed; therefore relatively fast sampling time should be used for the estimation (100 sus). Electric Vehicles - Modelling and Simulations

Efficient Sensorless PMSM Drive for Electric Vehicle Traction Systems

Fig. 7. Block diagram of the Back-EMF/ Hall-effect based position and speed estimator.

Fig. 8. Experimental setup for the in-wheel brushless motor drive validation and the IPM
oased power circuit.
An experimental set up was fabricated in the laboratory using a 48V/2kW in-wheel gearless
Brushless motor which is fed by a three-phase full bridge inverter built using compact
Intelligent Power Module (IPM) (Fig. 8). This system is powered by 48V//5AH battery pack.
For estimation and control tasks, an eZdsp2812 board has been used. To keep an eye on the
control mode of the motor, the Park frame currents (d-q) are measured. — Fig. 8. Experimental setup for the in-wheel brushless motor drive validation and the IPM oased power circuit. An experimental set up was fabricated in the laboratory using a 48V/2kW in-wheel gearless Brushless motor which is fed by a three-phase full bridge inverter built using compact Intelligent Power Module (IPM) (Fig. 8). This system is powered by 48V//5AH battery pack. For estimation and control tasks, an eZdsp2812 board has been used. To keep an eye on the control mode of the motor, the Park frame currents (d-q) are measured.

Vac is the DC Link voltage measured via LEM sensor that already exists for battery energy
management purposes, and D is the PWM duty cycle.

The compensation term Vcomp is associated with the inverter device losses and is determined
from non-linear tables relating inverter device voltage drop to phase current.

The block diagram of the developed control scheme is shown in Fig. 9. Inputs to the position
estimator are motor stator currents and voltages as well Hall sensor signals. The output is a
high resolution estimate of the rotor position and speed estimation.
Fig. 9. Overall scheme of the combined BLDCM/Sensorless PMSM control strategy. — Vac is the DC Link voltage measured via LEM sensor that already exists for battery energy management purposes, and D is the PWM duty cycle. The compensation term Vcomp is associated with the inverter device losses and is determined from non-linear tables relating inverter device voltage drop to phase current. The block diagram of the developed control scheme is shown in Fig. 9. Inputs to the position estimator are motor stator currents and voltages as well Hall sensor signals. The output is a high resolution estimate of the rotor position and speed estimation. Fig. 9. Overall scheme of the combined BLDCM/Sensorless PMSM control strategy.

Fig. 11. Commutation from BLDCM mode to PMSM mode at 50 rpm.
Fig. 10. Hall-effect signals and estimated position. — Fig. 11. Commutation from BLDCM mode to PMSM mode at 50 rpm. Fig. 10. Hall-effect signals and estimated position.

Fig. 12. Starting up with complete BLDCM/Sensorless PMSM control strategy.

Fig. 13. Dynamic behavior of the drive in Sensorless PMSM control mode

Electric Vehicles - Modelling and Simulations

The hybrid electric car has 8 working modes: idle stop, ICE drive, motor drive, serial mode,
parallel mode, serial & parallel mode, ICE drive & battery charge and regenerative brake.
Fig.2 shows power flowing in 4 modes of them. — Electric Vehicles - Modelling and Simulations The hybrid electric car has 8 working modes: idle stop, ICE drive, motor drive, serial mode, parallel mode, serial & parallel mode, ICE drive & battery charge and regenerative brake. Fig.2 shows power flowing in 4 modes of them.

Table 1. Basic parameters of the vehicle
The basic parameters of the vehicle are shown in the Table 1.

The power of 1.51 ICE is 43kW. The transmission is auto transmission. Power of G/M is 8kW.
Table.2 shows the simulation results of ICE drive only, with high base speed electric
machinery and with low base speed electric machinery. The simulation condition is the
same. With the highest torque, the drive assembly with low base speed electric machinery
obtains better dynamic performance. And its fuel consume is lower, because the ICE work
in more efficient area. These can be found in simulated ICE working points under 1 ECE-
EUDC cycle in Fig.3. — Table 1. Basic parameters of the vehicle The basic parameters of the vehicle are shown in the Table 1. The power of 1.51 ICE is 43kW. The transmission is auto transmission. Power of G/M is 8kW. Table.2 shows the simulation results of ICE drive only, with high base speed electric machinery and with low base speed electric machinery. The simulation condition is the same. With the highest torque, the drive assembly with low base speed electric machinery obtains better dynamic performance. And its fuel consume is lower, because the ICE work in more efficient area. These can be found in simulated ICE working points under 1 ECE- EUDC cycle in Fig.3.

1 - rotor core, 2 - stator core, 3 - radial air gap, 4 - radial three phase excitation coil (armature), 5 -
magnetization permanent magnet, 6 - axial compensatory excitation coil core, 7 - axial compensatory
excitation coil, 8 - magnetized end cover, 9 - magnetized housing, 10 - axial air gap, 11 - axes, 12 - flux
path produced by axial compensatory excitation coil, 13 - flux path produced by permanent magnet
\ccording to the analysis above, a hybrid switched reluctance motor with flux adjustme:
leveloped to obtain low base speed torque-speed features of the main motor. Hy
witched reluctance motor is developed from Hybrid Stepper Motor. In the motor, there
xial excitation coils on both sides of end cover, which is different from HB stepper mc
‘hese coils produce axial flux which is controlled to compensate and adjust PM flux.

\ structure figure of a three-phase hybrid switched reluctance motor is shown in Figut
‘The motor consists of: rotor core 1, stator core 2, radial air gap 3, radial three pl
xcitation coil 4, axial magnetization permanent magnet 5, axial compensatory excita
oil core 6, axial compensatory excitation coil 7, magnetized end cover 8, magneti
.ousing 9, axial air gap 10, and axes 11. 12 indicates the path of flux produced by a
ompensatory excitation coil and 13 indicates the path of PM flux. Except for 6, 7 and 10,
tructure is same as HB stepper motor. Axial compensatory excitation coil and core
oaxial in line with PM through axial air gap. Controlling the value and direction of curr
lowing in axial coil will compensate and adjust axial flux mainly produced by
“onsequentially, the main flux is adjustable. — 1 - rotor core, 2 - stator core, 3 - radial air gap, 4 - radial three phase excitation coil (armature), 5 - magnetization permanent magnet, 6 - axial compensatory excitation coil core, 7 - axial compensatory excitation coil, 8 - magnetized end cover, 9 - magnetized housing, 10 - axial air gap, 11 - axes, 12 - flux path produced by axial compensatory excitation coil, 13 - flux path produced by permanent magnet \ccording to the analysis above, a hybrid switched reluctance motor with flux adjustme: leveloped to obtain low base speed torque-speed features of the main motor. Hy witched reluctance motor is developed from Hybrid Stepper Motor. In the motor, there xial excitation coils on both sides of end cover, which is different from HB stepper mc ‘hese coils produce axial flux which is controlled to compensate and adjust PM flux. \ structure figure of a three-phase hybrid switched reluctance motor is shown in Figut ‘The motor consists of: rotor core 1, stator core 2, radial air gap 3, radial three pl xcitation coil 4, axial magnetization permanent magnet 5, axial compensatory excita oil core 6, axial compensatory excitation coil 7, magnetized end cover 8, magneti .ousing 9, axial air gap 10, and axes 11. 12 indicates the path of flux produced by a ompensatory excitation coil and 13 indicates the path of PM flux. Except for 6, 7 and 10, tructure is same as HB stepper motor. Axial compensatory excitation coil and core oaxial in line with PM through axial air gap. Controlling the value and direction of curr lowing in axial coil will compensate and adjust axial flux mainly produced by “onsequentially, the main flux is adjustable.

Consider the axial coil on the side of 1-1 in Fig.5. When current flows in the coil, main flux
linking the coil is produced by coil itself and radial excitation coil on the stator poles.
Because of the drop of MMF on the sheet gap of rotor core, the effect of flux by PM is weak.
This effect is ignored in analysis. The equivalent circuit of flux path on side of 1-1 is shown

in Figure 6. — Consider the axial coil on the side of 1-1 in Fig.5. When current flows in the coil, main flux linking the coil is produced by coil itself and radial excitation coil on the stator poles. Because of the drop of MMF on the sheet gap of rotor core, the effect of flux by PM is weak. This effect is ignored in analysis. The equivalent circuit of flux path on side of 1-1 is shown in Figure 6.

Table 4. Parameters of the type motor
In the simulation, the motor parameters are as shown in the table 4. — Table 4. Parameters of the type motor In the simulation, the motor parameters are as shown in the table 4.

VVE CULIULUT LUO pete: GALILALULE CULIETIL Ao OLLUVVAIL LiL celia © Sie tilt Orr tat Ub ule AAA CUIDS La)
oe modulated into sinusoidal signal with frequency three times of that of the armature EMF
[he zero detect pulse of this signal is applied to control the armature current as shown it
rig.8. Two conditions above are satisfied.

Motor drive is developed. The torque control motor drive block diagram of it is shown it
Fig.9. Sensor processing section processes the axial coil signal and works as a rotor positio1
and speed sensor. Only controlling the magnitude and direction of the current in the axiz
-xiting coils, the flux strengthening ( giving positive current to the axial exciting coils) an
lux weakening (giving negative current to the axial exciting coils) control are realized
When large torque is needed, as starting the vehicle immediately, flux strengthening contro
s executed. Flux weaken control is executed when charging the battery in high speed. Th
armature current and EMF keep in the same angle phase.

simple armature current commuting and axial coil flux adjusting control are developed t
ychieve larce toraiue and hich cneed The contro! block diacram is chown in fioure 9. — VVE CULIULUT LUO pete: GALILALULE CULIETIL Ao OLLUVVAIL LiL celia © Sie tilt Orr tat Ub ule AAA CUIDS La) oe modulated into sinusoidal signal with frequency three times of that of the armature EMF [he zero detect pulse of this signal is applied to control the armature current as shown it rig.8. Two conditions above are satisfied. Motor drive is developed. The torque control motor drive block diagram of it is shown it Fig.9. Sensor processing section processes the axial coil signal and works as a rotor positio1 and speed sensor. Only controlling the magnitude and direction of the current in the axiz -xiting coils, the flux strengthening ( giving positive current to the axial exciting coils) an lux weakening (giving negative current to the axial exciting coils) control are realized When large torque is needed, as starting the vehicle immediately, flux strengthening contro s executed. Flux weaken control is executed when charging the battery in high speed. Th armature current and EMF keep in the same angle phase. simple armature current commuting and axial coil flux adjusting control are developed t ychieve larce toraiue and hich cneed The contro! block diacram is chown in fioure 9.

