Chương 06: Máy Phát Gió- Công Suất& Ngẫu Lực

(WE Generators Coupling Fields- Torque & Power)

Tháng 02.2019
GV Nguyễn Hữu Phúc
 Torque production
Let’s devise a way so that a similar rotating magnetic field is
produced by the windings on the rotor, such that it rotates at
the same speed as that from the stator. Thus, we now have
two rotating magnetic fields.

The two rotating magnetic fields, that from the rotor and
the composite field from the armature, are “locked in,”
and as long as they rotate in synchronism, a torque
(Torque=P/ωm=Force×radius, where Force is tangential
to the rotor surface), is developed.

This torque is identical to that which would be

developed if two magnetic bars were fixed on the same
pivot as shown on the next slide.
 Torque production
In the case of generator operation, we can think of bar
A (the rotor field) as pushing bar B (the armature field),
as in Fig. a. In the case of motor operation, we can
think of bar B (the armature field) as pulling bar A (the
rotor field), as in Fig. 11b. To compute the torque,
we must make use of
coupling fields and co-
energy, according to:
Wc
S S
S N
Te 
Bar A
Bar B
Bar A
Bar B

N N
N S

Fig a: generator operation Fig b: Motor operation

 Torque production
One can make conceptual use of a version of the Lorentz
Force Law, below, in computing torque exerted on a current
-carrying conductor in the presence of a magnetic field.
  
F  il  B
However, most electromechanical energy-conversion
devices contain magnetic material; in these systems, forces
of the B-field act magnetically on the magnetic material, and
also on any current-carrying conductors. So the above
relation is not enough for us to compute force.

Instead, we turn to an energy method, which makes use of

d
W   Pdt P  T  T
dt
The above two relations will allow us to express torque via energy relations.
 Coupling fields
In electromechanical systems, we have an electrical
system, a mechanical system, and a coupling field. We
assume in our work that the coupling field is magnetic.
Most of the coupling Coupling
field energy is stored in field
airgap because it has
very large “resistance”
(reluctance) to the flux. Electrical Mechanical
system system

Example of an elementary translational-motion electromechanical system with

one electrical input, one mechanical input, and one coupling field.
P. Krause, O. Wasynczuk, & S. Sudhoff, “Analysis
of Electric Machinery,” IEEE Press, 1995.
 Coupling fields
Mechanical system
Coupling
field
Coupling
field

Electrical system
Example of an elementary rotational-motion electromechanical system with
one electrical input, one mechanical input, and one coupling field.
A. Fitzgerald, C. Kingsley, and A. Kusko,
“Electric Machinery, 3rd edition, 1971
 Coupling fields
Mechanical system is the rotational motion of the rotor.
Coupling field is
Electrical in the flux path
systems are the of the iron
two coils. path/air gap

Example of an elementary rotational-motion electromechanical system with

two electrical inputs, one mechanical input, and one coupling field.
P. Krause, O. Wasynczuk, & S. Sudhoff, “Analysis
of Electric Machinery,” IEEE Press, 1995.
 Coupling fields: torque
The importance of the coupling field is that this is where the
energy conversion takes place. If losses in the coupling
field (eddy current, hysteresis) are neglected or assumed
to be modeled external to the coupling field, then

Energy transferred to Energy transferred to

Energy stored in
the coupling field = the coupling field by + the coupling field by
the electrical system the mechanical system

OR W f  We  Wm
In differential form, dW f  dWe  dWm
The last equation enables computation of torque.

We investigate dWm and dWe next. We consider a rotational

system with just 1 electrical input and 1 mechanical input.
 Coupling fields: torque
Generalization of systems with 1 electrical input,
1 mechanical input, and 1 coupling field.
This is neglected, or
it is represented
within the model of
the electrical or the
mechanical system.

A. Fitzgerald, C. Kingsley, and A. Kusko,

“Electric Machinery, 3rd edition, 1971
 Coupling fields: torque dW f  dWe  dWm

From previous slide, we observe the electrical terminals at

the coupling field receive current i with voltage e. Thus power
is ei and the total energy supplied by the electrical source is:
We   eidt dWe  eidt
In electromagnetic systems, the coupling element is a
machine winding, and e will be an induced voltage according
to Faraday’s Law: d
e
dt
Therefore:
d d
We   idt   id dWe  idt  id
dt dt
For the following development, see (a) P. Krause, O. Wasynczuk, & S. Sudhoff, “Analysis of Electric
Machinery,” IEEE Press, 1995; (b) A. Fitzgerald, C. Kingsley, and A. Kusko, “Electric Machinery, 3 rd
edition, 1971; (c) J. Meisel, “Principles of electromechanical-energy conversion,” McGraw-Hill, 1966.
 Coupling fields: torque dW f  dWe  dWm
The mechanical torque Tm and the electromagnetic torque Te
are assumed positive in the same rotational direction. We
neglect torque due to shaft twist.
d 2 d J: moment of inertia
Tm  Te  J 2  D D: damping
dt dt
d
2
d
Tm  J 2  D  Te
dt dt
Then the total energy supplied by the mechanical source is
d 2 d
This equation is the rotational WM   Tm d   J 2 d   D d   Te d
analogue to the one for linear motion,
dt
  dt 

which is WM  Fmdx
Term1 Term 2
Term 3

Term 1 on RHS is stored inertial energy; Term 2 is losses.

