See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/280087488

A GENERALIZED DYNAMIC MODEL FOR BRUSHLESS DOUBLY-FED

INDUCTION MACHINES

Data · July 2015

CITATIONS READS

0 33

1 author:

Ahmed Mohamed Azmy

Tanta University
131 PUBLICATIONS   783 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Optimal DG Deployment to Improve Voltage Stability Margin Considering Load Variation View project

Power System Operation View project

All content following this page was uploaded by Ahmed Mohamed Azmy on 16 July 2015.

The user has requested enhancement of the downloaded file.

 ERJ
Engineering Research Journal
Faculty of Engineering
Minoufiya University

A GENERALIZED DYNAMIC MODEL FOR BRUSHLESS

DOUBLY-FED INDUCTION MACHINES
S. M. Allam1, A. M. Azmy1, M. M. Khater2 and M. A. El-Khazendar1
1
Electrical Power and Machines Department, Faculty of Engineering,
Tanta University, Tanta, Egypt
2
Electrical Engineering Department, Faculty of Engineering
Minoufiya University, Shebin El-Kom, Egypt

ABSTRACT
This paper presents a detailed dynamic model, by which the dynamic behaviour of Brushless
Doubly-Fed Induction Machine (BDFIM) can be successfully predicted. The investigated doubly-
fed machine has two sets of three-phase stator windings with different pole numbers. The rotor is
a squirrel-cage type with a simple modification in order to support the two air-gap rotating fields
that are produced by the stator windings and have different pole numbers. The machine model is
derived in both natural-machine variables abc-axis and qdo variables axis. The qdo- axis is
attached to rotor and hence, it rotates at the rotor speed ( ωm ). The electromagnetic torque
expression is also obtained based on the two presented models. The winding function method is
presented as a convenient approach that can be used for machine-winding inductances
calculations. Sample case studies are introduced to examine the performance of the proposed
model.
‫ﻼ ﻳﻤﻜﻦ ﻣﻦ ﺧﻼﻟﻪ ﺗﺤﻠﻴ ﻞ اﻷداء اﻟﻌ ﺎﺑﺮ ﻟﻶﻟ ﺔ اﻟﺘﺄﺛﻴﺮﻳ ﺔ اﻟﻼﻓﺮﺷ ﻴﺔ ﻣﺰدوﺟ ﺔ ﻣ ﺪﺧﻞ‬ ً ‫ﻳﻘﺪم هﺬا اﻟﺒﺤﺚ ﻧﻤﻮذﺟًﺎ رﻳﺎﺿﻴًﺎ ﻣﻔﺼ‬
‫ واﻵﻟﺔ ﻣﺰدوﺟﺔ اﻟﺘﻐﺬﻳﺔ اﻟﺘﻰ ﺗﻢ دراﺳﺘﻬﺎ ﺗﺤﺘﻮى ﻋﻠ ﻰ زوج ﻣ ﻦ اﻟﻤﻠﻔ ﺎت ﺛﻼﺛﻴ ﺔ اﻷوﺟ ﻪ ﻓ ﻰ اﻟﻌ ﻀﻮ اﻟﺜﺎﺑ ﺖ ﻟﻜ ﻞ‬.‫اﻟﺘﻐﺬﻳﺔ‬
‫ أﻣﺎ اﻟﻌﻀﻮ اﻟﺪوار ﻓﻬﻮ ﻣﻦ ﻧﻮع اﻟﻘﻔﺺ اﻟ ﺴﻨﺠﺎﺑﻰ ﻣ ﻊ إﺟ ﺮاء ﺗﻌ ﺪﻳﻞ ﺑ ﺴﻴﻂ ﻋﻠﻴ ﻪ ﻟﻴﺘﻮاﻓ ﻖ ﻣ ﻊ‬.‫ﻣﻨﻬﻤﺎ ﻋﺪد أﻗﻄﺎب ﻣﺨﺘﻠﻒ‬
‫ و اﻟﻨﻤﻮذج اﻟﻤﻘﺪم ﺗﻢ اﺳﺘﻨﺘﺎﺟﻪ ﻣﻦ ﺧﻼل ﺗﻤﺜﻴ ﻞ اﻵﻟ ﻪ ﻓ ﻰ‬.‫ﻣﺠﺎل اﻟﺜﻐﺮﻩ اﻟﻬﻮاﺋﻴﻪ اﻟﻤﺰدوج اﻟﻨﺎﺗﺞ ﻣﻦ ﻣﻠﻔﺎت اﻟﻌﻀﻮ اﻟﺜﺎﺑﺖ‬
‫اﻟﻤﺤﺎور اﻟﻄﺒﻴﻌﻴﻪ )اﻟﻤﺒﺎﺷﺮة( وآﺬﻟﻚ ﻓﻰ اﻟﻤﺤﺎور اﻟﻤﺘﻌﺎﻣﺪﻩ واﻟﺘﻰ ﺗﻢ اﺧﺘﻴﺎرهﺎ ﻟﺘﻜﻮن ﻣﺘﺮاﻓﻘﻪ ﻣﻊ اﻟﻌﻀﻮ اﻟﺪوار وﻣﻦ ﺛ ﻢ‬
‫ ﺗﻢ اﻟﺤﺼﻮل ﻋﻠﻰ ﻋﻼﻗﺔ اﻟﻌﺰم اﻟﻜﻬﺮوﻣﻐﻨﺎﻃﻴﺴﻰ‬،‫ واﻋﺘﻤﺎدًا ﻋﻠﻰ اﻟﻨﻤﻮذﺟﻴﻦ اﻟﻤﻘﺪﻣﻴﻦ‬.‫ﻓﺈﻧﻬﺎ ﺗﺪور ﺑﺴﺮﻋﺔ اﻟﻌﻀﻮ اﻟﺪوار‬
‫ آ ﺬﻟﻚ ﺗ ﻢ ﺗﻘ ﺪﻳﻢ أﻣﺜﻠ ﻪ‬.‫ أﻳﻀًﺎ ﺗﻢ ﺗﻘﺪﻳﻢ ﻃﺮﻳﻘﺔ "داﻟﺔ اﻟﻠﻒ" آﻄﺮﻳﻘﻪ ﻣﻨﺎﺳﺒﻪ ﻳﻤﻜﻦ اﺳﺘﺨﺪاﻣﻬﺎ ﻟﺤ ﺴﺎب ﻣﻤﺎﻧﻌ ﺎت اﻵﻟ ﻪ‬.‫ﻟﻶﻟﻪ‬
.‫ﻣﻦ ﻧﺘﺎﺋﺞ أداء اﻵﻟﻪ ﺗﺤﺖ ﻇﺮوف ﺗﺸﻐﻴﻞ ﻣﻌﻴﻨﻪ ﺑﺎﺳﺘﺨﺪام اﻟﻨﻤﻮذج اﻟﺮﻳﺎﺿﻰ اﻟﻤﺴﺘﻨﺘﺞ‬
Keywords: Brushless doubly-fed machine- Dynamic modelling- Generalized theory- Induction
machine- Self-cascaded, Speed control- Variable speed drives