Fig.11, Fig.12 and Fig.13 are the measured wave form of armature EMF (upper one) and
axial coil EMF when rotor speeds are 300 rpm, 1200rpm and 2400rpm. It is seen that there
are 6 zero points in the axial coil EMF signal corresponding in 1 cycle of armature EMF
signal which can be the commuted signals to drive motor.
‘ig. 11. Measured wave form of axial coil emf and armature emf when 300rpm — Fig.11, Fig.12 and Fig.13 are the measured wave form of armature EMF (upper one) and axial coil EMF when rotor speeds are 300 rpm, 1200rpm and 2400rpm. It is seen that there are 6 zero points in the axial coil EMF signal corresponding in 1 cycle of armature EMF signal which can be the commuted signals to drive motor. ‘ig. 11. Measured wave form of axial coil emf and armature emf when 300rpm

Fig. 12. Measured wave form of axial coil emf and armature emf when 1200rpm

In this chapter an accurate mathematical model for high performance three phase
permanent magnet motor drive systems, including interaction with the servoamplifier
power conditioner, based on physical principles is presented (Guinee et al, 1999) for
performance related prediction studies in embedded systems, through comparison with
actual drive experimental test data for model fidelity and accuracy, and for subsequent
dynamical parameter identification strategies where required. The BLMD system (Moog
GmbH, 1988, 1989), which is modelled here as an example, can be configured for either
torque control operation or as an adjustable speed drive in high performance EV
applications (Emadi et al, ibid; Crowder, 1995). The motor drive incorporates two feedback
loops for precision control with (a) a fast tracking high gain inner current loop, which forces
the stator winding current equal to the required torque demand current via pulsewidth
modulation and (b) an outer velocity loop for adjustable speed operation of the motor drive
shaft in high performance applications.
When configured for adjustable speed drive (ASD) operation the outer BLMD velocity loop ot
low bandwidth encloses the inner wideband current loop and tends to partially obscure its
operation as a result of outer loop coupling. It is for this reason that the BLMD is initially
modelled with a separate torque loop, uncoupled from the outer velocity feedback loop, for
complete visibility of its high frequency PWM current control loop operation. The most
difficult aspect of the BLMD modelling exercise for torque control operation that has to be
addressed concerns the simulation of the current controlled PWM output voltage, from the
three phase inverter to the motor stator windings, with sufficient accuracy to incorporate the
effects of inverter dead-time. This issue arises when the modulating control signal to the
pulsewidth modulator is non deterministic during the transient phase of motor operation fot
random step changes in command input that may occur during normal online operation of the
embedded drive in industrial applications eventhough the modulation employed is sinusoidal
PWM. It could be argued that a simplified model of the PWM process is adequate in this
instance in that only the low frequency filtered components of current feedback and speed are — In this chapter an accurate mathematical model for high performance three phase permanent magnet motor drive systems, including interaction with the servoamplifier power conditioner, based on physical principles is presented (Guinee et al, 1999) for performance related prediction studies in embedded systems, through comparison with actual drive experimental test data for model fidelity and accuracy, and for subsequent dynamical parameter identification strategies where required. The BLMD system (Moog GmbH, 1988, 1989), which is modelled here as an example, can be configured for either torque control operation or as an adjustable speed drive in high performance EV applications (Emadi et al, ibid; Crowder, 1995). The motor drive incorporates two feedback loops for precision control with (a) a fast tracking high gain inner current loop, which forces the stator winding current equal to the required torque demand current via pulsewidth modulation and (b) an outer velocity loop for adjustable speed operation of the motor drive shaft in high performance applications. When configured for adjustable speed drive (ASD) operation the outer BLMD velocity loop ot low bandwidth encloses the inner wideband current loop and tends to partially obscure its operation as a result of outer loop coupling. It is for this reason that the BLMD is initially modelled with a separate torque loop, uncoupled from the outer velocity feedback loop, for complete visibility of its high frequency PWM current control loop operation. The most difficult aspect of the BLMD modelling exercise for torque control operation that has to be addressed concerns the simulation of the current controlled PWM output voltage, from the three phase inverter to the motor stator windings, with sufficient accuracy to incorporate the effects of inverter dead-time. This issue arises when the modulating control signal to the pulsewidth modulator is non deterministic during the transient phase of motor operation fot random step changes in command input that may occur during normal online operation of the embedded drive in industrial applications eventhough the modulation employed is sinusoidal PWM. It could be argued that a simplified model of the PWM process is adequate in this instance in that only the low frequency filtered components of current feedback and speed are

necessary, since these are uncoupled from the actual PWM process except for the dead time,
for accurate BLMD simulation with minimal run time. This simplified low frequency model
strategy, based on the fundamental component of the PWM process, can only be used when
there is negligible inverter delay and is the approach that is adopted in such circumstances for
simulation purposes as the ‘average’ BLMD model. The presence of inverter dead time,
however, requires additional BLMD model processing in that the current flow direction must
checked in each phase, during every PWM switching period, in order to determine whether a
delay pulse or correction term is to be added or subtracted to the fundamental signal
components. Consequently the modulated pulse edge transitions have to be accurately known
to include the exact instances of fixed delay triggering of the basedrives controlling power
transistor inverter ON/OFF switching. Once a satisfactory BLMD model of sufficient
functional accuracy has been generated and ‘mapped’ to an actual embedded drive system,
through parameter identification of the motor dynamics, the addition of the outer velocity
control loop can then be completed in a holistic BLMD model for ASD simulation. Correlation
accuracy of this complete model with an actual ASD is established through subsequent step
response simulation and comparison with experimental shaft velocity test data. — necessary, since these are uncoupled from the actual PWM process except for the dead time, for accurate BLMD simulation with minimal run time. This simplified low frequency model strategy, based on the fundamental component of the PWM process, can only be used when there is negligible inverter delay and is the approach that is adopted in such circumstances for simulation purposes as the ‘average’ BLMD model. The presence of inverter dead time, however, requires additional BLMD model processing in that the current flow direction must checked in each phase, during every PWM switching period, in order to determine whether a delay pulse or correction term is to be added or subtracted to the fundamental signal components. Consequently the modulated pulse edge transitions have to be accurately known to include the exact instances of fixed delay triggering of the basedrives controlling power transistor inverter ON/OFF switching. Once a satisfactory BLMD model of sufficient functional accuracy has been generated and ‘mapped’ to an actual embedded drive system, through parameter identification of the motor dynamics, the addition of the outer velocity control loop can then be completed in a holistic BLMD model for ASD simulation. Correlation accuracy of this complete model with an actual ASD is established through subsequent step response simulation and comparison with experimental shaft velocity test data.

The brushless motor consists of a 12-pole PM rotor, a wound multiple pole stator, a 2-pole
transmitter type pancake resolver and a ntc thermistor embedded in the stator end turns with
a typical cross-section sketched in Figure 4. Stator current is provided by a 3® power cable
with a protective earth while a signal cable routes rotor position information from the pancake
resolver located at the rear side of the motor structure. The outer motor casing (stator) houses
the 3® stationary winding in a lamination stack. The Y-connected floating neutral winding is
embedded in slots around the air gap periphery with a sinusoidal spatial distribution. This has
the effect of producing a time dependent rotating sinusoidal MMF space wave centred on the
magnetic axes of the respective phases, which are displaced 120 electrical degrees apart in
space. The inner member (rotor) contains the Samarium-Cobalt magnets, which have a high
holding force with an energy product of 18 MGO, (Demerdash et al, 1980), in the form of arc
segments assembled as salient poles on an iron rotor structure. The fixed radially directed
magnetic field, produced by the rotor magnets, is held perpendicular to the electromagnetic
field generated by the stator coils and consequently yields maximum rotor torque for a given
stator current. This stator-to-rotor vector field interaction is achieved by electronic
commutation, which processes rotor position information from the shaft resolver to provide a
balanced three phase sinusoidal stator current. The high peak torque achievable, which is — The brushless motor consists of a 12-pole PM rotor, a wound multiple pole stator, a 2-pole transmitter type pancake resolver and a ntc thermistor embedded in the stator end turns with a typical cross-section sketched in Figure 4. Stator current is provided by a 3® power cable with a protective earth while a signal cable routes rotor position information from the pancake resolver located at the rear side of the motor structure. The outer motor casing (stator) houses the 3® stationary winding in a lamination stack. The Y-connected floating neutral winding is embedded in slots around the air gap periphery with a sinusoidal spatial distribution. This has the effect of producing a time dependent rotating sinusoidal MMF space wave centred on the magnetic axes of the respective phases, which are displaced 120 electrical degrees apart in space. The inner member (rotor) contains the Samarium-Cobalt magnets, which have a high holding force with an energy product of 18 MGO, (Demerdash et al, 1980), in the form of arc segments assembled as salient poles on an iron rotor structure. The fixed radially directed magnetic field, produced by the rotor magnets, is held perpendicular to the electromagnetic field generated by the stator coils and consequently yields maximum rotor torque for a given stator current. This stator-to-rotor vector field interaction is achieved by electronic commutation, which processes rotor position information from the shaft resolver to provide a balanced three phase sinusoidal stator current. The high peak torque achievable, which is

Mathematical Modelling and Simulation of a PWM Inverter Controlled Brushless
Motor Drive System from Physical Principles for Electric Vehicle Propulsion Applications — Mathematical Modelling and Simulation of a PWM Inverter Controlled Brushless Motor Drive System from Physical Principles for Electric Vehicle Propulsion Applications

The MMF standing wave, which is wrapped around the air gap periphery, is effectively
produced by a sinusoidally distributed current sheet located on the inner stator
circumference as shown in Figure 6 for phase-a. The standing space wave components are
modulated by the time varying balanced 3®@ stator current, with electrical angular frequency
@e, represented by — The MMF standing wave, which is wrapped around the air gap periphery, is effectively produced by a sinusoidally distributed current sheet located on the inner stator circumference as shown in Figure 6 for phase-a. The standing space wave components are modulated by the time varying balanced 3®@ stator current, with electrical angular frequency @e, represented by

rotor displacement has been explained. The raison d’étre of this simplifying assumption is
that the permeability of the magnetically hard Samarium-Cobalt (6mCos) material is almost
equal to that of air. As a consequence of this property the SmCos material has some desirable
features from a BLMD modelling perspective in terms of its intrinsic demagnetization
characteristic. The PM rotor air gap line (Matsch, 1972) is a design feature which is
optimized in terms of the maximum energy product of 160 kJm (Crangle, 1991) for a given
machine configuration and magnet geometry. In Figure 7 the locus of operation of the air
gap line, due to changes in gap width, is a minor hysteresis loop (Match, ibid) with axis
tangent to the magnetization curve through the retention flux ®p. — rotor displacement has been explained. The raison d’étre of this simplifying assumption is that the permeability of the magnetically hard Samarium-Cobalt (6mCos) material is almost equal to that of air. As a consequence of this property the SmCos material has some desirable features from a BLMD modelling perspective in terms of its intrinsic demagnetization characteristic. The PM rotor air gap line (Matsch, 1972) is a design feature which is optimized in terms of the maximum energy product of 160 kJm (Crangle, 1991) for a given machine configuration and magnet geometry. In Figure 7 the locus of operation of the air gap line, due to changes in gap width, is a minor hysteresis loop (Match, ibid) with axis tangent to the magnetization curve through the retention flux ®p.

