AC Winding Analysis Using Winding Function Approach
AC Winding Analysis Using Winding Function Approach
AC Winding Analysis Using Winding Function Approach
Gojko Joksimovi
Department of Electrical Engineering
University of Montenegro, 20000 Podgorica
Montenegro
[email protected]
Abstract One of the crucial parts of every electrical machines course is the introduction to the basic
elements of alternating current (AC) windings design. After introductory part regarding definition
of the full-pitch coil magnetomotive force (MMF), usually continues analysis based on fundamental
MMF harmonic. This paper presents winding function approach to the AC winding analysis.
Although this approach is well known and intensively used in the scientific circles, especially last
few years, authors impression is that this approach is not sufficiently present in the education
process. This approach has at least two principal advantages over the conventional one. Firstly, it
takes into account exact MMF waveform i.e. all of the MMF space harmonics simultaneously.
Secondly, this approach is mainly computer oriented and on such a manner it promotes very
important aspect of electrical engineering education - an interdisciplinary approach. All of the basic
terms in the field as distribution factor, pitch factor, synchronous speed could be and are derived in
this paper, using winding function definition. Additionally, basic elements of Fouriers analysis
using built-in MATLAB functions are described and its applications are illustrated.
Keywords Turns function, Winding function, MMF, distribution factor, pitch factor, Fourier
analysis
INTRODUCTION
With the advent of low-cost personal computers and various easily accessible software packages,
computer-aided teaching tools have become an essential part of both classroom lectures and
laboratory experiments in electrical machinery education [1]-[5]. These tools have particular
importance in those cases when it is not convenient to organize laboratory exercises as in the case
of AC winding design and analysis.
The fundamentals of Electric machines are taught at the University of Montenegro in two main
courses. The first course is an introductory course which covers the fundamental principles of
electromechanical energy conversion and transformers. The second one deals with rotational
electrical machines in depth. In the first course, one of the crucial parts is AC winding design and
analysis, one of the most difficult topics in electric machines for students to understand. This is why
a special attention must be given to this topic with extensive use of graphics and animation.
Presented approach to this complex matter at the Department of Electrical Engineering at
University of Montenegro was carried out last two years. After learning the materials in the
traditional manner, by presentation of slides and animation of rotating MMF wave, the students take
active participation in the modeling of the concrete winding in the MATLAB environment. During
the classroom lectures students have a chance to explore and look at real winding machine, too, Fig
1. The quality of this way of active participation in the educational process describes a high
percentage of success at the mid-term test: almost 86%.
The described approach facilitates students understanding of a very important concept in electric
machinery: spatial harmonics of MMF. It makes a good basis for distinguishing these from the
others, later introduced time harmonics of voltage or current. The described approach has the
2
following additional good side: it provides an animation of the rotating MMF wave. Spectral
analysis of rotating MMF wave at different instants of time assures students of a constant harmonic
content of the wave regardless of its different waveforms in different instants of time.
A
3
2
4
1
Assuming uniform air-gap length i.e. neglecting existence of stator and rotor slots, as well as
assuming infinitely permeable iron, Amperes circuital law, in common shape,
H dl = J dS
C
(1)
(2)
where H() is radial component of magnetic field intensity in the air gap at position described by
angle , g is air gap length, i is coil current and n() is turns function, [12].
From Fig 2 it is obvious that by changing the position of the closed loop side 2 (this closed loop
takes part in the Amperes circuital law), turns function could be defined on following manner:
N
n( ) = n
0
1 < 2
for other
(3)
B dS = 0
(4)
(5)
or,
l 2
H rddz = 0
(6)
0 0
Assuming that H does not depend on the axial coordinate, above expression is equal to,
2
H d = 0
(7)
1
H d = 0
2 0
(8)
1
i
H (0 ) + n( ) d = 0
2 0
g
H (0 ) +
i
n( )d = 0
2g 0
H (0 ) =
(9)
(10)
i
n()
g
(11)
where n() is the mean value of the turns function. From (11) and (2) follows:
H () g = [n( ) n( ) ] i
(12)
(13)
where,
N () = n( ) n()
(14)
is winding function, [13], [14]. Obviously, winding function is nothing else but MMF per unit
current. In further text terms winding function and MMF per unit current will be used
interchangeable. Winding function could be alternatively written as,
m
N n 1 2
N () =
N m
n
2
1 < 2
(15)
for rest of
3
<
2
2
for other
N
n A () = n
0
(16)
3
<
2
2
Nn
2
N A () =
N n
2
(17)
for other
Fig 3 show placement of coil along the stator circumference, turns function profile, winding
function profile together with two the most significant space-harmonics as well as coefficients of
the Fouriers expansion of winding function.
