FOC and Flux Weakening for Traction Drive with

Permanent Magnet Synchronous Motor

Jiri Simanek1, Jaroslav Novak2, Ondrej Cerny1, Radovan Dolecek1
1
University of Pardubice, Jan Perner Transport Faculty, Studentska 95, Pardubice 532 10, Czech Republic
[email protected], [email protected], [email protected]
2
Czech Technical University, Faculty of Mechanical Engineering, Technická 4, Praha 166 07, Czech Republic
[email protected]
Abstract–This paper analyses control algorithm for permanent brushless PM motors have been avoided due to its worse
magnet synchronous traction motor (traction PMSM) usable for dynamic behavior [2], [3].
light rail vehicle. The algorithm is based on field oriented control.
The field oriented control and carrier based pulse width This paper is organized as follows: Section II deals with
modulation allow simply implementation of flux weakening theoretical background of PMSM and its flux weakening.
control algorithm. The flux weakening provides wider speed Proposed algorithm is presented in section III. Simulations and
operation of traction drive with PMSM. Simulations and experimental results are given in sections IV and V. Section V
experimental results are given. For experiments we have used a discusses results and behavior of control algorithm from
special experimental stand with a prototype of tram traction
PMSM. Power of the PMSM is 58 kW. Description of the stand is traction drive point of view.
given in the paper too. II. MATHEMATICAL MODEL OF PMSM
Index terms–Field oriented control (FOC) PMSM means that motor is excited by permanent magnets.
Due to flux and dumping winding absence the mathematical
I. INTRODUCTION
model is simplified considering classical synchronous motors.
Traction drives of rail vehicles with permanent magnet Fig. 1 shows equivalent circuit. A non-salient pole motor is
synchronous motor (PMSM) are not traditional. However supposed in our application thus we can suppose similar direct
usage of PMSM brings several advantages. Thanks to high and quadrature inductance [4]. Thanks to this, the electric
torque-weight ratio the PMSM enables so-called direct drive of torque of the PMSM is proportional only to a torque
wheel set or single wheel. The direct drive has higher component. PMSM model is given as in (1) - (3) [5]:
efficiency then traditional system (motor – gearings – wheel
set) and makes possible 100 % low floor vehicle. d 1 R Lq
Traction drive with PMSM has to meet with specific id = ud − id + pωr iq (1)
dt Ld Ld Ld
requirements. Instead of industry drives the traction drive has
to operated in wide voltage range (e.g. from -33% to +20 % of d 1 R L Ψ pω
iq = uq − iq − d pωr iq − PM r (2)
nominal overhead line voltage) and voltage surges are dt Lq Lq Lq Lq
M i = 1,5 p ⎡⎣ Ψ PM iq + ( Ld − Lq ) id iq ⎤⎦
common. Also flux weakening operation in wide speed range
(3)
is supposed. The speed range of flux weakening can be up to ⅔
of whole speed range. where id, iq are flux and torque current component, ud, uq are
PMSM drive also brings several disadvantages or problems. terminal voltages in d-q coordinates, R is stator winding
Parallel coupling of PMSM is not possible. A position sensor resistance, Ld, Lq are direct and qudrature inductance, p is pairs
such as an encoder or a resolver is required to synchronize the of poles number, ΨPM is permanent magnet flux and Mi is
phase currents to the rotor position. A sensorless control is not electric torque.
suitable due to problems at zero speed of rotor [1]. If a flux For non-salient PMSM, we can use simplified equivalent
weakening is used, DC bus and converter should be designed circuit (Fig. 1.). Phasor diagram of constant flux operation can
to higher voltage due to high emf (electromotive force) of be derived from the equivalent circuit (Fig. 2).
PMSM in case of converter failure during flux weakening
operation.
Aim of the paper is analysis of simply torque control
algorithm for PMSM traction drive for tram which is based on
field oriented control and which enables flux weakening
operation. The control algorithm avoids mathematical models
of PMSM to reach motor parameters independence. The field
oriented control reach excellent dynamic behavior and it
enables simply implementation of flux weakening algorithm. Fig. 1. PMSM equivalent circuit where U is terminal voltage, I
Control algorithms based on sinusoidal commutation of is phase current, L is inductance (L = Ld = Lq), R is winding
resistance and Ui is induced voltage.

978-1-4244-1666-0/08/$25.00 '2008 IEEE 753

 We decided to use flux oriented control (FOC) [6] thanks to
its robust behavior in wide speed (frequency) range. FOC for
PMSM is simplified considering asynchronous motor thanks to
fact that no mathematical models are needed to determinate
flux space vector position. The position we can derive from
angular position sensor (e.g. resolver). This is great advantage
– the control is not dependent on PMSM characteristics
(especially resistance which rise with temperature of winding).
Block scheme of FOC is shown in Fig. 4.