1. NOMENCLATURE g −1(θ m , θ) The inverse air-gap function

B viscous coefficient Ni (θ m , θ) The winding function of winding (i)
iabck 3-ph currents of either stator windings λ abck 3-ph flux linkages of either stator windings
irki ith-rotor loop current of either rotor circuit
J Rotor inertia L k Leakage inductance of either stator
Nk Turns per phase of either stator windings windings
p The differential operator "d/dt" Lm k Magnetizing inductance of either stator
Pk Pole-pairs of either stator windings windings
R Rotor radius Mr Mutual inductance between any two rotor
rsk Per-phase resistance of either stator windings loops
Tem Total electromagnetic developed torque L mr Rotor loop magnetizing inductance
TL Load torque ωm Mechanical rotor speed
Vabck 3-ph voltages of either stator windings
rb , L b Rotor bar resistance and leakage
µo Permeability for vacuum inductance respectively.
Rotor axial length re , L e End ring segment resistance and leakage
θm Rotor mechanical angular position with inductance respectively
respect to phase (a) reference axis of the two
sets of stator windings
θ Angular position along the stator inner surface

Engineering Research Journal, Vol. 31, No. 4 October 2008, PP 345-354 345
© Faculty of Engineering, Minoufiya University, Egypt
 S. M. Allam, A. M. Azmy, M. M. Khater and M. A. El-Khazendar, "A Generalized Dynamic Model for … "

pole-pair numbers) exclusively. An extended model,

2. INTRODUCTION
which is appropriate for any number of pole-pairs,
Squirrel-cage induction motors are the most widely was introduced in [2]. However, the structure of the
used motors in AC-drive applications. This is BDFM used is similar to that used in [1]. The main
attributed to their ruggedness, low cost, and few problem of these models is the complicated structure
maintenance requirements. The number of industrial of both stator and rotor windings that these models
applications, which utilize AC adjustable speed based upon.
drives (ASDs), is increasing rapidly. The primarily
A new model based on a dual stator winding
reason for this development is the significant advent
induction machine is presented in [8]. The dual stator
of power electronics. The main disadvantage of using
machine consists of a conventional squirrel-cage
singly-fed machines for ASDs is the need for large
induction motor having two three-phase stator
and expensive power conditioning units that should
windings wound for two and six poles. The two-
withstand the full motor power. This increases both
stator windings are simultaneously fed from two
the capital cost and the harmonic current pollution of
independent variable frequency power supplies. This
the overall system [1]. The Brushless Doubly-Fed
results in increasing the capital cost of the overall
Induction Machines (BDFIMs) shows a great
system and thus, reducing the applications, in which
promise in reducing the rating of power electronic
this system can be used.
converters needed for ASDs application [2]. The
BDFIM represents a development of self-cascaded The aim of this paper is to present a generalized and
machines with two three-phase stator windings detailed dynamic model valid for predicting the
having different number of poles accommodated in a dynamic performance of BDFIMs under different
single machine frame. In addition, it is required to loading conditions. The proposed modelling process
introduce a modification to the cage rotor structure starts by describing all machine equations in its
[3]-[6]. natural variables abc-axis. Then, the model is
transformed to qdo variables axis using appropriate
The two stator windings are referred to as: the power
transformation matrices. The machine-winding
winding, connected directly to the main power
inductances are calculated using winding function
supply, and the control windings connected to the
method. In addition, the paper aims at investigating
main power supply via power electronic converters.
the possibility of simpler structure of the BDFIM
In a BDFIM, the rotor structure basically determines
than that used before. The investigated BDFIM
the magnetic coupling between the two stator
consists of a simply modified squirrel-cage rotor and
windings, which in turn determines the machine
a stator with two separate windings wound for a
behaviour [6]. BDFIM is a special form of slip
dissimilar number of poles. Each stator winding is
recovery machines that reduce the capacity of the
fed from an independent power supply. Only one of
required inverter to be used if the required speed-
the two stator windings is fed from a variable-
control range is limited [1]-[7]. This will lead to a
frequency variable-voltage inverter.
significant reduction in the drive cost. Owing to its
brushless property, the BDFIM is mechanically
3. SYSTEM DESCRIPTION
robust and reliable. It is found that the BDFIM has
the ability to provide precise speed control and good Fig. 1 shows the schematic diagram of the
power factor with reduced converter capacity, as low investigated BDFIM. There are two three-phase
as 25% of the machine rating [3]-[4]. This results in a windings with different number of poles, in order to
greatly reduced capital cost of the overall system. A avoid magnetic mutual coupling, placed in the stator
further advantage of the BDFIM is derived from the slots [1]. These windings are simultaneously fed
fact that the power electronic converters do not from two independent power supplies and each one
process the bulk power entering or leaving the produces a separate torque component. One set of
machine directly. stator windings is called the power windings;
through which full machine power must pass, and is
The dynamic model of Brushless Doubly-Fed
fed directly from a constant voltage constant
Machines (BDFMs) was first presented in [1].
frequency power supply. The other set is called the
However, the presented model is based on a special
control windings, through which partial controlled
configuration of both stator and rotor winding
power passes, and is fed from a variable voltage
connections. The stator-winding configuration
variable frequency power supply.
appears to be far more complicated than the three-
phase winding of a conventional induction or The rotor is a squirrel-cage type with a simple
synchronous machine. In addition, a more complex modification in its structure to support the two air-
rotor configuration than the conventional cage rotor gap rotating fields that have different pole numbers
is used. The dynamic model introduced in [1], has produced by the two-sets of stator windings [6]. In
addressed the 3/1 (power winding to control winding [7], it is stated that the essential requirement for