A
A
3
C
2
4
1
a)
b)
Fig. 3. a) Cross section view, b) Turns function, winding function along with two the most significant higher spaceharmonics and Fourier expansion of winding function for a full-pitch coil in a two pole machine (p=1). Coil has 10
turns.
Obviously, only odd higher space-harmonics are present in Fourier series of winding function. The
fundamental space-harmonic amplitude is 6.37 turns, Fig 2b, i.e. 4/ higher than the winding
function amplitude, so fundamental MMF wave amplitude is,
MMFA max 1 ( ) =
4 Nn
I 2
2
(18)
where I is the alternating current rms value. Higher space-harmonic of order has times smaller
amplitude than the fundamental one:
MMFA max ( ) =
4 1 Nn
I 2
2
(19)
Turns and winding function for full pitch coil as well as Fourier expansion of winding function
given on Fig 3 could be easily defined using following few rows written in MATLAB:
H=1024;
dx=2*pi/H;
nn=10;
It should have in mind following two facts: 1) in order to obtain exact discrete Fourier transform,
number of samples in 2 must be power of two, i.e. H=2n where n is an integer; 2) first element in
Fourier expansion is coefficient equal to the mean value of the winding function and this element is
not shown on Fig 3.
Fig 4 shows cross section view of the machine as well as turns and winding function for a full-pitch
coil winding in a four pole machine, p=2. Two full-pitched coils (in a four pole machine full-pitch
coil has a pitch m=/2 rad) are series connected. Total turns and winding function of phase
winding is simply the sum of the turns and winding functions of every single coil in the winding.
By other words, superposition is allowed because infinitely permeable iron was assumed. Now, the
fundamental harmonic is second order harmonic (p), third harmonic is sixth order harmonic (3p)
etc. The amplitude of the fundamental MMF wave now is,
4 NA
I 2
2p
MMFA max 1 () =
(20)
where NA is the total number of turns in the phase winding. The amplitude of the MMF higher
space-harmonic of order is:
MMFA max () =
4 1 NA
I 2
2p
(21)
A1
A1
A2
3
C
4
A2
a)
b)
Fig. 4. a) Cross section view, b) Turns function, winding function along with two the most significant higher-space
harmonics and Fourier expansion of winding function of two series connected full-pitch coils (A1-A1 and A2-A2) in four
pole machine (p=2). Every single coil has 10 turns.
components. In the design of the AC windings, serious efforts are made to distribute the coils
making up the windings so as to minimize the higher-order harmonic components and to produce an
air-gap MMF wave which consists predominantly of the space-fundamental sinusoidal component.
On this way, additionally, machine is better utilized. Therefore, in practice, AC winding of
induction machine is always designed as a distributed winding. Distributed phase winding of a
three-phase machine occupies one third of stator slots along the circumference of the machine. Fig
5 shows such one phase winding in a two pole machine which stator has S=18 slots. It means that
one phase winding occupy six slots, three under one and three under other pole.
A3
A2
A1
A1
A2
A3
a)
b)
Fig. 5. a) Cross section view, b) Turns function, winding function along with the two most significant higher-space
harmonics and the Fourier expansion of distributed winding consists of three series connected full-pitch coils (A1-A1,
A2-A2 and A3-A3 ) in two pole (p=1) machine. Every single coil has Nn=10 turns, NA=30.