Only term 3 is energy transferred to coupling field. So…
Wm    Te d dWm  Te d
 Coupling fields: torque
dW f  dWe  dWm (a) dWe  id (b) dWm  Te d (c)
Substituting expressions (b) and (c) into (a) results in
dW f  id  Te d
Solving for the torque term results in:
Te d  id  dW f (*)
Assume independent variables of our system are i, θ. Then:
 
   i,  d  di  d
i 
W f  W f i,  Taking the total W f W f
differential
dW f  di  d
i 
Substitute these last equations into (*)…
  
d  di  d
Coupling fields: torque i
W f

W f
dW f  di  d
i 

  W f W f
Te d  id  dW f  i di  i d  di  d
i  i 
Gather like terms in di and dθ:
  W f    W f 
Te d   i  di   i  d
 i i     
Equate coefficients:
  W f    W f 
Te d   i  d 0   i  di
     i i 
The first expression provides a way to compute torque:
 W f
Te  i  (T1)
 
13
 Coupling fields: torque
Define co-energy: Wc  i  W f
Recalling that independent variables of our system are i, θ:
Wc  W f
i 
  
Solve for the last term on the right:
W f  Wc
i 
  
Substitute into (T1):
  Wc  W f
Te  i i  Te  i 
     (T1)
0
Wc
Te  (T2)

14
 Coupling fields: torque
What is co-energy? Wc  i  W f
An expression for which torque computation is convenient
Wc
Te  (T2)

To better understand co-energy, consider the relation for the
coupling field energy:
W f  We  Wm
Set the system so that initially, Wf=0 (no stored energy), and
fix θ so that no energy can be added to the coupling field via
mechanical means, i.e., Wm=0. Then increase the current to
a value ia, establishing a corresponding flux linkage of λa.
There is then energy in the coupling field, but only via
electrical means, i.e.
W f  We
15
 Coupling fields: torque
But recall that:
d
We   eidt   idt   id
dt
And since Wm=0, this is also the
energy of the coupling field:
W f   id
We observe the corresponding area
in the λ-i curve in the figure.
Recall the definition of co-energy: Wc  i  W f
Since iλ is the area of the (shaded) box, then Wc must be the
area below the curve. Wc is therefore given by
Wc   di
16
 Coupling fields: torque
If the medium is magnetically linear
(no saturation), then the λ-i curve is
just a diagonal through the iλ
shaded box, as shown. In this case,
the area above the λ-i is the same
as the area below it, and we have:
W f  Wc
Wc
Recall (T2), which is Te 

Therefore, under condition of magnetic linearity, we have:
W f
Te  (T3)


17
Coupling fields: energy
Recall from basic physics that the energy stored in a
winding of self-inductance Lpp carrying current i is given by:
1
W f  L ppi 2
2
with Lpp in henries and defined by Lpq=λp/Iq=NpNq/Rpq for
linear medium; Rpq is path reluctance (like resistance).
Generalization: For a linear electromagnetic system with J
electrical inputs (windings), the total field energy is given by:
1 J J
W f   L pqi p iq
2 p 1 q 1
where Lpq is the winding’s self inductance when p=q and
when p≠q, it is the mutual inductance between the two
windings. Derivation: Given pp 22-24 of Krause, Wasynchzuk, and
Sudhoff, “Analysis of electric machinery,” 1995. 18
 Example
This device consists of
two conductors, #1 is on
the stator; #2 is on the
rotor. The magnetic
system is assumed linear.
So Wf is given by (with J=2):
1 J J
W f   L pqi p iq  L11i1i2  2 L12i1i2  L22i1i2   L11i1i2  L12i1i2  L22i1i2
1 1 1
2 p 1 q 1 2 2 2
The self-inductances, given by Lpp=λp/ip, are constant,
independent of θ, because the reluctance of the path seen by
the winding does not change as the rotor turns.
But the mutual inductances are not constant.

19
 Example
The mutual inductances,
given by Lpq=λp/iq, i.e., the
amount of flux seen by
winding p due to a current
in winding q, are not
constant. The extreme
conditions of maximum & minimum linkages are given below.

θ=0° condition θ=0° condition

θ=90° condition
(maximum (maximum
(zero linkages)
positive negative
linkage) linkage)

The mutual inductance goes from max positive to 0 to max

negative to 0 and back to max positive. Thus, L12, L21 are:
L12  L21  M cos  20
 Example

θ=0° condition θ=0° condition

θ=90° condition
(maximum (maximum
(zero linkages)
positive negative
linkage) linkage)

The mutual inductance goes from max positive to 0 to max

negative to 0 and back to max positive. Thus, L12, L21 are:
L12  L21  M cos  21
 Example
Summarizing, we have.
L11, L22 are constant.
L12  L21  M cos 
1 1
W f  L11i1i2  L12i1i2  L22i1i2
2 2
1 1
Substituting, we obtain W f  L11i1i2  M cosi1i2  L22i1i2
2 2
W f
Recalling Te  , we obtain Te   Mi1i2 sin 


22
 Summary
The previous procedure can be applied to a three-phase
induction machine to obtain an expression for its torque.
The effort requires a coordinate transformation which is
involved, and so we will not do it.

The resulting relation for torque (see Meisel, pp. 321-

322, and chapter 12) may be used to derive the torque
expression for steady-state conditions. We will use this
relation in the next set of material.