346 Engineering Research Journal, Minoufiya University, Vol. 31, No. 4, October 2008
 S. M. Allam, A. M. Azmy, M. M. Khater and M. A. El-Khazendar, "A Generalized Dynamic Model for … "

brushless doubly-fed machine action is that the

frequency and distribution of currents induced in the 4. INDUCTANCE CALCULATIONS
rotor by the first air-gap field match that induced by The machine-winding inductances can be calculated
the second. The two stator windings will be then by a particular convenient approach called winding
coupled via the rotor, even though they have functions method [9]-[10]. The winding function is
different pole numbers and may be excited at defined as the spatial distribution of MMF due to a
different frequencies. This implies that the number of unit current flowing in the winding. According to
rotor bars should be equal to the summation of pole- this method, the machine inductances are calculated
pair numbers of both stator windings. Thus, the by an integral expression representing the placement
number of rotor bars will be small for any realistic of winding turns along the air-gap periphery. This
combination of pole-pair numbers [7]. method is particularly convenient for the analysis of
Such low number of rotor bars results in a very high- unusual machines since it assumes no symmetry in
referred rotor leakage reactance [7]. Therefore, more the placement of any motor coil in the slots [10].
work has been done to increase the number of rotor According to the winding function theory, the mutual
slots and distribute each bar between several slots. inductance between any two windings (i) and (j) in
The most common practice, by which this can be any machine can be easily computed from the
accomplished is to connect the rotor bars selectively following expression:
to produce a set of concentric or nested loops. Each 2π
nested loop has to be isolated from the others, and L ij = µ o R ∫ g −1 (θ m , θ ) N i (θ m , θ ) N j (θ m , θ ) dθ (1)
the number of nested loops is equal to the required 0
number of rotor bars. As an alternative, the 4.1. Stator Winding Function
outermost loop of each nest has to form the required Each stator phase is assumed to be sinusoidally
number of rotor bars whilst the other loops are joined wound along the air gap. Hence, the winding
at a common end ring. These two ways used in rotor function of each one of the two sets of stator
bars arrangements are not easily fabricated. windings can be written as:
Furthermore, the cost of this type of machines is high
since both stator and rotor have to be redesigned. N ak (θ ) = (N k 2Pk ) cos(Pk θ ) ⎫
⎪
In this paper, a very simple modification in the rotor N bk (θ ) = (N k 2Pk ) cos(Pk θ − 2π 3) ⎬ (2)
N ck (θ ) = (N k 2Pk ) cos(Pk θ + 2π 3) ⎪
cage is introduced. The rotor bars is divided into ⎭
number of groups that equal the required number of It should be noted that all equations used for the two-
rotor bars (summation of pole-pairs of the two-stator stator windings are similar except for their different
windings). Separating each group from the others can number of poles, turns per phase, resistance, and
be easily done at one side only of the two end rings. inductance. So, instead of writing theses equations
This means that one end ring is divided into a twice, the subscript "k" is used denoting either the
number of sections; each connecting a number of power or the control windings. Hence, the same
bars, while the other end ring connects all rotor bars. equation can be used to describe the two stator
It should be noted that this modification in the end windings just by defining "k" with suitable number
ring structure results in zero currents in some rotor related to each one of the two sets of stator windings.
loops at which the end ring is separated. For 6-pole
and 2-pole stator windings, the number of required 4.2. Rotor-Loops Winding Function
rotor bars or groups equals four. Therefore, one side Squirrel-cage rotor has "n" identical and equally
of the end rings should be divided into four groups or spaced bars, which are shorted together by the two
sections as shown in Fig. 1. end-rings. Therefore, it can be modelled as "n"
identical magnetically coupled circuits or loops [8].
Main Supply Each loop encompasses two adjacent rotor bars and
the connection portions of the end rings between
them. The normalized winding function for the ith-
Inverter rotor loop is given by the MMF distribution produced
ωm by one ampere flowing through this loop. For all
a6 a2 other loops, the normalized winding functions are
similar, with only difference in the phase angles,
b6 b2
which are defined by the position of each loop with
c6 cage rotor c2 respect to the reference point. Fig. 2 shows the
normalized winding function for the ith-rotor loop,
6-pole stator winding 2-pole stator winding which can be written as follows:
Fig. 1. Schematic diagram of the BDFIM

