154

Chapter 5

Decoupled PWM Algorithm Based Open-End

Winding Induction Motor Drive

5.1 Introduction:
A simple generalized PWM algorithm has been presented in the
previous chapter for a diode-clamped multilevel inverter fed DTC-IM
drive. Nowadays, in medium and high power drive applications, the
open-end winding induction motor drives are becoming popular due to
their numerous advantages. This chapter presents a simplified
decoupled PWM algorithm for open-end winding induction motor
drive. In the proposed method, the open-end winding induction motor
fed by two 2-level inverters at either end which, produces space vector
locations, identical to those of a conventional 3-level inverter. The
proposed PWM algorithm does not employ any look-up tables and time
consuming task of sector identification. The proposed algorithm has
been developed by using the concept of imaginary switching times,
which are proportional to the instantaneous phase voltages. Thus, the
proposed algorithm reduces the complexity when compared with the
conventional SV approach.
5.2 Open-End Winding Induction Motor Drive:
Fig.5.1 shows the basic open-end winding induction motor drive
operated with a single power supply. The symbolsV AO , V BO and VCO

155

,
denote the pole voltages of the inverter-1. Similarly, the symbols V AO
and VCO
denote the pole voltages of inverter-2. The space vector
V BO
locations from individual inverters are shown in Fig. 5.2. The numbers
1 to 8 denote the states assumed by inverter-1 and the numbers 1
through 8 denote the states assumed by inverter-2 (Fig. 5.2).

Vdc/4

O
Vdc/4

S1

A
S4

Open-End wdg.
Induction Motor

S5

S3

S5l

S1l

S3l

B
A
S2

S6

S2l

S6l

S4l

Inverter 2

Inverter 1

Fig.5.1 The primitive open-end winding induction motor drive.

3(-+-)

4(-++)

7(+++)

1(+--)

4(-++)

2(++-)

7(+++)

1(+--)

8(---)

8(---)

5(--+)

3(-+-)

2(++-)

6(+-+)

5(--+)

Vdc/2

6(+-+)
Vdc/2

Fig. 5.2 Space vector locations of inverter-1 (Left) and

inverter-2 (Right).

Table 5.1 summarizes the switching state of the switching

devices for both the inverters in all the states.

In Table 5.1, +

indicates that the top switch in a leg of a given inverter is turned on

156

and - indicates that the bottom switch in a leg of a given inverter is

turned on. As each inverter is capable of assuming 8 states
independently of the other, a total of 64 space vector combinations are
possible with this circuit configuration. The space vector locations for
all space vector combinations of the two inverters are shown in Fig.
5.3. In Fig.5.3, |OA| represents the DC-link voltage of individual
inverters, and is equal to Vdc 2 while |OG| represents the DC-link
voltage of an equivalent single inverter drive, and is equal toVdc .
(36)
K

(25)
I

(35, 26)
J
10

12

(31, 46) L
14

45
21 38
C
86
76
37

85
34
B
28 75
27

1
11
22, 77 66, 88
O
55, 87
33, 78
44
6
4

H (15, 24)
8
7

3
71
81

16

16

15

(41) M

11

13

47
D
32

48
56

18

65
18
A
74
23
84

17

(51, 42) N

17

57
58
43
E
61
72
82

20

19

(52) P

G (14)
24

23

83
12
73
F
67
68
54

S (13, 64)
22

21

Q
(53, 62)

R
(63)

Fig. 5.3 Resultant space vector combinations in the dual-inverter

scheme.

Fig.5.1 shows the basic open-end winding induction motor

drive. It cannot be operated with a single power supply, due to the
presence

of

zero-sequence

voltages

(common-mode

voltages).

Consequently, a high zero-sequence current would flow through the

157

motor phase windings, which is deleterious to the switching devices

and the motor itself. To suppress the zero-sequence components in
the motor phases, each inverter is operated with an isolated dc-power
supply as shown in Fig. 5.4.

Table 5.1 Switching states of the individual inverters.

Vdc/4

O
Vdc/4

State of
inverter 1
1 (+--)

Switches
Turned ON
S 6, S 1, S 2

State of
inverter 2
1 (+--)

Switches
Turned ON
S6, S1, S2

2 (++-)

S 1, S 2, S 3

2 (++-)

S1, S2, S3

3 (-+-)

S 2, S 3, S 4

3 (-+-)

S2, S3, S4

4 (-++)

S 3, S 4, S 5

4 (-++)

S3, S4, S5

5 (--+)

S 4, S 5, S 6

5 (--+)

S4, S5, S6

6 (+-+)

S 5, S 6, S 1

6 (+-+)

S5, S6, S1

7 (+++)

S 1, S 3, S 5

7 (+++)

S1, S3, S5

8 (---)

S 2, S 4, S 6

8 (---)

S2, S4, S6

S1

S4

S3

S5
C

Open-End wdg.
Induction Motor

S5l

S3l

S1l

Vdc/4

C
B

O
A

S6

S2

Inverter 1

S2l

S6l

S4l

Vdc/4

Inverter 2

Fig. 5.4 The open-end winding induction motor drive with two
isolated power supplies.

