Direct Torque Control

 Main Features:
o Direct control of flux and torque vectors (by
selection of optimum inverter switching vectors);
o Indirect Control of stator currents and voltages;
o Approximately sinusoidal stator fluxes and stator
currents;
o Possibility for reduced torque oscillations;
o Torque oscillation depends upon the duration of
zero switchig vectors;
o High dynamic performance;
o Inverter switching frequency depends on the flux
and torque hysteresis bands
1
 Direct Torque Control
 Main Advantages:
 Absence of coordinate transformation;
 Absence of separate voltage modulation block;
 Absence of voltage decoupling circuits;
 Absence of several controllers;
 Only sector position where the flux linkage
vector is located, has to be determined with
60º accuracy;
 Minimum torque response time;

2
 Direct Torque Control
 Main disadvantages:
 Possible problem during starting and low
speed operation and during changes in torque
command;
 Requirement of flux and torque estimator;
 Changing switching frequency;
 High ripple torque;

3
 Operation of DTC Drive
 The DTC controls the switching of the inverter
directly based on table of optimal switching
(TOS);
 The TOS generates switching signals for the
inverter depending upon the flux and torque error
and sector of operation (θ);
 The detailed vector positioning can be observed
using flux hodograph.

4
 Flux Vector Estimation
 The Flux vector can be estimated using sensed voltage and
currents;
 The sensed stator currents can be converted to d-q components as:
isqs = (1/3){2ia-ib-ic}=ia and isds=(1/√3){ic-ib}= -(1/√3){ia+2ib}
 The sensed stator voltages can also be converted to d-q component
as:
vsqs = (1/3){2va-vb-vc}= (1/3){vab+vac} and
vsds=(1/√3){vc-vb}= -(1/√3) vbc
 Using these values, the d-q components of Stator flux can be
calculated as:
λsds = ∫(vsds-Rsisds)dt and λsqs = ∫(vsqs-Rsisqs)dt
Therefore λs = √ (λsds)2+(λsqs)2
5
 Torque & Sector Estimation
The torque can be estimated using the
relation:
T = (3P/4)(λsdsisqs-λsqsisds)
The sector of operation can be estimated by
θ = sin-1 (λsqs / λs)

6
Table of Optimal Switching (TOS)
ΔTe Δ ψs θ1 θ2 θ3 θ4 θ5 θ6
1 1 V2 V3 V4 V5 V6 V1
-1 V3 V4 V5 V6 V1 V2
-1 1 V6 V1 V2 V3 V4 V5
-1 V5 V6 V1 V2 V3 V4
0 1 V0 V7 V0 V7 V0 V7
-1 V7 V0 V7 V0 V7 V0
7
 Flux Hodograph and Voltage Vector
θ3 θ2
V4 V3

V5
V4

V5
V3 V2 V4 V3 V1 = 1,0,0
V2 = 1,1,0
V6
λs V2
V3 = 0,1,0
θ4 θ1
V4 = 0,1,1
V5 V4 V1 V3
V5 = 0,0,1
V2
V6 = 1,0,1
V5 V6 V0 = 0,0,0
V6
V1
V7 = 1,1,1
V1 V6
V2
V1
θ5 θ6 8
 VSI based DTC
HTe
Converter
Hψs TOS

T*
λ*
Controller

VSI
Speed

q
λ Vb
0 T Flux & Ib
Torque Va
Estimator Ia

ω r* wr d qr Motor
dt

9
Cycloconverter based DTC

TOS

T*
Cycloconverter
λ
Controller

*
Speed

q
λ Vb
0 T Flux & Ib
Torque Va
Estimator Ia

ωr* wr d qr Motor
dt

10
Matrix Converter based DTC

SVM
TOS
T*

Converter
Matrix
λ*
Controller
Speed

q
λ Vb
0 T Flux & Ib
Torque Va
Estimator Ia

ωr* wr d qr Motor
dt

11
 CSI based DTC
λr* Flux Converter
Controller

T*
TOS
Inverter
θ
Speed λr
Controller Rotor Flux
& Torque
T
Estimator

w Motor

ω*
12
 CSI based DTC…. continued
(
The stator flux linkage can be obtained as λs = ∫ V s - Rs i s dt )
where Rs is stator resistance and V s , i s , λs are the space vectors
of stator voltage, stator current and stator flux linkage respectively,
and T (=  
3 / 2 ) P Re  j λ s i s  ( 3 / 2 ) P λ s is sin θ
" * " *

