XIII International scientific and technical conference "Dynamics of Systems, Mechanisms and Machines" (Dynamics)

05–07 November 2019, Omsk, Russia

Direct Torque Control of the Axial Flux Permanent

Magnet Motor

Andrew Malyshev Alexander Ivanov

Department of Electrical Engineering, Electric Drive and Department of Open Cast Mining and Electrical
Industrial Electronics Engineering
Siberian State Industrial University Siberian State Industrial University
Novokuznetsk, Russia Novokuznetsk, Russia
E-mail: [email protected] E-mail: [email protected]

Abstract — This paper presents the direct torque control of switching semiconductor elements and frequency drives:
Axial Flux Permanent Magnet motor. The also presents a frequency and vector [4].
mathematical model of the type of motor under study, reflecting
its characteristics. The Direct torque control (DTC) is considered The AC motor control method, called "vector control",
a principle in which we are trying to integrate magnet flux and appeared in the late 70s - early 80s of the last century [5]. The
torque control in one switching algorithm, using the fact that the Vector Control (VC) is the control of a motor element,
voltage can change very quickly when the inverter transistors separately changing the amplitude and frequency of the current
change state. or supply voltage [4]. Based no longer on stationary modes, but
on a mathematical model in differential equations [6],
Keywords — Axial Flux Permanent Magnet (AFPM) Motor, describing the operation of a converter, both in stationary
Direct Torque Control (DTC), Vector Control, Space Vector, modes and in transient behaviors. In turn, vector control is also
control system. divided into two subtypes [7]:
1) Direct Torque Control (DTC): This is a control mode in
I. INTRODUCTION
which the motor torque on the shaft is controlled by measuring
The very first electric machine invented by humankind was and calculating the magnet flux linkage vector and torque on
an Axial Flux Permanent Magnet (AFPM) machine. In 1821, the shaft, and comparing with reference values and sequential
Michael Faraday introduced a primitive disk-shaped machine,
control adjustment [4]. Direct torque control includes two
similar to the AFPM machine. After that, in 1837, Thomas
Davenport received a patent for a Radial Flux Permanent subtypes:
Magnet (RFPM) machine [1], which later became applied both a) Direct Self-Control (DSC): This is a control mode in
in general purpose industrial version and in special application
which harmonic currents or voltages are not approximated by
systems (space technology, medical industry, robotics industry,
a sinusoid, but a meander is created, which intermittently
and defense industry). Recently, due to AFPM have higher
torque to inertia ratio and thus, faster dynamic torque response controls the torque [4]. This method is used for high-power
with respect to RFPM, AFPM machines began to be used for drives with low-frequency-oriented keys.
electric drives, in which previously only RFPM machines were b) Space Vector Modulation (SVM): This is a control
used [2, 3]. mode in which the motor element is three-phase, and the
The AFPM machines are actively developing, and we can control creates a sinusoidal alternating current from a direct
speak about the formation of a new class of rectifier drive current by simultaneously controlling by power electronic
competitive with traditional rectifier drive. There is a process transistor switches, where each turns the controlled vector 120
of transition from the design of single products to the degrees.
development of an industrial series of electric machines of this 2) Field Oriented Control (FOC): This is a control mode
type. in which direct monitoring and control of the magnet flux
In this regard, there is a throw into sharp relief question linkage vector occurs [7].
about the choice of the optimal control system for these DTC technique has faster dynamic torque response relative
machines. There are two main ways to control AC drives using to FOC. In DTC, the corresponding configuration of the
 XIII International scientific and technical conference "Dynamics of Systems, Mechanisms and Machines" (Dynamics)
05–07 November 2019, Omsk, Russia

inverter is selected from the switching table in accordance with where p is the number of pole-pairs and ωR is rotor rpm.
the error signals between the reference values of torque and
stator flux and their actual values to maintain the torque and After analyzing Figure 1, we can write the following vector
stator flux within the hysteresis range. equation:
    
This article discusses the DTC technique with a fixed U S   E  I S RS  j ( I Sd LSd  I Sq LSq ) , (3)
switching frequency for the AFPM motor in order to reduce
pulsations of magnetic flux and torque.

where U S is stator voltage vector;