Engineering Research Journal, Minoufiya University, Vol. 31, No. 4, October 2008 347
 S. M. Allam, A. M. Azmy, M. M. Khater and M. A. El-Khazendar, "A Generalized Dynamic Model for … "

⎧ − α r 2π 0 ≤ θ ≤ θi ⎫ λ abck abck : is the flux linkage of any of the two stator

⎪ ⎪ (3)
N i (θ i , θ) = ⎨1 − α r 2π θi ≤ θ ≤ θi + α r ⎬ windings due to the corresponding current of the
⎪ − α 2π θ i + α r ≤ θ ≤ 2π ⎪ same windings, and
⎩ r ⎭
λ abck rki : is the flux linkage of any of the stator
Where: θ i = θ m + (i − 1)α r
windings due to ith-rotor loop current of the
N i (θ ) corresponding rotor circuit.
The total flux linkage for both stator windings can be
expanded as:
1 - (α r / 2π )
αr λ abck = [L abck abck ][i abck ] + [L m abck rki ][i rki ] (5)
- (α r /2π )
θ Where:
θi θi + α r 2π
⎡L k + L mk − Lmk 2 − L mk 2 ⎤
Fig. 2. The ith-rotor loop winding function [Labck abck ] = ⎢⎢ − L mk 2 L k + L mk − L mk 2 ⎥
⎥
⎢⎣ − L mk 2 − Lmk 2 L k + L mk ⎥⎦
4.3. Inverse Air-Gap Function , the matrix [L m abck rki ] is given in the appendix,
For a conventional squirrel-cage rotor, the air-gap is
assumed to be uniform and hence, the inverse air-gap [i rki ]T = [i rk1
i rk2 ... i rkn ] , and
function may be considered to be constant without "n" is the total number of squirrel-cage rotor bars
contribution in the integration shown in (1). (loops)
4.4. Inductance Formulas It should be noted that each column in the matrix
Magnetizing and mutual inductance formulas of both [L m abck rki ] corresponds to a certain loop. In
stator windings and n-rotor loops can be easily
obtained using (1), (2), and (3). The final expressions addition, each element in the matrix [i rki ]T
or formulas of these inductances are given in the corresponds to a certain loop. For a standard squirrel-
appendix. cage rotor, without any modification in its structure,
all rotor loops represent closed loops through which
5. DYNAMIC MODEL IN NATURAL the rotor loop current can pass. Therefore, the order
MACHINE VARIABLES AXIS of the two matrices [L m abck rki ] and [i rki ]T is related
5.1. Stator Winding Voltage Equations to the total number of rotor loops. Thus, the order of
The 3-phase voltage equation of each one of the two these two matrices, for a standard cage rotor, is '3*n'
stator windings can be written in a matrix form as and '1*n' respectively. However, for a modified cage
follows: rotor, which is presented in this paper and due to the
[v abck ] = [rsk ][i abck ] + p [λ abck ] (4) discontinuity of one of the two end rings, some rotor
loops represent open loops. For the open loops at
It should be noted that in the flux linkage which the end ring is separated, the rotor loop current
calculations, there is no component of flux linkage of is set to zero and thus, the corresponding columns or
any of the two stator windings due to the current of elements are eliminated from the two matrices. The
the other windings. This is attributed to the mutual number of these columns equals the number of end
decoupling of the two stator windings since they ring sections (i.e. summation of pole-pairs of both
have different number of poles. As a result, the rotor stator windings). This reduces the order of the
can be modelled as two independent circuits. Each matrices to be '3*m' and '1*m' respectively, where 'm'
one has "n"-rotor loops and is related to one of the is the total number of closed rotor loops (i.e.
two-stator windings. So, the subscript 'k' may be also difference between total number of rotor bars or
used to define the two different rotor circuits related loops and open loops at which the end ring is
to the corresponding two stator windings. Hence, the separated). Mathematically "m" can be given by:
flux linkage in each of the stator windings is
m = n − ∑ Pk , where ∑ Pk is the summation of the
established due to only two components. The first is
due to the stator current in the same windings, and pole-pairs of the two stator windings.
the second is due to the current of the corresponding These notes can be easily concluded by putting a
rotor circuit, which has the same number of poles. condition for the subscript "i", which used in [i rki ]T
The flux linkage can be rewritten as:
matrix as follows:
λ abck = λ abck abck + λ abck rki
i = (1, 2, …, n), i ≠ q np
Where:
Where:

348 Engineering Research Journal, Minoufiya University, Vol. 31, No. 4, October 2008
 S. M. Allam, A. M. Azmy, M. M. Khater and M. A. El-Khazendar, "A Generalized Dynamic Model for … "