as shown in Figure 9 before evaluation of the drive torque. Since the flux linkages are
functions of the stator winding current, complex and lengthy numerical integration of (XXX)
would be required over the nonlinear A-i magnetization characteristic in Figure 9, which
must be known, if saturation effects are to be included. However if magnetic nonlinearity is
neglected, with the assumption that the flux linkages and MMFs are directly proportional
for the entire magnetic circuit as in air, the resulting analysis and integral expression (XXXI)
is greatly simplified. In this case the flux linkages are assumed to be linear with current
magnitude, which is often done in the analysis of practical devices, in the winding
inductances as in (XXII). However a simpler and more convenient alternative (Krause, 1986)
than obtaining the EM torque as a function of As via W;(A,,0,) in (XXXI), relies on the
coenergy state function W-(I;,8;) to determine the applied torque I, in terms of the stator
currents I; as the independent PWM controlled state variables in BLMD system operation.
This methodology is more effective during BLMD model simulation as the motor winding
currents are immediately available for motor torque computation. — as shown in Figure 9 before evaluation of the drive torque. Since the flux linkages are functions of the stator winding current, complex and lengthy numerical integration of (XXX) would be required over the nonlinear A-i magnetization characteristic in Figure 9, which must be known, if saturation effects are to be included. However if magnetic nonlinearity is neglected, with the assumption that the flux linkages and MMFs are directly proportional for the entire magnetic circuit as in air, the resulting analysis and integral expression (XXXI) is greatly simplified. In this case the flux linkages are assumed to be linear with current magnitude, which is often done in the analysis of practical devices, in the winding inductances as in (XXII). However a simpler and more convenient alternative (Krause, 1986) than obtaining the EM torque as a function of As via W;(A,,0,) in (XXXI), relies on the coenergy state function W-(I;,8;) to determine the applied torque I, in terms of the stator currents I; as the independent PWM controlled state variables in BLMD system operation. This methodology is more effective during BLMD model simulation as the motor winding currents are immediately available for motor torque computation.

The physical manifestation of motor torque production, which is expressed as the product of
stator current and the rotor air gap flux given in (XLIII), results from a tendency of the stator
and rotor air gap flux fields to align their magnetic axes with minimum energy configuration.
Consequently maximum torque is developed by the motor as in (XLIII) when the torque angle
ar, defined as the angle between the armature reaction A,,(J,8;) and PM flux Asm(8;) space
vectors in (XVII), is maintained at 90 degrees. Optimum torque angle control can be achieved
in a BLMD system by means of self synchronization, via a rotor shaft position resolver as
shown in Figure 1, in which a current controlled PWM inverter ensures an orthogonal spatial
relationship between the stator and rotor flux vectors. This can be visualized with the aid of
the phasor diagram in Figure 10 where the torque angle a, given by
This is identical to electrical power P, provided to each of the stator phase windings
ignoring losses, by the PSU, in sustaining the magnetic coupling field from collapse during
mechanical energy transfer. The electric power is accomplished by means of phase current jj;
injection against coupling field back-emf reaction v,;. The PSU contribution P, is expressed

by means of (XIV) and (XLIV) as V,7, with — The physical manifestation of motor torque production, which is expressed as the product of stator current and the rotor air gap flux given in (XLIII), results from a tendency of the stator and rotor air gap flux fields to align their magnetic axes with minimum energy configuration. Consequently maximum torque is developed by the motor as in (XLIII) when the torque angle ar, defined as the angle between the armature reaction A,,(J,8;) and PM flux Asm(8;) space vectors in (XVII), is maintained at 90 degrees. Optimum torque angle control can be achieved in a BLMD system by means of self synchronization, via a rotor shaft position resolver as shown in Figure 1, in which a current controlled PWM inverter ensures an orthogonal spatial relationship between the stator and rotor flux vectors. This can be visualized with the aid of the phasor diagram in Figure 10 where the torque angle a, given by This is identical to electrical power P, provided to each of the stator phase windings ignoring losses, by the PSU, in sustaining the magnetic coupling field from collapse during mechanical energy transfer. The electric power is accomplished by means of phase current jj; injection against coupling field back-emf reaction v,;. The PSU contribution P, is expressed by means of (XIV) and (XLIV) as V,7, with

VACLE LELtE Littell Cilwvi.

\ mathematical model of 3-phase inverter operation can be obtained by describing the
ontrol action of the pulsewidth modulator in each phase, fed with a fixed voltage dc supply
Iz, on the amplitude and frequency adjustment of the BLMD stator winding voltage ir
esponse to the current compensator error output. Modulation proceeds in two steps ir
ccordance with the current controller output as shown in Figure 1. A symmetrical double
dge width modulated pulse train is generated for each phase by means of a voltage
omparator as the first step. A triangular carrier waveform V;,i(t), of fixed frequency f; and
ommon to all three phases, establishes the switching period T; and the current control erro1
eference v(t) then modulates the switch duty cycle as shown in Figure 11 for phase-a.
The modulated bipolar switch pulse train v,(t) is then used as the firing signal, at the
modulator o/p as shown in Figure 12 for phase-a, to control the “ON” and “OFF” periods
of the power transistors, via the base drive circuitry (BDA), in each half-bridge of the 30
inverter as the subsequent step. Similar switching sequences, with a phase angle separation
of 120 degrees, are obtained for control of the other two phases of the inverter bridge
resulting in a 30 PWM voltage supply to the motor stator windings. In practical inverters
the finite turn-off time of the power transistor bridge necessitates the use of a finite blanking
switch time in the PWM process to avoid short circuiting the dc busbar to ground (Murai,
1985; Evans et al, 1987; Dodson et al, 1990). This fixed ‘interlock delay’ 5, which is typically
20 uS, is conservatively chosen for slow switching Darlington transistors in medium power
motor drives in the low kilowatt range. — VACLE LELtE Littell Cilwvi. \ mathematical model of 3-phase inverter operation can be obtained by describing the ontrol action of the pulsewidth modulator in each phase, fed with a fixed voltage dc supply Iz, on the amplitude and frequency adjustment of the BLMD stator winding voltage ir esponse to the current compensator error output. Modulation proceeds in two steps ir ccordance with the current controller output as shown in Figure 1. A symmetrical double dge width modulated pulse train is generated for each phase by means of a voltage omparator as the first step. A triangular carrier waveform V;,i(t), of fixed frequency f; and ommon to all three phases, establishes the switching period T; and the current control erro1 eference v(t) then modulates the switch duty cycle as shown in Figure 11 for phase-a. The modulated bipolar switch pulse train v,(t) is then used as the firing signal, at the modulator o/p as shown in Figure 12 for phase-a, to control the “ON” and “OFF” periods of the power transistors, via the base drive circuitry (BDA), in each half-bridge of the 30 inverter as the subsequent step. Similar switching sequences, with a phase angle separation of 120 degrees, are obtained for control of the other two phases of the inverter bridge resulting in a 30 PWM voltage supply to the motor stator windings. In practical inverters the finite turn-off time of the power transistor bridge necessitates the use of a finite blanking switch time in the PWM process to avoid short circuiting the dc busbar to ground (Murai, 1985; Evans et al, 1987; Dodson et al, 1990). This fixed ‘interlock delay’ 5, which is typically 20 uS, is conservatively chosen for slow switching Darlington transistors in medium power motor drives in the low kilowatt range.

This dwell time, between successive power transistor switching instants, is effected through
an integrate and dump RC ‘lockout’ network connected to a voltage threshold in the base
drive circuitry. A switching delay of this magnitude, for inverter frequencies in the audio
range (5kHz), has to be included in the motor drive modelling process for accurate
simulation studies. The effect of the switching delay can be modelled as follows by
considering phase-a only with a similar procedure for the other two phases (Guinee et al,
1998, 1999). — This dwell time, between successive power transistor switching instants, is effected through an integrate and dump RC ‘lockout’ network connected to a voltage threshold in the base drive circuitry. A switching delay of this magnitude, for inverter frequencies in the audio range (5kHz), has to be included in the motor drive modelling process for accurate simulation studies. The effect of the switching delay can be modelled as follows by considering phase-a only with a similar procedure for the other two phases (Guinee et al, 1998, 1999).

The duty cycle of the bipolar switch control voltage vsa(f), during one switching period T; in
Figure 12 is determined by the relative magnitude comparison and accurate crossover
evaluation, from simulation via the regula-falsi iterative search method shown in Figure 16,
of the current control voltage v.a(t) with the dither reference v;,;(t). The switch control pulse
sequence can be described as — The duty cycle of the bipolar switch control voltage vsa(f), during one switching period T; in Figure 12 is determined by the relative magnitude comparison and accurate crossover evaluation, from simulation via the regula-falsi iterative search method shown in Figure 16, of the current control voltage v.a(t) with the dither reference v;,;(t). The switch control pulse sequence can be described as

Fig. 19. Inverter o/p Voltage (ins <0)
Fig. 18. Inverter o/p Voltage (iss >0) — Fig. 19. Inverter o/p Voltage (ins <0) Fig. 18. Inverter o/p Voltage (iss >0)

The adoption of a single iteration of the regula falsi method along with a small simulation
time step At simplifies the search problem of the pulse edge transition with sufficient
accuracy without the expenditure of considerable computational effort for a modest gain in
accuracy by comparison with the other iterative methods available. An indication of the step
size required for accurate resolution of PWM inverter operation with delay can be obtained
from consideration of the anticipated signal ‘curvature’ due to (a) the signal bandwidth and
amplitude at the current controller o/p v,j in the magnitude comparison with the triangular
carrier shown in Figure 16 in the comparator modulator and (b) the rate of exponential
voltage ramp up to the base drive threshold Vin, which controls the inverter dead time, in
the RC delay circuits shown in Figure 20.