Fig. 2. Phasor diagram of constant flux operation of PMSM

A. Flux Weakening of PMSM
In constant flux operation mode, torque is maximal when we
keep the flux current component to zero. That is possible until
induced voltage is higher then maximal terminal voltage. The
induced voltage is proportional to speed, thus the constant flux
operation is possible to so-called nominal speed. Over the
nominal speed the torque decreases rapidly to its minimum
value. To expand speed range above the nominal speed, flux
reducing is necessary. Flux of permanent magnets is constant
and we cannot reduce it (comparing with a classical
Fig. 4. Scheme of FOC with flux weakening
synchronous motor). However we can reduce excitation flux in
air gap and stator when we produce negative flux current
Clarke and Park transformation is used for the control
component id, as shown in Fig. 3.
algorithm. Forward transformation is performed by abc/dq
block and the inverse is performed by dq/abc block. The d-q
coordinates are synchronized with rotor flux position thus the
phase currents is performed as constant values. Thanks to this
the behavior of the control is independent on generated
frequency.
Desired torque M* is proportional to torque component iq*.
Desired flux component is generated by flux weakening
controller. This controller also reduces torque component
during flux weakening operation.
B. Flux weakening controller
There is a few flux weakening control strategies [7]. We
used so-called “Optimum current vector control” based on
Fig. 3. Phasor diagram during flux weakening operation desired voltage feedback. The strategy enables maximum
torque profile during whole flux weakening operation.
Note: The phase current can have capacitive power factor – Principle of flux weakening control is obvious as shown in Fig.
it is similar to overexcited synchronous motor. 5.
III. PMSM CONTROL STRATEGY
Torque control is base for common traction drive. If we
supposed non-salient PMSM, equation (3) can be simplified as
in (4).
M i = 1,5 pΨ PM iq (4)
We can reach the best utilization of current and also the
highest efficiency of drive if the flux current component is set
to zero. In fact, it means the current space vector is
perpendicular to flux. This is the aim of control strategy.
Fig. 5. Flux weakening control block

754
 R = 0.087 Ω
If desired voltage magnitude |U| reach limit which is set by Ld = 0.8 mH
Umax, the PI regulator begins generate negative flux component Lq = 0.8 mH
id*. The flux in air gap is weakened by this component thus the ΨPM = 0.2 Wb
induced voltage gets lower. To avoid generation of positive p = 22
flux component, limitation of the PI controller output is J = 2 kg.m2
necessary (also integrator output should be limited). The flux
weakening controller keeps constant voltage output during flux Inertia torque of PMSM was simulated lower to reach faster
weakening operation of the drive. simulations. There is no influence to simulated control
If the flux component is non-zero, the current magnitude algorithms.
increases. To avoid current overload, the flux weakening Control system model includes influence of dead times of
controller also reduces desired torque component iq*’. converter and influence of angular position sensor resolution.
The drive was simulated with conditions similar to real
IV. SIMULATION OF THE DRIVE experimental drive. Conditions were: converter switching
frequency was 4 kHz, DC bus voltage was 560 V, dead times
The application MATLAB Simulink was used for were 2 μs and sensor resolution was 8,192 pulses per
simulations of control algorithms and PMSM drive. The model revolution.
consists of these models: PMSM model, converter model and Simulation sequence of drive with rated current is shown in
control system model. Fig. 6. Transient effects of flux weakening during setpoint
Parameters of PMSM were the same like real PMSM steps and voltage surges are shown in Fig. 7. – 10.
described in section V. PMSM model parameters were:

Fig. 6. Simulation sequence of PMSM drive – starting, breaking and speed reversal. From above: speed of drive, inner torque of
PMSM, desired current components (black) and actual current components (torque component – red, flux component – green)
and desired voltage (absolute value).

755
Fig. 7. Step down of desired torque from 100 % to 15 % during Fig. 8. Step up of desired torque from 15 % to 100 % during
flux weakening operation flux weakening operation

Fig. 9. Step down of DC bus voltage from 600 V to 420 V Fig. 10. Step up of DC bus voltage from 420 V to 720 V during
during flux weakening operation flux weakening operation

TABLE I
V. EXPERIMENTAL RESULTS AND EXPERIMENTAL SETUP
MOTOR PARAMETERS
Experiments have been carried out using a special stand with Nominal power 58 kW
a 58 kW traction PMSM. A photo of the stand and auxiliary Nominal torque 852 Nm
systems is shown in Fig. 11. The stand consists of PMSM, tram Nominal speed 650 rpm
wheel and “endless” rail. The PMSM is a prototype for low Nominal current 122 A
floor trams. Motor parameters are given in Tab. 1. The stand Maximal torque 2000 Nm
was loaded by an asynchronous motor. Maximal speed 1000 rpm
An IGBT inverter was used for feeding of PMSM. A photo Maximal current 368 A
is shown in Fig. 12. For control, a DSP TMS320F240 by Texas Number of poles 44
Instruments was used.