II. AFPM MOTOR EQUATIONS


I S is stator current vector;
A. Phasor diagram and Electromagnetic torque of AFPM  
motor I Sd and I Sq are the d- and q-axis components of the stator
Figure 1 shows a phasor diagram explaining the processes current vector;
occurring in the motor. We will consider them in the dq frame,
the d-axis of which is oriented along the axis of the rotor. LSd and LSq are the d- and q-axis stator synchronous
inductances;
The motor rotor creates a flux-linkage ( f ) equal to the
product of the rotor flux and the number of turns of the stator RS is stator resistance.
winding. This flux-linkage vector is directed along the rotor d-
The EMF vector and the drop in voltage across the active
axis from the positive to the negative pole.
(RS) and reactive (LSq, LSq) resistances of the stator winding
balance the stator voltage vector (Fig. 1).
The motor torque is created due to the interaction of the
current and the stator flux-linkage (ψS). Given the phase and
the number of pole-pairs, it is equal to the product of these
vectors:
 mp  
M I S  S , (4)
2
where m is the number of phases.
Therefore, for the modulus of torque, we can write the
following equation
mp  
M I S  S sin  , (5)
2
where ε is angle between the vectors of stator current and stator
flux-linkage. Or, moving to the d- and q-axis, the torque is
equal to the equation
mp
M (iSq d  iSd q ) , (6)
Fig. 1. Phasor diagram of the AFPM motor 2
The flux-linkage vector  f is behind the current vector by where d and q are the d- and q-axis components of the
the twist angle φ. The constant flux of the rotating rotor creates stator flux-linkages, which given by equations:
in the stator windings a vector EMF (E) 90 degrees behind it.
 d  LSd i Sd  f ;  q  LSqiSq . (7)
The amplitude of the EMF vector is determined by the
equation: Substituting (7) into (6) gives equation
E  f  , (1) mp
M (iSq f  iSd iSq ( LSd  LSq )) . (8)
where ω is a frequency equal to 2

  p R , (2) For a surface-magnet AFPM motors LSd  LSq , and

 XIII International scientific and technical conference "Dynamics of Systems, Mechanisms and Machines" (Dynamics)
05–07 November 2019, Omsk, Russia

mp The magnetic flux-linkage  is a complex number that is

M  iSq f . (9) constant in the steady state. The value of  leave over in the
2
fixed position in Fig. 2.
After analyzing the expression (9), we can conclude that if
the magnet flux-linkage is fixed, so the torque depends only on Rewrite (14) in x, y axes
the q- axis component of the stator current. That is, the d- axis
component of the current is parasitic, not creating a torque, and    x  j y , (15)
only cause’s incidental losses in the motor.
where the coordinate origin xy coincides with the coordinate
Controlling the q-axis component of stator current iSq origin dq, while the x-axis is rotated relative to the d-axis by
signifies controlling both the magnitude and phases of the an angle  . If  is chosen as Fig. 2 (where the x-axis
current. This phase is defined relative to the rotor position, collocates along the space vector ), then
which also defined the phase of fundamental flux-linkage  f
  x ;  y  0 . (16)
and the EMF.
Fig. 2 also shows that
B. Equilibrium equations of the stator of the AFPM motor
If we neglect a value of magnetic losses, then in a fixed  LSqi Sq
frame of axis for the AFPM motor, as well as for a salient pole   arctan .

(17)
synchronous machine without any rotor winding, the voltage

 f  L i
Sd Sd 
equations for the stator circuit are as follows [1]

d S
u S   iS RS ; (10)
dt
d S
u S   iS RS . (11)
dt
Putting equations (7) into equations (10) and (11), the stator
voltage equations in the frame of axis dq can be written as