The maximum harmonic o/p voltage from the high gain current compensator G; is
determined by the carrier amplitude A, at the onset of overmodulation (m; = 1) in PWM
inverter control with a frequency that is limited bv the 3dB bandwidth me = 1/te (~3kH7 in
The accuracy with which the estimated width modulated pulse transition instants tx are
determined can be gauged by comparing the deviation error of the actual intersection time ft’
of the triangular carrier with the control signal v., due to its curvature, to that tx obtained
with the piecewise linear chord approximation of the signal in the regula-falsi method as
lustrated in Figure 16. The ‘curvature’ of the signal in (LXVIIJ) with time, determined
‘(Kreyszig, 1972) from — The adoption of a single iteration of the regula falsi method along with a small simulation time step At simplifies the search problem of the pulse edge transition with sufficient accuracy without the expenditure of considerable computational effort for a modest gain in accuracy by comparison with the other iterative methods available. An indication of the step size required for accurate resolution of PWM inverter operation with delay can be obtained from consideration of the anticipated signal ‘curvature’ due to (a) the signal bandwidth and amplitude at the current controller o/p v,j in the magnitude comparison with the triangular carrier shown in Figure 16 in the comparator modulator and (b) the rate of exponential voltage ramp up to the base drive threshold Vin, which controls the inverter dead time, in the RC delay circuits shown in Figure 20. The maximum harmonic o/p voltage from the high gain current compensator G; is determined by the carrier amplitude A, at the onset of overmodulation (m; = 1) in PWM inverter control with a frequency that is limited bv the 3dB bandwidth me = 1/te (~3kH7 in The accuracy with which the estimated width modulated pulse transition instants tx are determined can be gauged by comparing the deviation error of the actual intersection time ft’ of the triangular carrier with the control signal v., due to its curvature, to that tx obtained with the piecewise linear chord approximation of the signal in the regula-falsi method as lustrated in Figure 16. The ‘curvature’ of the signal in (LXVIIJ) with time, determined ‘(Kreyszig, 1972) from

The value of K, was subsequently used in a motor-generator electrical load test, at different
speeds as illustrated in Figure 24, to estimate the stator winding parameters L, and r, as a
cross check of the nominal catalogued (Moog, 1998) values. The difference AV between the
measured terminal voltage Vr, across the load resistance R., and the generated voltage Vc
using the fitted coefficient K, via (XXIV) is equated to the internal voltage drop of the
Thevenin equivalent circuit shown in Figure 24

xarath — The value of K, was subsequently used in a motor-generator electrical load test, at different speeds as illustrated in Figure 24, to estimate the stator winding parameters L, and r, as a cross check of the nominal catalogued (Moog, 1998) values. The difference AV between the measured terminal voltage Vr, across the load resistance R., and the generated voltage Vc using the fitted coefficient K, via (XXIV) is equated to the internal voltage drop of the Thevenin equivalent circuit shown in Figure 24 xarath

Fig. 25. Winding parameter estimation
Fig. 24. Motor - generator load test — Fig. 25. Winding parameter estimation Fig. 24. Motor - generator load test

Electric Vehicles - Modelling and Simulations

The friction coefficient Bm is obtained from a linear first order polynomial fit, displayed ir
Figure 26, based on expression (LXXXVI) with estimate B,, =2,141x10° Nm.rad™ as ir
Table II. Alternative confirmation of the accuracy of the damping factor estimate is obtainec
from consideration of the electrical power transfer P. from the coupling field expressed ir
(XLVII) and comparison with the resultant mechanical power dissipation P, associated with
dynamic friction via (XLVI). The continuous power supplied from the coupling field.
necessary to sustain motor rotation with frictional losses at various shaft speeds unde
steady state conditions, is determined from the rms values of reaction EMF using the
measured estimate K, from the o/c test and the experimental FC test data with lagging
power factor balanced over three phases as — Electric Vehicles - Modelling and Simulations The friction coefficient Bm is obtained from a linear first order polynomial fit, displayed ir Figure 26, based on expression (LXXXVI) with estimate B,, =2,141x10° Nm.rad™ as ir Table II. Alternative confirmation of the accuracy of the damping factor estimate is obtainec from consideration of the electrical power transfer P. from the coupling field expressed ir (XLVII) and comparison with the resultant mechanical power dissipation P, associated with dynamic friction via (XLVI). The continuous power supplied from the coupling field. necessary to sustain motor rotation with frictional losses at various shaft speeds unde steady state conditions, is determined from the rms values of reaction EMF using the measured estimate K, from the o/c test and the experimental FC test data with lagging power factor balanced over three phases as

The actual drive system with network structure as shown in Figure 28 was tested at critical
internal nodes with multiplexed sampled data waveforms acquired at rates corresponding
to the different inertial loaded shaft conditions (J; ) specified in Table II. The length of each
data record is fixed at 4095 sample points with a normalized duration of approximately 10
machine FC cycles for reference purposes during comparison with simulated motor
response for model validation and accuracy and also during system identification for
accurate extraction of drive motor model parameter estimates. — The actual drive system with network structure as shown in Figure 28 was tested at critical internal nodes with multiplexed sampled data waveforms acquired at rates corresponding to the different inertial loaded shaft conditions (J; ) specified in Table II. The length of each data record is fixed at 4095 sample points with a normalized duration of approximately 10 machine FC cycles for reference purposes during comparison with simulated motor response for model validation and accuracy and also during system identification for accurate extraction of drive motor model parameter estimates.

Both the simulated current transients iga(kT) and ig(kT) exhibit the characteristics of a frequency
modulated sinusoid with fixed amplitude and swept frequency due to the exponential
buildup of motor shaft speed during the acceleration phase. This can be visualized from the
amplitude spectrum shown in Figure 33, for the extended filtered feedback current displayed
in Figure 34, which appears constant over the electrical frequency band of 286 Hz
corresponding to the swept motor speed range from standstill to 3000 RPM. These simulated
waveforms provide an excellent fit in terms of frequency and phase coherence with test data
when correlated. The measure of fit in this instance is expressed by the trace response
correlation coefficients, listed in Table III, as — Both the simulated current transients iga(kT) and ig(kT) exhibit the characteristics of a frequency modulated sinusoid with fixed amplitude and swept frequency due to the exponential buildup of motor shaft speed during the acceleration phase. This can be visualized from the amplitude spectrum shown in Figure 33, for the extended filtered feedback current displayed in Figure 34, which appears constant over the electrical frequency band of 286 Hz corresponding to the swept motor speed range from standstill to 3000 RPM. These simulated waveforms provide an excellent fit in terms of frequency and phase coherence with test data when correlated. The measure of fit in this instance is expressed by the trace response correlation coefficients, listed in Table III, as

Furthermore the accuracy of fit of the simulated traces consisting of the shaft velocity and
current controller output with experimental step response test data, as indicated by the
correlation coefficients in Table III, confirms model integrity. The fidelity and coherence of
BLMD model trace simulation, when compared with drive experimental test data, is also
established for known inertial shaft loads (Guinee, 1998, 1999) which further substantiates
model accuracy and confidence. A number of BLMD transient waveform simulations, based
on established model accuracy and confidence, at strategic internal nodes provide insight
into and confirmation of motor drive operation during the acceleration phase. The filtered
feedback current from each phase of the motor winding to the compensators in the three
phase current control loop is illustrated in Figure 35. These waveforms show a reduction in
the period of oscillation, accompanied by a very slight decrease in amplitude due to the
impact of back emf reaction and machine impedance effects, as expected with an increase in
shaft speed.
Mathematical Modelling and Simulation of a PWM Inverter Controlled Brushless
Motor Drive System from Physical Principles for Electric Vehicle Propulsion Applications — Furthermore the accuracy of fit of the simulated traces consisting of the shaft velocity and current controller output with experimental step response test data, as indicated by the correlation coefficients in Table III, confirms model integrity. The fidelity and coherence of BLMD model trace simulation, when compared with drive experimental test data, is also established for known inertial shaft loads (Guinee, 1998, 1999) which further substantiates model accuracy and confidence. A number of BLMD transient waveform simulations, based on established model accuracy and confidence, at strategic internal nodes provide insight into and confirmation of motor drive operation during the acceleration phase. The filtered feedback current from each phase of the motor winding to the compensators in the three phase current control loop is illustrated in Figure 35. These waveforms show a reduction in the period of oscillation, accompanied by a very slight decrease in amplitude due to the impact of back emf reaction and machine impedance effects, as expected with an increase in shaft speed. Mathematical Modelling and Simulation of a PWM Inverter Controlled Brushless Motor Drive System from Physical Principles for Electric Vehicle Propulsion Applications

Fig. 40. Spectrum of phase voltage vs
Fig. 39. Stator phase voltages
Fig. 37. Basedrive command signals
Fig. 38. PWM inverter o/p voltage — Fig. 40. Spectrum of phase voltage vs Fig. 39. Stator phase voltages Fig. 37. Basedrive command signals Fig. 38. PWM inverter o/p voltage

The winding currents lag the reaction EMF as shown in Figure 41 by the internal power
factor angle g; ~ 66.6, obtained from statistical averaging of the estimated crossover
current required to sustain frictional torque in (LXXXVI) is minimal.

The normalized spectrum of the six step phase voltage, which has a sharp line structure
indicative of steady state motor operation close to rated speed, is displayed in Figure 40.
This amplitude spectrum, which is the characteristic signature of sub-harmonic PWM
inverter operation (Murphy et al, 1998), consists of the fundamental machine electrical
frequency f. (~312 Hz) and side frequency component pairs (kf; +f) associated with pulse
generation about the triangular carrier switching harmonics kf; The side frequency
distribution contains even order pairs symmetrically disposed about odd carrier harmonics
and odd order pairs about even harmonics with significant amplitudes dependent on the
index of modulation m; in (LIII). These extraneous component contributions are located well
outside the machine winding passband, which has a 3dB cutoff frequency f, determined
from the stator electrical time constant 7 = L;/1r; (~2.6ms) in Table I as — The winding currents lag the reaction EMF as shown in Figure 41 by the internal power factor angle g; ~ 66.6, obtained from statistical averaging of the estimated crossover current required to sustain frictional torque in (LXXXVI) is minimal. The normalized spectrum of the six step phase voltage, which has a sharp line structure indicative of steady state motor operation close to rated speed, is displayed in Figure 40. This amplitude spectrum, which is the characteristic signature of sub-harmonic PWM inverter operation (Murphy et al, 1998), consists of the fundamental machine electrical frequency f. (~312 Hz) and side frequency component pairs (kf; +f) associated with pulse generation about the triangular carrier switching harmonics kf; The side frequency distribution contains even order pairs symmetrically disposed about odd carrier harmonics and odd order pairs about even harmonics with significant amplitudes dependent on the index of modulation m; in (LIII). These extraneous component contributions are located well outside the machine winding passband, which has a 3dB cutoff frequency f, determined from the stator electrical time constant 7 = L;/1r; (~2.6ms) in Table I as

Fig. 44. Phasor diagram of brushless motor
Table IV. Evaluation of phasor magnitudes from steady state conditions in figure 42 — Fig. 44. Phasor diagram of brushless motor Table IV. Evaluation of phasor magnitudes from steady state conditions in figure 42

Fig. 45. Current lag compensation
Fig. 46. Commutation phase lead
using (XXIV) it is assumed that the stator winding back EMF v,j is in phase with the forced
stator current jj, in response to the equal magnitude current demand i,; from shaft sensor
position information, for maximum torque production via the applied and electronically
commutated stator terminal voltage v,. This assumption, however, is not accurate in that a
phase lag equal to the power factor angle @ develops with shaft velocity between the
injected current Jj; and voltage Vj; phasors as shown in Figure 44. During normal motor
operation current commutation is used in an attempt to maintain a virtual armature flux
phasor A*j,; in quadrature with the rotor flux, in accordance with fixed current demand, for
maximum motor torque production. As the motor reaches rated speed, for zero shaft load
torque conditions, the motor impedance angle @, in Figure 42 increases along with the back
EMF V,;. The cumulative effect of increased impedance voltage Vz with V.j result in further
current lag by the angle @ in order to comply with fixed torque current demand via the
applied stator phase voltage Vjs. — Fig. 45. Current lag compensation Fig. 46. Commutation phase lead using (XXIV) it is assumed that the stator winding back EMF v,j is in phase with the forced stator current jj, in response to the equal magnitude current demand i,; from shaft sensor position information, for maximum torque production via the applied and electronically commutated stator terminal voltage v,. This assumption, however, is not accurate in that a phase lag equal to the power factor angle @ develops with shaft velocity between the injected current Jj; and voltage Vj; phasors as shown in Figure 44. During normal motor operation current commutation is used in an attempt to maintain a virtual armature flux phasor A*j,; in quadrature with the rotor flux, in accordance with fixed current demand, for maximum motor torque production. As the motor reaches rated speed, for zero shaft load torque conditions, the motor impedance angle @, in Figure 42 increases along with the back EMF V,;. The cumulative effect of increased impedance voltage Vz with V.j result in further current lag by the angle @ in order to comply with fixed torque current demand via the applied stator phase voltage Vjs.