"q" = (1, 2, …, ∑ Pk ), and λ rki rkj : is the ith-rotor loop flux linkage of either
"np" ( = n ∑ Pk ): is the total number of squirrel-cage rotor circuits due to n- rotor loops current of the
rotor bars per pole-pair summation of the two-stator same circuit, and
windings. λ rki abck : is the ith-rotor loop flux linkage of either
5.2. Rotor Loops Voltage Equations rotor circuits due to the current of corresponding
stator winding.
In order to obtain the n-loops rotor voltage equations,
the rotor circuit can be divided into two circuits with The total n-rotor loops flux linkage is written in an
the same number of poles of the corresponding two- expanded matrix form as:
stator windings. Fig. 3 shows a schematic diagram of [λ rki ] = [Lr ][i rki ] + [Lm rki abck ][iabck ] (7)
n-rotor loops winding for any of the corresponding
two-stator windings just after defining 'k'. An open Where:
circuit rotor loop from one side of end rings is also ⎡L r + L mr M r −L b M r −L b ⎤
shown in Fig. 3. ⎢ M −L L r + L mr Mr ⎥
Open segment [L r ] = ⎢⎢ r b ⎥,
⎥
⎢ ⎥
Rotor bar Two-end ⎣ M r −L b Mr L r + L mr ⎦
irk 1 irk rings
L r = 2(L b + L e ) , and [L m rki ] = [L m abck rki ]T
irk n 2 irk n-1
abck
5.3. Electromagnetic Developed Torque Equation
Fig. 3. Rotor loops winding configuration
The electromagnetic developed torque of a BDFIM
After assigning 'k', it should be noted that the can be divided into two components. Each one is due
equivalent two rotor circuits have the same loop to the mutual coupling between one of the two-stator
resistance, loop leakage inductance, loop windings and the corresponding rotor circuit. Thus,
magnetizing inductance and mutual inductance the total electromagnetic developed torque is given
between any two rotor loops with values equal to that by the summation of these two components and can
of conventional squirrel-cage rotor. The main be written as:
difference between the two rotor circuits is the Tem = ∑ Temk (8)
mutual inductance between each rotor circuit itself
Where:
and the corresponding stator windings. Since the two
⎛ d ⎞
stator windings are mutually decoupled, the flux Temk = [i abck ]T ⎜⎜ [L m abck rki ]⎟⎟ [i rki ] (9)
linkage in each rotor circuit is created due to only ⎝ dθ m ⎠
two components: the first is due to the rotor current
in the same circuit and the second is due to the 5.4. The electromechanical equation
current of the corresponding stator winding, which The electromechanical equation of a BDFIM can be
have the same number of poles. easily written as:
The rotor-loops voltage equation of each one of the Tem = J * pω m + B * ω m + TL (10)
two rotor circuit is written in a matrix form as
Using (4) through (10), the system dynamic
follows:
performance can be analyzed in natural machine
[0] = [rr ][irki ] + p [λ rki ] (6) variables axis.
Where:
6. DYNAMIC MODEL IN QDO-AXIS ROTOR
⎡r - rb - rb ⎤
⎢ ⎥ REFERENCE FRAME
[] ⎢- rb r
rr = ⎢
0 ⎥
⎥ and r = 2(rb + re ) .
The main problem of the previous model, which is
derived in natural machine variables axis, is the time-
⎢ ⎥
⎢- rb 0 r ⎥ varying mutual inductances. This results in a very
⎣ ⎦ complex model. In order to eliminate the time-
In addition, the corresponding n-rotor loops flux varying inductances in voltage equations, the
linkage of either rotor circuits can be rewritten as: previous dynamic model is transformed to qdo-axis.
λ rki = λ rki rkj + λ rki abck , j = (1, 2,…, n) The qdo-axis is fixed in the rotor, which rotates at
the rotor speed ( ω m ). Fig. 4 shows the phase-axis
Where:
relationship for phase "a" of any of the two stator
windings axis and the nth-rotor loops axis.

Engineering Research Journal, Minoufiya University, Vol. 31, No. 4, October 2008 349
 S. M. Allam, A. M. Azmy, M. M. Khater and M. A. El-Khazendar, "A Generalized Dynamic Model for … "

Pk ω m ⎡rsk 0 0⎤ ⎡ 0 ωm 0⎤
[rsk ] = ⎢⎢ 0 0 ⎥⎥ , and [ω k ] = Pk ⎢- ω m 0⎥⎥
Q- axis
rsk 0
⎢
ki
k2 ⎢⎣ 0 0 rsk ⎥⎦ ⎢⎣ 0 0 0⎥⎦
k1
Pk θ r
(i-1)Pkαr Pkαr B. QDO-Axis Rotor Loops Voltage Equations
-Pkαr
kn The transformation matrix of either rotor circuits of
n-rotor loops axis to the corresponding qdork-axis
variables can be written based on Fig. 4 as:
a k - axis
⎡f ⎤= k f
⎢⎣ qdork ⎥⎦
[ ][ ]
rk rki