 
The Rotor flux linkage can be obtained as

(
− Lsc i s + ∫ V s - Rs i s dt )
"
λ =
r

"
where λ is the space vectors of rotor flux linkage,
r

and T = ( 3 / 2 ) P Re  j λ r i s 
 " *

 

13
Flux Hodograph with Current Vector
θ2

i3 i4
i3
i4 i2
θ3 θ1
i5 i2 λs
i4 i3
i1
i5

i4 i6
i6

i5 i5
θ4 i2 θ6

i6 i6
i1 i1

θ5 14
Table of Optimal Switching (TOS)

T θ1 θ2 θ3 θ4 θ5 θ6
1 I3 I4 I5 I6 I1 I2
0 I0 I0 I0 I0 I0 I0
-1 I5 I6 I1 I2 I3 I4

15
 Modeling of Direct Torque
Controlled Induction Motor Drive
(DTCIMD)

16
 Modeling of DTCIM Drive
 The modeling of direct torque controlled induction
motor drive consists of following:
 Modeling of Speed Controllers
 Modeling of Flux and Torque Hysteresis
Comparators
 Modeling of Estimator
 Modeling of Induction Motor

17
Modeling of Speed Controller

18
 Modeling of Speed Controller
 A PI controller is normally used as a closed loop speed
controller;
 The output of the PI controller at kth instant will be :
T(k) = T(k-1) + KP{ωe(k) – ωe(k-1)} + KIωe(k)
where KP and KI are the proportional and integral gains
of the PI controller.
 The output of the speed controller (T) after application
of limit, is considered reference torque (T*);
 The speed error at kth instant of time is given as :
ωe(k) =ω*r(k)-ωr(k) where ω*r(k) is reference speed
ωr(k) is rotor speed and ωe(k) is speed error 19
Conventional PI Controller

w*(n) we(n) Te
+- Kp +
+

w(n)
Ki /(z-1)

20
 Modeling of Speed Controller
 Other than PI controller, Sliding mode (SM), fuzzy
pre-compensated PI (FPPI) and hybrid fuzzy PI
controllers etc., can also be used depending upon
accuracy and range of control;
 Sliding mode (SM) controller:
The SM controller operates on the basis of switching
function. The detailed design is discussed next.

21
 Modeling of SM Controller
 The switching functions at the kth sampling instant are
defined as:
X1 = 1 if ZY1(k)>0 and X1 = -1 if ZY1(k)<0;
X2 = 1 if ZY2(k)>0 and X2 = -1 if ZY2(k)<0; where
Y1(k) = {ω*r(k)-ωr(k)} and Y2(k) = d {Y1(k)}/dt;
Z is switching hyper plane function on the speed locus
of the drive and given by Z = C1Y1 + Y2; and C1 is
the parameter of the switching hyper plane.
 The output of the SM controller at kth instant will be :
T(k) = C2Y1(k) X1(k)+ C3Y2(k) X2(k)
where C2 and C3 are the gains of the SM controller.
22
 Modeling of FPPI Controller
 Fuzzy Pre-compensated PI Controller:
 The speed error is ωe(k) = ω*r(k)-ωr(k); and the rate of
change of speed error Δ ωe(k) = ωe(k) – ωe(k-1);
 the reference speed ω*r(k) is modified according to
fuzzy intelligence to compensate the non-linearities
present in the model.
 The modified reference input given to PI speed
controller shall be:
ω*’r(k) = ω*r(k) + γ(k) where γ(k) is the output of the
fuzzy logic controller and given by :
γ(k) = F{ωe(k), Δωe(k)}
23
 Modeling of FPPI Controller
 The compensation term F{ωe(k), Δωe(k)} is used to
change the reference speed.
 This improves the transient response of the output
speed and reduces the effects of non-linearities.
 The compensated reference speed ω*r(k) is applied to PI
controller to obtain the modified torque reference
T’ref(k) at kth instant as :
T’ref(k) = T’ref(k-1) + KP Δ ω’e(k) + KI ω’e(k)
where ω’e(k) = ω*’r(k) - ωr(k); and
Δω’e(k) = ω’e(k) – ω’e(k-1);
24
Fuzzy Pre-compensated PI
Speed Controller