Fig. 48. Phase ripple current
Fig. 47. Simulated airgap torque I’,
The internal power factor angle adjusts towards 90° to reduce the torque angle q@ in
(XLVIII) with reaction EMF increase in compliance with load torque requirements as shown
in Figure 44. The increase of current lag @, with impedance angle @, due to motor speed, can
be compensated for with power factor correction by electronically advancing the current
command phase angle in accordance with (LXXXIV), in the current commutator circuit of
Figures 1 and 28, as — Fig. 48. Phase ripple current Fig. 47. Simulated airgap torque I’, The internal power factor angle adjusts towards 90° to reduce the torque angle q@ in (XLVIII) with reaction EMF increase in compliance with load torque requirements as shown in Figure 44. The increase of current lag @, with impedance angle @, due to motor speed, can be compensated for with power factor correction by electronically advancing the current command phase angle in accordance with (LXXXIV), in the current commutator circuit of Figures 1 and 28, as

Fig. 49. Simulated shaft velocity
This results in the smooth mechanical motion illustrated as the simulated motor shaf
velocity in Fig. 49. As the motor reaches rated speed the generated torque decreases to tha
necessary in (LXXXVI) to sustain motion with frictional torque retardation. The simulatec
power transfer and the filtered version derived from the developed torque characteristics
depicted in Figure 47 using expression (XLVI), are shown in Figure 50. The net motive
power required under steady state conditions, at a motor speed of @, ~ 310 rad.sec+, tc
overhaul mechanical losses is P, ~ 182 watts which correlates reasonably well with the
friction power estimate of P;~200 watts obtained from Figure 27. — Fig. 49. Simulated shaft velocity This results in the smooth mechanical motion illustrated as the simulated motor shaf velocity in Fig. 49. As the motor reaches rated speed the generated torque decreases to tha necessary in (LXXXVI) to sustain motion with frictional torque retardation. The simulatec power transfer and the filtered version derived from the developed torque characteristics depicted in Figure 47 using expression (XLVI), are shown in Figure 50. The net motive power required under steady state conditions, at a motor speed of @, ~ 310 rad.sec+, tc overhaul mechanical losses is P, ~ 182 watts which correlates reasonably well with the friction power estimate of P;~200 watts obtained from Figure 27.

which for a unit torque demand input with MI = 0.145 ( = 1/6.9), T; = 176s and 6 = 19.6u1s
in Figure 53 is estimated as 80.5%. The percentage ratio of the corresponding rms feedback
currents, with and without delay respectively in Figure 53 over the extended time span of
0.24 secs, is 78.5% (=1.555/1.981) which is almost identical to that from pulse time
considerations. The resultant torque ratio from Figure 51 is also approximately 80%, as it is
proportional to the current ratio, in the settled region which corresponds to the torque
necessary to overhaul frictional effects. Motor shaft speed exhibits a similar variation in
Figure 52, since it is proportional to the time integral of the torque, with delay of about
82.5%. — which for a unit torque demand input with MI = 0.145 ( = 1/6.9), T; = 176s and 6 = 19.6u1s in Figure 53 is estimated as 80.5%. The percentage ratio of the corresponding rms feedback currents, with and without delay respectively in Figure 53 over the extended time span of 0.24 secs, is 78.5% (=1.555/1.981) which is almost identical to that from pulse time considerations. The resultant torque ratio from Figure 51 is also approximately 80%, as it is proportional to the current ratio, in the settled region which corresponds to the torque necessary to overhaul frictional effects. Motor shaft speed exhibits a similar variation in Figure 52, since it is proportional to the time integral of the torque, with delay of about 82.5%.

3.5 Novel PWM Inverter dead-time compensation
Timing counter circuits with optocoupler isolation (Murai et al, 1985, 1992) have been used
for delay correction in each phase during PWM operation of induction motors. A novel and
simpler solution is proposed here for simultaneous delay compensation in all three phases
through amplitude adjustment of the triangular dither waveform. The technique relies on
the simple expedient of additional symmetrical double edge pulse widening during PWM,
via the signal magnitude comparison with a reduced carrier amplitude contribution, to
counterbalance the effects of inverter lag. The additional modulation index Am; required for
increased pulse duration using (CI) to nullify the effect of inverter delay in (CII) is — 3.5 Novel PWM Inverter dead-time compensation Timing counter circuits with optocoupler isolation (Murai et al, 1985, 1992) have been used for delay correction in each phase during PWM operation of induction motors. A novel and simpler solution is proposed here for simultaneous delay compensation in all three phases through amplitude adjustment of the triangular dither waveform. The technique relies on the simple expedient of additional symmetrical double edge pulse widening during PWM, via the signal magnitude comparison with a reduced carrier amplitude contribution, to counterbalance the effects of inverter lag. The additional modulation index Am; required for increased pulse duration using (CI) to nullify the effect of inverter delay in (CII) is

A detailed and accurate reference model, based on physical principles, of a typical
embedded BLMD system used for EV propulsion and high performance motive power
industrial applications has been presented for the express purpose of computer aided design
and simulation of EV propulsion systems where performance prediction and evaluation are
a necessity before fabrication. Model fidelity is confirmed by extensive numerical simulation
with particular emphasis at critical internal observation nodes when contrasted with
measured data from a high performance PM drive system. Model validation for
identification purposes is provided by frequency and phase coherence of simulated data
with step response transient current feedback test data possessing FM attributes. A novel
and effective delay compensation solution, based on carrier voltage level adjustment for
multi-phase operation, is provided to counterbalance the effect of inverter blanking in
torque reduction which is substantiated by BLMD model simulation. — A detailed and accurate reference model, based on physical principles, of a typical embedded BLMD system used for EV propulsion and high performance motive power industrial applications has been presented for the express purpose of computer aided design and simulation of EV propulsion systems where performance prediction and evaluation are a necessity before fabrication. Model fidelity is confirmed by extensive numerical simulation with particular emphasis at critical internal observation nodes when contrasted with measured data from a high performance PM drive system. Model validation for identification purposes is provided by frequency and phase coherence of simulated data with step response transient current feedback test data possessing FM attributes. A novel and effective delay compensation solution, based on carrier voltage level adjustment for multi-phase operation, is provided to counterbalance the effect of inverter blanking in torque reduction which is substantiated by BLMD model simulation.

Table 1. Specifications of an EV
where frwheei is the drive wheels radius and Qunee is the rotational wheels speed.

Based on the specifications of an urban EV (Ehsani et al., 2005), summarized in Table 1,
and on the above-described EV traction system, the requirements of one in-wheel motor
can be easily computed. All the results are presented in Table 2. Note that, in addition to
provide its requirements, the in-wheel motor must also respect some constraints. The
main constraints are the total weight of each of the four in-wheel motors Minotor (imposed
by the maximal authorized “unsprung” wheel weight) and the imposed outer radius Rou:
of the motor (imposed by the rim of the wheel). Those are also specified in Table I and
Table 2. — Table 1. Specifications of an EV where frwheei is the drive wheels radius and Qunee is the rotational wheels speed. Based on the specifications of an urban EV (Ehsani et al., 2005), summarized in Table 1, and on the above-described EV traction system, the requirements of one in-wheel motor can be easily computed. All the results are presented in Table 2. Note that, in addition to provide its requirements, the in-wheel motor must also respect some constraints. The main constraints are the total weight of each of the four in-wheel motors Minotor (imposed by the maximal authorized “unsprung” wheel weight) and the imposed outer radius Rou: of the motor (imposed by the rim of the wheel). Those are also specified in Table I and Table 2.

Fig. 2. (a) Stator and rotor of an AFPM machine and (b) doubled-sided AFPM machine with
internal slotted stator
where R;,, and R,,; are respectively the inner and outer radius of the machine (see Fig.

2(a)).
The use of the average radius as a design parameter allows evaluating motor parameters
and performances based on analytical design methods (Gieras et al., 2004).

The air gap flux density B, is calculated using the remanence flux density B, (in the order of
1.2 T for a NdFeB type PM) and the relative permeability fi; of the PM as well as the
geometrical dimensions of the air gap and the PM (thickness and area) according to: — Fig. 2. (a) Stator and rotor of an AFPM machine and (b) doubled-sided AFPM machine with internal slotted stator where R;,, and R,,; are respectively the inner and outer radius of the machine (see Fig. 2(a)). The use of the average radius as a design parameter allows evaluating motor parameters and performances based on analytical design methods (Gieras et al., 2004). The air gap flux density B, is calculated using the remanence flux density B, (in the order of 1.2 T for a NdFeB type PM) and the relative permeability fi; of the PM as well as the geometrical dimensions of the air gap and the PM (thickness and area) according to:

Fig. 3. Per-unit electromagnetic toque T;,, as a function of the factor kg
3.2 AFPM motor model validation
eee eee ee SED SEE SESE EEE

In order to validate the analytical AFPM motor model presented in this chapter, analytical
and experimental results are compared. To do so, the proposed model is applied to a
5.5 kW, 4000 rpm AFPM motor. The calculated motor parameters are then compared with
parameters obtained by classical tests (test at dc level, no-load test, etc.) performed on an
existing AFPM pump motor. All the results are reported in Table 3. As can be seen, very
small differences are obtained between the analytical and experimental results, whatever the
parameters. According to this validation method, one can conclude that the proposed
analytical design process gives reasonable results in this particular case and can be used in
the optimization procedure of the in-wheel motor of an EV. — Fig. 3. Per-unit electromagnetic toque T;,, as a function of the factor kg 3.2 AFPM motor model validation eee eee ee SED SEE SESE EEE In order to validate the analytical AFPM motor model presented in this chapter, analytical and experimental results are compared. To do so, the proposed model is applied to a 5.5 kW, 4000 rpm AFPM motor. The calculated motor parameters are then compared with parameters obtained by classical tests (test at dc level, no-load test, etc.) performed on an existing AFPM pump motor. All the results are reported in Table 3. As can be seen, very small differences are obtained between the analytical and experimental results, whatever the parameters. According to this validation method, one can conclude that the proposed analytical design process gives reasonable results in this particular case and can be used in the optimization procedure of the in-wheel motor of an EV.