158

From the Fig.5.4, when isolated DC power supplies are used for
individual inverters, the zero-sequence current cannot flow as it is
denied a path. Consequently, the zero-sequence voltage appears
across the points O and O'. The zero-sequence voltage resulting from
each of the 64 space vector combinations is reproduced in Table 5.2.

Table-5.2: Zero sequence voltage contributions in the difference

of the pole-voltages of the individual inverters.
0
Vdc 2
Vdc 3
V dc 6
Vdc 6
Vdc 3
Vdc 2
8-7

8-4

8-5, 8-3

8-8, 5-5

5-8, 3-8 4-8

8-6

5-4, 3-4

5-3, 3-5

4-5, 4-3 6-8

8-2

8-1, 5-6

3-3, 4-4

1-8, 6-5 2-8

5-7

5-2, 3-6

5-1, 3-1

2-5, 6-3 7-5

3-7

3-2, 4-7

4-6, 4-2,

2-3, 7-4 7-3

1-7

1-4, 1-6

1-5, 1-3

4-1, 6-1 7-1

1-2, 6-7

6-4, 2-4

2-1, 7-6

2-7

1-1, 6-6

7-2

7-8

6-2, 2-6
2-2, 7-7

In Fig. 5.5, the vector OT represents the reference vector (also called
the reference sample), with its tip situated in sector-7 (Fig. 5.3). This
vector is to be synthesized in the average sense by switching the space
vector

combinations

situated

in

the

closest

proximity

(the

combinations situated at the vertices A, G and H in the present case)

159

using the space vector modulation technique. In the work reported in

reference [87], the reference vector OT is transformed to OT in the
core

hexagon

ABCDEF

by

using

an

appropriate

coordinate

transformation, which shifts the point A to point O.

B-Ph axis

B
W

T
V

A-Ph axis

C-Ph axis

Fig. 5.5: Resolution of the reference voltage space vector in the

middle and outer sectors.

In the core hexagon, the switching timings of the active vectors

OA, OB and the switching time of the null vector situated at O to
synthesize the transformed reference vector OT are evaluated. The
switching algorithm described in reference [80] is employed to
evaluate these timings. These timings are then employed to produce
the actual reference vector OT situated in sector-7 by switching

160

amongst the switching combinations available at the vertices A, G and

H. The latter step requires a lookup table in which the space vector
combinations available at each space vector location are stored. Thus,
it is evident that with this switching algorithm, the controller
negotiates a considerable computational burden primarily because of
sector identification and coordinate transformation. Also, there is a
need

requirement

for

look-up

tables,

enhancing

the

memory

requirement. Further, the zero-sequence voltage in the difference of

the respective pole voltages of individual inverters (which is dropped
across the points O and O in Fig. 5.4) is also high with this PWM
scheme.

5.3 Proposed Decoupled PWM Algorithm:

The proposed PWM strategy is based on the fact that the
reference voltage space vector Vref can be synthesized with two equal
and opposite components Vref / 2 and Vref / 2 , by subtracting the
latter component from the former. It is also based on the observation
that the effect of applying a vector with inverter-1 while inverter-2
assumes a null state is the same as that of applying the opposite
vector with inverter-2 while inverter-1 assumes a null state. Fig. 5.6
shows the method of this PWM strategy. It is worth noting that the
phase axes of the motor viewed with reference to individual inverters
are in phase opposition.
In Fig.5.6, the vector OT represents the actual reference voltage
space vector that is to be synthesized from the dual-inverter system

161

and is given by Vref . This vector is resolved into two equal and

opposite components OT1 Vref / 2 and OT2 Vref / 2180 0 + . The

vector OT1 is synthesized by inverter-1 in the average sense by
switching amongst the states (8-1-2-7) while the vector OT2 is
reconstructed by inverter-2 in the average sense by switching amongst
the states (8-5-4-7).

B-ph axis
T1

7,8

A-ph axis

A-ph axis
3'

4'
J

A-ph axis

2'

7',8'

1'

T1

C-ph axis

5'

6'

Fig. 5.6 The proposed decoupled PWM strategy.