Where:
⎡ ⎤
⎢ cos(0) cos(Pk α r ) cos((n - 1)Pk α r ) ⎥
D- axis
[k rk ] = 2 ⎢⎢− sin(0) - sin(Pk α r ) - sin((n - 1)Pk α r )⎥
⎥
Fig. 4. Phase axis relationship of the BDFIM m
⎢ 1 1 1 ⎥
⎢⎣ 2 2 2 ⎥⎦
6.1. QDO-Axis Voltage Equations (14)
A. QDO-Axis Stator Winding Voltage Equations The back transformation matrix of qdork-axis to the
Aided with Fig. 4, the transformation matrix of either corresponding rotor circuits of n-rotor loops winding
stator windings (abck-axis variables) to the axis is given by:
corresponding qdok-axis variables can be written as
follows:
[f rki ] = [ k rk back ] ⎡⎣fqdork ⎤⎦
⎡f [ ][ ]
⎤= k f
⎢⎣ qdok ⎥⎦ sk abck
Where:
⎡ cos(0) − sin(0) 1⎤
Where 'f' represents either voltage, current or flux ⎢ cos(P α ) - sin(P α ) 1⎥
linkage, and [k rk back ] = ⎢⎢ k r k r ⎥ (15)
⎥
⎡ 2π 2π ⎤ ⎢ ⎥
⎢cos(Pk θ r ) cos(Pk θ r − 3 ) cos(Pk θ r + 3 )⎥ ⎣cos((n - 1)Pk α r ) - sin((n - 1)Pk α r ) 1⎦
⎢ ⎥
[k sk ] = 2 ⎢ sin(Pk θ r ) sin(Pk θ r − 2π ) sin(Pk θ r + 2π ) ⎥ It should be noted that the absence of some columns
in the [i rki ]T matrix, corresponding to open rotor
3⎢ 3 3 ⎥
⎢ 1 1 1 ⎥
⎢ 2 2 2 ⎥ loops, necessitates the absence of corresponding
⎣ ⎦
(11) columns in the transformation matrix [k rk ] and rows
Where θ r = θ m + δ , and δ = α r 2 in the back transformation matrix [k rk back ] . In
addition, the number of these absent columns or rows
The inverse of this transformation matrix that is equal to summation of pole-pairs of both stator
transforms the qdok- axis variables to either stator windings.
winding abck- axis variables is given by:
Appling the transformation matrix and the back
[f abck ] = [k sk ]−1 [f qdok ] transformation matrix to (6) results in qdork-axis
rotor voltage equation. The final qdork-axis rotor
Where:
voltage equation can be written in a matrix form as:
⎡ ⎤
⎢ cos(Pk θ r ) sin(Pk θ r ) 1⎥ 0 = ⎡⎣ rqdo rk ⎤⎦ ⎡⎣i qdo rk ⎤⎦ + p ⎡⎣ λ qdo rk ⎤⎦ (16)
⎢ ⎥
[k sk ]−1 = ⎢cos(Pk θ r − 2π ) sin(Pk θ r − 2π ) 1⎥ (12)
⎢ 3 3 ⎥ ⎡rrk 0 0⎤
⎢cos(P θ + 2π ) sin(P θ + 2π ) 1⎥ Where: rqdo rk = ⎢⎢ 0 rrk 0 ⎥⎥
⎢⎣ k r
3
k r
3 ⎥⎦
⎢⎣ 0 0 ror ⎥⎦
Applying the transformation matrix and its inverse to
(4), results in the representation of the stator voltage rrk = r − 2rb (1 - ∑ Pk m )cos(Pk α r )
equation in the qdok-axis. The final qdok-axis stator ror = (r − 2rb )
voltage equation can be written in a matrix form as:
⎣⎡ vqdok ⎤ = [ rsk ] ⎡i qdok ⎤ + p ⎡ λ qdok ⎤ + ωk ⎡ λ qdok ⎤ (13)
⎦ ⎣ ⎦ ⎣ ⎦ ⎣ ⎦
6.2. QDO-Axis Flux Linkage Relations
Appling the transformation matrix and its inverse to
Where:
the flux linkage equations of both stator windings

350 Engineering Research Journal, Minoufiya University, Vol. 31, No. 4, October 2008
 S. M. Allam, A. M. Azmy, M. M. Khater and M. A. El-Khazendar, "A Generalized Dynamic Model for … "

and the corresponding rotor circuits results in the ⎡L rk 0 0 ⎤

final qdo-axis flux linkage relations. Where: L qdo rk = ⎢⎢ 0 L rk 0 ⎥⎥
A. QDO-Axis Stator Flux Linkage Relations ⎢⎣ 0 0 L or ⎥⎦
The total qdok-axis equation of the stator flux linkage
can be written as: L rk = L r + L mr − M r − 2L b (1 - ∑ Pk m )cos(Pk α r )
λ qdok = λ qdok qdok + λ qdok qdork L or = L r + L mr + (n − 1)M r − 2L b
Where: The qdork-axis rotor flux linkage due to the
λ qdok qdok : is the qdok-axis flux linkage of either corresponding qdok-axis stator winding current is
given by:
stator winding due to the corresponding qdok-axis
stator winding current, and ⎡⎣ λ qdork qdok ⎤⎦ = ⎡⎣ L m qdork qdok ⎤⎦ ⎡⎣i qdok ⎤⎦ (20)
λ qdok qdork : is the qdok-axis flux linkage of either
Where:
stator winding due to the corresponding qdork-axis
⎡L mrk 0 0⎤
rotor current.
The qdok-axis stator flux linkage due to the [L m qdork qdok ] = ⎢⎢ 0 L mrk 0⎥⎥ , and
corresponding qdok-axis stator winding current is ⎢⎣ 0 0 0⎥⎦
given by: L mrk = (3 2 ) M kr
⎡⎣ λ qdok qdok ⎤⎦ = ⎡⎣ L qdok qdok ⎤⎦ ⎡⎣i qdok ⎤⎦ (17) The final form of qdo-axis system model in the rotor
reference frame is given in the appendix
Where:
⎡L sk 0 0 ⎤ 6.3. Electromagnetic Torque in terms of QDO-
[L qdok qdok ] = ⎢⎢ 0 L sk
3
0 ⎥⎥ , L sk = L k + L mk
2
Axis currents
⎢⎣ 0
It is convenient to describe the developed
0 L k ⎥⎦
electromagnetic torque from both stator windings as
The qdok-axis stator flux linkage due to the a function of qdo-axis currents. This can be easily
corresponding qdork-axis rotor current is given by: obtained after applying the transformation matrices
⎡⎣ λ qdok qdork ⎤⎦ = ⎡⎣ L m qdok qdork ⎤⎦ ⎡⎣i qdo rk ⎤⎦ (18) given in (12) and (15) to (9). This yield;