fable 3. Analytical and experimental results for the 5.5 kW, 4000 rpm AFPM pump motor
3.3 VSI model
—e_a_ SK ee

The total loss of the semiconductor devices (IGBTs and diodes) of the VSI (employing
sinusoidal pulse-width-modulation) consists of two parts: the on-state losses and the
switching losses. The on-state losses of the devices are calculated from their average Ine and
rms I; currents by the well-known expression (Semikron, 2010): — fable 3. Analytical and experimental results for the 5.5 kW, 4000 rpm AFPM pump motor 3.3 VSI model —e_a_ SK ee The total loss of the semiconductor devices (IGBTs and diodes) of the VSI (employing sinusoidal pulse-width-modulation) consists of two parts: the on-state losses and the switching losses. The on-state losses of the devices are calculated from their average Ine and rms I; currents by the well-known expression (Semikron, 2010):

their ranking. Thus, the solutions of Fz are chosen next, followed by solutions from F3.
However, as shown in Fig. 4, not all the solutions from F3 can be inserted in population Py1.
Indeed, the number of empty slots of P+: is smaller than the number of solutions belonging
to F3. In order to choose which ones will be selected, these solutions are sorted according to
their crowding distance (in descending order) and, then, the number of best of them needed
to fill the empty slots of Pj+1 are inserted in this new population. The created population P}+1
is then used for selection, crossover and mutation (see below) to create a new population

Q+1, and so on for the next generations. — their ranking. Thus, the solutions of Fz are chosen next, followed by solutions from F3. However, as shown in Fig. 4, not all the solutions from F3 can be inserted in population Py1. Indeed, the number of empty slots of P+: is smaller than the number of solutions belonging to F3. In order to choose which ones will be selected, these solutions are sorted according to their crowding distance (in descending order) and, then, the number of best of them needed to fill the empty slots of Pj+1 are inserted in this new population. The created population P}+1 is then used for selection, crossover and mutation (see below) to create a new population Q+1, and so on for the next generations.

Fig. 5. Flowchart of the design procedure using GAs

Electric Vehicles - Modelling and Simulations

In the first approach, the solutions obtained from the optimization technique are analyzed to
choose a particular solution whereas, in the second approach, the optimization technique is
directed towards a preferred region of the Pareto front. Therefore, only the techniques
belonging to the first category are helpful in this chapter. Among these techniques, the
Compromise Programming Approach (CPA) (Yu, 1973) is often used in multiobjective
problems. The CPA picks a solution which is minimally located from a given reference point
(e.g. the ideal point which is a nonexistent solution composed with the minimum value of
the two objectives). Note that other techniques, such as the Marginal Rate of Substitution
Approach (Miettinen, 1999), the Pseudo-Weight Vector Approach (Deb, 2002) or a method
based ona sensitivity analysis (Avila et al., 2006), can also be used. — Electric Vehicles - Modelling and Simulations In the first approach, the solutions obtained from the optimization technique are analyzed to choose a particular solution whereas, in the second approach, the optimization technique is directed towards a preferred region of the Pareto front. Therefore, only the techniques belonging to the first category are helpful in this chapter. Among these techniques, the Compromise Programming Approach (CPA) (Yu, 1973) is often used in multiobjective problems. The CPA picks a solution which is minimally located from a given reference point (e.g. the ideal point which is a nonexistent solution composed with the minimum value of the two objectives). Note that other techniques, such as the Marginal Rate of Substitution Approach (Miettinen, 1999), the Pseudo-Weight Vector Approach (Deb, 2002) or a method based ona sensitivity analysis (Avila et al., 2006), can also be used.

Fig. 8. Evolution of the percentage of individuals violating the constraints during the
optimization procedure — Fig. 8. Evolution of the percentage of individuals violating the constraints during the optimization procedure

Fig. 7. Evolution of the percentage of individuals belonging to the first nondominated front
during the optimization procedure — Fig. 7. Evolution of the percentage of individuals belonging to the first nondominated front during the optimization procedure

Fig. 9. Evolution of kg along the Pareto front

Fig. 11. Evolution of lpyalong the Pareto front

Fig. 14. Evolution of p along the Pareto front
Fig. 13. Evolution of f;along the Pareto front — Fig. 14. Evolution of p along the Pareto front Fig. 13. Evolution of f;along the Pareto front

Fig. 12. Evolution of g along the Pareto front

Fig. 15. Evolution of qalong the Pareto front
Electric Vehicles - Modelling and Simulations — Fig. 15. Evolution of qalong the Pareto front Electric Vehicles - Modelling and Simulations

Fig. 16. Correlation between the optimization variables and the objective functions

Fig. 17. Distribution of the weight among the AFPM motor and the VSI

Fig. 18. Distribution of the power loss among the AFPM motor and the VSI

Fig. 19. Distribution of the weight among the several parts of the AFPM motor

In solution #2, the stator and rotor weights largely dominate the winding and PMs weights
compared with the other solutions. This can be explained regarding the value of the flux
density in the air gap. Indeed, this value is the greatest among the three solutions (in the
order of 0.90 T). So, the stator and rotor must be thick enough to avoid saturation of steel. As
these thicknesses are computed using the air gap flux density, the stator and rotor of the
solution #2 are thicker and, therefore, heavier. Moreover, the windings of the solution #2 are
lighter than the windings of the two other solutions since the frequency is greater. Indeed,
according to Faraday’s law (Mohan et al., 2003), the windings turns necessary to obtain a
same electromotive force is less than in the other solutions which, therefore, yields lighter
windings realization.

a a we aa) a . - eo ma as . de a — In solution #2, the stator and rotor weights largely dominate the winding and PMs weights compared with the other solutions. This can be explained regarding the value of the flux density in the air gap. Indeed, this value is the greatest among the three solutions (in the order of 0.90 T). So, the stator and rotor must be thick enough to avoid saturation of steel. As these thicknesses are computed using the air gap flux density, the stator and rotor of the solution #2 are thicker and, therefore, heavier. Moreover, the windings of the solution #2 are lighter than the windings of the two other solutions since the frequency is greater. Indeed, according to Faraday’s law (Mohan et al., 2003), the windings turns necessary to obtain a same electromotive force is less than in the other solutions which, therefore, yields lighter windings realization. a a we aa) a . - eo ma as . de a

Fig. 1. Electric vehicle drive system.
directly connect two devices in parallel, (FC/battery, FC/SC, or battery/SC). However, in
this way the power drawn from each device cannot be controlled, but is passively
determined by the impedance of the devices. The impedance depends on many parameters,
e.g. temperature, state-of-charge, health, and point of operation. Each device might therefore
be operated at an inappropriate condition, e.g. health and efficiency. The voltage
characteristics also have to match perfectly of the two devices, and only a fraction of the
range of operation of the devices can be utilized, e.g. in a fuel cell battery configuration the
fuel cell must provide almost the same power all the time due to the fixed voltage of the
battery, and in a battery/supercapacitor configuration only a fraction of the energy
exchange capability of the supercapacitor can be used. This is again due to the nearly
constant voltage of the battery. By introducing DC/DC converters one can chose the voltage
variation of the devices and the power of each device can be controlled (Schaltz &
Rasmussen, 2008). — Fig. 1. Electric vehicle drive system. directly connect two devices in parallel, (FC/battery, FC/SC, or battery/SC). However, in this way the power drawn from each device cannot be controlled, but is passively determined by the impedance of the devices. The impedance depends on many parameters, e.g. temperature, state-of-charge, health, and point of operation. Each device might therefore be operated at an inappropriate condition, e.g. health and efficiency. The voltage characteristics also have to match perfectly of the two devices, and only a fraction of the range of operation of the devices can be utilized, e.g. in a fuel cell battery configuration the fuel cell must provide almost the same power all the time due to the fixed voltage of the battery, and in a battery/supercapacitor configuration only a fraction of the energy exchange capability of the supercapacitor can be used. This is again due to the nearly constant voltage of the battery. By introducing DC/DC converters one can chose the voltage variation of the devices and the power of each device can be controlled (Schaltz & Rasmussen, 2008).

Variation of level of signal at the output of LISN versus frequency is the spectrum of
interference. The electromagnetic compatibility of a device can be evaluated by comparison
of its interference spectrum with the standard limitations. The level of signal at the output of
LISN in frequency range 10 kHz up to 30 MHz or 150 kHz up to 30 MHz is criteria of
compatibility and should be under the standard limitations. Converting the results to dBuV
(Equation 1) makes it possible to compare the spectrum of interference with standard
requirements. In practical situations, the LISN output is connected to a spectrum analyzer
and interference measurement is carried out. But for modeling and simulation purposes, the
LISN output spectrum must be calculated using appropriate software. — Variation of level of signal at the output of LISN versus frequency is the spectrum of interference. The electromagnetic compatibility of a device can be evaluated by comparison of its interference spectrum with the standard limitations. The level of signal at the output of LISN in frequency range 10 kHz up to 30 MHz or 150 kHz up to 30 MHz is criteria of compatibility and should be under the standard limitations. Converting the results to dBuV (Equation 1) makes it possible to compare the spectrum of interference with standard requirements. In practical situations, the LISN output is connected to a spectrum analyzer and interference measurement is carried out. But for modeling and simulation purposes, the LISN output spectrum must be calculated using appropriate software.

Fig. 4. The RST canonical structure of a digital controller
The canonical structure of the RST controller is presented in Fig. 4. This structure has two
degrees of freedom, ie., the digital filters R and S are designed in order to achieve the
desired regulation performance, and the digital filter T is designed afterwards in order to
achieve the desired tracking and regulation. This structure allows achievement of different
levels of performance in tracking and regulation. The case of a controller operating on the
regulation error (which does not allow the independent specification of tracking and
regulation performance) corresponds to T=R. Digital PID controller can also be represented
in this form, leading to particular choices of R, S and T (Landau, 1998). — Fig. 4. The RST canonical structure of a digital controller The canonical structure of the RST controller is presented in Fig. 4. This structure has two degrees of freedom, ie., the digital filters R and S are designed in order to achieve the desired regulation performance, and the digital filter T is designed afterwards in order to achieve the desired tracking and regulation. This structure allows achievement of different levels of performance in tracking and regulation. The case of a controller operating on the regulation error (which does not allow the independent specification of tracking and regulation performance) corresponds to T=R. Digital PID controller can also be represented in this form, leading to particular choices of R, S and T (Landau, 1998).

Fig. 3. Block diagram of control mode.
The modeling of studied converters is done by using the Simpower tools of
Matlab/Simulink, and it takes into account the IGBT and Diodes parameters (real
components) and the inductors and capacitors losses. To achieve accurate voltage
regulation, two control loops are used as shown in Fig. 3. This control mode (current mode
control) requires knowledge of the inductor current, which is controlled via the inner loop.
The outer loop manages the output voltage error by commanding the necessary current. The
control was done using RST controllers. — Fig. 3. Block diagram of control mode. The modeling of studied converters is done by using the Simpower tools of Matlab/Simulink, and it takes into account the IGBT and Diodes parameters (real components) and the inductors and capacitors losses. To achieve accurate voltage regulation, two control loops are used as shown in Fig. 3. This control mode (current mode control) requires knowledge of the inductor current, which is controlled via the inner loop. The outer loop manages the output voltage error by commanding the necessary current. The control was done using RST controllers.