ω∗r ωe Fuzzy γ ω∗’r DTCIM

PI
Logic Drive
ωr ωr

25
 Modeling of HFPI Controller
 Hybrid Fuzzy PI Controller:
 The speed error is as usual ωe(k) = ω*r(k)-ωr(k); and
the rate of change of speed error for a specified
sampling period shall be Δ ωe(k) = ωe(k) – ωe(k-1);
 The output of the fuzzy logic controller (FLC) can
be given as:
TFLC(k) = F{ωe(k), Δωe(k)}
 The FLC output along with PI speed controller
modifies the torque reference as:
T*r(k) = T*r(k-1) + KPI TPI + KFLC TFLC
where KPI and KFLC are membership functions and
TPI and TFLC are the output of PI and FLC. 26
 Fuzzy Logic Controller (FLC)
 The fuzzy logic controller mainly comprises of three
functional blocks namely, the fuzzifier, the decision-
maker and the defuzzifier.
 The two input signals to FLC are the speed error ωe and
the rate of change of the speed error Δωe.
 Fuzzifier: The inputs are scaled to lie in the universe of
discourse U, which is, chosen as [-1 1] and for each of
these scaled inputs a linguistic value is assigned
according to the associated membership functions
chosen for the two inputs.
 These linguistic values are chosen for the computation of the
output of the fuzzy logic controller. Triangular membership
functions are chosen for the inputs and the output are rule base. 27
 Block diagram of fuzzy logic
Controllers
Rule Base

we= w(*n) − w(n)

(n) Decision γ
Fuzzifier Making Defuzzifier
ce=[ we ( n ) − we ( n−1) ] / T Block

Data Base

28
Logic Rules for Fuzzy Logic Controller

ωe NB NM NS ZO PS PM PB

Δωe
NB NB NB

NM NB NB NB

NS NB NM NM NM PM

ZO NB NM NS ZO PS PM PB

PS NM PS PS PM

PM PM PB PB

PB PM PM PB

29
 Fuzzy Logic Controller (FLC)
 Decision making : This process is associated with a set
of fuzzy logic rules.
 In accordance with the linguistic rules and the linguistic values
of the inputs from the fuzzifier, the linguistic value of the
output is computed.

30
 Fuzzy Logic Controller (FLC)
 Defuzzifier: The defuzzifier employs the centroid
method, which is simple and widely used, as it gives
relatively good results. The defuzzification method is
explained below in steps.
 α = Min [µ (input1), µ (input2)], where µ represents
the “membership of.”
 Latest instant α should be maximum of all the
previous instant values.
 Crisp value = Cf {∑ p(m). α}/ ∑ α, Where ‘p(m)’ is
the location of peak of the membership function and
Cf is the scaling factor.