Where: [ ] ⎛ d
Temk = [k sk ]−1 i qdok T ⎜⎜
⎞
[
[L m abck rki ] ⎟⎟ [k rk back ] iqdo rk ]
⎝ dθ m ⎠
⎡L mkr 0⎤
0
[L m qdok qdork ] = ⎢⎢ 0 L mkr 0⎥⎥
(21)
The final expression of the total electromagnetic
⎢⎣ 0 0 0⎥⎦ developed torque corresponding to qdo-axis currents
L mkr = (m 2)M kr , and M kr is given in the can be given by:
appendix Tem = ∑ Temk = Tk 1 + Tk 2
Where:
Tk 1 = (3 2 ) L mkr Pk i qk i drk
B. QDO-Axis Rotor Loops Flux Linkage Relations
The total qdork-axis rotor flux linkage equation can
be written as: Tk 2 = (- 3 2) L mkr Pk i dk i qrk

λ qdork = λ qdork qdork + λ qdork qdok Equations (10) and (13) through (21) provide a
useful dynamic representation of the machine. Thus,
Where: this model can be used to describe the dynamic
λ qdork qdork : is the qdork-axis flux linkage of either performance of the machine in the qdo rotating axis.
rotor circuit due to the corresponding qdork-axis rotor
current, and 7. SAMPLE CASE STUDIES
λ qdork qdok : is the qdork-axis flux linkage of either A sample case study is presented using the proposed
rotor circuit due to the corresponding qdok-axis stator model. The presented samples of the simulation
current. results are obtained using the calculated parameters
of an actual set-up system. The stator and the rotor of
The qdork-axis rotor flux linkage due to the a conventional three-phase squirrel-cage induction
corresponding qdork-axis rotor current is given by: motor, rated at 1.1 kW, 460 V and 4-poles, are
⎡⎣ λ qdork qdork ⎤⎦ = ⎡⎣ Lqdo rk ⎤⎦ ⎡⎣i qdork ⎤⎦ (19) modified to fit the nature of BDFIM. The original
stator winding is replaced by two sets of three-phase
windings having different number of poles. One set

Engineering Research Journal, Minoufiya University, Vol. 31, No. 4, October 2008 351
 S. M. Allam, A. M. Azmy, M. M. Khater and M. A. El-Khazendar, "A Generalized Dynamic Model for … "

of the two stator windings, the power windings, has Then, the control winding is connected to the main
six poles, while the other, the control windings, has supply via an inverter. In Fig. 6a, the control winding
two-poles. One side of the two-end rings of the rotor is excited with the same sequence of the power
is separated to a number of sections that equals the winding, while in Fig. 6b, the control winding is
summation of pole-pairs of both stator windings. All excited with the reverse sequence of the power
mutual and magnetizing inductances are calculated winding.
using winding function method. Stator resistance and
leakage inductance are measured using standard
techniques that used for the conventional induction
machines. Dimensions of both rotor bar and end-ring
are used to calculate their resistance and leakage
inductance.
In this section of the paper, only two samples of
different mode of operations are simulated and
presented using the proposed model. Simulation
results are obtained using Matlab/Simulink software.
This paper does not aim to discuss or analyse the
presented results. Instead, it presents sample results
to confirm the presented model in different modes of
operation.
7.1. Singly-Fed Operation
(a) Control winding is excited with the same sequence
Fig. 5 shows the run-up response of the machine for
a singly fed mode. In this mode of operation, the
power winding is connected to the main supply while
the control winding is left open. First, the power
winding started with no-load until 0.5 second where
a 0.75 N.m load torque is applied. The machine gives
response similar to that of the conventional induction
machine. Thus, the obtained behaviour confirms that
the proposed model is valid to describe the transient
behaviour of conventional induction machine
operating with only one stator winding.

(b) Control winding is excited with a reverse sequence

Fig. 6. Run-up response of a doubly fed mode of a
BDFIM

8. CONCLUSIONS
This paper has presented a detailed and accurate
analysis that is valid for predicting the dynamic
performance of brushless doubly-fed induction
machines. A detailed dynamic model in both natural
machine variables axis and qdo-axis has been derived
Fig. 5. Run-up response of a singly fed mode of a from the basic relations of the machine. The winding
BDFIM function method has been introduced as a viable
method that can be used for calculating the machine
7.2. Doubly-Fed Operation inductances. In order to confirm the validity of the
proposed model, Sample case studies have been
Fig. 6 shows the run-up response of a doubly- fed introduced using actual experimental set-up
mode of a BDFIM under no-load condition. First, the parameters.
power winding is connected to the main supply while
the control winding is left open until 0.5 second.