A boost DC/DC converter (step-up converter shown in Fig. 5.) is a power converter with an
output DC voltage greater than its input DC voltage. It is a class of switching-mode power
supply containing at least two semiconductor switches (a diode and a switch) and at least
one energy storage element (capacitor and/or inductor). Filters made of capacitors are
normally added to the output of the converter to reduce output voltage ripple and the
inductor connected in series with the input DC source in order to reduce the current ripple.
The smoothing inductor L is used to limit the current ripple. The filter capacitor C can
restrict the output voltage ripples. The ripple current in the inductor is calculated by
neglecting the output voltage ripple. The inductance value is given by the following
equation: — A boost DC/DC converter (step-up converter shown in Fig. 5.) is a power converter with an output DC voltage greater than its input DC voltage. It is a class of switching-mode power supply containing at least two semiconductor switches (a diode and a switch) and at least one energy storage element (capacitor and/or inductor). Filters made of capacitors are normally added to the output of the converter to reduce output voltage ripple and the inductor connected in series with the input DC source in order to reduce the current ripple. The smoothing inductor L is used to limit the current ripple. The filter capacitor C can restrict the output voltage ripples. The ripple current in the inductor is calculated by neglecting the output voltage ripple. The inductance value is given by the following equation:

The structure of the regulator is a RST form. The polynomials R, S and T are calculated
using the methodology explained above. The bandwidth of the current loop w; should be ten
times lower than the switching frequency.
The output voltage loop was designed following a similar strategy to the current loop. To
define the voltage controller, it is assumed that the current control loop is perfect. The
capacitor current and voltage models are given by Equation 37 and Equation 38, respectively. — The structure of the regulator is a RST form. The polynomials R, S and T are calculated using the methodology explained above. The bandwidth of the current loop w; should be ten times lower than the switching frequency. The output voltage loop was designed following a similar strategy to the current loop. To define the voltage controller, it is assumed that the current control loop is perfect. The capacitor current and voltage models are given by Equation 37 and Equation 38, respectively.

The output voltage control loop is shown in Fig. 7.

Fig. 8. Boost converter efficiency versus current load

Fig. 10 shows a basic interleaved step-up converter of 4 identical levels where the
inductances L1 to L4 are built by a separate magnetic core. The gate signals to the power
switching devices are successively phase shifted by T/N where T is the switching period
and N the number of channels. Thus, the current delivered by the electric source is shared
equally between each basic step-up converter level and has a ripple content of period T/N
(Destraz et al., 2006).
The design of the 4-channels converter is the same like the boost one. The output voltage is
idjustable via the duty cycle a of the PWM signal switching the IGBTs as given in the
‘ollowing expression:
5.3 Interleaved 4-channel DC/DC converter — Fig. 10 shows a basic interleaved step-up converter of 4 identical levels where the inductances L1 to L4 are built by a separate magnetic core. The gate signals to the power switching devices are successively phase shifted by T/N where T is the switching period and N the number of channels. Thus, the current delivered by the electric source is shared equally between each basic step-up converter level and has a ripple content of period T/N (Destraz et al., 2006). The design of the 4-channels converter is the same like the boost one. The output voltage is idjustable via the duty cycle a of the PWM signal switching the IGBTs as given in the ‘ollowing expression: 5.3 Interleaved 4-channel DC/DC converter

Fig. 13. EMI simulation results of interleaved 4-channels DC/DC converter.
Fig. 12. 4-channels converter efficiency versus current load. — Fig. 13. EMI simulation results of interleaved 4-channels DC/DC converter. Fig. 12. 4-channels converter efficiency versus current load.

The output voltage control loop is shown in Fig. 16
The bandwidth of the voltage loop , should be ten times lower than the current loop
bandwidth ; which means hundred times lower than the switching frequency. — The output voltage control loop is shown in Fig. 16 The bandwidth of the voltage loop , should be ten times lower than the current loop bandwidth ; which means hundred times lower than the switching frequency.

Table 4 recapitulates the volume, weight, efficiency and the EMI of each converter. The
inductor volume and weight were approximated. It can be noticed that the full-bridge
converter has the biggest volume and weight due to the output inductance. This inductance
value can be reduced by increasing the switching frequency of the converter. We can notice
that the best candidate for the application is the Interleaving multi-channel topology which
has the higher efficiency and lower weight and volume. Weight and volume estimation
takes into account only the IGBT, DIODE, Inductor and capacitor (transformer for full
bridge) and it doesn’t take into account the cooling system and the arrangement of
components in the casing of the converter. — Table 4 recapitulates the volume, weight, efficiency and the EMI of each converter. The inductor volume and weight were approximated. It can be noticed that the full-bridge converter has the biggest volume and weight due to the output inductance. This inductance value can be reduced by increasing the switching frequency of the converter. We can notice that the best candidate for the application is the Interleaving multi-channel topology which has the higher efficiency and lower weight and volume. Weight and volume estimation takes into account only the IGBT, DIODE, Inductor and capacitor (transformer for full bridge) and it doesn’t take into account the cooling system and the arrangement of components in the casing of the converter.

Fig. 19. Efficiency and approximated weight and volume of each converter

This part is devoted to an analytical sizing allowing calculation of geometrical sizes of the
two SMPMM configurations which are the PMSMER and the PMSMIR.

Figure 1 represents the PMSMER and the PMSMIR with the number of pole pairs is p=4 and
a number of principal teeth is 6, between two principal teeth, an inserted tooth is added to
improve the wave form and to reduce the leakage flux (Ben Hadj, N. et al., 2007). The slots
are right and open in order to facilitate the insertion of coils and to reduce the production
cost (Magnussen, F. et al., 2005; Bianchi, N. .et al., 2003; Libert , F. et al., 2004).
2.2 Geometrical structures of PMSMIR and PMSMER — This part is devoted to an analytical sizing allowing calculation of geometrical sizes of the two SMPMM configurations which are the PMSMER and the PMSMIR. Figure 1 represents the PMSMER and the PMSMIR with the number of pole pairs is p=4 and a number of principal teeth is 6, between two principal teeth, an inserted tooth is added to improve the wave form and to reduce the leakage flux (Ben Hadj, N. et al., 2007). The slots are right and open in order to facilitate the insertion of coils and to reduce the production cost (Magnussen, F. et al., 2005; Bianchi, N. .et al., 2003; Libert , F. et al., 2004). 2.2 Geometrical structures of PMSMIR and PMSMER

The analytical study of motor sizing is based on the schedules data conditions parameters
(Table 1), the constant characterizing materials (Table 2), the expert data and configurations
of the two motors.
2.3 Analytical sizing of the two SMPMM structures — The analytical study of motor sizing is based on the schedules data conditions parameters (Table 1), the constant characterizing materials (Table 2), the expert data and configurations of the two motors. 2.3 Analytical sizing of the two SMPMM structures

Expert data
The expert data are practically represented by three sizes which are, the magnetic induction
in the air gap Be, the magnetic induction in the stator yoke B.y and the magnetic induction in
the rotor yoke Byy. It should be noted that the zone of variation of these three parameters
varies between 0,2 to 1,6T (Ben Hadj et al., 2007). — Expert data The expert data are practically represented by three sizes which are, the magnetic induction in the air gap Be, the magnetic induction in the stator yoke B.y and the magnetic induction in the rotor yoke Byy. It should be noted that the zone of variation of these three parameters varies between 0,2 to 1,6T (Ben Hadj et al., 2007).

where B,,o, is the magnetic induction in the tooth.

Fig. 3. Simplified stator for the thermal study in the PMSMER

Fig. 5. Thermal model of the PMSMIR in transient behaviour

Figure 6 shows the thermal model in transient behaviour of the PMSMER.

Table 3. The thermal conductivities of materials
The below table presents the different thermal conductivities of materials (Jérémi, R., 2003) — Table 3. The thermal conductivities of materials The below table presents the different thermal conductivities of materials (Jérémi, R., 2003)

Fig. 7. Various temperatures in different parts of the PMSMER in transient state.

Electric Vehicles - Modelling and Simulations

‘ig. 1. Illustration of the posed problem in the DFIM control system with constant excitation. — Electric Vehicles - Modelling and Simulations ‘ig. 1. Illustration of the posed problem in the DFIM control system with constant excitation.

The resolution of (32) gives place to the arbitrary integration constant C from which the
TOF-relationship (36) can be easily tuned. This one can be adjusted by a judicious choice of
the integration constant, while figure 2 presents TOF effect on armature DFIM currents with
C-tuning. Note that this method offers the possibility to reduce substantially the magnitude
of the armature currents into the machine and we can notice an increase in energy saving.
Hence using TOF strategy, we can avoid the saturation effect and reduce the magnitude of
machine currents from which the DFIM efficiency could be clearly enhanced. — The resolution of (32) gives place to the arbitrary integration constant C from which the TOF-relationship (36) can be easily tuned. This one can be adjusted by a judicious choice of the integration constant, while figure 2 presents TOF effect on armature DFIM currents with C-tuning. Note that this method offers the possibility to reduce substantially the magnitude of the armature currents into the machine and we can notice an increase in energy saving. Hence using TOF strategy, we can avoid the saturation effect and reduce the magnitude of machine currents from which the DFIM efficiency could be clearly enhanced.

Realizing emission reduction in EVs is a crucial step toward emission free vehicle
However, it requires a better understanding of emission free EVs and the ensuing energ
sources. These topics are addressed in the following section: EV Emission.
2. EV emission — Realizing emission reduction in EVs is a crucial step toward emission free vehicle However, it requires a better understanding of emission free EVs and the ensuing energ sources. These topics are addressed in the following section: EV Emission. 2. EV emission

Fig. 4. EV Charging Energy Flow and Efficiency Diagram
Fig. 3. Electric Vehicle Model — Fig. 4. EV Charging Energy Flow and Efficiency Diagram Fig. 3. Electric Vehicle Model

Fig. 5. EV Operating Energy Flow and Efficiency Diagram
Generally the efficiency of the EV Electrical Motor (EM) is exceptionally high ~ 85 %
compared with an Internal Combustion Engine (ICE) ~ 25 %. Power losses in an EV are
negligible, in this section we will focus on power losses from key components occurring in
an electrical propulsion system during driving mode due to power conversion, operation
and propulsion. As illustrated in Figure 5 approximately 81.3 % of the energy stored in the
HV battery is utilized to propel the EV. Combining the EV overall charging efficiency with
the EV overall operational efficiency, the EV efficiency becomes ~ 67.9 % around four times
more efficient than an ICE propelled vehicle with an overall efficiency of ~ 14 %. — Fig. 5. EV Operating Energy Flow and Efficiency Diagram Generally the efficiency of the EV Electrical Motor (EM) is exceptionally high ~ 85 % compared with an Internal Combustion Engine (ICE) ~ 25 %. Power losses in an EV are negligible, in this section we will focus on power losses from key components occurring in an electrical propulsion system during driving mode due to power conversion, operation and propulsion. As illustrated in Figure 5 approximately 81.3 % of the energy stored in the HV battery is utilized to propel the EV. Combining the EV overall charging efficiency with the EV overall operational efficiency, the EV efficiency becomes ~ 67.9 % around four times more efficient than an ICE propelled vehicle with an overall efficiency of ~ 14 %.