diSd
u Sd  LSd  iSd RS  LSqiSq ; (12)
dt
diSq
u Sq  LSq  iSq RS  LSd iSd   f . (13)
dt
In this case, uncompensated components appear in the Fig. 2. Phasor diagram showing dq and xy axes
equations, which are related to the rotational rate of the frame
of axis dq (ω). The component parts of space vector x and y can be
obtained from (7) by the reexpression
III. DTC CONCEPT
Direct torque control (DTC) is based on motor equations
 x   cos  sin    d 
     . (18)
written in a coordinate system that is tied to the air gap flux,
 y   sin  cos    q 
which includes the magnet flux in the d-axis and the armature
reaction fluxes LSd i Sd and LSqi Sq in the d- and q-axis. From (18) can be obtained
Classical theory mainly uses scalar equations for components  d  cos   sin    x 
along the d and q axes, as in (7). DTC are most often expressed      . (19)
using space vectors. These vectors are physical quantities that  q   sin  cos    y 
not only move in time, but also rotate in space [8].
To determine these space vectors, we write (7) as follows Similar to (15), the current is i  i x  ji y  id  ji q ,
and then the torque (8) can be rewritten as follows
   d  j q . (14)
mp mp
M  ( x i y  y i x )    iy , (20)
2 2
 XIII International scientific and technical conference "Dynamics of Systems, Mechanisms and Machines" (Dynamics)
05–07 November 2019, Omsk, Russia

where  is the amplitude

Fig. 3. Solid State AC-AC converter and equivalent 3-phase AFPM motor
  x . connection

This equation is identical in form to (9). However now  The inverter shown in Figure 3 has 6 power electronic
is the total flux linkage, which can be adjusted. This makes it transistor switches. The switches are usually used Insulated-
possible to control the torque by regulation either  or i y , or Gate Bipolar Transistors (IGBT) or MOS field effect
transistors (MOSFET).
both of them.
The complete power-supply circuit shown in Figure 3 can
Using the backward transformation (19), the current is be simplified and shown as in Figure 4.

iy 
1
LSd
 
sin   f   ( LSq  LSd ) cos  . (21)

For the Surface mounted AFPM motor with

LSd  LSq  LS this equation will be rewritten as follows

f
iy  sin  . (22)
LSd
With all that said, the torque of the AFPM motor is

mp  f
M   sin  . (23) Fig. 4. Simplified representation an inverter-fed AFPM motor
2 LSd
After analyzing equation (23), we can conclude that torque If the inverter is connected to the AFPM motor stator
windings, as shown in Figure 4, voltage vectors in system of
can be increased or decreased by increasing or decreasing  - axes α -β can be determined at the position of the 3 switches
that is, by control the phase of the total flux-linkage. The (Sa, Sb, Sc), so that their respective parts are
principle of the DTC concept judge by from the voltage
equation Vdc
vS( )  (2S a  S b  S c ) , (25)
d 3
u  Ri  . (24)
dt Vdc
v S(  )  (S b  S c ) . (26)
In this equation, the resistance R can be neglected, 3
because it is usually small. Then it is hard to escape a
conclusion that the rate of change of flux linkage  is equal to According to the position Sa, Sb and Sc, the inverter has 8
the terminal voltage u . This means that the tip of the magnetic switching states u0, u1 … u7 including 2 zero-voltage states u0
and u7, Fig. 5. By quickly switching of the voltage status of the
flux linkage vector moves in a direction parallel to the spatial
inverter, the flux linkage  can be limited so that its path is a
voltage vector.
more-or-less circular trajectory. The angle between each pair of
Figure 3 shows the complete power-supply circuit for a 3- six nonzero vectors (u1, u2 …u6) is 600 as shown in Fig. 5 [9].
phase AFPM motor. In this circuit, the 3-phase AC supply is
first rectified to obtain a fixed DC voltage, and then rectified
output is inverted to a 3-phase AC voltage for AFPM motor.
 XIII International scientific and technical conference "Dynamics of Systems, Mechanisms and Machines" (Dynamics)
05–07 November 2019, Omsk, Russia

   (u  Ri)dt . (27)

If we take into account that during the switching period of

inverter the voltage vector is constant, equation (27) can be
written as

ut  R  idt   t  0 . (28)

Because the stator resistance is usually small, it can be

neglected. Then, according to (28), the trajectory of the stator
flux vector will move in the direction of the vector of the
applied voltage (Fig. 5). At the instant of switching the stator
flux linkage is  t  0 .