23
Torque and Power

J. McCalley
 Torque in abc quantities
The electromagnetic torque of the DFIG may be evaluated according to
Wc
Tem 
 m
where Wc is the co-energy of the coupling fields associated with the various windings;
ϴm is the angle in mechanical degrees between the main rotor axis and fixed reference.
We are not considering saturation here, assuming the flux-current relations are linear, in
which case the co-energy Wc of the coupling field equals its energy, Wf, so that:
W f See Slides called “WE
Tem 
 m Generators – Coupling Fields”
We use electric rad/sec by substituting ϴm=θm/p where p is the number of pole pairs.
W f
Tem  p
 m
We learned in the above-mentioned slide-set (slide 35), that for a linear electromagnetic
system with J electrical inputs (windings), the total field energy is given by:
1 J J
W f   L pqi p iq
2 p 1 q 1
where Lpq is the winding’s self inductance when p=q and when p≠q, it is the mutual
inductance between the two windings.
Derivation: Given in pp 22-24 of Krause, Wasynchzuk, and Sudhoff, “Analysis of electric machinery,” 1995.
25
 Torque in abc quantities
1 J J
W f   L pqi p iq
2 p 1 q 1
The stored energy is the sum of
• The self inductances (less leakage*) of each winding times
one-half the square of its current and
• All mutual inductances, each times the currents in the two
windings coupled by the mutual inductance
Observe that the energy stored in the leakage inductances is not
a part of the energy stored in the coupling field*.

Consider the abc inductance matrices given in slide 7 of “dq

transformation” slides.

* See pg. 178 of Krause.

26
 Torque in abc quantities
 1 1 
 L r  Lmr  Lmr
2
 Lmr
2  The stored energy is given by:
 
L r    Lmr 
1 1 1 T 1 T
L r  Lmr  Lmr W f  i abcs ( L s  Ls U )i abcs  i abcs L sr i abcr  i abcr ( L r  Lr U )i abcr
T
 2 2  2 2
 1 1 
  Lmr  Lmr L r  Lmr  Applying the torque-energy relation
 2 2 
W f
  Tem  p
 m
1 1
 L s  Lms  Lms
2
 Lms
2 
  to the above, and observing that dependence
L s    Lms 
1 1
L s  Lms  Lms on θm only occurs in the middle term, we get
 2 2 
 1 
  Lms 1
 Lms L s  Lms  W f 

T
 2 2  i abcs L sr i abcr
 m  m
 cos  m cos  m  120  cos  m  120   So that
 
L sr  Lsr  cos  m  120  cos  m cos  m  120  

cos  m  120  cos  m  120  Tem  p
T
cos  m  i abcs L sr i abcr
 m
 cos  m cos  m  120  cos  m  120   But only Lsr depend on θm, so
 
L rs  Lsr cos  m  120  cos  m cos  m  120    L sr
T

 cos  m  120  cos  m  120 

 L sr
 Tem  pi abcs
T
cos  m i abcr
27  m
 Torque in abc quantities
 L sr
Tem  pi abcs
T
i abcr
 m
We may perform the above differentiation and associated matrix multiplication to show
that the above evaluates to
  1 1   1 1   1 1 
Tem   pLm ias  iar  ibr  icr   ibs  ibr  iar  icr   ics  icr  ibr  iar  sin  m
  2 2   2 2   2 2 
 Negative value for

3
ias ibr  icr   ibs icr  iar   ics iar  ibr cos m 
2  generation
To complete our abc model we relate torque to rotor speed according to:

J is inertia of the rotor

in kg-m2 or joules-sec2 J dm
Tem   Tm
p dt

Inertial Mech
torque torque (has
negative
value for
28 generation)
 Torque in qd0 quantities
The previous torque expression is a bit complicated; it is not obvious how to control abc
stator and rotor currents in order to achieve a certain torque.

So…let’s express torque in qd0 quantities, to see if there are any good control strategies
that result.

To this end, recall that we may write the abc quantities in terms of the qd0 quantities
using our inverse transformation, according to:
1
i abcs  K s i qd 0 s
1
i abcr  K r i qd 0 r

Substitute the above into our torque expression:

Tem  pi
T
abcs

 m

L sr i abcr  p K i
1
s qd 0 sT 
 m
1
L sr K r i qd 0 r

29
 Torque in qd0 quantities
Tem  p K i  1
s qd 0 s
T
 
 m
1
L sr K r i qd 0 r

 cos sin  1  cos  sin  1

K s  cos(  120) sin(  120) 1 K r  cos(  120) sin(   120) 1
1 1

cos(  120) sin(  120) 1 cos(  120) sin(   120) 1

 cos  m cos  m  120  cos  m  120  

 
L sr  Lsr  cos  m  120  cos  m cos  m  120  
cos  m  120  cos  m  120  cos  m 
  cos  m  120  cos  m  120    cos  1 iqr  
T
  cos  sin  1  iqs   cos  m sin 
      
Tem  p   cos(  120) sin(  120) 1 ids    Lsr cos  m  120  cos  m cos  m  120   cos(   120) sin(   120) 1 idr  
 cos(  120) sin(  120) 1 i    m  cos   120  cos   120  cos  m  cos(   120) sin(   120) 1 i0 r  
   0s     m m 