$\dding a predictive and an intelligent component to the BMS design can make the ichitecture of the EV more energy and emission efficient, as it would facilitate acquiring raffic condition; offer a dynamic response to all future stochastic traffic flow situations hrough travel route and drive profile advisement. The integration of the predictive and ntelligent components with the BMS led to the concept of the Predictive Intelligent Battery Vianagement System (PIBMS). Nith the rapid advances in wireless communication, global positioning svstem and the$

\dding a predictive and an intelligent component to the BMS design can make the ichitecture of the EV more energy and emission efficient, as it would facilitate acquiring raffic condition; offer a dynamic response to all future stochastic traffic flow situations hrough travel route and drive profile advisement. The integration of the predictive and ntelligent components with the BMS led to the concept of the Predictive Intelligent Battery Vianagement System (PIBMS). Nith the rapid advances in wireless communication, global positioning svstem and the

To evaluate the performance of the PIBMS a simulation tool integrating traffic, vehicle and
network models shall be employed. The simulation platform PIBMST (Predictive Intelligent
Battery Management Simulation Tool) consists of three main building blocks described
below and illustrated in Figure 11 The traffic, vehicle and network models are to allow for
bidirectional coupling among each other, thus resulting in enhanced simulation accuracy
however at the cost of increased computation time. — To evaluate the performance of the PIBMS a simulation tool integrating traffic, vehicle and network models shall be employed. The simulation platform PIBMST (Predictive Intelligent Battery Management Simulation Tool) consists of three main building blocks described below and illustrated in Figure 11 The traffic, vehicle and network models are to allow for bidirectional coupling among each other, thus resulting in enhanced simulation accuracy however at the cost of increased computation time.

Fig. 12. Applied Forces to the EV
The EV model represents a series of mathematical equations representing the characteristics
of the identified EV components in Figure 3 and the forces applied to the vehicle as depicted
in Figure 12
4.2 Vehicle model — Fig. 12. Applied Forces to the EV The EV model represents a series of mathematical equations representing the characteristics of the identified EV components in Figure 3 and the forces applied to the vehicle as depicted in Figure 12 4.2 Vehicle model

Fig. 1. Structure of Diesel Hybrid Electric Vehicle
(HCU). The controllers of lower lever include Engine Management System (EMS), which is a
diesel controller in this system, Battery Packets Control Module (BPCM), Automation
Disconnect Module (ADM), Drive Motor Control Module (DMCM). And the main
communication between controllers is CAN communication. Main CAN nodes of HEV
powertrain are shown Fig. 2. HCU receives information form other lower ECU in order to
know the whole vehicle state, at the same time HCU sends control massage to them. — Fig. 1. Structure of Diesel Hybrid Electric Vehicle (HCU). The controllers of lower lever include Engine Management System (EMS), which is a diesel controller in this system, Battery Packets Control Module (BPCM), Automation Disconnect Module (ADM), Drive Motor Control Module (DMCM). And the main communication between controllers is CAN communication. Main CAN nodes of HEV powertrain are shown Fig. 2. HCU receives information form other lower ECU in order to know the whole vehicle state, at the same time HCU sends control massage to them.

Design and Analysis of Multi-Node CAN Bus for Diesel Hybrid Electric Vehicle

Table 1. Information of CAN frames
Data combination and data split are applied, when data’s addresses were defined. For
example, the detail data definition is shown in Table2. Some variables which have two or
three bits are combined in one same byte, while some variables which occupy two bytes
were split high byte and low byte which in two bytes. These methods can improve the
ability of CAN transmitting, which can save the space of transmitting. In addition, the load
rate of CAN bus could be reduced. — Table 1. Information of CAN frames Data combination and data split are applied, when data’s addresses were defined. For example, the detail data definition is shown in Table2. Some variables which have two or three bits are combined in one same byte, while some variables which occupy two bytes were split high byte and low byte which in two bytes. These methods can improve the ability of CAN transmitting, which can save the space of transmitting. In addition, the load rate of CAN bus could be reduced.

4.2 Design of CAN communication software algorithmic
4.2.1 Buffer time-sharing — 4.2 Design of CAN communication software algorithmic 4.2.1 Buffer time-sharing

The new structure of CAN communication is shown in Fig.6. Priority assignment is applied
in this new structure. Each task is triggered by event interrupt. If there are two or more
tasked triggered at the same time, the task of higher priority is responded firstly. Multi-task
method can optimizes the resource of HCU, and supplies an effective means to extend CAN
communication program. — The new structure of CAN communication is shown in Fig.6. Priority assignment is applied in this new structure. Each task is triggered by event interrupt. If there are two or more tasked triggered at the same time, the task of higher priority is responded firstly. Multi-task method can optimizes the resource of HCU, and supplies an effective means to extend CAN communication program.

4.4 CAN Calibration Protocol (CCP)
standardization.

CCP driver is needed for ECU which parameters can be calibrated. A CCP driver software is
integrated in the infra program of HCU. A basic implementation of CCP driver needs only a
few control unit resources such as RAM, ROM, and CPU time. The CCP driver occupies two
CAN buffers, which should be assigned low bus priority to avoid influencing other bus
communication [8]. The trigger mode of CCP driver is interrupt mode. CCP driver is the
foundation of calibration platform. The calibration platform structure is shown as Fig.7. CCP
driver achieves data upload, download and calibration of controlled parameters. This
calibration system provided reliable, accurate and quickly CAN communication between
calibration platform and ECU. It has been used successfully in HEV controlled system.
The maximum bits of each flame TR can be calculated by following equation [7] — 4.4 CAN Calibration Protocol (CCP) standardization. CCP driver is needed for ECU which parameters can be calibrated. A CCP driver software is integrated in the infra program of HCU. A basic implementation of CCP driver needs only a few control unit resources such as RAM, ROM, and CPU time. The CCP driver occupies two CAN buffers, which should be assigned low bus priority to avoid influencing other bus communication [8]. The trigger mode of CCP driver is interrupt mode. CCP driver is the foundation of calibration platform. The calibration platform structure is shown as Fig.7. CCP driver achieves data upload, download and calibration of controlled parameters. This calibration system provided reliable, accurate and quickly CAN communication between calibration platform and ECU. It has been used successfully in HEV controlled system. The maximum bits of each flame TR can be calculated by following equation [7]

Design and Analysis of Multi-Node CAN Bus for Diesel Hybrid Electric Vehicle

5. Verify — Design and Analysis of Multi-Node CAN Bus for Diesel Hybrid Electric Vehicle 5. Verify

Fig. 9. Block diagram of hardware in the loop

Fig. 1. Generalised Membership Functions for the j-th sensor
The polynomial for each Sugeno rule is given from the following expression (3):
Based on the general Sugeno rule description the resulting defuzzyfied output is given
from the following expanded equation (4) for time t=t which represents the nominal
system response: — Fig. 1. Generalised Membership Functions for the j-th sensor The polynomial for each Sugeno rule is given from the following expression (3): Based on the general Sugeno rule description the resulting defuzzyfied output is given from the following expanded equation (4) for time t=t which represents the nominal system response:

Where the term 68,, is the perturbation for the n-th rule for the given antecedent conditions.
Where f;(p), fo(p), f3(p) are the parametric functions with respect toa vector p. u,,U, ar
the fuzzy-hybrid system inputs. And g(u>) is a function of the input wu. The mathematica
expression (9) will be associated to the electric propulsion equation. The (H%) tern
represents the fuzzy Sugeno non-singleton type system which will associate to senso
perturbations 58,,=¢,, and thus observe how key variables can potentially drift from thei
expected nominal value for given conditions. Hence, by incorporating the work i
Economou et al. (2007), the Singleton type inference is provided from the followin,
equation:
3 Static error bound models
3.1 Class of Static Isotropic Error Bounds (SIEB) — Where the term 68,, is the perturbation for the n-th rule for the given antecedent conditions. Where f;(p), fo(p), f3(p) are the parametric functions with respect toa vector p. u,,U, ar the fuzzy-hybrid system inputs. And g(u>) is a function of the input wu. The mathematica expression (9) will be associated to the electric propulsion equation. The (H%) tern represents the fuzzy Sugeno non-singleton type system which will associate to senso perturbations 58,,=¢,, and thus observe how key variables can potentially drift from thei expected nominal value for given conditions. Hence, by incorporating the work i Economou et al. (2007), the Singleton type inference is provided from the followin, equation: 3 Static error bound models 3.1 Class of Static Isotropic Error Bounds (SIEB)

For this system the following expression exists:
2.3.3 Class of Static Clustered Isotropic Error Bounds (SCIEB) — For this system the following expression exists: 2.3.3 Class of Static Clustered Isotropic Error Bounds (SCIEB)

For the SIEB type of perturbation equations (10a) and (10b) hold, while for the case of SAEB
the perturbations are unequal for each sensor and therefore the following expression holds:
Where 68, #58, corresponds to the asymmetry of the perturbations in the consequent
fuzzy component. Subject to the constraint (12d):
2.5 Relation of sugeno perturbation and error type classification — For the SIEB type of perturbation equations (10a) and (10b) hold, while for the case of SAEB the perturbations are unequal for each sensor and therefore the following expression holds: Where 68, #58, corresponds to the asymmetry of the perturbations in the consequent fuzzy component. Subject to the constraint (12d): 2.5 Relation of sugeno perturbation and error type classification

The proposed simulation block diagram for the UAV which incorporates a perturbation and the fuzzy-hybrid model for the UAV thrusters is presented as shown next in Figure 5: Figure 6 shows the system’s operational requirements for a near to sea level UAV altitude thus having overall a constant armature resistance. Figure 6. shows a similar block diagram which includes the UAV Altitude Profile, UAV “dry/moist” profile which provide an estimate for the perturbed temperature via a Sugeno-type Fuzzy Inference System (FIS). The armature resistance variation with temperature and the electrical thrusters are based on physical system modelling. The fuzzy Sugeno system produces a nominal armature resistance variation which is related to the UAV altitude and air moisture conditions. Lastly figure 8 shows the Sugeno FIS which produces the perturbed armature resistance values for the electrical thruster. Two inputs, the UAV armature voltage and UAV current profile are Fig. 3. UAV operator and EPS performance for varying altitudes

Electrical Vehicle Design and Modeling

Sign up for access to the world's latest research

Abstract

Figures (424)

Related papers

Related papers