756
 weakening was 65 % instead of simulations. Also torque was
reduced during steps by safety reasons.
Break chopper
Testing stand with PMSM

=
=
PMSM
Converter

3 x 400 V / 50 Hz
= ASM
~
Inverter
=
~

Fig. 13 Layout of the experimental setup

Experimental data were downloaded from the DSP. The data

are shown in Fig. 14 – 18. Experimental conditions were
Fig. 11. Experimental setup with experimental stand
similar to simulation (converter switching frequency 4 kHz,
DC bus voltage 560 V, dead times 2 μs).

Fig. 14. No-load start of drive and full flux to flux weakening
transition, desired torque of 20 %

Fig. 12. IGBT inverter for PMSM (Semikron)

Inverter was connected to DC bus of loading machine

converter. It enables effective operation of stand. A layout of
the experimental setup is shown in Fig. 13.
Four-pole resolver is used for angular position sensing of
PMSM. Resolver output signals are processed by integrated
resolver-to-digital converter (R/D converter) [8]. The converter
resolution is 12 bits it means 4069 positions per half a
mechanical turn or 372 positions per electric turn of motor.
Maximum operation speed of drive was up to nominal speed
of the motor to avoid dangerous overvoltage in DC bus during
faults of inverter at higher speed than nominal. To reach flux Fig. 15. Torque component limiting during flux weakening
weakening operation, the maximum voltage threshold for flux operation, desired torque of 60 %

757
 VI. CONCLUSION
The paper analyses behavior of FOC during field weakening
operation. Transient effects during setpoint steps and voltage
surges were especially examined. No mathematical model of
PMSM is the greatest advantage of proposed control algorithm.
A utilization of current during flux weakening is also highest
instead of other strategies based on keeping of constant active
power.
However disadvantages of proposed field weakening are
noticeable from simulation and experimental results. The flux
weakening control is relative slow. It causes overshoots in
actual values during setpoint steps and voltage surges. To
avoid overshoots during setpoint steps, setpoint ramps should
be used. Transient effect during voltage surges cannot be
Fig. 16 Torque reversal from 50 % to -50 %, speed 76,5 rpm stifled. Step up of DC bus voltage is especially critical due to
step up of torque. The torque step up could cause wheel slip on
rail and also make a ride uncomfortable. The torque steps are
caused by the torque component limit block in the flux
weakening controller. To avoid the torque steps, a prediction of
torque component could be better then pure limitation derived
from flux component value.
Output power of PMSM is not constant during flux
weakening operation since torque decrease more rapidly then
nominal torque / speed ratio. It can be disadvantage especially
for traction drives with flux weakening in wide speed range.
The proposed control algorithm principle is suitable for
traction drive regarding necessary modifications mentioned
above.
REFERENCES
[1] H. A. Toliyat, L. Hao, D. S. Shet, and T. A. Nondahl, “Position
Sensorless Control of Surface Mount Permanent Magnet AC (PMAC)
Fig. 17. Step down of desired torque (30 % to 5 %) during flux Motors at Low Speeds,” IEEE Transactions on Industrial Electronics,
weakening operation, speed 437 rpm vol. 49, pp. 157 – 164, Feb 2002.
[2] M. Novak, M. Cambal, and J. Novak, “Application of Sinusoidal Phase
Current Control for Synchronous Drives,” in IEEE International
Symposium on Industrial Electronics, 2006, pp. 2260 – 2265.
[3] J. Simanek, J. Novak, R. Dolecek, and O. Cerny, “Control Algorithms for
Permanent Magnet Synchronous Traction Motor,” in IEEE International
Conference on Computer as a Tool, 2007, pp. 1839 – 1844.
[4] J. F. Gieras, and M. Wing, Permanent Magnet Motor Technology,
Eastern Hemisphere Distribution, Basel, 2002.
[5] Matlab Help, The MathWorks, Inc., 2007
[6] E. Simon, “Implementation of a Speed Field Oriented Control of 3-phase
PMSM Motor using TMS320F240,” Texas Instruments, Inc., App. Rep.
SPRA588, 1999.
[7] Z. Mihailovic, “Modeling and Control Design of VSI-Fed PMSM Drive
Systems with Active Load,” M. S. Thesis, Virginia Polytechnic Institute
and State University, Virginia, ON, United States, 1999.
[8] M. Cambal, M. Novak and J. Novak, “Study of Synchronous Motor
Rotor Position Measuring Methods”, in 13th International Conference on
Electrical Drivers and Power Electronics, 2005, pp. 62 – 66.

Fig. 18. Step up of desired torque (5 % to 30 %) during flux

weakening operation, speed 437 rpm

758