I will not go through this differentiation and matrix multiplication, but I will instead just
provide the result (using M=(3/2)Lsr) which is surprisingly simple (see eq. 4.6-2 on
pg. 154 of Krause 2002 edition):
Tem  pM  iqs idr  idsiqr 
3
2
30
 Torque in qd0 quantities
 pM  i i  i i 
3 (1a) Eq. 3.16(a) on p. 59 in Wu’s book;
Tem qs dr ds qr Eq. 4.6-2 on p. 154 of Krause’s 2002 edition.
2
By using the below expressions (see slide 39 of “dq transformations”),

qs   L s  M  iqs  Miqr qr  Miqs   L r  M  iqr

ds   L s  M  ids  Midr dr  Mids   L r  M  idr

making appropriate substitutions into (1a), and then performing algebraic manipulations,
some other useful expressions may be derived from the above, as follows:

Tem  pqr idr  dr iqr 

3 (1b) Eq. 4.6-3 on p. 154 of Krause’s 2002 edition.
2
Tem  pds iqs  qs ids 
3 (1c) Eq. 3.16(a) on p. 59 in Wu’s book;
Eq. 4.6-4 on p. 154 of Krause’s 2002 edition
2
pdr iqs  qr ids  (1d) Eq. 3.16(d) on p. 59 in Wu’s book.
3 Lm
Here, Wu defines on p. 56:
Tem 
Ls=Lls+Lm; Lr=Llr+Lm. Thus, the
Lm here is what I have called 2 Lr  Lm
Lms and Lmr (see slide 7 of
notes on dq xfrmation.) I have

p  qs idr  ds iqr  (1e)

also used M=(3/2)Lms. Thus, Lm 3 Lm
here could be replaced by
Tem  Eq. 8.56 on p. 257 in Wu’s book.
(2/3)M.
31 2 L s  Lm
 Torque in qd0 quantities

Comment: We will work with the torque expressions on the previous slide to show that
the electromagnetic torque Tem can be directly controlled by the rotor quadrature
current iqr.

At the same time, we will also show that the stator reactive power Qs can be directly
controlled by the rotor direct-axis current idr.

This will provide us the necessary means to control the rotor-side converter.

As a last step before moving into the control, let’s obtain real and reactive powers in
qd0 quantities. We do that next….

32
 Power in qd0 quantities
We will develop power expressions using a space vector approach, taking
the following steps:
1. Show that our space vectors have same magnitude as abc quantities.
2. Obtain power expressions for space vectors.
3. Substitute qd0 quantities into space-vector power expressions.
(Alternative approach would be to express
power in terms of abc quantities and then use
transformation to convert to qd0 quantities.)

33
 Power in qd0 quantities: step 1
1. Show that our space vectors have same magnitude as abc quantities.
You will recall in our slides called “Machine Transformations,” in slide 8, we
expressed our current space vector as: i  i 
j
s i i a b c 
3
This space vector came about as a result of a scaling choice we made, of 2/3. Let’s
see here if we can understand why we made that scaling choice.
Consider that the three phase currents may be expressed as:
ia  I p cos t ; ib  I p cost  120 ; ic  I p cost  120 
where Ip is the peak amplitude of the waveform.
Substitution into the space vector expression above yields:
is  I p cos t 
j
I p cost  120  I p cost  120
3
Factor out the Ip and also multiply and divide the second term by 2 (to get it to a
form recognizable as a trig identity) results in:

is  I p cos t 
2j 1
cost  120  cost  120
 32 
Recalling that sinu sinv = (1/2)[cos(u-v)-cos(u+v)], we can write the above as:
   2j  3 
is  I p cos t 
2j
sin t sin 120   I p cos t   sin t    I p cos t  j sin t 
 3   3 2 
Taking magnitudes:  2 2

is  I p cos t  sin t  I p  i a

34 Therefore, we can write: is  I p e jt e j i

 Power in qd0 quantities: step 2
2. Obtain power expressions for space vectors.
Write down the familiar three-phase power expressions for P and for Q.
P  3Vrms I rms cos  Q  3Vrms I rms sin 
where φ=θv-θi, and with θv=0 (the reference), φ=-θi.
Change rms values to peak (Xrms=Xp/sqrt(2)); use Re, Im of rectangular expression
Vp I p Vp I p
P3 Recos  j sin   Q3 Imcos  j sin  
2 2 2 2
Take the √2√2=2 to the denominator out front; change the rectangular to exponential
3
P  V p I p Re e j  3
Q  V p I p Im e j
2
 
2
Bring the peak magnitudes inside the Re and Im operator; multiply by ejωt and by e-jωt
3

P  Re V p e jt I p e  jt e j  3

Q  Im V p e jt I p e  jt e j
2

2
Replace φ by –θi.    
3   3  
P  Re V p e jt I p e jt e ji  Q  Im V p e jt I p e  jt e  ji 
2     2    
 v s i *
s   vs *
is 

 
Observe the space vectors inside the curly brackets. Therefore:

35
3
P  Re vi
2
*
  3
Q  Im vi
2
* (Note: The 3/2 is because our
transformations are all amplitude
invariant but not power invariant.)
 Power in qd0 quantities: step 3
3. Substitute qd0 quantities into space-vector power expressions.
On slide 18 of our slides called “Machine Transformations,” we wrote that a space
vector may be decomposed into our d-q transformation variables according to
a
x  xd  jx q
Applying this to our v and i space vectors on the previous slide results in (assuming
they are stator quantities) :
is  ids  jiqs v s  vds  jv qs
Now substitute into our power expressions from the previous slide:
3
2
 