Turns and winding function of phase A winding could be easily obtained by summing turns and
winding functions of individual coils. As it is clear from Fig 5, resultant MMF profile of distributed
full-pitch winding has trapezoid like shape what is much closer to the sinusoid than before, Figs 3
and 4. However, amplitude of the resultant winding function is not three times as greater than
before, but somewhat smaller as it could be easily concluded observing the coefficients of the
Fourier series. The reason for that is in the space displacement of individual coils. The winding
distribution factor Kd just define this ratio: per definition, distribution factor is ratio of the resultant
MMF space-harmonic amplitude and sum of the MMF space-harmonics amplitudes of the
constituent coils (ratio of vector and algebraic sum of constituent coils MMFs):
K d =
MMFA max
MMFA max
=
+ MMFA A' max + ... + MMFA A' max n MMFcoil max
2 2
(22)
For the fundamental space harmonic, =1, from Fourier expansions, Figs 3 and 5:
K d1 =
MMFA max 1
18.3344
=
= 0.96
3 MMFcoil max 1 3 6.3662
MMFA max 5
0.8418
=
= 0.22
3 MMFcoil max 5 3 1.2731
These results are in good correlation with results obtained from the well known analytical
expression for distribution factor, [15]-[18],
K d
sin q1
sin
2m
2
=
=
q1 sin q1 sin
2
2mq1
(23)
where m is number of phases, q1 is number of slots per pole per phase and is an electrical angle,
=pm=p2/S. For analyzed phase winding,
sin
sin
23 =
6 = 0.9598
Kd1 =
3 sin
3 sin
233
18
Kd5
5
sin 5
sin
2
3
=
6 = 0.2176
=
5
3 sin 5
3 sin
233
18
10
Having in mind definition of distribution factor, resultant distributed phase winding MMF spaceharmonic amplitude could be defined as:
MMFA max () =
4 1 N A K d
I 2
2p
(24)
C2
C3
C1
B1
B2
A3
A2
B3
A1
A1
B3
A2
B2
A3
B1
C1
C3
C2
a)
b)
Fig. 6. a) Cross section view, b) Winding functions of phase windings. Every phase winding consists of three series
connected full-pitch coils. Two pole machine, p=1. Every single coil has 10 turns.
Winding functions i.e. MMFs per unit current for such machine is given on Fig 6. In order to obtain
the resultant MMF wave, care must be taken about instantaneous value of phase currents. As it is
well known, through the three-phase windings flows phase shifted currents. At one instant of time,
say t1=5ms, Fig 7, phase current ia has a value ia=1A while other two phase currents are
11
ib=ic=0.5A. Resultant MMF wave could be obtained by multiplying phase winding functions with
adequate instant values of the phase currents,
(25)
i.e.
(26)
(27)
i.e.
MMFS _ t2 =10ms =
3
(N b N c )
2
(28)
Fig. 7. Three-phase, 50Hz, unit currents. Phase current rms value is 1/2=0.707A.
The resultant MMF profiles at these two instant of time are shown on Fig 8. Obviously, positions of
the maximal values of resultant MMFs are shifted in space, i.e. resultant MMF wave is not fixed in
space: actually it is rotating MMF wave. Rotating speed could be easily found from the following
considerations: at t1=5ms maximal value of the rotating MMF is at position described by the axes of
phase winding A, i.e. t1=2000. At t2=10ms, maximal value of the rotating MMF is at position
t2=2900. Rotating MMF speed, i.e. synchronous speed, is:
12
[rad ] 4 rad
angle 2900 2000
900
rad
ns =
=
=
= 2
=
= 2f
=
T
time
t 2 t1
s
[s ] T [s ] 2T s
4
4
2rad
rev
f
= 60 f
1
min
min
60
(29)
More generally, in case of machine with p pole pairs, synchronous speed is p times smaller:
ns =
2f rad 60 f rev
=
p s
p min
(30)
Hence, resultant MMF wave rotate with synchronous speed, having different waveforms in every
different instant of time. However, regardless of shape of the MMF profile at different instants of
time, Fouriers expansion of the MMF is always the same. Spectral content of the presented MMF
waveforms is also shown on Fig 8.
Fig. 8. Resultant rotating MMF wave in two different instants of time, t1=5ms and t2=10ms and Fourier expansion of
MMF wave. Two pole machine.