The simplified switching algorithm, which is described in

chapter-4 for the classical case of a 2-level inverter feeding an
ordinary induction motor is extended for the dual-inverter system to
compute the switching timings for individual inverters. The proposed
algorithm uses only the instantaneous phase reference voltages and is

162

based on the concept of effective time as follows:

In the proposed decoupled PWM algorithm, when the reference voltage
vector falls in the first sector of inverter-1, the imaginary switching
time which is proportional to the a-phase ( Tan ) has a maximum value,
the imaginary switching time which is proportional to the c-phase
( Tcn ) has a minimum value and the imaginary switching time which is
proportional to the b-phase ( Tbn ) is neither minimum nor maximum
switching time. Thus, in general to calculate the active vector
switching times, the maximum and minimum values of imaginary
switching times are calculated in every sampling time as given in (5.1)
(5.2).
T max = Max (Tan , Tbn , Tcn )

(5.1)

T min = Min (Tan , Tbn , Tcn )

(5.2)

The effective time Teff can be defined as the time difference between
T max and T min and can be given as in (5.3).

Teff = T max T min

(5.3)

The effective time means the duration in which the effective

voltage is supplied to the machine terminals. In the actual switching
instants, there is one degree of freedom that the effective time can be
located anywhere within one sampling interval. To generate actual
switching pattern which preserves the effective time, the zero
sequence time is subjoined to the phase voltage time. In order to
locate the effective time in centre of the sampling interval, the zero

163

sequence voltage has to be symmetrically distributed at the beginning

and end of one sampling period. Therefore, the actual switching times
for each inverter leg can be simply obtained by the time shifting
operation as below.

T ga = Tas + Toffset
T gb = Tbs + Toffset
T gc = Tcs + Toffset

(5.4)

To distribute zero voltage symmetrically during one sampling

period, the offset time Toffset is achieved using a simple sorting
algorithm. The zero voltage vector time duration can be calculated as
given in (5.5).

T zero = Ts Teff

(5.5)

And, T min + Toffset T zero / 2

(5.6)

Therefore, Toffset = Tzero / 2 Tmin

(5.7)

In order to generate symmetrical switching pulse pattern within

two sampling intervals, when the switching sequence is ON sequence,
the actual switching time will be replaced by the subtraction value
with the sampling time as fallows.

Tga ,gb,gc = Ts Tga ,gb,gc

(5.8)

As described above, the effective time implies the applied time of a

certain active vector. Therefore, with the effective vector concept, the
actual switching time can be obtained directly from the stationary
frame reference voltage without sector identification, effective time

164

calculation and recombination. the similar procedure is adopted for

inverter-2 also.
In the context of a dual inverter drive, there exist two sets of
phase switching times, one for each inverter. The timings Tga , Tgb and

Tgc

'
'
'
correspond to inverter-1 while the timings Tga
, Tgb
and Tgc

correspond to inverter-2. The instantaneous reference phase voltages

Va* ,Vb* andVc* correspond to the actual reference space vector Vref of
the dual-inverter system. As individual inverters operate with the
references Vref / 2 and Vref / 2 respectively, it follows that the
corresponding phase references are given by Va* / 2,Vb* / 2 and Vc* / 2
for inverter-1 and Va* / 2,Vb* / 2 and Vc* / 2 for inverter-2. These
references are then employed to determine the phase switching
timings

of

each

inverter

using

the

aforementioned

switching

algorithm. Thus, both inverters are operated with the same sequence
so that the null vector combinations are 88 and 77. From Table 5.1,
it may be noted that these two combinations result in the zerosequence voltage that is zero. If one inverter is operated with onsequence

and

the

other

with

off-sequence,

the

null

vector

combinations would be 87 or 78. From Table 5.2 it is evident that

the zero-sequence voltage of the difference of the pole-voltages is
maximum for these two combinations. It is interesting to note that
this zero-sequence voltage is much lesser with this algorithm than the
lookup table approach used in [83]. This is because the combinations

165

87 and 78are used extensively with that approach [83]. The merit of
the decoupled control is that there is no computational burden on the
controller and is therefore amenable to be used with slower controllers
(processors) and possibly the reduced zero-sequence voltage in the
difference of pole-voltages. However, in this approach, both inverters
are to be switched.
The conventional d-q model of a normal 3-phase induction
motor is modified to compute the motor phase current of the open-end
winding induction motor drive as shown in Fig. 5.7.
+

V00'

+
+
+

Va0 +
-

Inverter-1

Vb0
Vc0

+
-

Van
-

Vbn
-

Induction
Motor

Vcn
-

V'a0
Inverter-2

V'b0
V'c0

Fig. 5.7 d-q model of an open-end winding induction motor.