* 3

Ps  Re v s i s  Re (vds  jvqs )(ids  jiqs )
2
 Qs 
3
2
Im v  
s i
*
s 
3
2
 
Im (vds  jvqs )(ids  jiqs )

Perform the algebra:

3
 
Ps  Re (vdsids  jvds iqs  jvqs ids  vqs iqs )
3
 
Qs  Im (vds ids  jvdsiqs  jvqs ids  vqs iqs )
2 2
Collect real and imaginary terms:
3
 
Ps  Re (vdsids  vqs iqs )  j (vqs ids  vdsiqs )
3
 
Qs  Im (vds ids  vqsiqs )  j (vqsids  vdsiqs )
2 2
Therefore:
3 3
Ps  (vds ids  vqs iqs ) Qs  (vqs ids  vds iqs )
2 2
36
 Control concepts
We want to obtain our control relations. To do this, we make the following remarks:

1. Reference frame orientations: We have a choice to make in terms of how to align

our d-q rotating reference frame. There are three questions about this which you
may find helpful:
a. What does it mean to “align our d-q rotating reference frame”?
 The d-q coordinate reference frame rotates at synchronous speed. But at
any given moment, it has a particular angular position. We align the d-q
reference frame when we choose the d-q reference frame to have a particular
angular position at a particular time (e.g., t=0),

b. Why do we have a choice about this?

The d-q reference frame is entirely “made-up.” There is nothing real about it.
The d-q transformation is a mathematical decomposition or projection of real
(rotating electric and magnetic) quantities. We are free to make that projection
θ
any way we like.
d-axis
Put another way, we recall the definition of θ, q-axis
ia
which is given below and illustrated to the right.
t
    ( )d  (0)
iq id a'
a
0

37
We are free to make θ(0) whatever we want.
 Control concepts
We want to obtain our control relations. To do this, we make the following remarks:

1. Reference frame orientations: We have a choice to make in terms of how to align

our d-q rotating reference frame. There are three questions about this which you
may find helpful:

c. What difference does the choice make?

Our stator-oriented space vectors rotate at synchronous speed which is the
same rotational speed as our d-q coordinate reference frame. Thus,
• if we choose to align our d-axis with a particular stator-oriented space
vector, the q-axis component of that space vector will always be zero.
• if we choose to align our q-axis with a particular stator-oriented space
vector, the d-axis component of that space vector will always be zero.
Forcing certain d-axis or q-axis quantities to be permanently zero can simplify
the control design.

38
 Control concepts
2. Stator voltage-oriented control (SVOC): SVOC is a reference frame orientation very
convenient for wind turbine DFIGs since the grid fixes the stator voltage. SVOC is
achieved by aligning the d-axis of the synchronous reference frame with the stator
voltage space vector vs. (We are free to align the reference frame however we like).
Therefore the resultant d- and q-axis stator voltages are
vqs  0 and vds  vs
The below figure illustrates (ref, pg. 256 of Wu). Observe the location of is indicates
that the machine is operating as a generator with unity power factor. Also observe
the location of Vr indicates it is operating supersynchronously (see slide 9 of DFIG
Set #2).

39
 Control concepts

3. Steady-state: We would like to identify control values of torque and power that
would result in appropriate steady-state operation. This means we can use steady-
state relations, obtainable by setting derivatives to zero in our machine equations.
From slide #44 in our slides called “dqTransformation” we saw:

 vqs   rs 0 0 0 0 0   iqs   L s  M 0 0 M 0 0   iqs   0  0 0 0 0   qs 

v   0  
 ds   rs 0 0 0 0   ids   0 L s  M 0 0 M 0   ids   0 0 0 0 0   ds 

 v0 s   0 0 rs 0 0 0  i0 s   0 0 L s 0 0 0  i0 s   0 0 0 0 0 0   0 s 
       
 vqr   0 0 0 rr 0 0  iqr   M 0 0 L r  M 0 0  iqr   0 0 0 0 (  m ) 0   qr 
 vdr   0 0 0 0 rr 0  idr   0 M 0 0 L r  M 0  idr   0 0 0   m 0 0   dr 
          
v0 r   0 0 0 0 0 rr  i0 r   0 0 0 0 0 L r  i0 r   0 0 0 0 0 0  0 r 

All of this goes away in the steady-state,

when the current derivatives are zero.