Having studied the presented Fig. 6, we can conclude that

to control the magnetic flux linkage vector of the AFPM motor
Fig. 5. The movement of stator flux in the α – β plane stator, it is necessary to adjust the along voltage vector, and to
control the rotation speed of the flux linkage vector, it is
necessary to adjust the vertical component of the voltage
vector.
In the case of AFPM motor (as contrasted with induction
motor), flux linkage  will non-constant value when zero
space vectors u0 and u7 are applied since the permanent
magnets rotate with rotor. Therefore, the use of zero-voltage
states to control the AFPM motor is impractical. The rotation
sense of the magnetic flux linkage  is determined by the
output signal of the hysteresis regulator for torque.
In order for the flux linkage  rotate all the time, need to
select the voltage vectors in the manner shown in Table 1. The
outputs of the hysteresis controllers for flux linkage and torque
are dψ and dM, respectively. Region numbers for the flux
linkage position are θ1, θ2 … θ6.
It will be observed that for the k-th sampling instant the
flux linkage   and   are calculated by discrete time model
Fig. 6. Action of components of voltage space vector on amplitude and
rotational speed of stator flux
as [9]
  (k )    k 1  (v k 1  Ri  )TS , (29)
TABLE I. OPTIMAL SWITCHING TABLE FOR DTC OF AFPM MOTOR
  (k )    k 1  (v k 1  Ri  )TS , (30)
θ
ψ M
Θ(1) Θ(2) Θ(3) Θ(4) Θ(5) Θ(6)
where TS is the quantization interval; k – 1 are preceding
dM = 1 u2 u3 u4 u5 u6 u1
dψ = 1 quantization intervals.
dM = 0 u6 u1 u2 u3 u4 u5
dM = 1 u3 u4 u5 u6 u1 u2  (k )   2 (k )  2 (k ) . (31)
dψ = 0
dM = 0 u5 u6 u1 u2 u3 u4 Given the above equations, the torque for the 3-phase
AFPM motor is equal to
As a rough approximation of the flux linkage  path to a
p  (k )i (k )   (k )i (k ) .
3
circular locus, an example of an 18-sided polygon shown in M (32)
Fig. 5. 2
Integrating equation (24), the stator flux linkage of the After analyzing equations (29-32), it can be seen that they
AFPM motor in the fixed frame of reference is do not depend on the parameters of the AFPM motor, except
for the stator resistance. It is interesting to note that for
 XIII International scientific and technical conference "Dynamics of Systems, Mechanisms and Machines" (Dynamics)
05–07 November 2019, Omsk, Russia

permanent magnet motors, the initial values  and  are The DTC block diagram is shown in Fig. 7. The calculation
of torque and magnetic flux is carried out according to
not zero. The values of these parameters can be obtained from equations (29-32) in the fixed frame of reference α – β. Then
the encoder. You can also do without the encoder, for example, the active values of torque and magnetic flux are compared
when using an external magnet. Then the rotor is located in a with their set values in the hysteresis controllers. In accordance
known position and we can estimate the initializing values   with dM and dψ (which are the output signals of the hysteresis
controllers), as
and   .

IV. SIMULATION RESULTS

To test the behavior of the proposed Direct Torque Control
of the AFPM motor, a Simulink model was created, the
simulation parameters of which are listed in Table 2.

Fig. 7. DTC block diagram of AFPM motor

well as θ, the voltage space vector is selected, and thus the Parameter Value
switches state will be selected from the switching table 1.
Rated torque MN (N·m) 11
TABLE II. PARAMETERS OF THE AFPM MOTOR Power switches of voltage source inverter IGBT

Parameter Value
For the current model, it is assumed that the bandwidths of
Poles p 4 torque and magnet flux hysteresis controllers are 0.01·M* and
Stator resistance R (Ω) 0.2 0.01 ·ψ* respectively. When modeling, it is also assumed that
when starting up, the rotor is in the starting position (   0 );
Stator inductance LSd, LSq (mH) 8.5
therefore, the initial flow area is equal to 1 . The variation
Rotor magnet flux ψf (Wb) 0.175
range of the reference torque is from 11 to -11 N·m at t = 0.175
2
Rotor inertia j (kg·m ) 0.089 s, and from -11 to 11 N·m at t = 0.175 s.
Damping coefficient B 0.005

DC voltage VDC (V) 250

Rated speed ωr (rpm) 300

 XIII International scientific and technical conference "Dynamics of Systems, Mechanisms and Machines" (Dynamics)
05–07 November 2019, Omsk, Russia

Fig. 8. Torque characteristics

Fig. 11. 3-phase stator currents of AFPM motor

The torque characteristics is shown in Fig. 8. This figure

shows that when a set torque changes, the reference torque
very quickly follows the set torque. The reference torque is
controlled within the bandwidth when the set torque is
Fig. 9. Reference and stator flux characteristics constant, i.e. when the system is in steady state.