The inputs for this model are the PWM signals of the individual
inverters and their DC link voltages. The pole voltages of the
individual inverters are then computed. Subtracting the pole voltages

166

of inverter-2 from those of inverter-1, the difference of pole voltages is

obtained. If the individual inverters are operated off isolated DC power
supplies, the zero-sequence content of the difference in pole voltages
is subtracted as shown in Fig. 5.7, to obtain the actual motor phase
voltage. It may be noted that the zero-sequence voltage, in this case,
appears across the points O and O'. The actual motor phase voltages
thus computed are impressed onto the conventional d-q model of
induction motor to compute the motor phase currents.

5.4 Results and Discussions:

Matlab-Simulink based simulation studies have been carried

out to validate the proposed decoupled based direct torque controlled
induction motor drive. Various conditions such as starting, steady
state, step change in load and speed reversal are simulated. The
simulation parameters and specifications of induction motor used in
this thesis are given in Appendix - I. The average switching frequency
of the inverter is taken as 3 kHz. For the simulation, the reference flux
is taken as 1wb and starting torque is limited to 40 N-m. The
simulation results for proposed decoupled PWM algorithms based
DTC-IM drive are shown in from Fig 5.8 to Fig 5.21.

Fig 5.8 and Fig 5.9 show the no-load starting transients of speed,
currents, torque, flux and phase and line voltages for proposed
decoupled PWM algorithm based DTC-IM drive. The no-load steady
state plots of speed, torque, stator currents, flux, phase and line

167

voltages at 1200 rpm are given in Fig 5.10-Fig.5.11. The harmonic

distortion in the steady state stator current along with THD value is
shown in Fig 5.12. From Fig 5.10 to Fig 5.12, it can be observed that
the steady state ripple in torque, flux and current is very less
compared to conventional DTC. Also, the proposed decoupled PWM
algorithm based DTC provides constant switching frequency of the
inverter. The locus of the stator flux is given in Fig 5.14. From which it
can be observed that the locus is almost is a circle of constant radius.
The transients in speed, torque, currents and flux during the step
change in load torque and corresponding phase and line voltages are
shown in Fig. 5.15-Fig.5.16. Also, the transients in speed, torque,
currents, flux, and voltages during the speed reversals (from +1200
rpm to -1200 rpm and from -1200 rpm to +1200 rpm) are shown from
Fig. 5.17 to Fig. 5.20. The four-quadrant speed-torque characteristic
of the proposed drive is shown in Fig. 5.21.

168

Fig. 5.8 Starting transients of speed, torque, stator currents and

stator flux for proposed decoupled PWM based DTC-IM drive.

Fig. 5.9 Starting transients in phase and line voltages for

proposed decoupled PWM based DTC-IM drive.

169

Fig. 5.10 Steady state plots of speed, torque, stator currents and
stator flux for proposed decoupled PWM based DTC-IM drive at
1200 rpm.

Fig. 5.11 The phase and line voltages for proposed decoupled
PWM based DTC-IM drive during the steady state.

170

Fig. 5.12 Harmonic Spectrum of stator current along with THD.

Fig. 5.13 Harmonic Spectrum of stator voltage along with THD.

Fig. 5.14 Locus of stator flux in proposed decoupled PWM based

DTC-IM drive.

171

Fig. 5.15 Transients in speed, torque, stator currents and stator

flux during step change in load: a 30 N-m load is applied at 0.5 s
and removed at 0.6 s.

Fig. 5.16 The phase and line voltages during a step change in load
torque: a 30 N-m load torque is applied at 0.5 s and removed at
0.6 s.

172

Fig. 5.17 Transients in speed, torque, stator currents and stator

flux during speed reversal: speed is changed from +1200 rpm to
-1200 rpm at 0.7 s.

Fig. 5.18 The phase and line voltage variations during the speed
reversal (speed is changed from +1200 rpm to -1200 rpm at 0.7s).

173

Fig. 5.19 Transients in speed, torque, stator currents and stator

flux during speed reversal: speed is changed from -1200 rpm to
+1200 rpm at 1.35 s.

Fig. 5.20 The phase and line voltage variations during the speed
reversal (speed is changed from -1200 rpm to +1200 rpm at
1.35s).

174

Fig. 5.21 The torque and speed characteristics in four quadrants

for proposed decoupled PWM based DTC-IM drive.

5.5 Summary:
A simple decoupled PWM algorithm has been presented in this
chapter for direct torque controlled open-end winding induction motor
drive. The proposed algorithm has been developed by using the
concept of imaginary switching times. The proposed algorithm
generates the voltages similar to the three-level inverter. To validate
the proposed algorithm. The numerical simulation studies have been
carried out and results are presented. From the simulation results, it
can be observed that the proposed algorithm gives reduced harmonic
distortion when compared with the two-level inverter fed drive.