And so our “steady-state”

voltage equations become…

40
 Control relations – torque
 vqs  rs 0 0 0 0 0  iqs   0  0 0 0 0  qs 
v   0 rs 0 0 0 0  ids   0 0 0 0 0  ds 
 ds  
v0 s   0 0 rs 0 0 0  i0 s   0 0 0 0 0 0 0 s 
       
vqr   0 0 0 rr 0 0  iqr   0 0 0 0  (   m ) 0 qr 
vdr   0 0 0 0 rr 0  idr   0 0 0   m 0 0 dr 
       
v0 r   0 0 0 0 0 rr  i0 r   0 0 0 0 0 0 0 r 
Extracting the relations for vqs and vds, we have:
vqs  rs iqs  ds vds  rs ids  qs
Solving for λds and λqs, respectively, we get
vqs  rs iqs  (vds  rs ids )
ds  qs 
 
Now substitute these relations for λds and λqs into our torque relation (1e) from slide 8:
  (vds  rs ids ) vqs  rs iqs 
Tem 
3 Lm

p qsidr  ds iqr 
3 Lm
p idr  iqr 
2 Lr  Lm 2 Lr  Lm    
p (vds  rs ids )idr  (vqs  rs iqs )iqr   p vds idr  rs ids idr  vqs iqr  rs iqsiqr 
3 Lm 3 Lm
Tem 
2 Lr  Lm 2 Lr  Lm
p vds idr  rs ids idr  rs iqs iqr 
3 Lm
Use vqs=0 (SVOC): em T 
2 Lr  Lm
 3 Lm pvds idr Electromagnetic torque is a function of d-axis
Assume rs≈0: em T 
2 Lr  Lm rotor current and d-axis stator voltage only!
41
 Control relations – power
Return to our power relations of slide 13 (now we denote them as stator powers):
3 3
Ps  (vds ids  vqs iqs ) Q s  (vqs ids  vds iqs )
2 2
3v i  3vds iqs
Use vqs=0 (SVOC): P  ds ds Q
2 2
From slide #39 of “dqTransformation” slides, we have that:
qs   L s  M  iqs  Miqr qr  Miqs   L r  M  iqr
Eqts. (*)
ds   L s  M  ids  Midr dr  Mids   L r  M  idr

Now Wu in his book (pg. 59) writes the above equations as:
qs   L s  Lm  iqs  Lmiqr qr  Lmiqs   L r  Lm  iqr Eqts. (**)
ds   L s  Lm  ids  Lmidr dr  Lmids   L r  Lm  idr
implying that Lm=M. However, he also writes (pg. 56) that:
Ls  L s  Lm ; Lr  L r  Lm Eqts. (***)
defining these previous two quantities (Ls and Lr) as the stator self inductance and the
rotor self inductance, respectively. This is a contradiction, because the “M” in Eqts. (*) is
3/2 times Lms=Lmr=Lrs, yet the “Lm in Eqts. (***) is definitely the mutual inductance, which
is the same as Lms=Lmr=Lrs. I will proceed here assuming that Eqts. (**) and (***) are
correct (this needs review). This implies that the equations on next slide are correct.
42
 Control relations – power

qs  Ls iqs  Lmiqr qr  Lmiqs  Lr iqr

ds  Ls ids  Lmidr dr  Lmids  Lr idr

43
 Control relations – power
3v i  3vds iqs
Ps  ds ds Qs 
2 2
Solve the below for ids and iqs : qs  Lmiqr
iqs 
qs  Ls iqs  Lmiqr Ls
ds  Ls ids  Lmidr ds  Lmidr
ids 
Ls
Substitute into the power relations above:
3vds  ds  Lmidr  3vds  qs  Lmiqr 
Ps  Q   
  s
2  L 
2  Ls  s

Now solve for idr and iqr, respectively:

2L 2Qs Ls
 s Ps  ds  qs
3vds 2 Ls ds 3vds 2 Ls qs
idr   Ps  iqr   Qs 
Lm 3vds Lm Lm Lm 3vds Lm v
Recall our stator flux equations from slide 18 (but with vqs=0):
vqs  rs iqs  rs iqs  (vds  rs ids )
ds   qs 
  
Substitute them into our idr and iqr equations above:
2 Ls rs iqs 2 Ls (v  r i )
idr   Ps  iqr  Qs  ds s ds
Assume rs≈0: 3vds Lm Lm 3vds Lm  Lm
For a given stator voltage, Ps
2 Ls 2 Ls vds
idr   Ps iqr  Qs  and Qs can be controlled by
44 3vds Lm 3vds Lm  Lm the dq-axis rotor currents!
Block diagram of DFIG wind turbine system w/ SVOC

Note that the

ωm/p converts
T to Ps. Also,
there should
be minus sign
here.

Diagram from p. 260 of Wu.

2 Ls m 2 Ls v
idr  
2 Ls
Ps   Tem iqr  Qs  ds
45 3vds Lm 3vds Lm p 3vds Lm  Lm
Conceptual control sequence of rotor-side converter
1. MPPT block generates the reference torque Tem* based on some method for max
power point tracking. The simplest one to understand is use of the Turbine Power
Profile, Fig. 1 below, but a better one is use of optimal torque control, Fig. 2 below.
The asterisk in all below equations indicates that the quantity is a reference quantity.

Fig. 1 Fig. 2
The desired power or torque are identified by the MPPT scheme and then one of the
below computations are made.
2 Ls m *
2 Ls i *
  Tem
idr  
* * dr
Ps 3vds Lm p
3vds Lm
2. For a given stator reactive power reference Qs*, the q-axis rotor current reference iqr*
is computed according to: * 2 Ls v
iqr  Qs*  ds
3vds Lm1.5 Lm1.5
3. The reference currents idr* and iqr* are compared to the measured values, and the
errors passed through PI controllers which are then transformed back to abc
46
quantities for use in the PWM scheme to control the RSC (see previous slide).
 Obtaining the transformation angle
The figure shows that the control scheme requires
the abc/qd0 transformation and the qd0/abc
transformation. To get these for both rotor and
stator quantities, we need to have the angles θ(t)
and β(t), to perform the transformations.