352 Engineering Research Journal, Minoufiya University, Vol. 31, No. 4, October 2008
 S. M. Allam, A. M. Azmy, M. M. Khater and M. A. El-Khazendar, "A Generalized Dynamic Model for … "

9. REFERENCES No. 5, September/October 2000, pp. 1369-

[1] R. Li, A. K. Wallace, R. Spee, and Y. Wang, 1379.
"Two-axis model development of cage-rotor [9] F. liang, L. Xu, and T. A. Lipo, "d-q analysis of
brushless doubly-fed machines," IEEE a variable speed doubly ac excited reluctance
Transaction on Energy Conversion, Vol. 6, No. motor," Electrical Machines and Power
3, September 1991, pp. 453-460. Systems, Vol. 19, No. 2, Mar. 1991, pp. 125-
[2] M. S. Boger, A. K. Wallace, R. Spee, and R. Li, 138.
"General pole number model of the brushless [10] L. Xu, F. liang, and T. A. Lipo, "Transient
doubly-fed machines," IEEE Transaction on model of a doubly excited reluctance motor,"
Industry Application, Vol. 31, No. 5, September IEEE Transaction on Energy Conversion, Vol.
/October 1995, pp. 1022-1028. 6, No. 1, March 1991, pp. 126-133.
[3] M. S. Boger, A. K. Wallace, R. Spee, and R. Li,
"Investigation of appropriate pole number 10. APPENDIX
combination for brushless doubly-fed machines A. Magnetizing inductance of each phase of any of
applied to pump drives," IEEE Transaction on the two stator windings
Industry Application, Vol. 32, No. 1, L m abck = L mk = µ o R g −1 π(N k / 2Pk )2
January/February 1996, pp. 189-194.
B. Mutual inductance between phases of any of the
[4] M. S. Boger, and A. K. Wallace, "Performance
two stator windings
capability analysis of the brushless doubly-fed
machine as a wind generator," Seventh L abck bcak = - L mk 2
International Conference on Electrical C. Mutual inductance between the two stator
Machines and Drives, 11-13 September 1995, windings having different pole number (6-pole
No. 412, pp. 458-461. and 2-pole as an example)
[5] P. Zaiping, Z. Yanfeng, and M. H. Nagrial, L m abc6 abc2 = 0
"Review and future application of brushless D. Mutual inductance between any of the two stator
doubly-fed machine," The 1998 International windings and the corresponding n- rotor loops
Conference on Communication, Computer & circuit
Power, Muscat December 7-10, 1998, pp. 306- Lm ak rki = M kr cosPk (θ m + (i − 1)α r + δ )
311.
Lm bk rki = M kr cos(Pk (θm + (i − 1)α r + δ ) - 2π 3)
[6] F. Wang, F. Zhang, and L. Xu, "Parameter and
performance comparison of brushless machines L m ck rki = M kr cos(Pk (θ m + (i − 1)α r + δ ) + 2π 3)
with cge and reluctance rotors," IEEE Where: δ = α r 2 , and
Transaction on Industry Application, Vol. 38,
No. 5, September /October 2002, pp. 1237-
( )
M kr = µ o R g −1 N k (Pk )2 sin (Pk δ )
1243. E. Magnetizing inductance of each loop
[7] S. Williamson, A. C. Ferreira, and A. K. L m ri = L mr = µ o R g −1α r (1 − α r 2π )
Wallace, "Generalized theory of the brushless
F. Mutual inductance between any two rotor loops
doubly-fed machine. Part 1: Analysis," IEE
(i, j)
Proc.-Electr. Power Appli., Vol. 144, No. 2,
March 1997, pp. 111-122. (
L m ri rj = M r = µ o R g −1 - α 2r 2π )
[8] A. R. Munoz, and T. A. Lipo, "Dual Stator G. Mutual inductance between any n-rotor loop
Winding Induction Machine Drive,” IEEE circuits and the corresponding stator windings,
Transaction on Industry Applications," Vol. 36, L m rki ak = L m ak rki

H. The matrix [L m abck rki ] described in section (5.1)

⎡ ⎤
⎢ cos(Pk (θ m + δ )) cos(Pk (θ m + δ + α r )) … cos(Pk (θ m + δ + (n − 1)α r )) ⎥
⎢ 2π 2π 2π ⎥
L m abck rki = M kr ⎢cos(Pk (θ m + δ ) − ) cos(Pk (θ m + δ + α r ) − ) … cos(Pk (θ m + δ + (n − 1)α r ) − ) ⎥
⎢ 3 3 3 ⎥
⎢cos(P (θ + δ ) + 2π ) cos(P (θ + δ + α ) + 2π ) … cos(P (θ + δ + (n − 1)α ) + 2π )⎥
⎢⎣ k m
3
k m r
3
k m r
3 ⎥⎦

Engineering Research Journal, Minoufiya University, Vol. 31, No. 4, October 2008 353
 S. M. Allam, A. M. Azmy, M. M. Khater and M. A. El-Khazendar, "A Generalized Dynamic Model for … "

I. Final form of qdo-axis system model in rotor reference frame

⎡ v qk ⎤ ⎡ rsk + L sk p Pk ω m L sk 0 L mkr p Pk ω m L mkr 0 ⎤ ⎡ i qk ⎤
⎢ v ⎥ ⎢- P ω L r + L p 0 - P ⎥⎢ i ⎥
⎢ dk ⎥ ⎢ k m sk sk sk k ω m L mkr L mkr p 0 ⎥ ⎢ dk ⎥
⎢ v ok ⎥ ⎢ 0 0 rsk + L k p 0 0 0 ⎥ ⎢ i ok ⎥
⎢ ⎥=⎢ ⎥⎢ ⎥
⎢ v qrk ⎥ ⎢ L mrk p 0 0 rrk + L rk p 0 0 ⎥ ⎢i qrk ⎥
⎢ v drk ⎥ ⎢ 0 L mrk p 0 0 rrk + L rk p 0 ⎥ ⎢i drk ⎥
⎢ ⎥ ⎢ ⎥⎢ ⎥
v
⎣⎢ ork ⎦⎥ ⎣⎢ 0 0 0 0 0 ror + L ork p ⎦⎥ ⎣⎢i ork ⎦⎥

354 Engineering Research Journal, Minoufiya University, Vol. 31, No. 4, October 2008

View publication stats