31
Modeling of Flux and Torque
Hysteresis Comparators

32
Modeling of Flux Hysteresis Comparator

 The inputs are the estimated and the reference

torques for the torque comparator and the
reference and estimated stator flux modulus for
the flux hysteresis comparator
 Two level flux hysteresis comparator :
Hψ s =−1 if | ψ s |≤| ψ s ref | − | ∆ψ s |
Hψ s =+1 if | ψ s |≥| ψ s ref | + | ∆ψ s |
Where Hψ s is the output of flux hysteresis comparator and
ψ s is the stator flux linkage
33
 Modeling of Torque Hysteresis
Comparator
 Three level torque hysteresis comparator :
For anticlockwise rotation (forward rotation)
H T e 1 if | Te |≥| Teref | +∆Te
HT e 0 if | Te |≥| Teref |
For clockwise rotation (backward rotation)
H T e = −1 if | Te |≤| Teref | −∆Te
HT e 0 if | Te |≥| Teref |
Where H T e is the output of torque hyeteresis comparator and
Te is the torque
34
Modeling of Estimator

35
 Modeling of Estimator
 The modeling of estimator consists of following:
 Computation of d-q stator voltages
 Estimation of flux linkage and torque
 Estimation of flux linkage vector location
 Estimation of rotor speed

36
Computation of d-q Stator Voltages
 The three phase voltages being applied to the motor at
any instant, which can be calculated using the
switching functions Sa, Sb, Sc and the DC link voltage
Vdc as
= v (1 / 3)v (2 S − S − S )
as dc a b c

=
vbs (1 / 3)vdc ( − S a + 2 Sb − S c )
=
vcs (1 / 3)vdc ( − S a − Sb + 2 S c )
The d-q axis stator voltages can be obtained
from three phase voltage equations as
1 −1
vsD = (vba − vac ) and vsQ = (vac + vba )
3 3
where vba = vbs − vas , vac =
vas − vcs
37
Estimation of Flux Linkage & Torque
 The stator flux linkages in the DTC drives have to be
estimated for two reasons.
 For optimum switching vector selection.
 For the estimation of the electromagnetic torque.
 The flux linkages can be obtained directly form the
stator voltages and currents in the stationary reference
frame as
∫ (vsD − isD Rs ) and ψ sQ =
ψ sD = ∫ (vsQ − isQ Rs )
Where i=
sD isa , and i=
sQ (isa + 2isb ) / 3
The Electromagnetic Torque is given by
= Te (3 / 2) P(ψ sD isQ −ψ sQ isD ) 38
Estimation of Flux Linkage Vector Location
 The stator flux linkage vector should satisfy
following condition for a particular sector :

If ψ sD ≥ 3ψ sQ and ψ sD ≥ 0, then Sector=I

If ψ sD < 3 ψ sQ and ψ sD ≥ 0, then Sector=II
If ψ sD < 3 ψ sQ and ψ sD < 0, then Sector=III
If ψ sD ≥ 3 ψ sQ and ψ sD < 0, then Sector=IV
If ψ sD ≤ 3 ψ sQ and ψ sD < 0 and ψ sQ < 0 then Sector=V
If ψ sD < 3 ψ sQ and ψ sD ≥ 0 and ψ sQ < 0 then Sector=VI

39
 Estimation of Rotor Speed
 The equation of the rotor speed in terms of the
stator and the rotor flux linkages is given below :
w=
(n) wmr − wsl
ψ rD pψ rQ − ψ rQ pψ rD Lm
{ }− (−ψ rQisD + ψ rDisQ )
|ψ r |
2
Tr | ψ r |
2

Lr Lr
ψ rD = (ψ sD − LsisD ), and ψ rQ =
'
(ψ sQ − L'sisQ )
Lm Lm
Where, Tr (=Lr/Rr) is the time constant of the rotor, wmr is the synchronous
speed, wsl is the slip speed, ψrD and ψrQ are d and q axis rotor flux
respectively, Lr and Ls are rotor and stator self inductances respectively
referred to the rotor, Lm is the magnetizing inductance, p = (d/dt) is the
derivative operator
40
MATLAB Model of Rotor Speed
Estimator