Transformations for stator:

 
cos cos(  120) cos(  120)  cos sin  1
 cos(  120) sin(  120) 1

2
Ks  sin  sin(  120) sin(  120)  K
1
3 1 1 1  s
  cos(  120) sin(  120) 1
 2 2 2 
t
 (t )    ( )d  (0)
0

Transformations for rotor:

   cos  sin  1
cos  cos(  120) cos(  120)
 cos(  120) sin(   120) 1

2 1
Kr  sin  sin(   120) sin(   120)  K r
3 1 
 1 1  cos(  120) sin(   120) 1
 2 2 2 
t
 (t )    ( )  m ( )d  (0)   m (0)
0 

 
r  (0)

Observe the “θs calculator.” What is this?

47
 Obtaining the transformation angle
We want to align the d-axis with the stator space vector vs (as this will make vqs=0).
From the figure below, angle θ is defined as the angle of the q-axis wrspt the a-axis
stator reference.
q-axis

d-axis

What if we wanted to align the q-axis with the stator space vector vs?
This means we would want to define
• the location of θ,
to be the same as
• the location of vs (θs).

That is, we want θ=θs. But how to know θs?

48
 Obtaining the transformation angle
Here is the vector we draw to represent vs at t=0.

θs We want to obtain θs, which is the angle of vs. To get θs,

vs
we observe that we can obtain the horizontal and
a
a' vertical components of vs via the α-β transformation.

 1 1
 v  2  1     va 
2 2 v 
v   3  3 3  b 
  0    vc 
α-axis
 2 2 
θs
vs
vβ
a'
1 v
a
vα β-axis
 s  tan
v

The above calculation is very possible since it is easy

to measure the instantaneous phase voltages.

In fact, we need only to obtain two of the instantaneous

phase voltages, as shown on the next slide.
49
 Obtaining the transformation angle
Under balanced conditions, we need measure only the a-phase and b-phase quantities,
as shown below:
 1 1
 v  2  1     va 
2 2 v 
v   3  3 3  b 
  0   v
 2 2   c 

2 1 1  2 1  2 v 
v   va  vb  vc    va  vb  vc    va  a   va
3 2 2  3 2  3 2
 
2 3 3  2 3  3  3
v    vb  vc     vb  vc     v   v  v   3  2vb  va 

3 2 2  3 2  3
b   
a 
 b
 vc 

50
 Obtaining the transformation angle
α-axis
 1 1  v 
θs
 v  2 1  2   a
2 v 
vβ
vs
v   3  3 3  b 
a
a'   0     vc 
vα β-axis  2 2 
1 v
 s  tan
v
When aligning the q-axis with vs, we set θ=θs. But what do we do if we want to
align the d-axis with vs?

Recalling that the d-axis is 90°

behind the q-axis, then we want to q-axis
have θ-90 positioned at θs, i.e.,

  90   s
d-axis
   s  90
51
 Obtaining the transformation angle
So the procedure for aligning the d-axis with vs is as follows:
1. Measure the instantaneous phase voltages.
2. Compute α-β components via
 1 1  v 
 v  2  1   
a
2 2 v 
v   3  3 3  b 
  0     vc 
 2 2 
3. Compute the angle of vs

1 v
 s  tan
v
4. Compute the initial angle θ(t=0) according to:

   s  90

52
 Obtaining the transformation angle
We also need β for the rotor transformation, which
is the angle between the rotor a-axis and the q-
axis of the synchronously rotating reference frame.
Recall from the below diagram (slide 25 of
“dqTransformation” slides) that β=θ-θm.
θ
ωm d-axis
q-axis θm

ω β ia

iq id a'
a

53
 Obtaining the transformation angle
Note the diagram on the right uses θsl instead of β.
To understand this, from the below diagram, we
observe that θsl is the angle between the rotor a-
axis and the d-axis of the synchronously rotating
reference frame, where θsl = θs -θr. Note this
diagram uses θr which we have called θm so we
have that θsl = θs-θm.

From previous slide, we have that β=θ-θm and with

SVOC, θ=θs+90, we have β=θs+90-θm. Therefore,
we have θsl = β-90. So θsl and β differ by 90 deg.
This is because the d-q transformation used in the
book from which these diagrams came differs from
the transformation used in these notes (see slide θm is measured by an encoder
34). mounted on the shaft.

(Reverting back to our own

nomenclature), then β=θ-θm.

54
 Obtaining the transformation angles
So the procedure for aligning the d-axis with vs is as follows:
1. Measure the instantaneous phase voltages.
2. Compute α-β components via
 1 1  v 
 v  2  1   
a
2 2 v 
v   3  3 3  b 
  0     vc  These two
 2 2 
measurements
3. Compute the angle of vs
need to
1 v synchronize.
 s  tan
v
4. Compute the initial angle θ(t=0) according to:
   s  90
The procedure for obtaining the angle β is
1. Measure the rotor angle θm
2. Compute:

    m
55
 Comment 1
What happens to the d-q transformation with the angle θ=0?
 1 1
   1  
2
cos  cos(  120) cos(  120) 2
2  2 3 3
K s (  0)  sin  sin(  120) sin(  120)  0  
3 1 1 1  3 2 2 
  1 1 1 
 2 2 2  0  2 2 2 

 1 1
 x  2  1     xa 
2 2 x 
x   3  3 3  b 
  0   x
 2 2   c 

Implication: the α-β transformation is the d-q transformation if we permanently

fix θ=0.
So when we do the α-β
transformation in step 2, we may
also think about it as doing the d-
q transformation with θ=0 (fixed
on the stator a-phase axis).