After examining the Fig. 6 and (28), it is clear that the The magnetic flux of the stator, controlled within the
smaller the sample time, the smaller the flux and torque. For bandwidth, is shown in the Fig. 9.
the model under study, the sample time TS = 10 μs was chosen. The trajectory of ratio of   and   is a circle as expected
that is shown in Fig. 10.
The 3-phase stator currents of AFPM motor is shown in the
Fig. 11. This figure shows a pulsating current, which can lead
to the appearance of harmonics and additional losses in the
stator windings. In the following papers, methods for reducing
pulsation currents will be studied.

V. CONCLUSION
Currently AFPM motors are becoming increasingly popular
due to its energy efficiency, as well as reducing the cost of
active materials of this type of motors. For example, the prices
Fig. 10. The ratio of the magnetic flux axis of rare-earth permanent magnets in the 90s of the 20-th century
followed a downward curve with a sharp decline over the past
three years. Also contributes to the development of this type of
motors and the rapid development of microprocessor
technology used in control systems.
In this paper, the DTC algorithm for the AFPM motor has
been proposed. The simulation results confirm that the
vibrations of the stator magnetic flux and torque are
significantly reduced compared with the basic FOC.
The proposed DTC control method can be integrated with
the speed evaluation scheme, and then control without speed
transducer will be achieved.

REFERENCES
 XIII International scientific and technical conference "Dynamics of Systems, Mechanisms and Machines" (Dynamics)
05–07 November 2019, Omsk, Russia

[1] Gieras, J.F., Wang R-J, Kamper M.J. Axial Flux Permanent Magnet [5] Kalachev YU.A. Vektornoye regulirovaniye, Zametki praktika:
Brushless Machines. Springer Science, 2008. DOI: 10.1007/978-1-4020- uchebnoye posobiye [Vector control, Practice notes: teaching guide].
8227-6 EFO Publ., 2013.
[2] K. Sitapati and R. Krishnan, "Performance comparisons of radial and [6] Apollonskiy S.M. Differentsial'nyye uravneniya matematicheskoy fiziki
axial field, permanent-magnet, brushless machines," in IEEE v elektrotekhnike [Differential equations of mathematical physics in
Transactions on Industry Applications, vol. 37, no. 5, pp. 1219-1226, electrical engineering]. St. Petersburg, Piter Publ., 2012.
Sept.-Oct. 2001. [7] Siami, M., & Gholamian, S. A. (2012). Predictive Torque Control of
[3] J. Jung and W. Hofmann, "Comparison of two concentrated winding Three Phase Axial Flux Permanent Magnet Synchronous Machines.
topologies applied on an axial flux permanent magnet machine," 17th Majlesi Journal of Electrical Engineering, 6(2). Retrieved from
European Conference on Power Electronics and Applications (EPE'15 http://mjee.iaumajlesi.ac.ir/index/index.php/ee/article/view/538
ECCE-Europe), Geneva, 2015, pp. 1-10. [8] Hendershot J.R. and Miller T.J.E., Design of Brushless Permanent-
[4] Vinogradov A.B., Gnezdov N.E., Glebov N.A., Zhuravlev S.V. Magnet Motors, Motor Design Books LLC Publications, 2010.
“Recordkeeping of steel saturation and loss in optimum vector control of [9] Siami, Mohsen & Asghar Gholamian, S & Yousefi, Mosayeb, “A
asynchronous traction electric drive,” Izvestiya ISPEU, 2012, issue 1, Comparative Study Between Direct Torque Control and Predictive
pp. 1-7. Torque Control for Axial Flux Permanent Magnet Synchronous
Machines,” Journal of Electrical Engineering, 64. DOI: 10.2478/jee-
2013-0052.