13
For difference from the phase windings, resultant rotating MMF wave does not contain any odd
space harmonic which is integer multiply of three. By other words, all MMF space-harmonics
belongs to the following series,
= 6k + 1
(31)
where k=0,1,2, From (31) follows that symmetrical three phase winding supplied from
symmetrical three-phase voltage supply contains fundamental
eleventh 11th etc. space harmonics. Minus sign means inverse rotating MMF wave. Synchronous
speed of th space-harmonic is times smaller than the synchronous speed of the fundamental one,
ns =
2f rad 60 f rev
=
p s
p min
(32)
As it one can conclude from the Fig 8, noteworthy higher space harmonics are 5th and 7th, so called
phase belt harmonics which are direct consequence of trapezoidal shape of the phase winding
MMF. However, the most significant higher space harmonics are, so called, slot harmonics, of
order S/p1. In analyzed case it is 17th and 19th space harmonics. They are direct consequence of the
discrete nature of the winding i.e. conductor placement in the slots. These harmonics on the best
manner fills the gaps in the stepwise shape of resultant MMF wave.
By comparison the amplitudes of the MMF space-harmonics from Fig 5 and Fig 8, it is easy to
conclude that the rotating MMF space-harmonic amplitude is 1.5 times higher than amplitude of the
phase winding MMF space harmonics:
MMFS max =
3
3 4 1 N A K d
3 1 N A K d
MMFphase max ( ) =
I 2=
I 2
2
2 2p
p
(33)
3 1 N A K d
3 1 30 0.96 1
3
I 2=
2 = 30 0.96 = 27.52 Aturns
p
1
1
14
MMFS max =
m
m 1 N A K d
MMFphase max () =
I 2
2
p
(34)
K p1
(35)
where is an electrical angle which describe the shortening of the coil pitch.
15
4
3
8
9
10
11
18
12
17
13
14
15
16
Fig. 9. Double layer three-phase winding. Two pole machine, p=1. Shortening of the coil is one stator slot pitch.
(36)
K p1 = sin
y
p 2
(37)
y
p 2
(38)
By taking into account just defined pitch factor, amplitude of the rotating MMF space-harmonic, in
distributed double-layer three-phase winding is,
MMFS max =
3 1 K p K d N phase
3 1 K N phase
I 2=
I 2
(39)
where K is winding factor which include distribution and pitch factor, K=KdKp. In many
electrical machines textbooks above expression is given in the following form,
MMFS max = 1.35
N eff I
p
(40)
16
K p1 = sin
8
= 0.9848
92
N eff I
p
= 1.35
28.36 1 / 2
= 27.15 Aturns
1 1
and somewhat is smaller than in a case of full-pitch windings, due to the pitch factor. This result is
in close agreement with results obtained numerically, from Fouriers expansion of the resultant
MMF wave, Fig 10. However, the main result of short-pitch coils using is in the attenuation of 5th
and 7th space harmonics in the rotating MMF wave. Namely, in order to cancel out 5th harmonic
from the rotating MMF wave, following condition must be satisfied,
K p 5 = sin 5
y
y
2
=05
= y = p
p 2
p 2
5
(41)
where is an integer, =0,1,2, and must be chosen so that coil pitch y is first number smaller of
the pole pitch expressed in number of stator slots. For analyzed winding, p=9 slots, so, the
reasonable solution for y is for =2:
y=
2
36
=2
9
= 7. 2
5
5
17
Fig. 10. Rotating MMF wave at two different instants of time and corresponding Fouriers coefficients. Two pole
machine. Every single coil has 5 turns.
As coil pitch must be an integer, possible solutions are y=7 or y=8. Obviously, for any choice, 5th
space harmonic could not be eliminated but for y=7 it will be significantly reduced. Similarly, for
canceling out 7th space harmonic, following condition must be satisfied:
K p 7 = sin 7
y=
y
y
2
=07
= y = p
p 2
p 2
7
(42)
2
54
=3
9
= 7.71
7
7
Again, y could be 7 or 8 and, again, 7th space harmonic could be only attenuated, especially for y=8,
but not absolutely eliminated from the MMF wave. If coils with 7 stator slot pitch are chosen, 5th
harmonic will be much attenuated than 7th. Inversely, if coils with 8 stator slot pitch are chosen, 7th
space harmonic will be much attenuated than 5th. In analyzed case study, coils with 8 stator slot
18
pitch are used (one stator slot pitch shortening), so 7th space harmonic will be much more attenuated
than 5th, as could be easily concluded by comparison of MMF wave spectrum from Figs 8 and 10.