56
 Comment 2
• The transformation in Wu’s book, p. 52, uses angles wrspt to d-axis.
• The transformation in Krause’s book, p. 135, uses angles wrspt q-axis.
This also affects the Clarke transformation.

This different results in slightly different transformation matrices (a sign change in row 2).
Trnsfrmtns from Wu Trnsfrmtns from Krause
   
 cos  cos(  120) cos(  120)  cos cos(  120) cos(  120)
2
sin(  120) sin(  120) 
2
K s   sin   sin(  120)  sin(  120) Ks  sin 
3 1 1 1  3 1 1 1 
   
 2 2 2   2 2 2 

   
 cos  cos(   120) cos(   120)  cos  cos(  120) cos(  120)
2
2
K r   sin   sin(   120)  sin(   120) Kr  sin  sin(   120) sin(   120) 
3 1  3 1 1 1 

1 1
  
 2 2 2   2 2 2 
The one I
 1 1   1 1 The one I have used
presented to you 1     
2 2 2  2
1
2 2 in these notes in
T   T 
3
before (in notes 3
3 3 3 3
0   order to maintain
called 0 
 2 2   2 2 
“Transformations”) consistency with the
d-q transformation
we have used.

57
 Function of grid-side converter
We see on slide 22 that the rotor-side converter
(RSC) functions control
• the direct-axis rotor current in order to achieve
a desired torque or stator power level;
• the quadrature-axis rotor current in order to
achieve a desired stator reactive power.
The grid-side converter (GSC) performs two main
functions:
• It provides reactive power to the grid when
required. The reactive power reference Q*GSC
can be set to zero for unity power factor
operation of the converter.
• It keeps the DC link voltage vdc constant.
We observe in the figure the reference quantities
v*dc and Q*GSC being fed to the GSC.

58
 Function of grid-side converter

59
 Function of grid-side converter – reactive power
To get GSC reference quantities, we write real & reactive power equations for the GSC
just as we wrote them in slide 13 for the RSC. The subscript “g” is used here to indicate
these equations are for the grid-side of the GSC. This means that our GSC voltages are
the same as we used for our RSC analysis (vs, vqs, vds) , but currents are not, i.e., the
currents are out of the GSC rotor circuit. This results in:
3 3
Pg  (vdg idg  vqg iqg ) Qg  (vqg idg  vdg iqg )
2 2
We deploy SVOC here just as we did for the
RSC (we use the same θ(t=0)), and we obtain
vqg=0. In this case, the above power
expressions become:
3 3
Pg  vdg idg Qg  vdg iqg  1.5vdg iqg
2 2
From the Qg equation (noting that vdg is the grid-
side voltage in our SVOC scheme), we see we
control the q-axis current, iqg, to achieve a *
Qg
desired Qg, according to: *
iqg 
1.5vdg
*
Q g can be set to 0 for unity power factor operation of the GSC, so that the overall power
factor of the DFIG is controlled by the RSC through its reference Q*s. Then:
• a negative value of Q*s is used in the iqr equation (slide 21) to supply vars to grid
• a positive value of Q*s is used in the iqr equation (slide 21) to absorb vars from grid.
60
 Function of grid-side converter – DC link voltage
We already saw on the previous slide that
3
Pg  vdg idg
2
The DC power crossing the DC link must equal
the active power flowing between the grid and
the GSC, as shown below. Therefore:
3
iDCr iDCg Pg  vdg idg  vDC iDCg
+ i 2
DCc 3
vDC  iDCg  vdg idg
- 2vDC

With the capacitance of C, then we have:

dv
C DC  iDCc  iDCr  iDCg
dt
Substituting for iDCg, and solving for dvDC/dt, yields:
dvDC 1  3 
  iDCr  vdg idg 
dt C 2vDC 
In the steady-state, the left-hand-side is zero, so:
3 3 So specification on vDC translates
iDCr  vdg idg  vDC  vdg idg
2vDC 2iDCr to a requirement on idg.
61
Wind turbine control levels
Rotor-side converter (RSC) is
controlled so that it provides
independent control of Tem
and Qs.
Grid-side converter (GSC) is
controlled so that it provides
independent control of Vbus
and Qg.

Level I: Regulates power flow

between grid and generator.

Level II: Controls the amount

of energy extracted from the
wind by wind turbine rotor.

Level III: Responds to wind-

farm or grid-central control
commands for MW dispatch,
voltage, or frequency control.
62
 Reactive Power at No Wind
S. Achilles, N. Miller, E. Larsen, and J. MacDowell, “Voltage and reactive power control, presentation to NERC ERSTF, June 11-12, 2014,
available at www.nerc.com/comm/Other/essntlrlbltysrvcstskfrcDL/VoltVarControl_Weaksys%20ERSTF%20JMM%20GE_0612.pdf.

“6. The network of variable speed wind

turbine generator systems of claim 1
wherein the individual generators operate
as static VAR regulators under any wind
condition.”

https://www.google.com/patents/US6924565
63