41
Modeling of Induction Motor

42
 Modeling of Induction Motor
 Three-phase induction motor is modeled using the d-q stationary
reference frame.
v =+ Ri pψ , v =+Ri pψ
qs s qs qs ds s ds ds
v' = R' i ' + pψ ' − w ψ ' , v' = R' i ' + pψ ' + w ψ '
qr s qr qr r dr dr s dr dr r qr
=
T Jpw + T
m r L
T (3 / 2) P(ψ i −ψ i )
m ds qs qs ds
ψ = L i + L i' , ψ = L i + L i'
qs s qs m qr ds s ds m dr
ψ' = L' i ' + L i , ψ ' = L' i ' + L i
qr r qr m qs dr r dr m ds
L = L + L , L' = L' + L
s ls m r lr m
i = i , i = ( 3i − i ) / 2, i = −(i + i )
sa sd sb sq sd sc sa sb
Where J is the combined inertia of the rotor and the load 43
MATLAB Model of Induction Motor
in Stationary Reference Frame

44
Performance of DTCIMD

45
 Performance of DTCIMD
 The performance of the sensorless direct torque
controlled induction motor drive is simulated under
different operating conditions.
 Behavior of the drive is achieved for the dynamic
conditions that include starting, sudden application and
removal of load and speed reversal.
 The obtained results are shown for the cases of direct
torque control of induction motor with mechanical
speed sensor, sensorless direct torque control and fuzzy
precompensated PI speed controller based sensorless
direct torque control. 46
Motor & Controller Specifications
Three phase squirrel cage induction motor
1hp (0.75kW), 3-phase, 2-pole, Y-connected,
415 V, 50 Hz, Rs = 11.72 Ω, Rr = 9.53Ω,
Xls = 11.04 Ω, Xlr = 11.04 Ω, Lm= 0.678 H,
J = 0.018 kg-m2, Torque Limit = 5.5 Nm.
Controller Parameters
Proportional gain (Kp) = 1.2,
Integral Gain (Ki) = 30.

47
Simulation Results of DTC IMD
with mechanical sensor and conventional PI speed controller

48
 Numerical Example - 1
Q. A three-phase, Y-connected, 220-V (line-line), 0.75 kW, 50 Hz,
two-pole squirrel cage induction motor has the following
parameters values in ohms/phase referred to the stator: Rs=1.294,
Rr= 1.412, Xs= 3.503, Xr= 3.209, Xm= 80.25 and full load
slip=0.075. The speed of the motor is controlled using direct
torque control (DTC) scheme through a single-phase PFC boost
converter fed 3-phase PWM VSI with an almost unity power
factor at ac mains of 220V, 50 Hz and dc bus voltage of 400V.
The motor is controlled with constant flux below base speed and
constant rated voltage above base speed. Calculate (a) supply
rms current, reference torque and reference flux at half the rated
synchronous speed and half of the rated torque, (b) supply rms
current, reference torque and reference flux at 125% of rated
synchronous speed and half the rated torque. Neglect core and
friction and windage losses.
49
 Numerical Example – 1 …Contd
Solution:
For 2 pole motor N s = 3000rpm;
=
so the rated speed at 0.075 =
slip N rated (1- 0.075) * N s 2775rpm
ωm rated =2*π *2775/60=290.597 rad/sec;
Trated P= rated / ωm rated =
750 / 290.597 2.581Nm
from the equivalent circuit of the induction motor at s=0.075
Z eq = (1.294 + j 3.503) +  j80.25 {(1.412 / 0.075 ) + j 3.209} / {(1.412 / 0.075 ) + j83.54}
Rated stator current I s rated =VL / Z eq 3 =6.162 −30.03A =5.335 − j 3.084 A;
= =
therefore I d 5.335 A and I q -3.08 A
λs* Ls is* − Lmir*
a)The torque reference T* =0.5Trated =1.2905Nm and flux reference =
To develop half the rated torque at half the rated synchronous speed
I*q = 0.5 × I q =−1.542 A; I d* = 5.335 A; ω * =
Id = 2π .1500 / 60 =
157.08rad / sec
so magnitude of I s* = I d*2 + I q*2 = 5.55 A; I r* = I s* ( j80.25 ) / {(1.412 / 0.075 ) + j83.54}
where I s* =5.335 − j1.542, therefore I r* =5.2 −3.42A =5.191 − j 0.3102 50
Numerical Example – 1 …Contd