Usually, coil pitch is chosen in such a way to attenuate 5th and 7th space harmonic simultaneously,
what is fulfilled for the following coil pitch,
y=
2
1
p = p
6
3
(43)
together with additional condition, that must be chosen on a such way that coil pitch y is slightly
smaller than the pole pitch. Condition (43) could be alternatively defined as,
y = p
p
6
5
p
6
(44)
However, from Figs 8 and 10 it should be observed that shortening of the coils have no any impact
on the intensity of the slot harmonics! Coil, regardless of its pitch, must begin in one slot and end in
the other, and therefore the coils spacing is an integral multiple of the basic spacing causing slot
harmonics in the first place.
CONCLUSIONS
A winding function approach for AC winding analysis is presented. This approach has at least two
principal advantages over conventional one. Firstly, it takes into account all of the MMF space
harmonics simultaneously, i.e. exact MMF waveform. Secondly, this approach is mainly computer
oriented and on such manner it promotes very important aspect of electrical engineering education an interdisciplinary approach. All of the basic terms in the field as distribution factor, pitch factor,
synchronous speed are derived in this paper using winding function approach. Additionally, basic
elements of Fouriers analysis using built-in MATLAB functions are described and its applications
are illustrated. This approach could be easily implemented for analysis of fractional slot windings as
well as any form of asymmetrical windings.
19
REFERENCES
[1] A.F. Zobaa, T.A. Boghdady, Integration into undergraduate courses of transformer tests using
MATLAB/Simulink, International Journal of Electrical Engineering Education, Vol. 44, No.4,
pp.319-332, December 2007.
[2] S. Ayasun, G. Karbeyaz, DC Motor Speed Control Methods Using MATLAB/Simulink and
Their Integration into Undergraduate Electric Machinery Courses, Computer Applications in
Engineering Education, Vol. 15, No.4, pp.347-354, 2007.
[3] S. Ayasun, C. Nwankpa, Induction Motor Tests Using MATLAB/Simulink and Their
Integration Into Undergraduate Electric Machinery Courses, IEEE Transactions on Education,
Vol. 48, No.1, pp.37-46, February 2005.
[4] Y.N. Anagreh, I.M. Al-Refaee, Teaching the self-excited induction generator using
MATLAB, International Journal of Electrical Engineering Education, Vol. 40, No.1, pp.55-65,
January 2003.
[5] J. Cathey, Electric Machines: Analysis and Design Applying MATLAB, McGraw Hill, 2001.
[6] H. Sequenz, The windings of electrical machines, Vol. 3, A.C. Machines, Springer Verlag,
Vienna, 1950 (in German).
[7] M.M. Liwschitz-Garik, Winding of alternating current machines, van Nostrand Publications,
1950.
[8] A. Demenko, Description of electrical machine windings in the finite element space,
COMPEL, Vol. 27, No. 4, pp. 711-719, 2008.
[9] J. Steinbrink, Design and analysis of windings of electrical machines, International
Symposium on Power Electronics, Electrical Drives, Automation and Motion, SPEEDAM, pp. 717720, 2008.
20
[10] D. A. Kocabas, A. F. Mergen, A novel method to reduce the effects of space harmonics in
alternating current machines, International Symposium on Industrial Electronics, ISIE, pp. 692697, 2008.
[11] H. Xueliang, D. Qiang, H. Minqiang, A Novel Exact and Universal Approach for Calculating
the Differential Leakage Related to Harmonic Waves in AC Electric Motors, IEEE Transactions
on Energy Conversion, Vol. 19, No. 1, pp. 1-6, March 2004.
[12] N.L. Schmitz, D.W. Novotny, Introductory electromechanics, Ronald Press, 1965.
[13] R.B. Robinson, Inductance coefficient of rotating machines expressed in terms of winding
space harmonics, Proc. Inst. Elec. Eng, Vol. 111, No. 4, pp. 769-774, 1964.
[14] T.A. Lipo, Theory and control of synchronous machines, ECE 511 lecture notes, University of
Wisconsin-Madison
[15] A.E.Fitzgerald, C. Kingsley, S.D. Umans, Electric machinery, McGraw Hill Professional, 2002
[16] I. Boldea, S.A. Nasar, The induction machine handbook, CRC Press, 2002
[17] P.C.Sen, Principles of electrical machines and power electronics, Wiley, 1996
[18] P.Cochran, Polyphase induction motors analysis, design and application, CRC Press, 1989
21