λs* = Ls is* − Lmir* =( 0.01115 ) . ( 5.335 − j1.542 ) − ( 0.2554 ) . ( 5.191 − j 0.3102 )

λs* =
−1.2644 + j 0.06203wb
Hence the power loss Ploss =3Is2 Rs = 119.2W
The motor power output Po = T*ω * =202.7W;
Input power to the motor=Po +Ploss =322.42W
This power has to be fed through the single phase supply, therefore
power input 322.42
=
supply rms current at upf I rms = = 1.465 A
Vrms 220
b) The torque reference T* =0.5Trated =1.2905Nm and
λs* Ls is* − Lmir*
flux reference =
To develop half the rated torque at 125% the rated synchronous speed,
the motor will operate in constant power and constant voltage region.
=
So ω * 1.25
= ωs 392.7rad / sec
51
 Numerical Example – 1 …Contd
The motor power output Po = T*ω * =506.778W;
In this region the flux is weakened in the
inverse ration of speed, so I d* =I d /1.25 =4.268 A; I*q = 0.5 × I q =−1.542 A;
so magnitude of I s* = I d*2 + I q*2 = 4.538 A;
I r* = I s* ( j80.25 ) / {(1.412 / 0.075 ) + j83.54}
where I s* = 4.268 − j1.542, therefore I r* = 4.253 −7.16A = 4.219 − j 0.5303
λs* = Ls is* − Lmir* =( 0.01115 ) . ( 4.268 − j1.542 ) − ( 0.2554 ) . ( 4.219 − j 0.5303)
λs* =
−1.03 + j 0.1182 wb
Hence the power loss Ploss =3Is2 Rs = 79.94W
input power to the motor=Po +Ploss =586.722W
This power has to be fed through the single phase supply, therefore
power input 586.722
supply rms current at =
upf I rms = = 2.67 A
Vrms 220
52
 References
1. I. Boldea and S.A. Nasar, “Electric Drives,” CRC Press, 1998
2. I. Boldea and S.A. Nasar, “Vector Control of AC drives,” CRC Press,
1992
3. B.K. Bose, “Modern Power electronics and AC drives,” Pearson
Education India, 2002.
4. P. Vas, “Sensorless Vector and Direct Torque Control,” Oxford
University Press, 1998.
5. A.E. Fitzgerald, C. Kingsley, Jr., S.D. Umans, “Electric Machinery,” Tata
McGraw-Hill, New Delhi, 2002.
6. W. Leonhard, “Control of Electrical Drives,” Springer-Verlag, 1990.
7. A. Kelemen and M. Imecs, “Vector Control of AC Drives,” Vol. 1,
Budapest, Omikk Publisher, 1991
8. M. Trzynadlowski, “The Field Orientation Principle in Control of
Induction Motors,” Netherlands, Kluwer Academic Publishers, 1994.
53
 References
9. D.W. Novotny and T.A. Lipo, “ Vector Control and Dynamics of AC
Drives,” Oxford, Oxford University Press, 1996.
10. P. Vas, “Vector Control of AC Machines,” Oxford, Oxford University
Press, 1990.
11. M. P. Kazmierkowski, R Krishnan and F Blaabjerg, “Control in Power
Electronics (Selected Problems),” Academic Press, 2002.
12. Chee-Mun Ong, “Dynamic Simulation of Electric Machinery,” Prentice
Hall, New Jersey, 1998.
13. P.C. Krause, “Analysis of Electric Machinery,” Mc Graw-Hill Book
Company, New York, 1986.
14. S. Heikkila, “Direct Torque Control Inverter Arrangement”, U.S. Patent 6
094 364, Jul. 25, 2000.

54